Fire Risk in Metro Tunnels

and Stations

Hyder Consulting

Presented by:
Dr. Leong Poon
Ir. Richard Lau
2 Dec 2005
© Hyder Consulting Pty Ltd
 Metro Tunnels and Stations
– General Characteristics
ƒ Limited to metropolitan area (hence the name)
ƒ Entire network is underground
ƒ Interspersed by stations every 500 – 800m
ƒ Predominantly one-way flow (ie single bore)

© Hyder Consulting Pty Ltd Route 8 Cheung Sha Wan to Shatin

2
 Rail tunnels

Tseung Kwan O Ext., HK New Southern Railway, Sydney West Rail, Mei Foo –
Parammatta Rail Link, Sydney GZ Metro Nam Cheong tunnel , HK

3
© Hyder Consulting Pty Ltd
 Stations and platforms, international

Stratford Station Concourse, UK

Berlin Hauptbahnhof Station, Germany

Federation Square, Melbourne

4
© Hyder Consulting Pty Ltd
 Stations and platforms, East Asia

Guangzhou Line 4 (Huangzhou Station) GZ Metro Line 1

KCRC West Rail DD400, HK Nam Cheong Station, HK Lai King Station, HK

5
© Hyder Consulting Pty Ltd
 Metro Tunnels and Stations
– Safety (or risk) characteristics
ƒ Traffic is well controlled, hence low accident rates
ƒ Combustible material is controlled, hence low fire hazard
ƒ Closely spaced stations allow train to continue to the
station to allow passenger evacuation and fire-fighting
ƒ Single bore tunnels lack escape passages unlike
twin bore tunnels, hence relatively higher risk
ƒ Large concentration of users, hence any incident places
many passengers at risk

6
© Hyder Consulting Pty Ltd
 Metro Tunnels and Stations
– Objectives (of risk assessment)
Risk assessment is used as a design tool to:
ƒ Identify relevant fire risks
ƒ What factors cause incidents/disasters
ƒ Determine key factors for improving safety
ƒ Determine recommendations for cost-effective
fire protection measures

7
© Hyder Consulting Pty Ltd
 Literature Review
– Statistics
ƒ Cause of fires in metro rails:
ƒ Ignition from mechanical/electrical failure, fuel from debris,
cabin material & baggage, terrorist activities?
Mechanical Arson
13% 13%

Cigarette
Station 10%
17%

Not specified
13%
Electrical Fault
34% 8
© Hyder Consulting Pty Ltd
 Literature Review
– Statistics
ƒ Rate of occurrence:
ƒ Small rail fire ~ a few a year
ƒ Severe rail fire ~ 0.5 a year worldwide (Anderson & Paaske)
ƒ 30 severe incidents 1970-1987
ƒ 43 fatalities in 5 incidents (King’s Cross = 31)
ƒ London underground, July 2005 (terrorist attack)
ƒ 50 fatalities (> sum of all past records)
ƒ Demand for rail metro usage increasing
ƒ Throughput of 26 billion passengers a year
ƒ Hence potential exposure higher – ie more at risk
9
© Hyder Consulting Pty Ltd
 Literature Review
– Fire Hazard
ƒ Carriage – main source of fuel + baggage
ƒ Fire size typically between 6-20 MW
ƒ Control of lining material will reduce likelihood of fire
development but not necessarily reduce the fire size
ƒ Terrorist factor ? Significant but highly indeterminate
– best handled through a risk assessment approach

10
© Hyder Consulting Pty Ltd
 Literature Review
– Fire Protection Systems
ƒ Purpose is to detect, warn and control
ƒ For stations, conventional building systems are provided
ƒ For carriages/tunnels, the following are provided:
ƒ Detection: – Smoke detectors in air-conditioned carriages
– Heat detectors/CCTV may be used in tunnels
ƒ Warning: – Communication systems include break-glass, intercom
phone or PA system for staff and passengers
ƒ Control: – Fire suppression systems in engine/equipment areas
– Portable systems in passenger area
ƒ Using risk assessment, the optimal combination of the
above systems can be determined
11
© Hyder Consulting Pty Ltd
 Literature Review
– Smoke control in tunnels
ƒ Smoke control is a key fire protection provision
ƒ Strategy is to take advantage of longitudinal ventilation
ƒ Force smoke downstream in the direction of
travel towards the ventilation shaft to be exhausted
ƒ Passengers take the smoke clear path upstream of air flow
ƒ Smoke control need to accommodate egress requirements:
ƒ Escape stairs may be required for long tunnel sections
ƒ Escape stairs also used by fire fighters to gain access
ƒ Train should continue to the next station to
facilitate egress and fire-fighting access
12
© Hyder Consulting Pty Ltd
 Basic smoke control strategy
– Schematics

Direction of longitudinal ventilation,

Direction of train travel
Exit Smoke clear path Downstream

Occupant evacuation Smoke exhaust

13
© Hyder Consulting Pty Ltd
 Risk assessment concept

ƒ Risk is a measure of the consequence of an event, i.e.

Risk = Probability × Consequence
ƒ Consequence is the estimated measure of the event
eg no of fatalities, cost of damage
ƒ This is a generic approach – can be readily applied to
assess situations where design is difficult to quantify

14
© Hyder Consulting Pty Ltd
 Risk assessment application

ƒ Main use of risk assessment is as a tool to determine a

cost-effective solution by:
ƒ Identifying important factors affecting life safety (or cost)
ƒ Identifying effective protection measures
ƒ Effectiveness of each system is measured by its:
ƒ Reliability – likelihood of the system operating, and
ƒ Efficacy – how well it performs its intended function.
ƒ A cost-effective solution is the least cost design meeting
acceptable level of safety requirements

15
© Hyder Consulting Pty Ltd
 Optimal solution using risk assessment
COST $

Optimal design
solution point

‘Balanced’
Min. cost solution point
solution

Acceptable
min. risk RISK IN DESIGN
16
© Hyder Consulting Pty Ltd
 Risk parameters

Any parameter having an impact on the objective

(ie life safety or cost) needs to be assessed.
Important categories for life safety are:
ƒFire scenarios – fire size, fire location (hard to predict)
ƒFire detection system – detect and warn
ƒFire protection systems – manage and control fire effects
ƒEgress provisions – provide safe egress passageway
human behaviour consideration important

17
© Hyder Consulting Pty Ltd
 Simple example using event tree
0.985 0.4925 0.5 0.00188
Train fire in tunnel Train fire in station Pedestrians
is controlled 0.5 0.00375 is controlled by FB evacuate safely
0.5 0.5 Train is brought to
station 0.5 0.00188 50 0.09375
Fire starts in tunnel
Train fire in station Pedestrians
0.015 0.0075 is not controlled threatened
Train fire in tunnel
is not controlled 0.3 0.00113 50 0.05625
Train fire in tunnel Pedestrians
1 0.5 0.00375 is controlled by FB threatened
Fire starts in metro Train is not brought
network to station 0.7 0.00263 200 0.525
Train fire in tunnel Pedestrians
is not controlled threatened

0.999 0.4995
Station fire is Pedestrians
controlled evacuate safely
0.5 0.5
0.8 0.0004
Fire starts in station
Train fire in station Pedestrians
0.001 0.0005 is controlled by FB evacuate safely
Station fire is not
controlled 0.2 0.0001 200 0.02
Train fire in station Pedestrians
complementary events is not controlled threatened

© Hyder Consulting Pty Ltd END 0.69518

 Simple example using event tree
0.985 0.4925 0.5 0.00188
Train fire in tunnel Train fire in station Pedestrians
is controlled 0.5 0.00375 is controlled by FB evacuate safely
0.5 0.5 Train is brought to
station 0.5 0.00188 50 0.09375
Fire starts in tunnel
Train fire in station Pedestrians
0.015 0.0075 is not controlled threatened
Train fire in tunnel
is not controlled 0.3 0.00113 50 0.05625
Train fire in tunnel Pedestrians
1 0.5 0.00375 is controlled by FB threatened
Fire starts in metro Train is not brought
network to station 0.7 0.00263 200 0.525
Train fire in tunnel Pedestrians
is not controlled threatened

0.999 0.4995
Station fire is Pedestrians
controlled evacuate safely
0.5 0.5
0.8 0.0004
Fire starts in station
Train fire in station Pedestrians
0.001 0.0005 is controlled by FB evacuate safely
Station fire is not
controlled 0.2 0.0001 200 0.02
Train fire in station Pedestrians
complementary events is not controlled threatened

END 0.695 19
© Hyder Consulting Pty Ltd
 The expected risk

ƒ Each unfavourable event has a potential consequence.

ƒ The consequence is the expected number of
passengers threatened by the fire event.
ƒ The expected risk of an unfavourable event is:
Riskevent = Probabilityevent × Consequenceevent
ƒ The expected risk of the scenario is the cumulative sum
of all the risks for unfavourable events:
ERL = ∑ Riskevent

20
© Hyder Consulting Pty Ltd
 Determining Consequences

ƒ The consequence of an unfavourable event is

determined by direct computation or modelling
ƒ For example, to determine the unfavourable event for
‘Train fire in tunnel is not controlled’:
ƒ A large fire is modelled, say 20MW, using CFD
ƒ Occupant egress is simulated under untenable conditions
ƒ Occupants threatened by the effects of high temperatures
ƒ Occupant movement is limited by reduced visibility

21
© Hyder Consulting Pty Ltd
 Results of CFD simulation
– FDS (Fire Dynamics Simulator)

Temperature SECTIONAL VIEW Visibility

Temperature PLAN VIEW Visibility

22
© Hyder Consulting Pty Ltd
 Occupant evacuation

ƒ Occupant movement speed affected by:

ƒ Crowding density
ƒ Visibility
ƒ Decision making
ƒ Time to exit depends on:
texit = tdetect + taware + tresponse + tmovement
where tdetect = time to detect and communicate fire cue
taware = time occupant becomes aware
tresponse = time to respond to cue
tmovement = movement time to exit
ƒ Simulation models available for simulating occupant
behavioural interaction with the environment.
23
© Hyder Consulting Pty Ltd
 Sensitivity study
Purpose is to:
ƒ Assess accuracy of assumptions (eg input values)
ƒ Identify key factors by varying important parameters
Parameter Base Min END,min Max END,max
Fire start in station 0.5 0.1 1.22 0.9 0.171
Tunnel fire does not sustain development 0.95 0.7 4.07 0.99 0.155
Tunnel fire controlled by extinguishers 0.7 0.4 1.37 0.9 0.245
Train fire brought to station 0.5 0.1 1.09 0.9 0.305
Tunnel fire controlled by Fire Brigade 0.3 0.1 0.808 0.8 0.414
Station fire does not sustain development 0.99 0.9 0.875 0.999 0.677
Station fire controlled by automatic sprink. 0.9 0.5 0.775 0.99 0.677
Station fire controlled by Fire Brigade 0.8 0.5 0.725 0.95 0.68

Note: The END for the Base case is 0.695 (values <0.3 and >1.0 are shown in bold)

24
© Hyder Consulting Pty Ltd
 Summary

ƒ Important aspects of a risk assessment requires a good

understanding of the potential hazards and scenarios
ƒ Many difficult design parameters can be assessed with
a simple risk concept: Risk = Probability × Consequence
ƒ A sensitivity analysis allows important parameters
to be identified and hence used to minimize risk in design
ƒ Various combinations of systems can be assessed to
determine an optimal cost-effective design solution.
ƒ This has been demonstrated for assessing fire risks in
metro tunnels and stations
25
© Hyder Consulting Pty Ltd
 Hyder Consulting

Thank you

www.hyderconsulting.com

26
© Hyder Consulting Pty Ltd