Structural behaviour of structures in fire

Structural behaviour of
structures in fire
Tomaž Hozjan
Faculty of civil and geodetic engineering
University of Ljubljana, Slovenia

Lesław Kwaśniewski
Warsaw University of
Technology, Poland
 PLAN FOR PRESENTATION
Structural behaviour of structures in fire

objectives and applications of thermal properties

structural fire analysis
fire loading
requirements of structural fire
analysis thermal conditions

types of FE analysis and fire resistance

solution methods mesh density study
types of structural analysis units
challenges a few words about ls-dyna
further developments example 1, 2, and 3
possible test scenarios verification and validation
temperature dependent of computer simulations
mechanical properties
Structural behaviour of structures in fire OBJECTIVES AND APPLICATIONS OF
STRUCTURAL FIRE ANALYSIS
to predict the effects of fires in buildings
fire resistance and the structure’s performance under heating and
cooling
can be applied in the design of fire protection systems
for evaluation of fire safety (safe evacuation and safe firemen
work)
addendum of experiments (large specimens, loading and
boundary condition, interpretation of results)
virtual testing
parametric studies
Structural behaviour of structures in fire FIRE PARTS WITHIN EC (part 1-2):
structural fire design
EC 1 : ACTIONS on STRUCTURES
EC 2 : CONCRETE STRUCTURES
EC 3 : STEEL STRUCTURES
EC 4 : COMPOSITE STRUCTURES
EC 5 : TIMBER STRUCTURES
EC 6 : MASONRY STRUCTURES
EC 9 : ALUMINIUM ALLOYS STRUCTURES
Structural behaviour of structures in fire REQUIREMENTS OF STRUCTURAL FIRE ANALYSIS
type of analysis
solution methods
thermal conditions
geometry representation
temperature dependent material properties
mechanical boundary conditions and loading

TYPES OF FE ANALYSIS AND SOLUTION METHODS

global vs. component analysis
structural, thermal or coupled structural-thermal
dominantly Finite Element (FE) Method
explicit or implicit methods for time integration
general purpose commercial programs and research oriented
specialized unique programs
Structural behaviour of structures in fire DIFFERENT DESIGN APPROACHES
Load bearing resistance of a structure during fire can be
analysed on the levels of member, part of structure and
entire structure.
Member analysis can be performed using experiments,
simple and advanced calculation models.
Advanced calculation model (ACM) is a Finite Element
(FE) model able to solve numerically, with reliable
approximation, the partial differential equations
describing member’s response for assumed fire
conditions.
Structural behaviour of structures in fire DIFFERENT DESIGN METHODS

Prescriptive approach Performance based approach

Thermal action defined Physically based Thermal
by standard fire curve Actions (natural fire curves)
Structural behaviour of structures in fire DIFFERENT DESIGN METHODS

Prescriptive approach
Thermal action defined
by standard fire

Tabulated data Simple Advanced

Type of analysis (critical calculation calculation
temperature) methods methods

Member
YES YES YES
analysis
Analysis of parts
X YES, if avaliable YES
of the structure
Global structural
X X YES
analysis
Structural behaviour of structures in fire DIFFERENT DESIGN METHODS

Performance based approach

Thermal action defined by
natural fire

Tabulated data Simple Advanced

Type of analysis (critical calculation calculation
temperature) methods methods

Member
X YES, if avaliable YES
analysis
Analysis of parts
X X YES
of the structure
Global structural
X X YES
analysis
Structural behaviour of structures in fire CHALLENGES

time and space variation of temperatures

temperature dependent axial and rotational restraints
material degradation due to heating
interactions with adjacent structural components
additional forces imposed due to thermal deformations
large scale models with local effects (connections)
nonlinearities (geometrical, material, BC, loading, interactions)
uncertainties and measurements (BC, temperature distributions)
repeatability (sensitivity on parameter variation)
multiphysics (thermo-hydro-mechanical interactions in concrete)
Structural behaviour of structures in fire FURTHER DEVELOPMENTS
most of the current research work on structures subjected to elevated
temperatures is dedicated to steel and concrete structures
the experimental and numerical studies show importance and complexity of
beam to column connections in structural analysis
thermo-hygro-mechanical phenomena in concrete structures resulting in
additional nonlinear effects:
1. transient creep strain,
2. load induced thermal strain,
3. shrinkage,
4. pore pressures
5. (explosive) spalling?.
prediction of behaviour of concrete structures and structural elements imposes
the main challenge for future research
future work in connections modeling
FEA model verification and validation
need for experimental benchmark problems which could be used for the FE
model validation
Structural behaviour of structures in fire POSSIBLE TEST SCENARIOS
1st scenario: increasing static loading in constant elevated temperature -
critical loading for selected temperatures.
2nd scenario: the structure is analyzed under constant loading but at increasing
temperature - critical temperature and time.
3rd scenario:, e.g. following experiment, both temperature and loading are time
depended. Loading due to constrained thermal elongation

FURNACE TESTS Temperature

Limitation of geometry and

dimensions.
Tcr
Artificial or undetermined boundary
conditions
Effects of continuity ignored.
Load capacity L tcr Time
Thermal expansion not restrained by
by surrounding structure
 TEMPERATURE DEPENDENT MECHANICAL
Structural behaviour of structures in fire PROPERTIES
STRUCTURAL STEEL (EC3)
Steel softens progressively from 100-200°C up.
Only 23% of ambient-temperature strength remains at 700°C.
At 800°C strength reduced to 11% and at 900°C to 6%.
Melts at about 1500°C.
Elastic modulus at 600°C reduced by about 70%.
Yield strength at 600°C reduced by over 50%.
 TEMPERATURE DEPENDENT MECHANICAL
Structural behaviour of structures in fire
PROPERTIES
CONCRETE (EC2)
Concrete loses strength and stiffness from 100°C upwards.
Does not regain strength on cooling.
High temperature properties depend mainly on aggregate type
used.
 THERMAL PROPERTIES
Structural behaviour of structures in fire

thermal expansion
thermal conductivity
specific heat
 THERMAL PROPERTIES
Structural behaviour of structures in fire
THERMAL EXPANSION

Thermal
expansion for steel
reduces to zero
due to cristal
change at 700-
800° C
Concrete seldom
reaches 700 ° C in
building fires
Uniform thermal
expansion is
assumed for light-
weight concrete
 THERMAL PROPERTIES
Structural behaviour of structures in fire
THERMAL CONDUCTIVITY AND SPECIFIC HEAT OF STEEL
 THERMAL PROPERTIES
Structural behaviour of structures in fire
THERMAL CONDUCTIVITY AND SPECIFIC HEAT OF CONCRETE
Structural behaviour of structures in fire
FIRE LOADING
STANDARD FIRE TEST CURVE (ISO834)
PARAMETRIC FIRE CURVES
Structural behaviour of structures in fire THERMAL CONDITIONS
Direct thermal loading (prescribed temperature fields)
Constant or time dependent prescribed temperatures
Full insulation
Prescribed flux
Heat transfer between a member and surroundings:
convection:
•
h net ,c = α c TS ( t ) − TM ( t )  [°C] or [°K]

radiation:
•
h net ,r = ε mσ SB TS4 ( t ) − TM4 ( t )  [°K]
•
h net ,r = ε mσ SB (TS ( t ) + 273) 4 − (TM ( t ) + 273) 4  [°C]
σ = 56.7×10−12 [kW /m2 K4] - Stefan–Boltzmann constant

εm = 0.8 – emissivity of the member (emissivity of the fire=1)

Structural behaviour of structures in fire
FIRE RESISTANCE
ACCORDING TO EUROCODES FIRE RESISTANCE CAN BE
ESTABLISHED IN 3 DOMAINS:

Time tfi,d > tfi,req - usually requires advanced calculation models

Load resistance Rfi,d,t > Efi,d,t - hand calculation methods used to find reduced
resistance at design temperature

Temperature: θcr,d > θd - the most common simple method used to find critical
temperature for loading and compare with design temperature
 MESH DENSITY STUDY
Structural behaviour of structures in fire
(based on Richardson extrapolation)
• Discretization error
. . .
.......
• Order of convergence

• Estimate of the asymptotic solution

≅
1

• E1 - the estimator of the relative error

$% "$&
!" , #
$%

NASA NPARC Alliance Verification and Validation, Examining

Spatial (Grid) Convergence, http://www.grc.nasa.
gov/WWW/wind/valid/tutorial/spatconv.html).
Structural behaviour of structures in fire
UNITS
Structural behaviour of structures in fire NOTE ON NUMERICAL MODELS!!!
GARBAGE
OUT
Garbage PERFECT
input Model

Perfect GARBAGE
input Model
Structural behaviour of structures in fire A FEW WORDS ABOUT LS-DYNA

History
LS-DYNA vs. ABAQUS
Implicit vs. Explicit
LSPREPOST
LSOPT
Examples
Verification and validation

Marta Sitek, Lesław Kwaśniewski

Warsaw University of
Technology, Poland

11:05 25
 A FEW WORDS ABOUT LS-DYNA
Structural behaviour of structures in fire History
LS-DYNA originated from the 3D FEA program DYNA3D,
developed by Dr. John O. Hallquist at Lawrence Livermore
National Laboratory (LLNL) in 1976.
DYNA3D was created in order to simulate the impact of the Full
Fusing Option (FUFO) or "Dial-a-yield" nuclear bomb for low
altitude release (impact velocity of ~ 40 m/s). At the time, no
3D software was available for simulating impact, and 2D
software was inadequate.
DYNA3D used explicit time integration to study nonlinear
dynamic problems.
In 1978 the DYNA3D source code was released into the public
domain without restrictions after a request from France.
At the end of 1988 Livermore Software Technology Corporation
(LSTC) was founded to continue the development of DYNA3D in
a much more focused manner, resulting in LS-DYNA3D (later
shortened to LS-DYNA). Since then, LSTC has greatly expanded
the capabilities of LS-DYNA in an attempt to create a universal
tool for most simulation needs.

11:05 26
 A FEW WORDS ABOUT LS-DYNA
Structural behaviour of structures in fire LS-DYNA vs. ABAQUS
LS-DYNA ABAQUS
Very efficient explicit part Efficient Implicit part (e.g. Riks
analysis)
Implicit part not as stable as in
ABAQUS Very good manuals
Solver updated several times a Expensive even for academic use
year
Limited usage (tokens)
Cheap for academic use ($1000 a
More popular at universities than
year (commercial very expensive)
LS-DYNA
MPP version well optimized for
Graphical interface ABAQUS/CAE
multi processor calculations
does not support all cards
Theory manual not as good as
ABAQUS/CAE – mesh development
ABAQUS manuals
is complex and limited
Used by 90% of automotive
Text input files *.inp have complex
industry
„encrypted” structure
Graphical LSPREPOST updated
every few weeks, supports all
cards but many times is unstable
Efficient contact algorithms 11:05 27
 A FEW WORDS ABOUT LS-DYNA
Structural behaviour of structures in fire Implicit vs. Explicit
EXPLICIT
IMPLICIT () * '" 31* 2* *
4
'() *+ ,(- *+ . / * ∆( 1 /* *+
2 /*
Only for dynamics
For static and dynamic calculations
Based on Central Difference Method
Predictive –adaptive algorithm
Requires inversion of only mass matrix
(increments and iterations)
Mass matrix must be diagonal (only
Based on Newtonian methods
linear finite elements)
Requires inversion of large matrices
Short time (integration) steps – ruled
Not feasible for multiprocessor by Courant criterion (the shorted time
calculations needed to cross a FE by stress wave)
7 :
Problems with convergence for highly ∆5 6 8 9
;
nonlinear problems (contact, failure,
complex material models) Only increments no iterations
Simple solution algorithm, very
feasible for highly nonlinear problems
Dedicated for parallel processing

11:05 28
 A FEW WORDS ABOUT LS-DYNA
Structural behaviour of structures in fire LSPREPOST

Old „skin” (F11)

New „skin” (F11)

11:05 29
 A FEW WORDS ABOUT LS-DYNA
Structural behaviour of structures in fire LS-OPT
Graphical optimization software LS-OPT that interfaces with LS-DYNA
Allows the user to structure the design process, explore the design
space and compute optimal designs according to specified constraints
and objectives.
The program is also highly suited to the solution of system
identification problems and stochastic analysis.
LS-OPT – freeware software, can work with ABAQUS

30
 EXAMPLE 1
Structural behaviour of structures in fire Transient heat transfer in the concrete beam (EC2 Annex A)

11:05 31
Structural behaviour of structures in fire EXAMPLE 1
Transient heat
transfer in a
concrete beam
(EC2 Annex A)

Temperature profiles for a

beam hxb=600x300 – R120
According EC2 Annex A (Fig.
A.8)

11:05 32
 EXAMPLE 1
Structural behaviour of structures in fire Transient heat transfer in a concrete beam (EC2 Annex A)

Calculated temperature profile for a beam hxb=600x300 – R120

11:05 33
 EXAMPLE 2
Structural behaviour of structures in fire Furnace test on a steel column

11:05 34
 EXAMPLE 2
Structural behaviour of structures in fire Furnace test on a steel column

11:05 35
 EXAMPLE 3
Concrete –steel balcony
Balcony 3x3 m
- Reinforced concrete
C25/30, tc= 15cm
- beams HEB260, S275
-

A two-step analysis:
-

-- thermal (to obtain Concrete slab Steel beam

temperature curves)
- coupled thermal –
structural (temperature
boundary condition)

FEM mesh

11:05 36
 EXAMPLE 3
Concrete –steel balcony
Quasi-static analysis using
explicit time integration
Time scaling
Large global viscous
damping
Temperature distribution
Material Model 172 (EC2)
for concrete, based on
Eurocode 2
Material model MAT 4 for
steel (elastic plastic
thermal)

Deflection at t=50min

11:05 37
 VERIFICATION AND VALIDATION OF COMPUTER
Structural behaviour of structures in fire SIMULATIONS
The sinking of the Sleipner A offshore platform
http://www.ima.umn.edu/~arnold/disasters/sleipner.html
The failure involved a total economic loss of
about $700 million.

Failure in a cell wall, resulting in a serious

crack and a leakage that the pumps were not
able to cope with. The wall failed as a result
of a combination of a serious error in the
finite element analysis and insufficient
anchorage of the reinforcement in a critical
zone.

The post accident investigation traced the

error to inaccurate finite element
approximation of the linear elastic model of
the tricell (using the popular finite element
program NASTRAN). The shear stresses were
underestimated by 47%, leading to
insufficient design. In particular, certain
concrete walls were not thick enough.
38
 VERIFICATION AND VALIDATION OF COMPUTER
Structural behaviour of structures in fire SIMULATIONS
FE model well replicates the experiment

FE model poorly replicates the experiment

Actual Mean
Simulated Mean
Simulated Mean
Probability of occurrence

Actual Mean
(number of samples)

Response

A Simulated Item An Actual Item

39
 VERIFICATION AND VALIDATION OF COMPUTER
Structural behaviour of structures in fire SIMULATIONS

Effect of calibration

Simulated Mean

Actual Mean
Simulated Mean
Probability of occurrence

Actual Mean
(number of samples)

Response

A Simulated Item An Actual Item

40
Structural behaviour of structures in fire

Thank you!