NFPA 92 guides smoke control system design

NFPA 92 is the starting point for any smoke control system, but too frequently it is assumed to be the only standard by which to judge smoke control calculations.
By Will Clay, PE, WSP, Houston May 11, 2018

This article has been peer-reviewed.Learning objectives:

  • Know the limits of NFPA 92: Standard for Smoke Control Systems.

  • Better understand the role engineering judgment plays in applying NFPA 92.

  • Learn common misconceptions when applying NFPA 92.


NFPA 92: Standard for Smoke Control Systems is the gold standard for the design of smoke control systems in the United States. Referenced by both the International Code Council and NFPA codes and standards, it’s the starting point for any smoke control system design.

However, sometimes NFPA 92 is used as a panacea to solve any number of problems for which the standard may not be the correct prescription. NFPA 92 should be the starting point for any smoke control system design, but it’s important to recognize the situations where using only NFPA 92 is inappropriate. In these situations, it may be necessary to rely on computer smoke modeling, the ASHRAE Handbook of Smoke Control Engineering, the Society of Fire Protection Engineers’ (SFPE) Handbook of Fire Protection Engineering, or basic engineering judgment to design smoke control systems.

To start, what is NFPA 92 appropriate in a broad sense? The 2015 edition of NFPA 92 reads as follows regarding the scope of the document: “This standard shall apply to the design, installation, acceptance testing, operation, and ongoing periodic testing of smoke control systems … ” It then goes on to list the purposes of the document including preventing smoke from entering safe areas, such as stairs and shafts; maintaining tenability in the means of egress; preventing migration between smoke zones; providing conditions outside of the smoke zone to assist with emergency response; and mitigating the risk posed to life and property.

So, NFPA 92 can be used to design smoke control systems. Simple enough, and on the surface, this covers a very broad scope. However, inside these boundaries are gaps where the standard alone is insufficient to address every aspect of a smoke control design and require the engineer to rely on engineering judgment or an entirely different standard/process.

Figure 1: This graph shows the wide range of heat-release rates based on time and differing growth factors. All graphics courtesy: WSP USAWhat NFPA 92 does not do

Even when NFPA 92 provides the appropriate path, there are still things the document does not do. Most importantly, it does not specify fire characteristics for design fire events. These scenarios should be selected by an engineer who has experience in evaluating/determining fire scenarios. Appendix B does provide some information for common fire sizes, but it is still up to the engineer to determine which of those, if any, are appropriate.

Furthermore, the engineer must determine the growth rate for the fire, though often this is avoided because a steady-state fire is assumed. Growth rates can vary widely (see Figure 1) and significantly impact the fire size.

NFPA 92 also does not state how tenable or safe an environment will be. It provides a set of prescriptive requirements and calculations, and by meeting these, it’s accepted that a sufficient level of safety is provided. NFPA 92 will not tell you where smoke is nor how dense, hazardous, or hot the smoke is in the zone. Things like temperature can be calculated, but these are boundary values for the purpose of use in calculations. In a real fire scenario, the calculated smoke layer temperature will likely differ significantly from the calculated value in addition to varying inside the smoke layer itself.

NFPA 92 does not address environmental effects. Criteria like winter and summer temperatures, wind speed, and stack effect can all have a significant effect on the operation of a smoke control system, especially when it comes to determining make-up air for smoke-exhaust systems.

Because of these factors, not just any engineer can pick up a copy of NFPA 92 or use a calculation spreadsheet to determine the performance criteria for a smoke control system. NFPA 92 should be considered a supplement to, not a substitute for, experience and engineering judgment.

Figure 2: This example shows a complex geometry where NFPA 92 calculations are not robust enough to provide reasonable conclusions. This model was generated using PyroSim.How NFPA 92 is misapplied

This section details real-world mistakes when applying NFPA 92. This is not intended to be an indictment of anyone who has made one of these mistakes before, but rather as a guide to prevent engineers from making these mistakes in the future. Every person has blind spots and gaps and misses things at times, but engineers should seek to at least minimize, if not eliminate, these faux pas.

Fire size is possibly the most important variable for smoke control calculations, but unfortunately, it is an area of great uncertainty. While NFPA 92 provides some equations for determining some characteristics of the fire, the most important piece—heat-release rate—is not established prescriptively. Whereas previous code editions (and some jurisdictions with this still in their DNA) specified a minimum fire size of 5 MW, the current International Building Code and NFPA 92 do not.

While engineers are always looking for prescriptive requirements to reduce personal liability, NFPA instead defers to the engineer’s judgment while providing some helpful, though limited, examples. Fire sizes of 100 to 500 kW are sometimes proposed for calculations, which are on the order of magnitude of a trash can fire or a wooden chair with minimal padding, but there are few to no situations where this is a reasonably conservative fire size without including sprinkler activation.

ASHRAE suggests a minimum fire size of 2,100 kW for a transient fire, which is a good starting point, but ASHRAE cautions against using this for every scenario. This heat-release rate is approximately that of a two-seater foam sofa, but other pieces (or arrangements) of furniture can readily exceed this, especially when sprinklers are not present or are too high to control the fire. Furthermore, while furniture is the common culprit for the worst-case fire scenario, it is not the only possible scenario, which can include sources like hazardous material spills, kiosks, art exhibits, and Christmas trees.

Often a fast-growth fire is assumed, regardless of the fire source, and the fire grows until it is controlled by sprinkler activation, at which point the heat-release rate for the fire remains constant for the duration of the evaluation. This is a reasonable, if not overly conservative, approach, but how is sprinkler-activation time determined?

Table 1: This demonstrates the significant differences between sprinkler-activating times when inappropriately using Alpert correlations for quasi-steady fires. Commonly, the Alpert correlation (detailed in NFPA’s Fire Technology, volume 8, but referenced in SFPE’s Design of Detection Systems) is used to calculate sprinkler-activation times, but given the fast-growth fire situation described above, this is a mistake. The Alpert correlation should only be used for steady-state fires. Either the Beyler correlation (detailed in “A Design Method for Flaming Fire Detection,” Fire Technology, volume 20, issue 4 but referenced in SFPE) or a quasi-steady stepped method should be used. An example comparing the results of Alpert and Beyler correlations is shown in Table 1.

Note that for smaller fires with a fast growth factor, if the time to sprinkler activation has been calculated using Alpert, the fire size will exceed the initial fire size used in Alpert, indicating the sprinklers will never activate. For larger fires, the fire does not have time to reach the specified heat-release rate used in Alpert.

Compare this with Beyler in Table 1, where time to activation is based on growth rates and not predicted peak heat-release rates. The quasi-steady stepped method is not illustrated in this table, but models a fire using a series of Alpert correlations with small time intervals, essentially modeling a curved growth with discrete, stepped increases.

Misapplying equations

NFPA 92’s equations are fairly straightforward and provide engineers with the boundaries of where these equations are appropriate, but in the end, the engineer must be familiar with these boundaries to use these equations effectively. Almost every atrium design will include an axisymmetric plume, but if there is any sort of balcony, overhang, or some sort of feature that involves two levels of horizontal construction in the atrium, a balcony spill-plume condition must be evaluated.

Furthermore, it’s sometimes assumed that the balcony width is solely based on the plume width at the height of the ceiling. This could not be further from the truth, which is specifically addressed by the NFPA 92 equation that states the width of the balcony (W) is equal to the width of the opening (w) (often the width of the plume at the ceiling height) plus the depth of the location of the opening/plume from the balcony (b). Where balconies are created by waiting areas, this can cause the exhaust rate to spiral out of control and require large (100,000+ cfm) exhaust rates for small atriums. This calculation cannot be ignored. Often, the best resolution is to run a fire model to show a lesser exhaust rate is required.

Opposed airflow can be used to keep smoke in a communicating space but should not be used in lieu of normal exhaust calculations. It calculates the amount of air needed to be injected—not exhausted—to maintain a boundary between two areas.

Design pressure differences is a feasible concept in smoke control, but this method is practically limited to smaller applications, like exit stairs. Sometimes this method is suggested in lieu of an exhaust calculation for atrium smoke control, but this denies the full scope of where the pressure differential must be maintained. If 0.05 in. wg must be provided inside an atrium to keep smoke from migrating to other spaces, this negative pressure must be maintained along the entire boundary and not just at the doors connecting the atrium to the rest of the building.

Furthermore, it is not merely the leakage across the separating wall that must be accounted for, but rather the leakage for the entire atrium, which grows the needed exhaust rate very quickly. This also does nothing to address maintaining the smoke layer inside the atrium, which is needed to ensure occupants inside the atrium are afforded the same level of safety. These calculations are best left to situations where smoke is being compartmentalized and is separate from required exit access, like in defend-in-place occupancies or exit enclosures.

Make-up air

Mechanical make-up air is often undesirable because it means that much more building space must be dedicated to ductwork in addition to the initial and maintenance costs for more fans. A common alternative is to use automatic opening doors and windows or louvers to the exterior to provide the needed make-up air. NFPA 92 provides little guidance on the locations of these openings, only requiring that they be accounted for. It’s common for engineers to locate these openings around the perimeter on multiple sides to mitigate the effects of wind.

However, this is not the best approach based on available literature. John H. Klote, PhD, PE, states the following in ASHRAE’s Smoke Control Handbook, Chapter 5, Fire Science and Design Fires:

When make-up air openings face in different directions, wind forces can result in velocities exceeding 200 fpm (1.02 m/second) inside the atrium. The wind can “blow” into openings facing one direction and out the other openings. A simple approach for minimizing wind effects inside an atrium is to have all the make-up air openings face in the same direction.

While high-wind speeds could still result in localized make-up airspeeds in excess of 200 fpm, if the openings are facing the same direction, the space will become pressurized, thus eventually mitigating the effects of the wind. However, if the openings are on opposite sites, the atrium could act as a wind tunnel, resulting in continuous, significant plume disruption.

Anyone who has opened multiple windows on a warm and windy spring day can attest to this phenomenon. While these velocities could be accounted for in a smoke model, NFPA 92 does not provide the ability to do this on its own, and merely requires that the make-up air velocity be limited to 200 fpm and that wind be accounted for. Without further justification, the make-up air openings should be located so that they face the same direction.

Furthermore, determining the area for make-up air openings is not as simple as taking the exhaust rate, dividing by 200 fpm. While the basic math is accurate, the practical effect of this is not obvious. This is a benefit of fire protection engineers (FPEs) working alongside mechanical, electrical, and plumbing (MEP) engineers within the same firm, as opposed to flitting in an out of a project as a consultant. The mechanical engineers generally are better at understanding the actual airflows.  

If the FPE determines that 100,000 cfm of exhaust is needed and the exhaust is preferred to provide make-up air via natural means, then a minimum of 500 sq ft of make-up air openings (ignoring take-offs for leakage) is required. However, this is not 500 sq ft of louvers. This is the amount of free area needed for the openings. This could be accomplished by 500 sq ft of automatic windows and doors that open at least 90 deg. But if louvers are used in lieu of windows and doors, the required area is going to increase because louvers are not 100% free area. It’s important to remember this when quoting the required area for make-up openings, as aesthetically there is a big difference between 500 and 1,000 sq ft of louvers.

Smoke layer height

A small but pivotal section in the beginning of NFPA 92 and its appendix language, both of which are often overlooked, reads as follows:

4.5.1.3 Minimum Design Smoke Layer Depth. The minimum design depth of the smoke layer for a smoke-management system shall be either of the following:

(1) Twenty percent of the floor-to-ceiling height.

(2) Based on an engineering analysis.

A.4.5.1.3 The depth of the smoke layer depends on many factors and generally ranges from 10% to 20% of the floor to ceiling height. An engineering analysis of the depth of the smoke layer can be done by comparison with full-scale experimental data, scale modeling, or CFD [computational fluid dynamics] modeling.

This means that if an atrium is 40 ft tall and the highest walking surface is at 32 ft, the calculations are not appropriate for maintaining the smoke layer at 38 ft, as this leaves a smoke layer depth of only 2 ft. If this is the case, another method, likely CFD modeling, must be the basis of the smoke control design.

Complicated geometries

It’s important to understand just what the calculations in NFPA 92 are trying to accomplish. They are not trying to describe exactly where smoke will be in every fire situation, nor how hazardous the smoke will be. The calculations are to provide estimations for fire protection/mechanical designs based on limited criteria to provide an acceptable level of life safety.

Because of their limited scope, these equations function on a concept similar to that of a zone model like the one used by the Consolidated Model of Fire and Smoke Transport (CFAST) program: At any point, there is either smoke or there is not. Smoke exists above the smoke layer interface, and smoke does not exist below it. Inside one compartment there is smoke, but across the pressurized boundary, there is not. Smoke is exhausted from the smoke layer and air is not, provided the exhaust openings are appropriately spaced. For simple situations, these calculations are robust and provide an acceptable, if not conservative, level of life safety.

However, these calculations do not address many situations: smoke impingement on multiple balcony levels, acceptable quantities of plugholing, make-up air exhaust speeds higher than 200 fpm, and acceptable exposure to smoke. Any of these situations makes NFPA 92 inappropriate in and of itself. This can be supplemented by engineering judgment, but that judgment ideally is based on more than a gut feeling.

Often, the best foundation for this judgment should be a computer model. Fire Dynamics Simulator and Smokeview software, both produced by the National Institute for Standards and Technology, have become the gold standard for any kind of modeling other than simple pressurization calculations. See Figure 3.

Figure 3: In this case, an example of the results in Smokeview can be obtained by using Fire Dynamics Simulator (FDS) to model a fire. NFPA 92 in a vacuum

Often, NFPA 92 is used in a vacuum. Engineers seek to open the standard and find every piece they need for preparing a rational analysis for a smoke control system, but this is not an appropriate use of the document. NFPA 92 does not specify that the smoke layer interface must be maintained 6 ft above walking surfaces or how long this condition must persist. It does not specify building leakage, though some examples are provided in the appendices.

If there is one point this article insists on, it is that any person cannot just pick up the standard and design a smoke control system. This standard is intended to be used by engineers and supplemented by their own judgment and experience. It is a guide and a tool, not a completely independent design method.

The intent of this document is not to condemn engineers misusing NFPA 92, but to push for a fully integrated FPE who is knowledgeable of NFPA 92 and its limits for projects involving smoke control. Generally, projects run more smoothly and there are fewer surprises during construction with fully integrated FPEs. This doesn’t necessarily mean that the FPE needs to work for the same company as the other consulting engineers or attend every single design meeting, but there needs to be a consistent dialogue not just between the FPE and the architect, but also between the FPE and the MEP designers.


Will Clay is a senior engineer at WSP USA, with 8 years of experience in the fire protection engineering and life safety consulting industry.

Want this article on your website? Click here to sign up for a free account in ContentStream® and make that happen.