Smoke in Atriums

Hot smoke from fire in a large volume space like an atrium can reach multiple levels of a building within mere minutes. The potential threat to life safety is why the National Building Code of Canada (NBC 1995) specifies the need for a protected zone on each floor and requires that egress routes should not force people to pass through atrium spaces that are connected to the fire zone. Under some circumstances, the NBC permits a smoke management system to be used to provide protection for building occupants as an equivalency to the specified protected zones and egress routes.

In contrast, U.S. codes based on the International Building Code (IBC) will allow for an atrium to be used for exiting in a fire emergency, provided a smoke management system is designed to provide a tenable environment for the evacuation or relocation of occupants. This provision is part of the stated objectives. The IBC then lists prescriptive requirements intended to support this goal. One requirement is that the designer must ensure the smoke management system has sufficient exhaust to control the height of the base of the smoke layer to at least 3048 mm (10 feet) above the required egress routes. The exhaust rate is determined by plume equations in that code.

In practice, we have found that, due to the architectural complexity of current atriums, it is often impractical to design smoke management systems that satisfy all the prescriptive requirements of the IBC. For example, often a design fire can be postulated to be located under a low ceiling at the base of the atrium, causing the smoke plume to spread as it spills upward. The balcony-spill plume equation in the code often predicts excessively high required exhaust rates. Conversely, for a design fire out in the middle of an atrium, the axisymmetric plume equation can underpredict the required exhaust, since in real buildings a fire plume interacts with a less than ideal distribution of makeup air and with the architecture.

For these and other situations it is worthwhile to present these challenges and the solution alternatives to the authority-having-jurisdiction early in the design process. It is through this dialogue that you may achieve approval of equivalencies, performance-based design approaches, and the use of computational fluid dynamic (CFD) computer model analysis.

In most cases, the authority will acknowledge the design team’s concerns about the large uncertainty in the exhaust rates calculated using prescriptive code methods. For these concerns and other issues, the authority and the design team might employ an engineering analysis that uses fire plume dynamics, building air flows, and calibrated CFD modelling techniques.

There is a risk associated with a lack of expertise in conducting CFD modelling: often incorrect results presented as colour graphics look convincing but disguise underlying mistakes. These mistakes can include: an under-prediction of smoke plume growth; a lack of resolution of ceiling jet dynamics; improper static pressure effects on external openings; improper accounting for thermal stratification; no accounting for delays in the activation of the emergency systems; and incorrect yield rates of soot and toxic chemicals used in tenability assessments.

Early input into the design

In consulting on many projects, we have found that the design of a smoke management system is primarily affected by architectural design and secondarily by the integration of the mechanical systems within the architecture.

The following points significantly affect the smoke management system design. Ideally, they should be considered during the conceptual and early schematic design phases of the project.

Height of the atrium. Tall open spaces with exposed exit routes on upper levels usually require significantly greater quantities of smoke exhaust. This need is because the rising hot smoke plume will entrain (mix with) more fresh air in the building, requiring more exhaust from the building to keep the upper occupied levels clear. However, if the smoke management system is designed such that exposed exit routes are at lower levels in the building, then the required smoke exhaust volumes (and associated costs) can be reduced substantially (perhaps as much as 50% in some atrium designs).

Smoke reservoir at the top of the atrium. It takes time to detect a fire event, turn off the normal ventilation systems and ramp up the emergency ventilation. Therefore, a volume of building space at the top of an atrium (above the highest level of occupancy open to the atrium) is always needed to collect smoke before the emergency ventilation system is activated. The required depth of the reservoir depends on a number of factors, but can be the equivalent of up to two storey heights for atriums less than 10 stories. The accumulation of smoke in the reservoir provides time for people on upper levels to exit. The reservoir also provides space for the momentum of the hot rising smoke plume to spread along the ceiling, thereby slowing down and reducing the severity of the initial impact in the highest occupied space. At the same time, the reservoir allows the smoke layer to be sufficiently deep so that the exhaust vents at the top of it are well within the layer, ensuring that smoke rather than fresh air can be efficiently extracted.

Complexity of the architecture. Multiple openings from floor to floor make it difficult to contain the spread of smoke unless permanent walls or deployable barriers are used.

Openings through floors. These should be aligned and large in cross-sectional area. If not, then smoke can affect each level, causing additional mixing which may in turn cause difficulties in exiting.

Open balconies, bridges or stairways. These features will likely be compromised by smoke impacts and also will cause extra unwanted mixing of the rising smoke plume. Alternate exit routes or physical protection (e.g. glass walls) may be required.

Narrow atrium. A narrow atrium will cause the smoke plume to have an impact on the adjacent unprotected occupied zones relatively quickly. For a very long narrow atrium, smoke exhaust systems need to be provided along the length of the atrium. This type of arrangement is not ideal as it may mean that air rather than smoke is being exhausted in some sections and that smoke is not being efficiently contained. However, there is little alternative with this architectural form.

Travel distances to two or more exits on each level. The design should allow for rapid exiting of occupants in the smoke zones. It is often desirable to restrict horizontal travel distances to less than 100 feet (30 metres) to multiple exits from the smoke zone.

Occupied spaces on upper levels of the building that require exiting through the atrium. In this situation, there can be concerns about greater delays in exiting and longer exit times. If, on the other hand, the only people who have to exit through an atrium during a fire are those initially in it, exiting times may be very short. This can be an opportunity to propose a cost saving design that controls the rate of descent of the smoke layer in the atrium.

Make-up air. This air should be widely distributed at relatively low air speeds. RWDI recommends speeds less than 0.86 m/s (170 fpm), but definitely not more than the IBC code limit of 1 m/s (200 fpm). As well, the make-up air delivery needs to be distributed so that it does not accelerate to higher speeds through narrow architectural openings into the atrium.

Most of the make-up air requirement is normally planned for the lower levels of the building. There are circumstances where exit routes at higher levels in the atrium can be protected with make-up air to either prevent intrusion of smoke down exit corridors or prolong tenable conditions in exit pathways.

In most cases, the smoke management system will perform best when the make-up air is supplied at or close to the average temperature of the space. If this air is not conditioned, then the worst-case challenge to the design will likely be a postulated fire event on a hot summer
day. But this supposed worst-case condition is counter to some codes (e.g. NBC) that suggest that worst-case conditions for design are during the winter months.

For some elements of the smoke management system the winter worst-case scenario may be appropriate. However make-up air/smoke interactions are aggravated by warmer make-up air. The problem of relatively hot outdoor make-up air occurs in the first few minutes where it can rise quickly in the initially cooler atrium air. The warm air can overwhelm the exhaust and push smoke down into occupied spaces causing adverse impacts. The problem diminishes once sufficient make-up air mixes in the atrium.

Exhaust stack and intake design. The concern is that once the smoke is exhausted from the building it will be carried by the wind to make-up air intakes where it will enter the building and diminish safe exiting. Solutions to this problem can be recommended by an air quality specialist or, in complex situations, with boundary layer wind tunnel testing.

A performance-based design

In one recent study, an atrium was being designed that consisted of multiple levels and connected spaces. The space was outfitted with a smoke management system developed by following the local code. Figure 1 shows a CFD computer model prediction that smoke would penetrate many of the occupied areas of the building, even with a relatively high exhaust rate.

To overcome this problem, RWDI recommended that the atrium be partially segregated and zoned into two smaller, simpler atriums by deployable smoke barriers when a fire is detected. This segregation, or zoning, reduced the required exhaust (make-up) air flow rate by more than a factor of three and significantly reduced the smoke spread, thus providing improved exiting from the building, as shown in Figure 2.

In closing, successful design of a smoke management system for an atrium starts with early input into the architectural design, then integration of mechanical and architectural features that will satisfy the design objectives. CFD computer modelling by experienced specialists can prove the success of a performance-based design, providing improved life safety and saving money.

Ray Sinclair, Ph.D., is a principal, and Duncan Phillips, P.Eng., Ph.D. is an associate, with Rowan Williams Davies & Irwin (RWDI) of Guelph, Ontario. www.rwdi.com