FUNDAMENTAL BEHAVIOR OF CFT BEAMCOLUMNS UNDER FIRE LOADING
Amit H. Varma, Sandgo Hong
Purdue University
2005 ASCE Structures Congress
New York City, NY
INTRODUCTION

Significant research has been conducted on the fire resistance
of composite CFT columns under standard fire loading.


Focus on the behavior of columns subjected to constant axial
loads, end conditions, and ASTM E119 fire time-temperature
loading



Researchers at NRC-Canada (Lie, Kodur, Irwin, etc.), China (LinHai Han), and Japan (Sakumoto using FR steel)
The results provide fire resistance rating (FRR) values and have
been used to develop standard fire resistant design
They do not provide knowledge of the fundamental forcedeformation-temperature behavior of the CFT column or the
critical failure segments
Limited research has been conducted on the fundamental
force-deformation-temperature behavior of composite CFT
beam-columns under combined axial and flexural loads and
elevated temperatures from fire loading
MOTIVATION

Why is it relevant? Four reasons
1. The section force-deformation-temperature (P-M-f-T) represents
the fundamental behavior of CFT beam-columns and it can be
used to investigate the effects of various geometric, material,
and insulation parameters on fire resistance.
2. These P-M-f-T responses can be used to calibrate beam-column
finite element models used to conduct structural analysis under
fire loading
3. The behavior and stability of moment resisting frames under fire
loading depends on the strength interaction P-M-T curve for and
the fire resistance of the connections
4. The stability of columns under fire loading also depends
eventually on the P-M-f-T response of the critical failure
segment at mid-span.
RESEARCH OBJECTIVES
The objectives of this research project are:



To analytically and experimentally investigate the fundamental
force-deformation-temperature (P-M-f-T) behavior of CFT
beam-columns under elevated temperatures from fire loading.
To evaluate the effects of various material (concrete strength,
steel yield stress), geometric (column size, width-to-thickness
ratio), and insulation (thickness, thermal conductivity) on the
fundamental P-M-f-T behavior of CFT beam-columns.
To develop (or calibrate) fiber-based finite element models for
modeling CFT columns and beam-columns while investigating
the fire behavior of CFT structures.
RESEARCH APPROACH




Development and validation of analytical approach for
simulating the thermal and structural behavior of CFT
members under structural loads and fire loading.
Preliminary analytical investigations of the fundamental P-M-fT behavior of CFT beam-columns under elevated temperatures
from fire loading. Evaluate the effects of geometric, material,
and insulation parameters.
Experimental investigations to measure the fundamental
P-M-f-T behavior of CFT beam-columns under combined axial
and flexural loads and elevated temperatures from fire loading.
Analytical model calibration. The experimental results will be
used to validate (or calibrate) the preliminary analytical
models. The experimental results and calibrated models will be
used to develop beam-column finite element models .
ANALYTICAL APPROACH



The analytical approach was developed and validated using
existing experimental data for CFT columns tested under fire
loading by researchers at NRC, China, and Japan.
Development and validation of the analytical approach was
presented in detail at the 2004 ASCE Structures Congress
The approach consists of three sequentially coupled analysis
steps, where the results from each step are required to
continue the analysis in the subsequent step:



Step I – Fire dynamics analysis
Step II – Nonlinear heat transfer analysis
Step III – Nonlinear stress analysis
Step 1
Step 2
Step 3
Fire Dynamics
Analysis
Nonlinear Heat
Transfer Analysis
Nonlinear
Stress Analysis
Step 1 - FIRE DYNAMICS ANALYSIS


Fire dynamics analysis is conducted to simulate the convection
and radiation heat transfer from the fire source (or furnace
walls) to the structural component by solving the simplified
Navier-Stokes equations numerically.
It is conducted using the NIST-BFRL developed software FDS,
and the results include the heat flux incident upon the surfaces
of the component or the surface T-t curves.
Step 2 - NONLINEAR HEAT TRANSFER ANALYSIS




Nonlinear heat transfer analysis is conducted to simulate the heat
transfer through the cross-section of the component and along its
length (and the associated convection and radiation losses).
The surface heat flux or T-t curves from the fire dynamics analysis
serve as thermal loading for conducting the heat transfer analysis.
The heat transfer analysis can be conducted using the FDM or
FEM. It is assumed uncoupled from the stress analysis, which is
adequate for structural materials.
We used FEM because it links more easily with step 3. The results
from the heat transfer analysis include the T-t curves for the nodes
of the FEM mesh and thermal contours.
Step 3 – NONLINEAR STRESS ANALYSIS




Nonlinear stress analysis is conducted to determine the
structural response of the component under applied structural
and thermal loads.
The nodal time-temperature (T-t) curves obtained from the heat
transfer analysis of step 2 define the thermal loads for the
nonlinear stress analysis
The stress analysis can be conducted using the finite element
method while using identical meshes for both steps 2 and 3.
The results from the analysis include the complete structural
response: deflections, strains, stresses, load-displacementtemperature relationships.
Step 3 – NONLINEAR STRESS ANALYSIS
For example, the behavior of CFT columns tested according to the
ASTM E119 was investigated using the 3-step approach

1200 30
10
20
Axial Displacement (mm)
Axial Displacement (mm)
Temperature (C)
1000
The analytical approach was validated for an assortment of CFT
columns with different material, geometric, insulation parameters
tested independently
by researchers in Canada,
China,
300 mm
CFT and Japan
250 mm CFT
10
800
600
0
-10
400 -20
200
-10
C2, surface, calculated
C2, surface, measured
C2, d=60mm, calculated
C2, d=60mm, measured
-20
-30
350x 350x 7.7mm CFT
Fy=285; f’c=19 MPa
L=3.8 m; P/Po=50%
SFRM Insulation
0
30
60
C9, calculated
C9, measured
20
10001000
10
800 800
0
600 600
-10
400
400
300x 300x 9mm CFT
Fy=360; f’c=37 MPa
L=3.5 m; P/Po=20%
FR steel, Ceramic
-20
200
200
-30
0
100
50
200 mm CFT
12001200
30
-50
-40
0
0
200x 200x 6.35mm CFT
Fy=350; f’c=47 MPa-40
L=3.8 m; P/Po=15%
-30
0
C1, calculated
C1, measured
(C) (mm)
AxialTemperature
Displacement

The sequentially coupled heat transfer and structural analysis were
conducted using the FEM and option in ABAQUS.
Temperature (C)

50
150
90
Time (min)
Time (min)
120
100
200
150
0
0
Time (min)0
00
150
40
30
30
60
C3, surface, calculated C1, surface, calculated
C3, surface, measured C1, surface,
measured
C5, calculated
C3, d=37mm, calculated C1, d=71mm, calculate
C5, measured
C3, d=37mm, measured C1, d=71mm, measure
20080
120 120
160150
90
Time
90
120 (min)
180
210
Time
(min)150
60
2
24
GENERAL FINDINGS

The analytical approach was developed and validated, but the
behavior of CFT columns under fire loading were found to be
very sensitive with respect to:




Temperature dependent steel and concrete material structural
properties, which are not measured or reported explicitly for most
CFT specimens.
Column end conditions (rotational and axial restraint). End
conditions obtained in the experiment may vary from those
assumed in the analysis.
Variations in axial load level (P/Po). Axial load level can vary due
to changes in axial load P due to restraint, or due to Po which
depends on steel and concrete strength variation.
Relative motion (slip) between steel tube and concrete infill at
ends. This slip occurs for some columns that were tested
individually and the slip was allowed to occur. This may not be
realistic.
PRELIMINARY ANALYTICAL INVESTIGATIONS

Preliminary investigations were conducted using the
(developed and validated) analytical approach to determine the
more fundamental force-deformation (P-M-f) behavior of CFT
sections under elevated temperatures from fire loading.


These P-M-f-T responses and the effects of various material,
geometric, and insulations parameters are the focus of the
research for reasons explained earlier.
CFT parameters:






Width b = 200 or 300 mm.
Width-to-thickness ratio = 32 or 48
Steel tube A500 Gr. B (300 MPa)
Concrete strength (f’c=35 MPa)
Axial load levels (P=0, 20%, 40%)
Thermal insulation thickness (0, 7.5, 13 mm thick)
PRELIMINARY ANALYTICAL INVESTIGATIONS


The analytical investigations were conducted on a segment of the
CFT beam-columns. The length of the segment was equal to the
cross-section width b.
It represents the critical segment of CFT column or beam-column
subjected to axial and flexural loads and elevated temperatures from
fire loading.
Step 1 – FDS analysis to simulate heat
transfer to the surface of the segment
Step 2 – Nonlinear heat transfer analysis
to simulate transfer through section and
along length
Step 3 – Nonlinear stress analysis for
constant axial load, monotonic flexural
loading moment), and nodal thermal
loading (T-t) from step 2
Steps 2 and 3 conducted using the finite
element method and ABAQUS
MATERIAL PROPERTIES – T Dependent



400
Temperature dependent thermal and structural material
properties were used along with the 3D finite element models
of the CFT failure segment.
These material properties were based on values generally
reported in the literature (Lie and Irwin 1995 etc.).
T-Thermal properties are given in a table in the paper
100oC
Steel s-e-T
500oC
T=500oC
700oC
T=700oC
100
900oC
500oC
30
Stress, MPA
T=300oC
200
Concrete s-e-T
400oC
300oC
300
Stress (MPa)
40
T=100oC
600oC
20
700oC
10
800oC
T=900oC
0
0
0.002
0.004
0.006
Strain (mm/mm)
0.008
0.01
0
0.000
0.010
0.020
Strain (mm/mm)
0.030
0.040
Thermal response- CFT without insulation
Step 1 – Results from FDS Analysis for ASTM
T-tfrom
curveheat transfer analysis
Step 2 – E119
Results
1200
Temperature
1000
800
600
400
200
0
0
30
60
90
120
150
180
oC
Surface Temperature =600oC
(minutes)
Surface TemperatureTime
=300
Time = 14.2 mts.
Time =ASTM
5.6E119
mts.
T-t
No Insulation
Structural Response – CFT without ins.
Step 3 – P-M-f-T curves for CFT without insulation
150,000.00
1.2
P/Po=0%, T=300oC
90,000.00
(M/M 20 )
o
P=20%, No Insul
P/Po=0%, T=20oC
P/Po=20%,
C
P=0,T=300
No oInsul
120,000.00
0.9
Moment/Moment @ 20 C
Moment (N-m)
P/Po=20%, T=20oC
0.6
P/Po=20%, T=600oC
60,000.00
P/Po=0%, T=600oC
0.3
30,000.00
P/Po=20%, T=900oC
0.0
0
0.00
200
0
2.5E-5
0.005
400
P/P800
C
o=0%, T=9001000
600
Temperature (T)
7.5E-5
10.0E-5
0.01
0.015
0.02
5.0E-5
o
Curvature (1/mm)
1200
12.5E-5
0.025
Findings for CFTs Without Insulation

For CFTs without insulation:




Fire loading results in quick heating of the steel tube (broiling)
while the concrete infill remains relatively cooler. Significant
portions remain at T< 100oC till much later
This relative heating causes rapid reduction in flexural stiffness
and strength of the CFT section under fire loading effects
This reduction depends primarily on the rise in steel temperature,
and is independent of axial load level, width, and other
parameters
This by itself, may not be a cause of concern unless the
demands placed on the CFT without insulation exceed the
reduced stiffness and strength at elevated temperatures
Thermal response of CFT with insulation

Consider similar CFTs with some insulation. Assume commonly
used insulation materials with properties given in the paper.
The presence of thermal insulation results in a slow increase in
the steel surface temperature.
1200
1000
Temperature

800
600
Steel surface
w/o insulation
Steel surface
with insulation
400
200
0
Insulation thick = 13.0 mm
Insulation
thick = 6.5 mm
0Time=180
30mts
60
90
120
150
180
Time=180
mts
Time (minutes)
The heatingASTM
of the
composite CFT section
becomes more
E119 T-t
No Insulation
uniform (notInsulation
broiling)
= 13 mm
Insulation=6.5 mm
Structural Response of CFT with Insulation
P-M-f-T
curves
forfor
CFT
with
b/t=32
P-M-f-T
curves
CFT
with
b/t=48
Normalized
Strength
P-M
Interaction
150,000
150,000
Moment
(N-m) P/P %
AxialMoment
Load Level
o
(N-m)
50
Ins=6.5 mm
120,000
120,000
40
T=20
C
P/P
o=20% Ambient
P/Po=0
o
Ins=13 mm
P/Po=40%
T=20 C
P/Po=20%
P/Po=0
Ins.Ambient
Thick
P/Po=20%
= 13
mmoC
T=20
30
90,000
90,000
b=200 mm, b/t=32
20
o
60,000
60,000
b=200 mm, b/t=48
10
b=300 mm, b/t=32
P/Po=40%
P/PoP/P
=40%
o=0
Ins. Thick
P/Po=20%
P/Po=20%
P/Po=0
13 mm
Ins.= Thick
P/P =0
o
P/Po=40%
P/Po=20% =
P/Po=40%
P/Po=0
P/Po=40%
6.5 mm
Ins. Thick
= 6.5 mm
30,000
30,000
0
00
0
0
0.000
0.2
2.5E-5
0.005
2.5E-5
0.005
0.4
0.6
0.8
0.02
0.015
0.01
10.0E-5
5.0E-5
7.5E-5
0.010
0.015
0.020
5.0E-5
7.5E-5
Moment
/Moment
@ 10.0E-5
P=0
Rotation
Rotation(rad.)
(Rad.)
Curvature
(1/mm)
Curvature
(1/mm)
1
12.5E-5
0.025
12.5E-5
0.025
1.2
Findings for CFTs with Insulation



The insulation thickness becomes the most important
parameter influencing P-M-f-T behavior and strength (P-M)
under elevated temperatures from fire loading.
As expected, CFTs with b/t =48 have greater increase in
moment capacity with increase in axial load (below the balance
point). This continues to be true at elevated temperatures also.
The tube width (b) and width-to-thickness (b/t) ratio do not
have significant influence on the P-M-f-T behavior of CFTs at
elevated temperatures from fire loading
EXPERIMENTAL INVESTIGATIONS

Experimental investigations will focus on measuring the P-M-f-T
response of CFT segments. Parameters included in the
experimental studies are:






Tube width (b) and b/t ratio
Concrete strength f’c
Axial load level
Heating (surface temperature)
Insulation thickness and type
Experimental test matrix is currently being finalized using results
of preliminary investigations
TEST SETUP


The test-setup will be similar to those used for measuring P-M-f
response of beam-column specimens at ambient temperature
It will be a cantilever column with axial force and lateral load applied
at the top (free) end and the bottom end clamped.
Hollow Core Jack
A custom-built portable furnace will be placed to
surround the plastic hinge region. It will subject
the surface to the selected T-t curve.
Thus, the specimen plastic hinge region will
be subjected to P, M, and T.
The deformations of the plastic hinge region will
be measured using close-range photogrammetry
and digital image processing techniques.
Axial Loading Beam
Axial Tension Rod
Hydraulic Ram Direction
C
F
T
Clevis
Steel Base Plate
Concrete Block
FURNACE DESCRIPTION

The portable furnace consists of ceramic fiber radiant heaters.
Four such panel heaters are assembled to form a box around
the hinge region.

Heaters have wattage density of 2.5 kW/ft2. Can provide surface
temperature approaching 1200oC. They use radiant heating which
is efficient and economic
FURNACE T-t CONTROL



Radiant heaters are fully controllable. Specify control surface Tt curve and the heater will provide it.
For example, in the experiment below, we are controlling steel
surface T-t curve under the insulation.
The insulation surface T-t curve can also be directly controlled.
This experiment is in progress.
500
Temperature (C)
400
300
Input temp
measured 1
measured 2
measured 3
measured 4
200
100
0
0
50
100
Time (min)
150
200
FURNACE T-t CONTROL

In this experiment we are directly controlling the steel surface
T-t curve without insulation to follow the ASTM E119 gas T-t
curve.
1200
Specified surface
T-t curve
Temperature (C)
1000
ASTM E119
T-t curve
800
Measured surface
T-t curves
600
ASTM E119
FDS1
steel 1
steel 2
steel 3
steel 4
400
200
0
0
20
40
60
80
100
Time (min)
120
140
160
180
200
DEFORMATION MEASUREMENTS
The deformation (or movement) of any point on the specimen
surface can be measured using close-range photogrammetry and
digital image processing fundamentals.

For example, the thermal expansion was measured as shown below.
1.4
1.2
Thermal Expansion (mm)

1
0.8
0.6
Predicted Thermal
Expansion at final 1.7mm
0.4
0.2
0
-0.2
0
20
40
60
80
100
Time (min)
120
140
160
180
200
TEMPERATURE MEASUREMENTS
Thermocouples bonded to steel and embedded in concrete to
measure temperatures
400
350
300
Temperature (C)

computed
measured
measured
computed
250
200
150
100
50
0
0
20
40
60
80
100
Time (min)
120
140
160
180
200
EXPERIMENTAL INVESTIGATIONS


Will be conducted this year. Results to be presented at next
ASCE Structures Congress.
Acknowledgments




National Science Foundation - funding
Purdue University
Dr. Jim Bethel – photogrammetry
Jarupat Srisa-Ard - student

THE END
MOTIVATION 2

Researchers around the world have developed finite element
method based computer programs to conduct structural
analysis under fire loading.


For example, researchers at Liege Univ. (SAFIR), Sheffield Univ.
(FEMFAN), Univ. of Manchester, Nat. Univ. of Singapore (SINTEF)
Most of these programs use fiber-based or concentrated hinge
based beam-column finite elements for modeling the behavior
of columns and beam-columns under fire loading

These finite elements must be validated (or calibrated) using
experimental data and realistic P-M-f-T behavior
MOTIVATION 3
Consider a 6-story structure with perimeter moment resisting frames for
lateral stiffness and stability.



Design for dead, live, wind loads.
Satisfy building and interstory drift requirements.
Consider state when subjected to D+0.5L+W+Compartment Fire load
1.0
Nomalized axial capacity P/PY

0.8
Section P-M at ambient
0.6
0.25 in. insulation
after 180 mts.
0.4
No insulation
after 35 mts.
0.2
0
0
0.25
No Insulation
after 15 mts. 0.5 in. insulation
after 180 mts.
0.50
0.75
1.0
Nomalized moment capacity Mn/Mp
1.25
MOTIVATION 4
The behavior and failure of columns under constant axial load and
elevated temperatures from fire loading also depends on the section
P-M-f-T response of the failure segment.
P
0
120
-2
100
-4
Moment (KN-m)
Axial Displacement (mm)
-6
-8
-10
-12
60
40
20
-14
-16
0
0
P
80
50
100
150
0
20
40
Time (min)
M=P d
60
80
Time (min)
6000
5000
Axial Force (KN)

4000
3000
2000
1000
P
P
0
0
100
200
300
Moment (KN-m)
400
500
100
120
Step 1 - FIRE DYNAMICS ANALYSIS
FDS model of the furnace, and the surface T-t curves for 200,
250, and 300 mm CFT columns that were tested shown below.
Symmetry
Symmetry
plane plane Symmetry plane
Symmetry plane
(b)
1200
1000
800
600
400
200
Hot air flow direction
(b)
Hot air flow direction
Symmetry plane
Symmetry plane
Hot air flow direction
Heated wall
Hot air flow direction
Heated wall
erQuarter
mevolume
Tof CFT
mncolumn
Hot air flow direction
Heated wall
Hot air flow direction
Heated wall
flow direction
t airHot
flowairdirection
flow direction
Hot airHot
flowairdirection
dHeated
wall wall
HeatedHeated
wall wall
(b)
Temperature (C)
(b)
(a)
Quarter volum e
of CFT
column
Hot air flow
direction
Quarter volum e
of CFT
column
Hot air flow
direction

During experiments, the furnace gas temperature was controlled
to follow the ASTM E119 T-t curve. The temperatures of the CFT
column surfaces were measured using thermocouples.
Quarter volum e
of CFT column
Quarter volum e
of CFT column

ASTM E119
FDS
Experiment
FDM
0
0
Figure
1. model
FDS
model
of
NRC
Furnace
with
CFT
column
gure
1. FDS
of
NRC
Furnace
with
CFT
column
Figure
1. model
FDS model
of NRC
Furnace
with CFT
column
Figure
1. FDS
of NRC
Furnace
with CFT
column
30
60
90
Time (min)
120
150
Step 2 - NONLINEAR HEAT TRANSFER ANALYSIS

For example, nonlinear heat transfer analyses of CFT columns
tested by other researchers were conducted.



The surface T-t curves from step 1 were used as thermal loading
3D finite element models were developed to conduct the heat
transfer analysis and analyzed using ABAQUS.
The results were compared with experimental results.
Stress analysis results for CFT beam-column with b/t=32, P/Po = 20%,
and insulation thickness=6.5 mm (Curvature = 12.5 x 10-5 1/mm)
Steel tube longitudinal strain
Concrete longitudinal strain
Steel tube longitudinal stress
Concrete longitudinal stress
(Pa)
(Pa)