You are on page 1of 81

Fire Safety Engineering & Structures in

Fire
Workshop at Indian Institute of Science
9-13 August, 2010
Bangalore
India

Material Behaviour in Fire

Organisers: CS Manohar and Ananth Ramaswamy


Indian Institute of Science

Speakers: Jose Torero, Asif Usmani and Martin Gillie


The University of Edinburgh

Funding and
Sponsorship:
Effect of fire on construction materials

 Chemical
decomposition, charring
 Physical
changes in density, softening, melting, spalling
 Mechanical
strength, stiffness, creep, thermal expansion
 Thermal
thermal conductivity, specific heat
Important material properties
 Propertiesof materials of interest in terms
of structural behaviour in fire
Thermal conductivity
Thermal
Specific heat
Density
Strength Mechanical
Stiffness
Creep
STEEL
Basics
 Structural steels and concrete reinforcing bars are
hot rolled low-carbon ferrite-pearlite steels
 Pre-stressing tendons are cold drawn high-carbon
pearlitic steels with elongated grain structure in
cold work directions
 Properties at room temperature
Density (r) = 7850 Kg/m3
Modulus of Elasticity (E) = 210 GPa
Coefficient of thermal expansion (a) = 11.5x10-6 K-1
Thermal conductivity (k) = 46-65 Wm-1 K-1
Specific heat (cp) = 450 J Kg-1 K-1
Steel Behaviour at ambient temperature
– a review
 Ductile
 Same in compression and tension
 Clear yield point
 Plastic range with lots of hardening
Stress-strain Behaviour of Steel
Steel-
Mechanical Properties at high temperature
 Loss of clear yield point / reduced yield point
 Reduction in UTS (above around 400C)
 Increased ductility
Obtaining stress-strain at ambient
temperature
 Single variable
P P  Easy to perform

Stress=
P/A

Strain = ΔL/L
Obtaining stress-strain-temperature
data

P T P
 Two variables (temp +
strain)
Stress=
P/A  Either isothermal test
 Or an-isothermal test
T+  Different results!
– minor in steel
– significant in concrete
 Creep
Strain = ΔL/L
Mild Steel: Stress-Strain behaviour

yield
Stress-Strain Behaviour at High
Temperatures
Mild Steel: Strength decline from
heating
Numerical Representation
 Eurocode (no plastic hardening)
– Linear range, E function of temperature
– Polynomial non-linear
– Plastic plateau
– Three parameters - E(θ), fy(θ), fp(θ)
 Eurocode (plastic hardening)
– As non-hardening model
– Then increased capacity
– Four parameters - E(θ), fy(θ), fp(θ), fu(θ)
 Osgood-Ramberg
– Non-linear elastic
– Can be made temperature dependent
Eurocode Stress-strain behaviour –Low
strains
Eurocode Stress-strain behaviour –Large
strains
Eurocode Stress-strain behaviour –Large
strains
Reduction Factors for Eurocode Steel
Behaviour
Thermal Expansion of Steel
th  a ( ) T
 Some variation with temperature
 Often taken as 1.2-1.4x10-5 ˚C-1 between
0-700 C
 Phase change at 700C results in shrinkage of
steel at this point
Steel: Thermal Conductivity
Steel: Specific Heat

ferrite-pearlite austenite
Cold-drawn pre-stressing steel: Stress-Strain behaviour
Cold-drawn steel: Strength decline
Steel properties: decline with heating
Steel: Proof stress

0.2%

0.2%
Steel: Creep

Also heating rate


dependent:

Higher heating rate


increases rate of creep
Steel: Connection failure
Steel: Conclusions

 Highly conductive, heats up quickly and loses


strength/stiffness very quickly in fire if left
unprotected
 Cold-drawn pre-stressing steels lose strength
quicker than hot-rolled mild steel
 Slender parts of steel sections under-
compression are susceptible to local buckling
Steel: Conclusions
 Steel expands considerably (> 1mm/m/100oC)
leading to further susceptibility to buckling
when restrained
L/L=14x10-6(T-20) => (EC3-1995)
 Plastic deformations in restrained steel
structural members lead to large tensile
stresses upon cooling
 Significant creep over 400-500oC
CONCRETE
Basics
 Hydrated paste forms only 24-43 % of volume,
therefore properties vary greatly with aggregate
 Divided into two major groups:
Dense or normal-weight concrete (2150-2450 Kg/m3)
Lightweight concrete (1350-1850 Kg/m3).
 Alsosiliceous and calcareous aggregates according
to aggregate composition
Basics
 Properties at room temperature (practical values)
E=5-35 GPa;
fcu = 20-50 MPa (NWC), 10-30 MPa (LWC)
k = 1.6 Wm-1K-1 (NWC-S), 1.3 Wm-1K-1 (NWC-C), 0.8 Wm-1K-1
(LWC)
cp = 1000 JKg-1K-1 (NWC) 840 JKg-1K-1 (LWC)
a = 18x10-6 K-1 (NWC-S), 12x10-6 K-1 (NWC-C), 8x10-6 K-1
(LWC) - But note LITS
Concrete: Thermal conductivity
Concrete: Specific heat
Concrete Expansion - LITS
Dense Concrete: Strength reduction
Dense Concrete: Relative strength reduction
Concrete: Strength reduction
(Design)
Concrete: Tension cracks in slab over a fire compartment
Concrete: Spalling
Concrete: Spalling (aggregate fracture)
Concrete: Spalling (explosive)
Concrete: Punching shear failure in flat slabs
Concrete: Conclusions
 Concrete loses strength and stiffness when heated
beyond 400oC, its fire behaviour however differs
considerably depending upon aggregate used (light-
weight aggregates give best performance)
 When used as a composite with steel, concrete
provides fire protection to steel
Concrete: Conclusions
 Highly insulating (heat propagation is ~50 times
slower than steel)
 Heat propagation is further delayed at ~100oC due to
the latent heat needed to evaporate large quantities
of moisture in the micro-pores
WOOD
Basics
 Used mainly in residential construction
 Properties at room temperature (of clear dry wood)
r (oven-dry) = 300-700 Kgm-3 (Douglas firs ~ 450, southern pines
~ 550)
True density of the solid forming the walls of wood cells ~ 1500
E=5-15 GPa; fcu = 13-70 MPa (along grain) ~ proportional to r
k ~ 0.5 W/moK (across grain)
cp = 1500 JKg-1 K-1
a = 3-4.5x10-6 K-1 (along grain), 21-40x10-6 K-1 (across grain)
Wood: Typical stress-strain relationship
Wood: Loading and failure mechanisms
Wood:
Char formation
Wood: Strength reduction below char
Wood: Strength reduction from heating
Wood: Char consideration in design
Wood: Conclusions
 Insulating material, but is also inflammable
 Low thermal conductivity is assisted by
charring
 Rate of char formation varies with wood
species and depends upon bulk density and
moisture content
 Strength and stiffness both reduce with
heating, compressive strength reduces at
faster rate than tensile strength
Wood: Conclusions
 Design strength of timber members is
calculated by omitting the charred portion of
sections
 Thermal expansion is low
MASONRY

 Properties at room temperature


r (oven-dry) = 1660-2270 Kgm-3 (based on raw materials,
moulding and firing techniques)
True density ~ 2600-2800kgm-3
E=10-20 Gpa; fcu = 9-110 MPa (~50 MPa average)
k ~ 1.0 W/moK (increasing to ~1.5 at 700 oC)
cp~ 700 J/Kg oK (increasing to ~1200 at 600 oC)
a = 5.5x10-6 /oK
MASONRY (BRICKS)
 Highly insulating and inherently stable in fire
 Often instability is caused by: movement of
connected structure
 Integrity failure can occur at very high fire loads
(bricks can resist temperatures up to a 1000oC but
melt at 1400oC)
 Masonry can spall
 Thermal properties are not of great interest due to
secondary and mainly non-load bearing role of brick
masonry in modern construction
GYPSUM
 Well known as plaster of Paris and used as interior
linings and in partition walls
 Low thermal diffusivity and large quantities of
chemically bonded water give it good insulating and
absorbing properties making it a cheap fire
retardant material
 Practically all strength is lost once the water of
crystallisation has evaporated
 Fibreglass reinforcement improves the strength in
fire
GLASS
 Main use is in glazing for doors and windows
 In this role, it has little resistance in fire and cracks
very quickly because of temperature gradient across
the surfaces and because of thermal expansion
 Double glazing does not help fire behaviour very
much
 Wire reinforcement can provide better performance
 In general, glazing cannot be relied upon to remain
intact in fire
PLASTICS
 Allplastics are inflammable and add fuel to fire
 Some treatments reduce combustibility but nothing
removes it
 A large amount of plastics can lead to very high
rates of heat release in fire leading to very high
temperatures
What is Load Induced
Thermal Strain?
How can we model it?
Angus Law, Martin Gillie,
Pankaj
Introduction
 What is Load Induced Thermal Strain (LITS)
– Description
– Terminology
 Why is it important?
 How can we model it?
– Existing models
– New model
What is LITS?
 Most materials….
– Get bigger when they are heated.
– Even under load
0.01
0.008
Strain (%)

0.006
0.004
0.002
0
0 100 200 300 400 500 600
20ºC 250ºC 500ºC
Temperature (oC)
What is LITS?
 Concrete is a bit different….
– With load…
0.015
0.02
–0.015
and with even more load…
0.015
(%)
(%)

0.01
0.01
0.01
(%)
Strain
Strain

0.005
Strain

0.0050
0.005
-0.005 0 200 400 600 800 1000
00
-0.01
-0.015 0 200 400
400 600
600 800
800 1000
1000
Temperature
Temperature(oC)
Temperature (oC)
20ºC 250ºC 500ºC
What is LITS
 More precisely…(I haven’t just made it up)

Schneider et. al (1988) Khoury et. al (2002)


What is LITS? – Terminology
 Lots of different terms…
– Elastic changes
– Shrinkage Elastic

– Transitional thermal creep


– Transient strain
Plastic
– Transient creep
– Creep

Time dependent
Why is it important?
 Larger strains during heating
 Locked in plastic strains…

Original Without With LITS


LITS
How can we model it?
– Elastic changes
– Shrinkage
– Transitional thermal creep
– Transient strain
– Transient creep

Elastic
How can we model it?
Elastic changes
– Shrinkage
– Transitional thermal creep
– Transient strain
– Transient creep

(LITS) Plastic Elastic


How can we model it?
– Elastic changes (LITS) Plastic
– Shrinkage
– Transitional thermal creep
– Transient strain
– Transient creep

Non-LITS plastic Elastic


How can we model it?
– Elastic changesNon-LITS plastic
– Shrinkage
– Transitional thermal creep
– Transient strain
– Transient creep

(LITS) Plastic Elastic

+ +
How can we model it?
 Build up strains to make full curve
– Apparent and or Actual modulus?
How can we model it?
 Column test….
How can we model it?
 Problem solved?
 What about 2 or 3D cases?
Plastic strain occurs in direction
orthogonal to yield surface…

f
What does this mean?!
 pl 

d

The trial stress (location


relative to the yield surface) is
dependent on the elastic
modulus…
How can we model it?
 The same constitutive curve will give us different
strains…
0.03

0.025

0.02

Deformation
0.015

0.01
Apparent U11
0.005
Actual U11
0
-0.02 -0.015 -0.01 -0.005 0
U22
How can we model it?
 This can be remedied with a new, two step,
approach…
Step 1)
How can we model it?
 This can be remedied with a two step
approach…
Step 2)
How can we model it?
 Does this work?
U22 0
0.03
Apparent U11 -0.02 -0.015 -0.01 -0.005
Actual U11 -0.002 0
Embedded U11 0.025 -0.004

Deformation (U11)
-0.006

Plastic strain (PE22)


0.02
-0.008
0.015 -0.01

-0.012
0.01
-0.014
0.005 -0.016

-0.018
0 Apparent PE22
-0.02 -0.015 -0.01 -0.005 0 Actual PE22 -0.02 Yes!
Embedded PE22
U22
Conclusions….
 LITS is an important factor when modelling
concrete;
 Representation of LITS needs a more thoughtful
approach than simple inclusion in the
constitutive curve;
 A new method for modelling LITS equations has
been successfully modelled.
Further work….
 Implementation in more complex material
model;
 Full scale validation from experimental results.
END

You might also like