Available via license: CC BY-NC-ND
Content may be subject to copyright.
International Journal of Technology (2017) 7: 1296-1305
ISSN 2086-9614 © IJTech 2017
MODELING OF SMOKE CONTROL IN UNDERGROUND PARKING-GARAGE
FIRES
Beline Alianto1, Nabilla Astari1, Darendra Nareshwara1, Yulianto Sulistyo Nugroho1*
1Department of Mechanical Engineering, Faculty of Engineering, Universitas Indonesia, Kampus UI
Depok, Depok 16424, Indonesia
(Received: February 2017 / Revised: July 2017 / Accepted: December 2017)
ABSTRACT
Smoke is the real threat in a fire in an enclosed, underground parking garage, and is a
significant issue and very critical for firefighters dealing with fire. Special attention should be
paid to fire safety, especially for those with multiple floors underground. A model of the smoke
movement in a basement was established using Fire Dynamics Simulator (FDS) 6.0 software. In
this paper, the study undertaken used a basement model of 60 m (length) × 30 m (width) × 3 m
(height) and has three typical floors. Smoke ventilation shafts were provided for the basement.
A well-controlled liquid pool fire with a heat release rate (HRR) of 2 MW was used as an input
parameter. The ventilation strategy was achieved through a mechanical exhaust fans and make-
up air fans. The required ventilation was based on the air changing ten times per hour. The
following parameters were varied: the location of the fire, the presence or absence of sprinklers,
the presence or absence of a smoke-extraction system, the presence or absence of openings for
incoming air, and the presence or absence of a jet fan and ducting. The impact of jet fans
(induction type) was also addressed in several simulations. Smoke modeling was investigated
under different fire scenarios. Simulations were conducted for smoke and heat control using
forced, mechanical, horizontal ventilation. When the combination of a jet fan and ducting was
applied, the fastest time for smoke removal was achieved compared to the other scenarios.
Keywords: Basement ventilation; Fire; Jet fan; Mechanical fan; Smoke exhaust
1. INTRODUCTION
Underground parking garages generally provide enclosed conditions. While the effectiveness of
fire protection systems is commonly associated with active water-based sprinkler and hydrant
protection systems, smoke movement is critical and is subject to failure because most design
engineers do not have a proper understanding of what happens when a fire breaks out. As
smoke flows through the building, it is restricted by the frictional forces that inevitably exist
due to the construction of the building. A large amount of smoke is generated, which leads to
unpredictable smoke movement and poor conditions for evacuation. Chow (1998) divides
underground fire protection systems into mechanical ventilation systems and fire protection
systems.
Various past experimental studies of single vehicle fires was done by Mangs and Keski-
Rahkonen (1994), Shipp and Spearpoint (1995), and Cheng and John (2002), for which the
main objective was to determine heat release rate (HRR) of vehicle fires. Further investigations
have been carried out in enclosed-basement fires, which focus on smoke movement (Hwang &
*Corresponding author’s email: yulianto@eng.ui.ac.id, Tel. +62-21-7270032, Fax. +62-21-7270033
Permalink/DOI: https://doi.org/10.14716/ijtech.v8i7.779
Nugroho et al. 1297
Edwards, 2005; Vauquelin & Telle, 2005) and fire development (Ingason & Lonnermark,
2005). Moreover, the full-scale experiments on modern cars reveal high HRR values for fires,
which exceed 16 MW when three cars are on fire (Horvath et al., 2013). Most of the building’s
utilities are located in the basement. Electrical cables are also categorized as a fire hazard.
Pangaribuan et al. (2016) concludes that the spread of fire in polyvinyl chloride (PVC)
insulated cables can be eliminated by reducing the oxygen concentration.
Engineering standards, regulations, and codes are always used; they are briefly explained in the
framework of prescriptive codes. However, different practices are carried out. Performance-
based fire safety will be applied in the near future; for example, a study on performance-based
fire safety was conducted by Sujatmiko et al. (2014).
Common practice concerns the use of mechanical, horizontal ventilation, which extracts hot
smoke from certain points in the parking garage and keeps it clear from smoke. This is similar
to longitudinal ventilation in tunnels (Gao et al., 2012). Parking garages, however, have a
substantially different height-to-width aspect ratio compared to tunnels. Consequently, the
principles longitudinal ventilation, which have been developed and predicted for tunnels,
becomes questionable for parking garages because of the complex flow patterns that might
occur.
Full-scale experiments are the best way to obtain valuable information. For instance, Horvarth
et al. (2013) performed a full-scale test and a reduced-scale test of parking garages on fire using
different heat release values (500–4,000 kW) with different inlet air velocities (0.3–2 m/s), and
derived the empirical critical velocity based on the heat rate. However, the huge costs and the
limited number of tests mean the study of this topic is challenging.
While most of the aforementioned investigations were done experimentally, Tilley et al. (2012)
conducted a large set (more than 350) of computational fluid dynamics (CFD) simulations have
been used as “numerical experiments.” In addition, Deckers et al. (2013a) also presents CFD
simulations for large parking garages and full-scale experiments for car-park fires. Wu et al.
(2017) applies natural underground ventilation in a study, and concludes that the total
ventilation area and the height of the shafts are the key factors influencing smoke control, and
that multistage shafts can be used for smoke control in the platform layer due to the formation
of the stack effect.
Common practice for ventilation is having a mechanical exhaust at each level with make-up air
provided by a fan and transported via the impulse ventilation from jet fans. An advantage of jet
fans is that they eliminate air-distribution ductwork within the parking garages, and thus reduce
supply and exhaust air-flow resistance. Various studies have been done on jet fans; for example,
Viegas (2010) ran a numerical CFD simulation involving the ventilation for covered parking
garages to investigate the control the smoke in the event of a fire, and to determine the
interaction between the fire ceiling jet and the flow driven by the jet fans.
A full-scale experiment on car-park fires with jet-fan ventilation was done by Deckers et al.
(2013b) reveals that the impact of the exact position of the extraction fans on the smoke pattern
is small when the extraction fans are not close to the fire source. This study also suggests that
the primary effect of jet fans (induction type, 50 N) does not have a significant impact on the
global flow pattern, and, for the cooling effect to be observed, the jet fans must not be in a
smoke-filled region. Other numerical simulations include the research conducted by Lu et al.
(2011) on impulse ventilation systems, the study on the impact of jet fans during the activation
of sprinklers in basement fires performed by Enright (2014), and an investigation of smoke
movement in basement structures by Arini (2017), which uses a reduced-scale model and FDS
simulation.
1298 Modeling of Smoke Control in Underground Parking-garage Fires
Most of the studies discussed previously include a model for a single-story underground
parking garage. It is also important to consider multi-story basements. Furthermore, only
limited fire scenarios were investigated. This paper discusses a case study undertaken on a
three-story underground parking-garage building of approximately 1800 m2, with each floor of
the underground car-park building being approximately 60 m (length) × 30 m (width) × 3 m
(height). Different locations of fires breaking out are studied. This paper focuses on smoke
modeling and the prediction of smoke movement in several fire scenarios. The time required to
clear the smoke from the basement is also discussed. The characteristics and prediction of
smoke movement when fires break out on different floors are elaborated.
2. SMOKE VENTILATION
Automobile parking garages can be either enclosed or partially opened. Partially open parking
garages are generally above ground with open sides. A combination of natural and forced
ventilation can be used for partially open parking garages. Ventilation in enclosed underground
spaces presents many concerns. American National Standards Institute (ANSI) / American
Society of Heating, Refrigerating and Air-Conditioning Engineers (AHSRAE) 62.1
recommends that a flat ventilation rate of 0.0075 m3/(s.m2) equal to 6 ACH is used for enclosed
parking garages (ASHRAE Handbook, 2015).
The use of mechanical intake fans can cause problems with mechanical extraction. This is
because the exhaust air that is extracted has a greater volume than the inlet air. As the fire
grows and declines, there is mismatch in volume between the inlet air and the extracted,
warmed air. A moving airstream that has a lower pressure than the still air will attract the
stationary air toward itself. The force of attraction increases as the velocity of the airstream
increases. The system should be designed to run in two parts; commonly, each part is arranged
so that 50% of the outlets are at high levels and 50% are at low levels, according to many
smoke ventilation codes (Chartered Institution of Building Services Engineers Guide B
[CIBSE], 2005) and BS 7346-7:2006 (British Standards Institute, 2006). Ramps can also serve
as natural ventilation openings.
3. SETUP OF THE SIMULATION
In the simulations, a basic configuration was chosen for the parking garage. In this
configuration, as shown in Figure 1, the underground parking garage has three floors below
ground containing the fire source and an array of sprinklers. The model’s size is 64 m (length) ×
30 m (width) × 13 m (height). Intake and exhaust ventilation shafts that are 2 m wide are
provided at the left and right sides of the basement. For expedience of calculation, the model’s
domain is deliberately set to discrete cells of size 50 cm × 50 cm × 25 cm in all simulations,
which results in 199,680 cells in this case. The simulation is performed using Fire Dynamics
Simulator (FDS) V.6.
The major assumption that was applied in this case is that the flow of smoke throughout the
building occurs steadily. It was also assumed that there was also negligible migration of smoke
between floors through the floor slabs compared to the amount of smoke that travels vertically.
The entire front and back wall were considered to be adiabatic, meaning there would be no
additional heat loss from the structure, while the two walls were modeled to have smoke
extraction on one side and air entering on the opposite side. The ventilation air-flow pattern was
essentially unidirectional.
The dimensions of the pool fire were 1 m × 1 m. The Heat Release Rate (HRR) of the fire is q =
2.000 kW/m2. In this case, a fast t2 growth rate was assumed. The CO yield (YCO) was taken to
be 0.006 and the soot yield (Ys) was 0.015. The fuel chosen was heptane is selected as a fuel.
Nugroho et al. 1299
Three locations of fire sources were simulated among those fire scenarios. A few sprinklers
were added above the fire source and separated to such a distance that each sprinkler covered a
protection area of 12 m2. Each sprinkler also operated at 0.5 bar pressure and had a K-value of
80, which met the conditions for the category of ordinary hazard (National Fire Protection
Agency [NFPA], 2013). Three sprinklers are then separated from each other by 4 m in groups
as indicated by the blue color dots in Figure 1. Table 1 shows the configuration of the
mechanical fans designed for use in the basement.
Figure 1 Building model and position of mechanical fans
The sizing of every fan is based on the air changing ten times per hour for the basement during
the fire, resulting in an air-flow quantity of 7.5 m3/s per unit for the exhaust and intake fans.
Moreover, the jet fans are necessary because of the advantages they provide through
eliminating the air-distribution ductwork within the parking garages. Therefore, in order to push
air from air the intake side to the negative-pressure smoke-extract side, the parking-garage
models contain three rows of jet fans. Each row consists of five jet-fan units, each with a
capacity of 2.0 m3/s.
The simulation was run for ten fire scenarios (FS). There were neither sprinklers nor
mechanical exhaust fans for FS1 and FS2, but both existed in FS3. Mechanical fans were
provided in FS4. Two units of make-up air fans were installed to deliver make-up air from the
atmosphere to the parking garages.
1300 Modeling of Smoke Control in Underground Parking-garage Fires
Table 1 Mechanical fans specification
Without ductwork (Fire Scenarios 4–9)
Capacity of each fan
Basement 3: SMF-B3/1,2; SMF-B2/1,2; SMF-B1/1,2
@7.5 m3/s
Basement 2: MAF-B3/1,2; MAF-B2/1,2; MAF-B1/1,2
@7.5 m3/s
Basement 1: Jet fans, 15 units, three rows with five fans/row
@2 m3/s
With ductwork (Fire Scenario 10)
Basement 3
Capacity of each fan
MAF-B3/1,2; MAF-B2/1,2
@7.5 m3/s
EF-B3/H-1; EF-B3/H-2
@3.75 m3/s
EF-B3/L-1; EF-B3/L2
@2.5 m3/s
Jet fans, five units/row, three rows
@2 m3/s
Half of the basement was treated as an extraction zone and the other was the make-up air zone.
The position of the mechanical fans can be seen in Figure 2.
Figure 2 Setup simulation for FS10
Table 2 Availability of the fire protection devices
Nugroho et al. 1301
Furthermore, the inclusion of jet fans was simulated in FS5 to FS9. Make-up air was delivered
by make-up air fans and then induced to the extraction zone by the jet fans. FS10 included the
advantage of having ducting, as illustrated in Figure 2; in the extraction zone, the jet fans were
removed and ducting was applied as the media of smoke extraction. The operation mode,
equipment specification, and simulation scenarios are shown in Table 2. Figure 3 illustrates the
variations in the location of the fire.
Figure 3 Location of the fire and sprinklers above fire source (red dots)
4. RESULT AND DISCUSSION
When the fires were ignited at each location, they were located a significant distance below the
ceiling sprinkler heads. There were three sprinklers in a row, each separated from each other by
4 m. The sprinkler in the middle was located 2.5 m above the fire source. The sprinklers are
generally assumed to activate immediately for a fire sizes of 2 MW. When the heat spreads in to
the adjacent areas, numerous sprinkler heads may activate due to the elevated ambient ceiling
temperatures. When too many sprinkler heads activate, the water pressure is significantly
reduced and the delivered water density decreases. This could result in ineffective sprinkler
operation and an inability to control the fire. In this experiment, the sprinklers operated quickly
for all intentional fire scenarios. The approximate sprinkler activation temperature after ignition
was 57°C
FS1: Neither sprinklers nor mechanical ventilation systems were provided. The simulation
shows smoke movement within a period of less than 1,400 sec. Smoke filled the volume of
basement 3 (B3) within less than 150 sec, and then rose upward through the ramp opening.
Figure 4 shows that smoke subsequently filled basement 1 (B1) and a small volume of smoke
started to fill basement 2 (B2). In a period of less than 1400 sec, B1 was completely filled with
smoke and most likely a large part of B2 was filled as well. Visibility in B3 was diminished to
less than 14 m. A total loss of visibility would be expected and would have been very unlikely
that any entry was available for firefighters to extinguish the fire since the basement was filled
with smoke.
FS2: When there is no sprinkler system activation, an intentional fire can grow out of control.
An HRR greater than 2 MW can occur. However, this study is limited to the spread of smoke,
the spread of flames is neglected, and the surface of the building structure is considered to be
inert and adiabatic. Smoke filled B3 within 157 sec after the ignition, and smoke subsequently
spread upstairs through the ramp opening. B1 was filled first before smoke started to fill B2 due
1302 Modeling of Smoke Control in Underground Parking-garage Fires
to the density of smoke, which is lighter than air. Unlike FS1, it took a longer time for the
smoke to fill the basement; for example, in t = 1800 sec the smoke density was less in B1, but
the room was completely filled with smoke at that time. Although less smoke was produced in
FS2 compared to FS1, there would have been limited ability to fight the fire that had occurred
due to reduced visibility and increased levels of toxicity caused by smoke, despite the fact that
the firefighters could enter the space. It is very unlikely that successful entry and aggressive
internal firefighting operations would be possible for such a fire.
Smoke spread at t = 565 sec
Smoke filled the basement at t = 1371 sec
Figure 4 Smoke spread over time and soot visibility for FS1
FS3: The smoke produced started to move upward to the outside through the ramp opening.
The activation of the sprinklers caused a reduction in HRR, meaning that the fire started to
decay. Without forced air circulation by mechanical fans, air rises naturally. At t = 412 sec, the
smoke was apparently cleared in B2. Because there was not a great amount of smoke produced
in the basement complete with sprinklers, the smoke is not likely to flow outdoor instead of
enter the above floors.
FS4: Fire occurred at the center of B3. Sprinklers and a central mechanical ventilation system
were provided. The air-flow pattern can be seen in Figure 5. The smoke was cleared by t = 250
sec, which is faster than in an unventilated basement.
Figure 5 Velocity profile for FS4
FS5: Fire occurred at the center of B3. The basement was provided with a complete, active fire
protection system, including sprinklers, jet fans, and similar configuration of mechanical fans to
FS4. Two smoke-extraction fans with a capacity of 7.5 m3/s each were provided at one side of
Nugroho et al. 1303
the basement. Two make-up air fans were provided at the opposite side, each with the same
capacity of 7.5 m3/s. Three rows of jet fans, with five fans in each row, distributed the make-up
air to the air-discharge zone. The flow rate of each jet fans was 2 m3/sec. The fire ignited for a
moment and was quickly extinguished approximately 20 sec after the growth of the flaming
fire. The simulation’s result shows that the smoke was apparently cleared within 216 sec, which
is faster than in the basement without jet fans. At the start of the fire, the smoke spread radially
in every direction. Because of the jet fans, the smoke-flow pattern then changed to be in one
direction, which was from the intake fan to the discharge zone.
FS6: This scenario is similar to FS5, except that the fire started in the center of B2. The first
sprinkler was located above the fire source, and the other two sprinklers were located to the left
and right sides of the first sprinkler. The distance between the sprinklers was 2 m, and they
were 2.5 m above floor level. The simulation shows that the smoke initially spread radially in
the first stage of the fire, but then finally changed to flow in one direction toward the extract
side. The smoke was cleared faster than in FS5, in a time of approximately 150 sec.
FS7: The fire was at the center of B1. The scenario is similar to both FS6 and FS5. The smoke
clearance time was longer than for the fire in the lower level. There was still a small amount of
smoke at t = 283 sec, and the smoke was finally cleared in a period of less than 5 min.
FS8: The same scenario as FS5 and FS6 was applied here, but the fire was set in a different
location: in B2 near the location of the smoke-extraction fan. The smoke was cleared faster, by t
= 283 sec.
FS9: The fire was in B3, located near the smoke-extraction fan, and consisted of a fire ignited
by a fuel pool of 1 m2 located near the side with the smoke extraction. The same scenario as
both FS5 and FS6 was also applied here. The smoke was cleared faster than for FS8, in
approximately 122 sec.
FS10: The fire broke out in the middle of B3. The jet fans in the exhaust zone were replaced
with high-level and low-level ductwork. The result of the simulation is shown in Figure 6. The
smoke was cleared within less than 90 sec, which is the fastest among the ten scenarios
discussed in this study.
Figure 6 Smoke spread over time and soot visibility for FF10 at t = 43.2 sec
1304 Modeling of Smoke Control in Underground Parking-garage Fires
5. CONCLUSION
Even though a fire in a basement is a rare event and such a fire may be initiated in numerous
ways, it is important to provide sprinklers and mechanical ventilation in underground parking
garages. Sprinklers are the vital components in an active fire protection system. When sprinkler
activation does not occur, such as in a space with no sprinklers, the ambient ceiling
temperatures are increased and the fire is allowed to continue to burn. In comparison, where
there are no sprinklers applied and proper ventilation, the production of smoke exceeds the
capacity for smoke extraction, which causes the whole basement to be filled by smoke. Even
when sprinklers are provided, the absence of a smoke exhaust fan might lead to smoke
remaining in the basement for a prolonged period of time, in spite of the fact that the fire is
completely extinguished by the sprinklers. The provision of jet fans may result in migration of
smoke that is faster than conventional ventilation without jet fans. This study showed that
having sprinklers, make-up air fans, smoke-extraction fans, jet fans, and ductwork gives the
fastest smoke-removal time in a basement fire.
6. ACKNOWLEDGEMENT
The authors would like to thank Fire Safety Engineering Research Group and Thermodynamics
Laboratory of Department of Mechanical Engineering, Universitas Indonesia for providing
supports and great discussion.
7. REFERENCES
Arini, D., Pancawardani, F., Santoso, M.A., Sugiarto, B., Nugroho, Y.S., 2017. Froude
Modelling of Fire Phenomena: Observation of Fire-induced Smoke Movement in Basement
Structure for Fire Fighting Purpose. Engineering Physics International Conference, EPIC
2016, Procedia Engineering, Volume 170, pp. 182–188
ASHRAE Handbook – Application, 2015. American Society of Heating, Refrigerating and Air
Conditioning Engineers. GA-Atlanta
BS 7346-7:2006. Components for Smoke and Heat Control Systems - Part 7: Code of Practice
on Recommendations and Calculation Methods on Smoke and Heat Control Systems for
Covered Car Parks, 2006. British Standards
Cheng, Y.P., John, R., 2002. Experimental Research of Motorcar Fire. Journal of China
University of Mining and Technology, Volume 31(6), pp. 557–560
Chow, W.K., 1998. On Safety Systems for Underground Car-parks. Tunnelling and
Underground Space Technology, Volume 13(3), pp. 281–287
CIBSE Guide B: Heating, Refrigerating, Air Conditioning and Refrigeration, 2005. The
Chartered Institution of Building Services Engineers. London, UK
Deckers, X., Haga, S., Sette, B., Merci, B., 2013a. Smoke Control in Case of Fire in a Large
Car Park: Full-scale Experiments. Fire Safety Journal, Volume 57, pp. 11–21
Deckers, X., Haga, S., Sette, B., Merci, B., 2013b. Smoke Control in Case of Fire in a Large
Car Park: CFD Simulations of Full Scale Configurations. Fire Safety Journal, Volume 57,
pp. 22–34
Enright, T., 2014. Impact of Jet Fan Ventilation Systems on Sprinkler Activation. Case Studies
in Fire Safety, Volume 1, pp. 1–7
Gao, R., Li, A., Hao, X., Lei, W., Deng, B., 2012. Prediction of the Spread of Smoke in a Huge
Transit Terminal Subway Station under Six Different Fire Scenarios. Tunneling and
Underground Space Technology Journal, Volume 31, pp. 128–138
Horvarth, I., Beeck, J.v., Merci, B., 2013. Full-scale and Reduced-scale Tests on Smoke
Movement in Case of Car Park Fire. Fire Safety Journal, Volume 57, pp. 35–43
Nugroho et al. 1305
Hwang, C.C., Edwards, J.C., 2005. The Critical Ventilation Velocity in Tunnel Fires – A
Computer Simulation. Fire Safety Journal, Volume 40, pp. 213–244
Ingason, H., Lonnermark, A., 2005. Heat Release Rates from Heavy Goods Vehicle Trailer
Fires in Tunnels. Fire Safety Journal, Volume 40, pp.646–668
Lu, S., Wang, Y.H., Zhang, R.F., Zhang, H. P., 2011. Numerical Study on Impulse Ventilation
for Smoke Control in an Underground Car Park. Procedia Engineering, Volume 11, pp.
369–378
Mangs, J., Keski-Rahkonen, O., 1994. Characterisation of the Fire Behaviour of a Burning
Passenger Car. Part II: Parameterization of Measured Rate of Heat Release Curves. Fire
Safety Journal, Volume 23, pp. 37–49
NFPA 13: Standard for the Installation of Sprinkler Systems, 2013. National Fire Protection
Association. 1 Batterymarch Park, Quincy, MA
Pangaribuan, A., Fadhil, Santoso, M.A., Nugroho, Y.S., 2016. Controlling Fire Growth in
Electrical Cable Compartment by Reducing Oxygen Concentration at Horizontal
Orientation. International Journal of Technology, Volume 7(2), pp.332–342
Shipp, M., Spearpoint, M., 1995. Measurements of the Severity of Fires Involving Private
Vehicles. Fire and Materials, Volume 19, pp. 143–151
Sujatmiko, W., Dipojono, H.K., Soelami, F.X.N., Soegijanto, 2014. Performance Based Fire
Safety Evacuation in High-rise Building Flats in Indonesia – A Case Study in Bandung.
Procedia Environmental Sciences, Volume 20, pp. 116–125
Tilley, N., Deckers, X., Merci, B., 2012. CFD Study of Relation between Ventilation Velocity
and Smoke Backlayering Distance in Large Closed Car Parks. Fire Safety Journal, Volume
48, pp. 11–20
Vauquelin, O., Telle, D., 2005, Definition and Experimental Evaluation of the Smoke
“Confinement Velocity” in Tunnel Fires. Fire Safety Journal, Volume 40, pp. 320–330
Viegas, J.C., 2010. The Use of Impulse Ventilation for Smoke Control in Underground Car
Parks. Tunnelling and Underground Space Technology, Volume 25, pp. 42–53
Wu, F., Jiang, J., Zhou, R., Zhao, D., Shi, L., 2017. A New Natural Ventilation Method for
Fire-induced Smoke Control in a Common Subway Station. International Journal of
Ventilation, Volume 0(0), pp.1–18
