NFPA Journal

Fire Dynamics Simulator
Ensure your software provides the safest atrium design for real world enforcement.

NFPA Journal®, January/February 2008

By Michael J. Ferreira, P.E.

Software modeling tools that understand and visually represent the dynamics of gas and fluid flow are very important to any engineering design project, and none more so than the Fire Dynamics Simulator (FDS) for designing atrium smoke control systems. A freely available product developed by the National Institute of Standards and Technology (NIST), the increased use of this computation fluid dynamics modeler is due in part to the architectural complexity of many new atria and in part to the increased acceptance of new alternatives in atrium smoke control system design. But there are other reasons more designers are turning to FDS. The use of a computer model may be necessary to evaluate how smoke will fill an atrium or to evaluate smoke conditions as a function of time. In addition, the software tool itself has grown more useful. Improvements in computer processing capabilities and simplified methods for entering building geometries into FDS make the model an attractive design tool.

Currently, the FDS software runs from a command-line interface and not a graphical interface like a Windows-based program such as Microsoft® Word or Excel. Third-party products that allow users to enter data using a familiar Windows application display are available, but may not be free. Information on the FDS and these third party tools are available at http://fire.nist.gov/fds/. NIST also offers "Smokeview", a freely available software program that displays FDS results visually.

When designing atrium smoke control systems, FDS is most often used to evaluate design alternatives. Until recently building codes contained prescriptive requirements for sizing atrium exhaust that were based on one objective: keeping the smoke at a specifi ed height within the atrium. This approach uses equations based solely on the fire size, and results in a design with a smoke layer interface height defi ned at a prescribed distance above the highest occupied level of the atrium or above unprotected openings into the atrium. The intent is to maintain smoke above the head-height of occupants within the atrium and to minimize the exposure of these occupants to smoke.

Now recent code developments provide the option of calculating the required atrium exhaust based on a performance-based approach that considers such factors as smoke tenability and occupant exposure times. This approach anticipates the exposure of occupants to smoke. FDS is used to evaluate tenability conditions, including smoke temperature, smoke toxicity, and visibility distance through smoke.

With each new design method brings the need for context. Using the tenable approach can often result in a significant reduction in smoke exhaust when compared to the defined smoke layer interface approach, which is highly conservative for some atrium geometries. However, when evaluating a design that uses the tenable-based approach, extra care must be taken to ensure that an adequate degree of safety is included.

Recent articles appearing in publications aimed at architects and engineers who are involved in the atrium smoke control system design process seem to offer a tantalizing promise: use FDS to greatly reduce the exhaust quantities required and overall system complexity. On the surface, this is quite positive. However, if designers are not careful in their selection of model input parameters or evaluation criteria, such as the design’s fire, smoke properties, tenability criteria, or egress considerations, their designs may lack sufficient conservatism. Results from designers who shun conservative parameter selection may not be recognized until there is an atrium fire involving loss of life and the atrium is later found to have insufficient exhaust due to the assumptions put forth in the original design.

The potential for atrium smoke control system designs that are not sufficiently conservative is a particular problem for fire marshals and other authorities having jurisdiction (AHJs) tasked with approving performance-based designs. These reviewers may lack familiarity with FDS or other advanced design tools that are used to substantiate the design or they may not be fully aware of the impact of design assumptions inherent in the analysis.

Evolving design requirements
Atrium smoke control system requirements have changed substantially over the years. Prior to 1991, atrium smoke control systems were required by building codes to provide smoke exhaust of four to six air changes per hour, based on the volume of the atrium space. Largely dependent on atrium geometry, little scientific basis existed for this requirement and system effectiveness was not assured. NFPA advanced the scientific basis for calculating atrium exhaust requirements by including empirical formulas for calculating the required exhaust that were validated using full-scale testing in the 1991 edition of NFPA 92B, Guide for Smoke Management Systems in Malls, Atria, and Large Spaces.

In 2005, NFPA 92B was revised to become the Standard for Smoke Management Systems in Malls, Atria, and Large Spaces. Rather than remaining a guide, with loose design guidelines and recommendations. the document was rewritten as a standard that other documents could reference. Also new to the standard was its recognition of systems designed to maintain a tenable environment in addition to those designed to maintain a defined smoke layer interface height. As recognized by NFPA 92B, Section 4.1, tenability calculations may now be performed in conjunction with a timed egress analysis to demonstrate adequate atrium smoke control system performance:

NFPA 92B [2005], Section 4.1 Design Objectives
"4.1.2 The design objectives to be achieved over the design interval time by a smoke management system shall include one or both of the following:
(1) Maintaining a tenable environment within all exit access and area of refuge access paths for the time necessary to allow occupants
to reach an exit or area of refuge.
(2) Maintaining the smoke layer interface to a predetermined elevation."

When the NFPA Technical Committee on Smoke Management Systems was considering the addition of the tenability approach to NFPA 92B, the main focus of the discussion was if enough was known about evaluating occupant tenability to allow a design approach that allowed occupants to be exposed to smoke within the atrium. The committee realized that even the smoke layer interface height method used calculations that resulted in occupant exposure to smoke.

The 2005 edition of NFPA 92B now recognizes designs for systems using the tenable environment maintenance and those using defined smoke layer interface height maintenance.

Using FDS as a design tool
FDS has been under development at NIST’s Building Fire Research Laboratory for over a decade. The model is public domain software available from the NIST Web site, making it a frequently used design tool by fire protection engineers. FDS is specifically designed with fire scenarios in mind and can be readily applied to fire protection engineering applications (see endnote 1). The model is intended to handle isolated and spreading fires in habitable spaces in the presence of obstacles such as furniture, overhead ceiling obstructions, and other structural members. It can handle both passive and forced vents (i.e., smoke exhaust), and users can define fuel properties and burning rates. The model is well validated for these applications, as described in the fire protection literature (see endnotes 2-4).

The first step in using FDS to evaluate a smoke control system design is to model the atrium geometry, including all architectural features that impact smoke filling in the space. Then a timed egress analysis is performed for those spaces within and adjacent to the atrium that will expose occupants to smoke when they travel through them. Finally, design fire scenarios representing the worst-case smoke conditions that occupants would be exposed to are evaluated against a specified smoke control system design, to discover if it maintains acceptable tenability conditions during the critical period of egress.

To illustrate a typical application of a tenability atrium smoke control system design, consider the proposed design for a college library building containing a four-story atrium.

The library atrium represents a straightforward application that meets NFPA 92B’s intent for a tenability smoke control system design. On the uppermost two floors of the library atrium, the only occupied space open to the atrium are balcony reading spaces and small conference rooms with glass walls separating these rooms from the atrium. If a fire occurred within the atrium, smoke filling the atrium would be immediately visible to the occupants of these adjacent areas. The occupants of the balcony and conference rooms have access to four sets of egress doors leading to corridors outside of the atrium, each having access to two enclosed exit stairs. No occupants other than those within the atrium space have to traverse the atrium space to exit the building.

For this example, an exhaust quantity of approximately 250,000 cfm (425,000 m3/h) would have to be provided to maintain a smoke layer interface 6 feet (1.8 meters) above the uppermost level of the atrium space (the fourth floor). This design approach is the same as that intended before NFPA 92B recognized tenability systems. The exhaust quantity for the library atrium was able to be reduced to 150,000 cfm (255,000 m3/h) by allowing the steady-state layer to descend to a height 6 feet (1.8 meters) above the second floor of the atrium space and evaluating smoke tenability versus timed egress for the upper two floors of the space.

The reduction in exhaust quantity also enables a corresponding reduction in required make-up air, and results in a significant reduction in cost and system complexity.

Each design consideration shapes the result
Multiple design considerations are involved when using FDS to evaluate atrium smoke control system performance. Design fire selection, selection of fuel properties, and boundary surface properties are just some. Other considerations are used to evaluate model outputs, including the selection of tenability criteria and performing egress calculations to determine critical times. The impact of other modeling related factors, such as grid definition, combustion and turbulence models, and the definition of boundary conditions, are discussed in fire protection engineering literature (see endnotes 5-6).

Many assumptions must be made when selecting critical design values in order to ensure that an adequate degree of conservatism is built into the analysis. The pass/fail assessment of FDS results are dramatically impacted by these underlying suppositions. The designer and the design reviewer should fully understand the impact of these assumptions.

Design fire selection
Fire modeling involves the selection of both a peak fire size and fire growth rate. The designer must decide whether to use values associated with the expected use of the atrium space, such as a furniture grouping or a kiosk, or whether to use maximum fuel loading values associated with a seasonal, transient, or one-time use of the space, which may greatly exceed the normal fuel loading conditions. Because NFPA 92B and other design standards are moving away from specifying a minimum fire size, this choice is left up to the smoke control system designer.

Design fires must be selected based on the maximum sustained heat release rate to ensure a conservative design, so a range of fire growth rates should be evaluated. Designers should evaluate a variety of fire locations, including applicable adjacent spaces and locations resulting in balcony spill plumes. They should also evaluate the impact sprinkler have on controlling the fire and on reducing the buoyancy of the rising smoke, as this may cause a greater degree of smoke stratification within the space.

Principles of Smoke Management, published by the American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) (see endnote 7), states that design fires used in the calculation of smoke control system requirements can be estimated using an appropriate heat release per unit area value, based on the type of fuel loading that is observed in typical buildings. The authors suggest that a value of 20 Btu/s-ft2 (225 kW/m2) is appropriate for restricted fuel spaces such as atria, where furniture combustibility is regulated by code. For these spaces, they suggest a design fire area of 100 ft2 (9.3 m2), resulting in a 2,000 Btu/s (2,111 kW) design fire for office building atria. The heat release rate per unit area method assumes that the area in question is a continuous fuel package, with the entire surface burning. This size design fire area is also consistent with the guidance provided in the 2000 edition of NFPA 92B for minimum design fire size for atrium design.

The authors also suggest a value of 44 Btu/s-ft2 (495 kW/m2) for mercantile or other high fuel load spaces. Use of this value results in a 4,400 Btu/s (4,642 kW) design fire over the same 100 ft2 (9.3 m2) burning area. This result is close to the 5,000 Btu/s design fire specified by the 2000 and 2003 editions of the IBC for use in atrium smoke control system design.

The fuel loading per unit area approach is one method for calculating a design fire. Fire test data for individual fuel packages may also be used as the basis for constructing a design fire’s heat release rate profile.

One problem with the use of fire test data to define a design fire is that designers continue to use relatively small design fires substantiated using administrative fuel controls or by the use of selective test data, despite the available guidance trending toward the use of larger, more conservative design fires. They often rely on a test performed by NIST on a small retail kiosk (see endnote 8) as the design fire for mall atrium smoke control system designs, despite the fact that larger mall kiosks and seasonal displays can easily exceed this design fire in terms of severity. This is a perfect example of a non-conservative approach to FDS-modeled atria design. While the instantaneous peak fire size from the kiosk test was slightly less than 1,800 Btu/s (1,900 kW) the average sustained heat release rate for the majority of the test was roughly 1,200 Btu/s (1,267 kW). This rate is much lower than the minimum 2,000 Btu/s (2,111 kW) design fire discussed in NFPA 92B.

Selection of tenability criteria
While the 2005 NFPA 92B now recognizes tenable environment-based designs, it does not provide guidance pertaining to the threshold tenability values to be used in the design. NFPA 92B, Annex G Additional Design Objectives, suggests that tenability analysis should include the evaluation of heat exposure, smoke toxicity, and visibility, yet Annex G specifically states, "methods for tenability analysis are outside the scope of this document."

Committee proposals for referencing ISO/TS 13571, "Life-Threatening Components of Fire—Guidelines for the Estimation of Time Available for Escape Using Fire Data (see endnote 9) in NFPA 92B are currently being considered. ISO/TS 13571 and Milke, et al., (endnote 10) provide a comprehensive overview of the tenability criteria that designers should be consider in an atrium smoke control system evaluation.

Heat exposure includes consideration of both convective exposure (direct immersion in hot smoke) and radiative exposure (indirect exposure to a hot smoke layer located above or at a distance from occupants). Smoke toxicity includes the consideration of both asphyxiant gases (e.g., carbon monoxide) and irritant gases (e.g., hydrogen chloride, hydrogen bromide, and sulphur dioxide). Visibility pertains to the distance occupants exposed to smoke can see through the smoke in negotiating their way to the nearest exit.

Experience has shown that for atrium spaces visibility through smoke is often the limiting tenability criteria for design, because for many atria the smoke is sufficiently diluted and cooled by the air entrained into the rising smoke plume such that visibility violates the minimum criteria well before temperature and toxicity become a concern. Therefore, the impact of other factors on the calculation of visibility distance deserves further discussion.

Selection of fuel properties
In order to calculate visibility distance through smoke, the designer must input into FDS fuel properties, such as soot, carbon monoxide yields, and heat of combustion. The relative impact of various parameters affecting the calculation of visibility distance is shown by relationships a) and b) below. The use of a higher soot yield may result in higher (and thus more conservative) smoke concentrations within a space and will reduce the calculated visibility distance.

Conversely, selection of a higher heat of combustion actually reduces the equivalent mass burned for a given heat release rate, which reduces the amount of smoke particulate in the air, resulting in a higher predicted visibility distance. Therefore, selecting a combination of a higher range soot yield and lower range heat of combustion from the available data for the type of fuel to be burned is the most conservative approach. The designer must also be appropriately conservative in selecting fuel properties based on the types of fuel that may be in the space. For example, the selection of a mostly wood combustible will create much less smoke (and thus higher predicted visibility distance) than a combustible containing primarily plastics.

Further complicating this selection is the limited data available in the literature to define the fuel properties accurately for a selected design fire, which will often contain a mix of fuel types rather than one specific fuel. In addition, even the most knowledgeable fire protection system designer may not have the experience to discriminate between similar fuel properties; for example, "is my fuel package expected to burn at 8,600 Btu/lb (20 kJ/g), or 17,200 Btu/lb (40 kJ/g)?" Given this limitation, once again the only choice is to be conservative in the selection of fuel properties.

Choosing a visibility constant
The selection of a visibility constant is an important consideration in the calculation of visibility distance. The visibility constant refers to the optical properties of the object or objects being viewed through smoke during the process of exiting the atrium. The typical values used are 3 (for a normally illuminated object) or 8 (for a backlit exit sign) (see endnote 7). For atria where the exit locations may not be obvious to the occupant unfamiliar with the building or where the occupant has to negotiate around objects or obstructions to reach the exit, such as malls or retail space atria, using the constant of 3 is appropriate.

Use of the constant 8 may be appropriate for confined exiting configurations, such as upper level atrium balconies or hotel corridors open to atria where the exit route is defined and free of obstructions and where there is an exit (e.g., a stairwell) in either direction.

Selection of the visibility constant of 8 for a backlit exit sign is one of the most frequent sources of non-conservatism in an atrium smoke control system design. The design reviewer must carefully weigh the use of this higher visibility constant. The use of the constant for a backlit exit sign (8) versus the constant for a normal object (3) allows over 2.5 times the amount of smoke to be present in the space and still be considered acceptable for an equivalent critical visibility distance.

Egress calculations
The calculation of egress time is a critical component of an atrium smoke control system design because the egress time establishes the point at which smoke conditions in the atrium need to be evaluated. Designers need to consider how occupants will travel and what will affect their travel speeds and flows through egress components. Considerations include occupant notification time needs and occupant travel delays due to impaired or slow-moving occupants. The fire protection literature suggests that a safety factor of at least 2 be applied to the calculated egress time. However, in residential occupancies, adding additional time to account for delays in the reaction of sleeping occupants to a fire alarm may be necessary.

An example — The impact of fuel properties
An analysis was performed to demonstrate how the level of life safety protection provided by an exhaust-type smoke management system designed per NFPA 92B can vary substantially, depending on the fuel properties selected to define the material being burned. Two simulations were conducted using FDS, reflecting greatly different fuel properties for the same design fire.

A FDS model domain was used that corresponded to a four-story hotel with a small lobby atrium — a typical small hotel built around the United States. This type of small atrium was selected because designers will often seek to limit atrium smoke exhaust quantities in these spaces due to space restrictions for the necessary equipment or cost restrictions that may exist for these small hotels.

A fast-growth rate T-squared design fire was assumed, having a peak heat release rate of 5,000 Btu/s (5,278 kW). This design fire is consistent with the 2000 and 2003 editions of the IBC. The required atrium exhaust quantity was calculated using the equations included in NFPA 92B to determine the minimum exhaust necessary to maintain the smoke layer interface 6 feet above the highest level open to the atrium, which was the corridor and elevator lobby on the fourth floor of the hotel.

This corresponds to a height of 36 feet (10 meters) above the base of a fire on the ground floor of the atrium. For a 5,000 Btu/s (5,278 kW) fire, the prescriptive exhaust quantity was calculated to be 170,000 cfm (290,000 m3/h).

An equal amount of supply make-up air was provided at the lowest portion of the atrium with sufficient inlet area as to ensure the make-up air velocity was between 200-300 fpm (1-1.5 m/s), so as not to disturb the fire.

In evaluating the atrium using FDS, two different types of fuel packages were considered. The first fuel package is representative of upholstered furniture and contained a mix of wood and polyurethane foam. The second fuel package consists entirely of plastic materials. The fuel properties used were:

For the purposes of this example, tenability was evaluated solely in terms of visibility distance. A critical visibility distance of 30 feet (10 meters) to a normal object (visibility constant = 3) was used. The results show that changes in fuel properties (soot yield and heat of combustion) greatly change the visibility conditions within the space. Visibility conditions are above the minimum threshold value (30 feet (10 meters)) for the duration of the run for the fuel package postulated for upholstered furniture. For the foam and plastics fire, unacceptable visibility conditions develop below the critical smoke layer interface. Visibility on the fourth floor is much lower than the critical threshold. Clearly, the selection of fuel properties directly impacts the assessment of the effectiveness of the system design.

A minimum basis for design and using a checklist
The need to make conservative assumptions when using FDS to evaluate atrium smoke control systems is even more important when the proposed design represents a drastic reduction in smoke exhaust quantity.

When exposing occupants to smoke, as is the basis of the tenability approach, the designer is responsible for building sufficient conservatism into the atrium’s design to ensure that an adequate degree of life safety protection is provided during worst-case conditions.

Fire marshals and other AHJs should make sure they aren’t fooled by a pretty FDS model simulation and should carefully review the design and be aware of the impact the various design assumptions have on the FDS model results. Where the design reviewer lacks familiarity with the model or has reservations as to the basis for the design, obtaining the opinion of a third-party reviewer having experience using FDS as part of the approvals process may be appropriate.

As a guide for the design reviewer, a basic checklist (see sidebar on Evaluating an Atrium Smoke Control System Design — A Checklist) of critical design assumptions aids in the review and approval of an atrium smoke control system designed with the aid of FDS. View the checklist as the starting point for discussion between the designer and the design reviewer. The designer may have perfectly valid reasons for varying from the criteria recommended by the checklist. As always, the goal of this discussion should be ensuring that the smoke control system design provides a sufficient degree of life safety protection for building occupants.

Endnotes

1. Satula, J., "Applications of the Fire Dynamics Simulator in Fire Protection Engineering Consulting,"
   Fire Protection Engineering, Spring, 2002.
2. Baum, H. R., McGrattan, K. B., & Rehm, R. G. (1996), "Large Eddy Simulations of Smoke Movement in Three Dimensions," in C. Franks & S. Grayson (eds.), Interflam ‘96 (pp. 189- 198), London: Interscience Communications Ltd.
3. McGrattan, K., "Fire Dynamics Simulator (Version 4) Technical Reference Guide," NIST SP 1018, National Institute of Standards and Technology,
   Gaithersburg, MD, 2005.
4. McGrattan, K. (2007), Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications,
   Volume 7: Fire Dynamics Simulator, U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research (RES), Rockville, MD, and Electric Power Research Institute (EPRI), Palo Alto, CA. NUREG-1824 and EPRI 1011999, May 2007.
5. Gobeau, N., et al., "Guidance for HSE Inspectors:
   Smoke Movement in Complex Enclosed Spaces and Assessment of Computational Fluid Dynamics," HSL/2002/29, Health and Safety Laboratory, Buxton, UK, 2002.
6. Gobeau, N., "Evaluation of CFD Methods for Predicting Smoke Movement in Enclosed Spaces," Fire Protection Engineering, Spring, 2007.
7. Klote, J. H., and Milke, J. A. (2002), Principles of Smoke Management, American Society for Heating
   Refrigeration and Air Conditioning Engineers,
   Atlanta, GA, 2002.
8. Mitler, H. E., Input Data for Fire Modeling, National
   Institute of Standards and Technology, Gaithersburg,
   MD, February 1996.
9. ISO, "Life Threatening Components of Fires—Guidance on the Estimation of Time Available for Escape Using Fire Data," ISO/TS 13571, Geneva, 2002.
10. Milke, J.A., Hugue, D.E., Hoskins, B.L., and Carroll, J.P. , "Tenability Analyses in Performance Based Design," Fire Protection Engineering, Fall, 2005.