SECAB Institute of Engineering and Technology

Bijapur, Karnataka
28 March, 2017

Structural Fire Behavior

Anil Agarwal
Assistant Professor, Civil Engineering
Indian Institute of Technology Hyderabad 1
MATERIAL BEHAVIOR AT ELEVATED
TEMPERATURES

2
 MATERIAL BEHAVIOR AT ELEVATED
TEMPERATURES
Degradation in material properties (Structural steel per Eurocode)

1 20
200
0.8
300
0.6 400
F(T)/Fy(20)

500
0.4 600
0.2 700
800
0
0 0.005 0.01 0.015 0.02
Strain
3
 MATERIAL BEHAVIOR AT ELEVATED
TEMPERATURES
Degradation in structural steel (Eurocode)

4
 MATERIAL BEHAVIOR AT ELEVATED
TEMPERATURES
Strength degradation in different types of steel
1.2

1 Stuctural Steel
Bolts
0.8
Welds
0.6

0.4

0.2

0
0 200 400 600 800 1000 1200 5
THERMAL STRAINS AND CREEP STRAINS
 Thermal elongation can introduce very large force
demands [Eurocode properties]

6
 CREEP STRAINS

 Steel exhibits significant

creep at elevated
(>400oC) temperatures

 Eurocode stress-strain
curve has inbuilt creep
effects for moderate rate
of heating

7
 EFFECTS OF THERMAL ELONGATION
• Internal forces produced by restraints against thermal
expansion (a=12x10-6 /oC, ey=0.0012)
• Complete length of a perfectly restrained beam will yield at
120 oC
• Extremely large forces act on connections and joints
• Cooling phase introduces stress reversal
• Arch action in heating and catenary action in cooling phase

8
 EFFECTS OF THERMAL ELONGATION
Other manifestation of the internal stresses.

9
 CASE STUDIES
STEEL BUILDINGS IN FIRE

10
 MAJOR STRUCTURAL FIRES
Alexis Nihon Plaza, Montreal, Canada (1986):
• 15 story steel building with a composite slab
• 30 ft spans, double angle connections
• 2 hour fire protection in beams and columns
• Fire broke out on the 10th floor and spread to other compartments and
floors
• After 2-3 hours partial collapse of the slab of the 10th floor
• NRC investigation suggests connection failure triggered the collapse
• Weld of the column to girder connections failed
• Lack of automatic detection and suppression in some areas was found
to be the major reason for rapid spread of fire
• Fire barriers were not adequate 11
 MAJOR STRUCTURAL FIRES
Broadgate Phase 8, London, UK (1990):
• 14 story building
• Lattice truss frame spanning 13 m supporting concrete slab
• Rigid transfer frame at the top story
• Building was under construction and the members were unprotected
• Large plastic shortening of some of the interior columns (Temperatures were below
600oC)
• No collapse reported, though trusses had permanent deformations > 500 mm

12
Courtesy: SCI, 1991
 MAJOR STRUCTURAL FIRES
One Meridian Plaza, Philadelphia, USA (1991)
• 38 story steel frame structure with composite slabs
• All the frames designed as moment resisting frames; fire protected ( 3 hr/2 hrs)
• Fire started at 22nd story and burnt for total 19 hours while the fire moved from
one floor to other
• Very large ( up to 1.4 m) permanent vertical deformation in beams
• No structural collapse or column failure
• Steel temperatures had exceeded the critical values

13

Courtesy: USFA, 1991

 MAJOR STRUCTURAL FIRES
World Trade Center Building 7 (2001):
• 47 story steel frame building with composite slab

• Shear tab, double angle shear connection, and seated moment

connections

• Fire protected beams and columns

• After several hours of fire, the building collapsed

14
 MAJOR STRUCTURAL FIRES
World Trade Center Building 7 (2001):
 NIST conducted detailed forensic investigation
 Columns did not reach the critical temperature
 Excessive elongation and shortening of beams lead to fracture / tear out
of shear-tab connections
 Failure of connections in several stories rendered column no. 79 unbraced
through 9 consecutive floors

15
Fire Behavior
Types, quantification, and estimation

16
 Contents - Part 1 – Fire ignition,
growth, and spread
 Fire initiation /ignition
 Fuel - material and quantity
 Compartment fire
 Pre-flashover fire
 Post-flashover fires
 Ventilation
 Enclosure properties
 Growth and spread of building fire
17
 Fire Ignition/Initiation
External source:
 Almost all solids need external source to start
ignition
 Matches, cigarettes, stove, candle, campfire,
short-circuit, etc.

Ignition:
 Sufficient heat and exposure duration is needed
 Object should have low thermal inertia, high
calorific value, and proper shape
 Thermal inertia = (density) x (specific heat) x 18
(conductivity)
 Fuel- Material and quantity
Types of fuel
 Domestic and office: wood, cotton, polyester, and plastics
 Industry: petroleum and other hydrocarbons; solvents, etc.
Fuel quantification
 Calorific value, DHc (MJ/kg)
 Cone calorimeter test: Generally based on oxygen consumption
 Fire Load Energy Density (FLED) (MJ/m2)
 FLED = DHc M/A [where, A is the room floor area and M is the fuel mass]
 Heat release rate (MW/m2 {compartment floor}) depends on
 Net calorific value (MJ/kg) ↑
 Density of the material (kg/m3) ↓ 19

 Exposed fuel surface area ↑

 Fuel- Material and quantity
Quantification of heat release rate
A simplistic model is used
 Rate of heat release per unit area is constant
 Fire is spreading at a constant speed in radial direction
 Sufficient amount of fuel surface is available
 Heat release rate is proportional to time squared
Q = t2/k2
Q = Heat release rate (MW)
k = Fire growth rate (s/MW0.5)

600 (slowly growing fire: dense wood) 20

150 (fast growing fire: cartons and
furniture)
 Fuel- Material and quantity
The assumptions of the t-squared model are true only in the
beginning of a fire
 Once the entire fuel surface starts burning Q becomes constant

21
 Compartment Fire
Compartment fires vs. open fires

 Heat radiations and hot gases  Heat radiations and hot

(combustion products) are not gases (combustion products)
able to easily escape the escape easily
compartment

 Hot gases accumulate in the  Open fire spreads only

upper half of the compartment through direct contact
 Magnified radiation can lead to
ignition without direct contact

22
 Compartment Fire
Pre-flashover fire
 Isolated plumes of fire and hot gases
(combustion products)
 A layer of hot gasses is formed near the
ceiling
 The layer keeps getting thicker till it
reaches an opening (door and window
soffits, etc.)
 Fire warning systems and active fire
protection systems should be triggered
 Two-zone model is most common
numerical model to predict temperatures
at this stage
 Assumes two homogenous layers and a
connecting plume 23
 “FASTlite” is a popular freeware that uses two-
zone model
 Compartment Fire
Flashover
 Fire starts as localized fire
 If enough fuel and ventilation are available
 Layer of hot gasses keeps getting hotter (~600°C)
 Emits enough radiation to ignite all combustible
material in the compartment (~20 kW/m2)
 This transition is flashover
 Flashover does not take place in open fires or in very
large compartments

24
 Compartment Fire
Post-flashover fire
 If enough fuel and ventilation are available
 All combustibles items burn
 Very high rate of radiation
 High intensity fire (T ≥ 1000 °C)
 High oxygen demand
 Requires large openings
 Flames extend out from the windows to grab oxygen
 Fresh air rushes in
 Very high level of turbulence 25

 Temperature is often considered to be uniform

 Compartment Fire
Ventilation controlled vs fuel controlled
 Window glasses break before or at flashover temperature
 Still for typical rooms, post-flashover fires are ventilation
controlled
 Wall openings control the amount of ventilation
 Rate of burning (kg/s) is proportional to:
(area of the opening x height of the opening0.5)
 Fires in large well-ventilated rooms can be fuel controlled
 All fires become fuel controlled in decay phase
26
 Compartment Fire
Enclosure properties
 Compartment enclosure is made of walls and ceiling
 Retains heat in the compartment
 High thermal inertia enclosure (such as concrete and
masonry)
 Absorb more heat from the compartment
 Reduce the fire intensity in the compartment

27
 Typical fire growth curve

 Pre-flashover phase can be modeled by t-squared curve

 Post-flashover, the rate of material burning is almost
constant while the temperature in the compartment
continues to rises
28
 When all the fuel is consumed, decay phase begins
 Spread of fire
Within the story of origin
 Fire rated doors and walls are used to stop the fire from
spreading
 Concealed spaces (such as false ceilings and HVAC ducts)
are most common paths
 Hot smoke and toxic gasses can travel between isolated
compartments through these ducts
 Can render the fire isolation strategy ineffective

29
 Spread of fire
To other stories
 Through cracks in the floor system
 Excessive heat transfer through the slab
 Through staircase
 Interconnected ducts and other concealed ducts
 Through the gap between the façade and the floor system
 Externally through windows

30
 Contents - Part 2 – Quantification
and Estimation
Nominal fire models
 Standard (compartment) fire curves
 External fire curves
 Hydrocarbon fire curves
Simple Fire Models (mostly empirical; account for factors such as
opening size, fuel density, etc.)
 Swedish Curves
 Parametric compartment fire
Advanced Fire Models (Based on energy balance)
 Zone models 31
 CFD based models
 Nominal Fire Models
Compares with a real fire as below
 Growth and decay phases are not included
 Areas under the t-T curves are often used to compare severity
of two fires

32
 Nominal Fire Models
Standard fire curves:
 Standard procedures for fire
resistant assessment
 ASTM E-119 and ISO 834 are
two common t-T curves

Other Eurocode (EC1)

curves:
 Hydrocarbon fire
 External fire
33
 Simple Fire Models
Swedish fire curves:
• Customized fire t-T curves that account for different levels of
ventilation and fuel load
• Represent only the post-flashover fires

 Ventilation factor, Fv = Av(Hv)0.5 /At

Av = total area of all openings
Hv = average (weighted) height of all openings
At = total enclosure area including openings

34
 Fuel load (MJ/m2) = Total calorific value / Total surface area
= (Hc*M)/At
 Simple Fire Models
Swedish fire curves
Effect of ventilation : Fire intensity increases and duration decreases

35
 Simple Fire Models
Swedish fire curves:
Effect of fuel load (25 – 500 MJ/m2): The growth part is NOT affected

36
 Simple Fire Models
Eurocode Parametric Curves:
Customized fire t-T curves that account for different levels of ventilation,
fuel load, and enclosure properties.
 Ventilation factor, Fv or O, is same as before
 Fuel load, qt,d (MJ/m2), is same as before
 Thermal inertia, b2, accounts for thermal properties of the enclosure,
where b = k r c p
Parametric Fires for qt,d = 160 MJ/m2 and
1200 b = 1000 J/m2s1/2K

1000
Fire #1 (O = 0.032)
800 Fire #2 (O = 0.0886)
Temperature (C)

Fire #3 (O = 0.02)
600

400

200 37
0
0 60 120 180 240 300 360
Time (min)
 Advanced Fire Models
Zone based models:
These are computer based models that solve heat balance equations
for the compartment
 One zone models: Popular for post-flashover fire modeling
 e.g., COMPF2 (Babrauskas), OZONE (Franssen et al.)

 Two zone models: Popular for pre-flashover or localized fire

modeling
 e.g., FASTLite (developed by NIST)

CFD based models:

These programs solve heat and mass transfer equations to calculate
temperatures in the compartment
38
 FDS (developed by NIST) is a common software.
Questions?

39