Tunnel and Station T l d St ti Design Train Fire ScenariosFundamental Issues
J. Greg Sanchez
Perspective
Galileo Galilei id G lil G lil i said: You cannot teach a person something he does not already know; we can only bring what he already knows to a higher level of awareness.
The search
Chesterton id Ch t t said: It is not that they do not see the solution; it is that they do not see the problem problem.
Types of Fires
A – ordinary combustibles B – flammable liquids C – electrical equipment D – combustible metals b tibl t l K – cooking oils and fats
Video
BLEVE
Video
Tunnel Fire Norway
Video
Tunnel pool fire test - Austria
Video
Frankfurt Train Fire Test
Video
PATH Train Fire
NYC Subway Cars
Video
Train seat and floor materials burning
Flashover
NIST ISO 9705 Results
NIST ISO 9705 Test
Pre-Flashover
Flashover
Statistics & Risk
Statistics in Germany
Statistics in Germany
Statistics in UK
Statistics in UK
Probability Tree
Fires Statistics
Fires Statistics
Fires Statistics
Fires Statistics
How to Use Risk?
• What could go wrong? • What can be done to prevent
the harm from occurring and in the aftermath of an accident? And, if something happens, how will we pay for it?
•
What Could We Learn? To Assess Fire Heat Release Rate for Train Fires •We must improve accuracy of p y physical models used ( (numerical or experimental) •And an extensive database of thermo-physical properties for train materials must be created
High Fire Resistant
Low Fire Resistant
Thermal Properties
Density, thermal conductivity, y y specific heat, heat of formation, ignition temperature, mass loss i iti t t l characteristics, characteristics and thickness of burning materials are essential g to improve fire modeling, fire growth, h t and mass t th heat d transfer f
Exclusive Properties
Heat of Combustion
Heat of Combustion is a function of temperature and eq i alence temperat re equivalence ratio, φ=AFRstoichiometric/AFRactual It is not constant!
Heat of Combustion
Heat of Combustion for Gasoline-C8H18
298K 10 500K 1000K 1500K 2000K 2500K
Endothermic 0
Exothermic -10 g ΔHc, MJ/kg
-20
f (T )
LEAN
f (T , φ )
RICH
-30
-40
-50 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Equivalence Ratio, φ
Mass Loss Rate
Mass Loss Rate
Mass loss rate is a function of temperature (e.g., Arrhenius Equation) Not of incident heat flux
α & m fuel
Ae
(− ( E / RT )
Mass Loss Experiments
Ignition
Ignition g
Causes of Ignition
• Chemical heat energy, such as heat of combustion in a flame, flame • Electrical heat energy, such as heat from resistance elements and from arching arching, • Mechanical heat energy, such as heat and friction, • Nuclear energy, such as heat of fusion. gy In general, only two sources of energy are used in tests for ease of ignition: • Ch i l h t energy i th f Chemical heat in the form of a di t fl f direct flame or heated object, and • Electrical heated energy in the form of resistance elements or arcs
Video
Igniting Floor Samples
Ignition Activation Energy
Ignition Temperature
-PILOT (225 – 645 °C typical) Good for material testing comparison -NON-PILOT Good for fire growth modeling (could be double pilot Tig)
Pilot Ignition
Non-Pilot Ignition
Ignition Depends on Packaging
… and proximity!
FHRR Measurements
Oxygen Consumption
• •
Has been common practice since late 1970’s Method recognized as the most accurate and practical technique to measure heat release rate from experimental fires Based on: most organic materials 13.1 MJ/kg of oxygen consumed Assumes full combustion A f ll b i
• •
Oxygen Consumption
Flame Types
very sooty, incomplete combustion, b ti high radiation, not suitable for oxygen consumption method no soot, complete combustion, low radiation, suitable for oxygen consumption p method
Testing
Testing
Sample Testing Flames
Fire Physics y
Perspective
S. E L bl S E. Leblanc said: id
Kinetics is nature’s way of preventing everything from happening all at once!
Fire Tetrahedron
Fuel and oxygen must be is gaseous state
Entropy, Enthalpy, and temperature must favor reaction
Fire Mechanism
Flammability Diagram
Expected FHRR Curves
FHRR Experiments
How Certain Are We About Our Predictions?
Uncertainty of predictions y p made by models derives from uncertainties in the p y physical models used
Can we sustain 400MW?
Assuming: Tunnel area=50m2 Tffire~1027C 1027C ΔHc=20MJ/kg AFR=13 AFR 13 Tunnel requirements: Mffuel=20kg/s Mair=963m3/s Vtunnel=19.3m/s 19 3 /
Velocity in the tunnel is around 3m/s. Chances are that something is over estimated.
FHRR Curves
10W/s2 100W/s2 10W/s2 300W/s2
10W/s2
300W/s2
FDS FHRR Predictions
FDS FHRR Predictions
FDS FHRR Predictions
FDS FHRR Predictions
FDS FHRR Predictions
FDS FHRR Predictions
FDS FHRR Predictions
Fire due to burning newspaper or flammable liquid placed on the sea in the center of train car (Arson)
FDS FHRR Predictions
Fire due to burning newspaper or flammable liquid placed in the corner next to face panel and driver console assembly (Arson)
FDS FHRR Predictions
Undercarriage fire due to electrical short circuit or overheating of light voltage (LV) equipment or battery charger (electric fault)
Train Undercarriage
FDS Model
• FDS uses mixture fraction approach - fuel and oxidizer always react regardless of react, temperature • If FHRR were prescribed accuracy 5 to 20% prescribed, with experimental measurements • If FHRR were to be predicted, uncertainty much higher (no quantity given) • Based on LES - requires small grid size (filter size), and a small time step l=1.5m, L=182m,H=12m,W=12m you would need 62 million cells
FDS Flammability Diagram
Turbulence Modeling Options
LES Requirements
Resolved TKE Ratio= Ratio Total T l TKE
> 0.8
LES-Resolved TKE/Total
JGS FHRR Predictions
JGS Advanced Fire Model
• • • • • • • • • Pseudo First-Order Reaction RANS - ke Pyrolysis Air-Fuel-Ratio Air Fuel Ratio Surface Mass Limits Surface Ignition T S f I iti Temperatures t Auto-Ignition Temperatures for fuel vapors Flammability limits Radiation
-Larger time steps -Coarser grids Coarser -Used in large industrial applications
Combustion Model
Fuel+a(O2+3.76N2) bCO2+cCO+dH2O+eH2+fO2+3.76aN2+Heat
If the mixture, is lean, then c=e=0 th i t i l th 0 If the mixture, is rich, then f=0
Enthalpy of Combustion
Enthalpy based on E th l b d equivalence ratio, i l ti temperature, temperature first and second laws of thermodynamics
JGS FHRR Predictions
JGS FHRR Predictions
Train Outside Ambient, Full Size, Tignition=300C Train Fuel Tavg % ignited area
3.5
700
3.0
600
2.5
Fire Heat Release R Rate (MW)
500
Average cabin temp perature (C), %fuel area inside ca abin ignited
2.0
400
1.5
300
1.0
200
0.5
100
0.0 0 50 100 150 200
time (s)
0 250 300 350 400 450
JGS FHRR Predictions
Train Outside Ambient, Full Size, Tignition=400C Train Fuel Tavg % ignited area
1.2
300
1.0
250
0.8
Fire Heat Release R Rate (MW)
200
Average cabin temp perature (C), %fuel area inside ca abin ignited
0.6
150
0.4
100
0.2
50
0.0 0 50 100 150 200
time (s)
0 250 300 350 400 450
JGS FHRR Predictions
JGS FHRR Predictions
Train in Tunnel, Full Size, Natural Ventilation, 3% Slope, Tignition=300C Train Fuel Tavg % ignited area
1.0
250
0.9
0.8
200
0.7
0.6
150
0.5
0.4
100
0.3
0.2
50
0.1
0.0 0 10 20 30
time (s)
0 40 50 60
Average cabin temp perature (C), %fuel area inside ca abin ignited
Fire Heat Release R Rate (MW)
Development and D l t d Application f Scenarios A li ti of S i
What to Account for
• Conduct an inventory of combustibles • Conduct a survey of potential ignition sources • Observe geometry of potential scenario i • Focus on your environment specifics y p • Don’t dream your system; observe it. It exists!
Observe Your System
Observe Your System
Observe Your System
Observe Your System
Observe Your System
Observe Your System
Observe Your System
What to Account for
•Use Pareto’s principle to determine top 20% of cause precursors to tackle 80% of the problems •Identify if combustible material is thermally thin or thick, and adjust ignition time accordingly. j g gy •Pyrolysis. •Mass loss (kg/s) as a function of wall temperature. •Material mass load distribution (kg/m²) for each surface expected to contribute to the fire heat release rate. •Material thickness (m) density (kg/m³) thermal (m), (kg/m³), conductivity (W/m-K), specific heat (J/kg-K), heat of combustion (J/kg) – lower heating value, and combustible load distribution for each combustible. •Ignition temperature (°C).
What to Account for
•Flammability limits. •Calculations should be made through the use a fire Calculations model. A three-dimensional formulation, such as computational fluid dynamics (CFD) is recommended recommended. Zone models represent a much simplified approach, and predictions may not be as accurate as required. y •Full three-dimensional geometry of fire compartment and the environment the smoke will occupy, including tunnel walls, ceilings, walkways, d t l ll ili lk doors, and windows, d i d should be taken into account. •Radiation through the wall surfaces and in the gas phase due to smoke.
What to Account for
•Full combustion (accounting for rich/ and lean combustion). combustion) •Conduction through the walls. •Transient analysis should account for incubation Transient incubation, growth, peak, and decay. •Determine fire duration (s). ( ) •Develop fire growth curve and α (W/s²) using least square curve fit method. •Determine peak fire h t and smoke release rates, (W) D t i k fi heat d k l t and (kg/s), respectively. •CO generation rate (kg/s) (kg/s). •Smoke generation rate (kg/s).
Implementation of Scenario
• Allow for incubation period, say 2 minutes. • Model growth rate as αt2 up to p g p peak. • AFR=12, or as estimated. • Keep track of total fuel mass available available, and mass loss to determine duration of fire. fire • Maintain peak heat release rate until spread i controlled. d is ll d • Fire volume based on 1.2MW/m3.
Recommendation
For train fires, α=11.723 W/ s², equivalent to a medium growth rate. Trains with additional amenities such as th t T i ith dditi l iti h inter-city trains, or train fires caused by accelerants, may have greater fire growth rates Allow for a two minute rates. incubation period. The peak heat release rate need to be approximated as realistically as p pp y possible. The duration of the fire should be determined from the total mass of combustibles available. A smoke yield (Ysmoke) of 0.165 (kg/kg), and a carbon monoxide yield (YCO) of 0.100 (kg/kg) should be assumed.
Recommendation
For luggage fires, α =46.892 W/s², equivalent to a fast growth rate. No incubation period is required, since th t N i b ti i di i d i most luggage carry polyester and cotton fabrics which ignite very quickly A typical luggage fuel package has a quickly. mass of 25 kg, and has a peak heat release rate of 1.0 MW, and a burning duration of 15 minutes. A smoke , g yield (Ysmoke) of 0.165 (kg/kg), and a carbon monoxide yield (YCO) of 0.100 (kg/kg) should be assumed.
Recommendation
For trash fires, α = 46.892 W/s², equivalent to a fast growth rate. No incubation period is required, since th t N i b ti i di i d i most trashes ignite very quickly. A typical trash fuel package has a mass of 10 kg and has a peak heat kg, release rate of 0.5 MW, and a burning duration of 15 minutes. A smoke yield (Ysmoke) of 0.165 ( g g), and a y ( (kg/kg), carbon monoxide yield (YCO) of 0.100 (kg/kg) should be assumed.
Recommendation
For concession fires, α = 46.892 W/s², equivalent to a fast f t growth rate. No incubation period is required, since th t N i b ti i di i d i most concessions carry newspapers and plastics which are thermally thin and ignite very quickly The fire quickly. duration should be determined from the total mass of combustion material available. A smoke yield (Ysmoke) of y ( 0.165 (kg/kg), and a carbon monoxide yield (YCO) of 0.100 (kg/kg) should be assumed. A peak heat release rate of 1.5 MW for unsprinkled concessions, or 0.5 MW for sprinkled, and 15 minutes minimum for the duration of the fire should be assumed assumed.
Recommendation
Do not design for short flashes of heat release, but consider a growth to a well defined peak heat release rate plateau. Unless full combustion is being modeled do modeled, not consider decay rates because decaying phase contributes l h t ib t less smoke th th peak k than the k heat release rate and does not change the fan performance requirements.
Recommendation
In accordance with NFPA-130, an emergency ventilation system must be provided (natural or mechanical) to maintain a tenable environment along the path of egress from a fire incident incident.
Recommendation
This means that if a fire scenario predicts conditions that create untenable conditions within the first two minutes - for example, while the egress would take six minutes, the ventilation system is not proper and must be redesigned to make sure tenable conditions are maintained d i egress as required b i t i d during i d by NFPA 130
Final Comment
What Wh t if NO fi incidents fire i id t are ever reported in the t d i th life of a metro system? What do we do?
Q Questions/Answers