Professional Documents
Culture Documents
Angela M. Kingsley
University of Washington
2005
University of Washington
Graduate School
Angela M. Kingsley
Committee Members:
_________________________________________________
Dawn E. Lehman
_________________________________________________
Charles W. Roeder
_________________________________________________
Peter Mackenzie-Helnwein
Date __________________________
In presenting this thesis in partial fulfillment of the requirements for a masters degree at the
University of Washington, I agree that the Library shall make its copies freely available for
inspection. I further agree that extensive copying of this thesis is allowable only for scholarly
purposes, consistent with fair use as prescribed in the U.S. Copyright Law. Any other
reproduction for any purposes or by any means shall not be allowed without my written
permission.
Signature ______________________________
Date __________________________________
Table of Contents
List of Figures
iii
List of Tables
Chapter 1: Introduction
17
32
44
52
85
99
102
104
107
ii
List of Figures
Figure 2.1 Free body diagram of CFT cross-section ....................................................................... 5
Figure 2.2 Strain distribution ........................................................................................................... 6
Figure 2.3 Ratio of predicted to experimental flexural strengths..................................................... 8
Figure 2.4 Types of column base connections for steel and CFT sections ...................................... 9
Figure 2.5 Load resisting mechanism in exposed base plate connection....................................... 10
Figure 2.6 Load resisting mechanism in embedded column base connection ............................... 11
Figure 2.7 Embedded CFT column base with stiffeners................................................................ 12
Figure 2.8 Encased structural steel column base connection......................................................... 13
Figure 2.9 Construction of semi-embedded connection ................................................................ 14
Figure 2.10 Force resisting mechanisms in semi-embedded connection ....................................... 15
Figure 3.1 Cross section of proposed CFT-footing connection ..................................................... 17
Figure 3.2 CFT-footing connection model .................................................................................... 19
Figure 3.3 Connection model and mesh used in elastic analyses .................................................. 20
Figure 3.4 Boundary conditions for analytical model.................................................................... 21
Figure 3.5 Locations of connection performance evaluation......................................................... 22
Figure 3.6 Deflected shape of baseline elastic analysis ................................................................. 23
Figure 3.7 Increasing stress at various locations of baseline tube ................................................. 24
Figure 3.8 Von Mises stress distributionin baseline tube .............................................................. 24
Figure 3.9 Maximum principle stress distribution in baseline footing .......................................... 25
Figure 3.10 Minimum principle stress distribution in baseline footing ......................................... 26
Figure 3.11 Load-drift response for models with varying annular ring dimensions...................... 28
Figure 4.1 Refined mesh used in inelastic analytical model.......................................................... 34
Figure 4.2 Deflected shape of baseline inelastic analysis .............................................................. 34
Figure 4.3 Progression of Von Mises stresses ............................................................................... 36
Figure 4.4 Progression of maximum principle stress..................................................................... 38
Figure 4.5 Progression of cracking strain ...................................................................................... 40
Figure 4.6 Progression of minimum principle stress ..................................................................... 42
Figure 4.7 Load-drift response of baseline inelastic analysis ........................................................ 43
Figure 5.1 Specimen geometry and reinforcement ........................................................................ 45
Figure 5.2 Modified construction sequence for Specimen IV ....................................................... 46
Figure 5.3 Experimental Test Rig .................................................................................................. 48
Figure 5.4 Lateral Load History..................................................................................................... 49
Figure 5.5 Calculation of specimen yield displacement ................................................................ 49
Figure 5.6 Instrumentation of CFT-footing specimens.................................................................. 50
Figure 5.7 Instrumentation mounting ............................................................................................ 50
Figure 5.8 Instrumentation of CFT-footing test apparatus............................................................. 51
Figure 6.1 Specimen I damage at low drift ratios .......................................................................... 53
Figure 6.2 Specimen I damage at moderate and high drift ratios .................................................. 54
Figure 6.3 Specimen I damage at final state .................................................................................. 55
Figure 6.4 Specimen I damage in tube base................................................................................... 55
Figure 6.5 Specimen I cyclic load-drift response .......................................................................... 55
Figure 6.6 Specimen II damage at low drift ratios......................................................................... 56
Figure 6.7 Specimen II damage at moderate and high drift ratios ................................................. 57
Figure 6.8 Specimen II damage at final state................................................................................. 57
iii
iv
List of Tables
Table 2.1 Comparison of predicted to experimental flexural strengths ........................................... 7
Table 3.1 Analyses performed for parametric study...................................................................... 18
Table 3.2 Preliminary analytical model definition......................................................................... 21
Table 3.3 Stress values at 1.5% drift for varying footing depths................................................... 27
Table 3.4 Stress values at 1.5% drift for varying embedment ....................................................... 27
Table 3.5 Stress values at 1.5% drift for varying annular ring dimensions ................................... 29
Table 3.6 Stress values at 1.5% drift for varying coefficients of friction ...................................... 29
Table 3.7 Stress values at 1.5% drift with presence of vertical reinforcement .............................. 30
Table 4.1 Nonlinear material parameters....................................................................................... 33
Table 5.1 Experimental variables and measured material properties ............................................ 47
Table 6.1 Maximum drift level recorded by embedded strain gages............................................. 74
Table 7.1 Comparison of experimental connection strength to design column strength ............... 85
Table 7.2 Comparison of experimental stiffness to design stiffness.............................................. 87
Table 7.3 Experimental footing stress demands ............................................................................ 90
Table 7.4 Experimental and moment-curvature response.............................................................. 91
Table 7.5 Specimen I response at identified performance states ................................................... 97
Table 7.6 Specimen II response at identified performance states.................................................. 97
Table 7.7 Specimen III response at identified performance states................................................. 98
Table 7.8 Specimen IV response at identified performance states ................................................ 98
Acknowledgements
This research was sponsored by the Army Research Laboratory (Cooperative Agreement Number
DAAD19-03-2-0036) and was accomplished with the financial support and assistance provided
by the Advanced Technology Institute and the Vanadium Technology Partnership. The views and
conclusions contained in this document are those of the author and should not be interpreted as
representing the official policies, either expressed or implied, of the Army Research Laboratory
or the U.S. Government.
The author would like to recognize Dawn Lehman, Peter Mackenzie-Helnwein, and Charles
Roeder for their influential guidance and support throughout the project.
The author would also like send a shout out to Ryan Thody and Travis Williams for their
assistance in the execution of the research.
vi
Dedication
My sincere thanks go to my family and each one of my friends for providing me with invaluable
support, patience, and inspiration. You have helped me realize my goal and imagine my future.
vii
Chapter 1: Introduction
Concrete-filled steel tube, CFT, structural members efficiently combine the tensile strength
and ductility of steel with the compressive strength of concrete. Lighter and more slender CFT
columns can replace traditional steel or reinforced columns with equivalent resistance. The tube
provides large buckling and bending capacity by placing the steel at the outer perimeter of the
section where the moment of inertia and radius of gyration are greatest. There, the steel can
perform most effectively in tension with the minimum amount of material. The concrete core
provides compressive strength and flexural stiffness to the section, and it delays and often
prevents local buckling of the steel tube. In addition, the steel tube enhances the shear resistance
and confines the concrete, increasing the compressive strength and strain capacity of the concrete,
and in turn the ductility of the member. Multi-story buildings develop large compressive loads
due to the accumulation of gravity loads over the height of the building, and the high axial
strength of CFT columns makes them particularly attractive for the lower story columns. In
seismic loading, columns may be subjected to significant drifts and large moment demands; an
extensive body of research has shown that CFT members provide favorable strength, ductility and
energy dissipation under cyclic loading.
In addition to reducing section sizes, CFT members provide economic benefits by reducing
costs associated with traditional steel or concrete construction. A CFT column providing
resistance equivalent to a steel column replaces a significant portion of the steel weight with
concrete, a much less expensive material in a dollar-per-strength comparison. In addition, CFT
construction can proceed rapidly, as erection of the tubes and framing elements in a building can
precede concrete pouring by several stories. CFT columns reduce time and costs associated with
reinforced concrete construction by eliminating the need for formwork and additional
reinforcement.
2
core reduces member deflections and restrains local buckling, while the full tensile strength of the
high-strength steel tube can still be utilized.
The addition of vanadium to the steel manufacturing process results in a high-strength lowalloy steel; the vanadium provides increased yield strength, while assuring weldability and yield
elongation (ductility) comparable to low carbon mild steels. In addition to permitting an economy
of materials, vanadium-alloy steel permits the use of spirally weld tubes. The steel is formed into
a coil rather than flat plate stock, and the coil is unwound to form the tube. The spiral-weld
manufacturing technique is the most rapid and economical method for producing large diameter,
thin wall tubes.
The analytical and experimental research described in this document is one component of a
comprehensive investigation of the complete behavior and design of CFT members and
connections to CFT constructed with spirally welded, high-strength vanadium alloyed steel tubes.
3
1.3 Overview of Report
This report is composed of eight chapters. Chapter 2: provides a summary of current
specifications for the design and construction of CFT members and a survey of previous research
studies of moment resisting column base connections. The literature review identifies various
column base connection types for both steel and CFT columns, and for each type describes the
construction, load resisting mechanisms and parameters affecting the connection behavior.
The proposed embedded connection detail is described in Chapter 3:, followed by a
description of a preliminary parametric study. The parametric study investigated the effect of
various connection parameters on the footing, column, and connection response, and was used to
determine connection parameters appropriate for experimental testing. A description of the finite
element model used in the parametric study is provided, followed by the results of the analyses
performed for the parametric study.
Chapter 4: provides a description of a nonlinear finite element analysis of the CFT-footing
connection. The analytical model used in the preliminary parametric study was modified to
include nonlinear, inelastic material behavior. The results of the nonlinear analysis were used to
form qualitative predictions of the inelastic behavior of the proposed connection.
Chapter 5: describes the experimental test program. The design and construction of four CFTfooting connection test specimens is presented, and the test apparatus, loading history, and
instrumentation used for testing are described.
The results of the experimental testing are presented in Chapter 6:. The behaviors observed
during testing are described for each specimen, including damage occurring in the footing and the
column, and the cyclic load-drift response. The behaviors and response of the specimens are
compared. Methods for data correction are described, and the measured connection response,
including the tube strain and curvature distribution, and flexural behavior of the column and
connection are discussed.
Chapter 7: provides assessments of the experimental connection performance by comparing
the measured response quantities to design expressions and theoretical and numerical predictions.
Performance states observed during testing are identified, described, and tabulated for each
specimen. The research program is summarized and conclusions are drawn in Chapter 8:.
D / t 0.15
Es
Fy
(2.1)
The effective bending stiffness, EIeff, of the composite cross-section is defined in Equations 2.2
and 2.3.
5
EI eff = Es I s + Es I sr + C3 Ec I c
(2.2)
As
C3 = 0.6 + 2
0.9
As + Ac
(2.3)
Where Es and Ec are the elastic moduli of the steel and concrete, respectively; Is, Isr and Ic are the
moments of inertia of the structural steel, reinforcing steel (if present), and concrete, respectively;
and As and Ac are the cross-sectional areas of steel and concrete, respectively. The effective
stiffness is used to determine the nominal compression and flexural strength of CFT members.
Two methods are provided in current AISC provisions to estimate the flexural strength of
CFT members subjected to bending or combined loading: the plastic stress distribution method
and the strain compatibility method. The plastic stress distribution method assumes that the steel
has reached its yield stress in tension and compression, and that a compressive stress of 0.95fc is
uniformly distributed in the compression region of the concrete for circular sections, as shown in
Figure 2.1. Enforcing equilibrium on the force diagram in Figure 2.1(b) enables the areas of steel
in tension and compression, Ast and Asc respectively, and the area of concrete in compression, Acc,
to be determined; and then the nominal moment capacity according to the plastic stress
distribution method, Mn(p.s.) can be calculated.
Asc
neutral axis
Acc
0.95fc
AscFy
0.95fc Acc
P
Ast
(a) CFT cross-section
Mn(p.s.)
AstFy
(b) force equilibrium
Figure 2.1 Free body diagram of CFT cross-section used in plastic stress distribution method
The strain compatibility method assumes a linear distribution of strain across the section, and
a limiting concrete compression strain of 0.003 in/in, as shown in Figure 2.2. AISC provisions
allow the strength to be estimated using any constitutive relations for the concrete and steel
considered reasonable by the engineer. Both methods assume the tensile capacity of the concrete
is zero.
c = 0.003
neutral axis
st
(b) strain diagram
(2.4)
The nominal flexural strength of each experimental CFT specimen was calculated using both
of the AISC strength determination methods and compared to the measured capacity. The results
of the comparison are listed in Table 2.1 and presented graphically in Figure 2.3. The moment
calculated using the plastic stress distribution method, Mn(p.s.), was calculated using the free body
diagram in Figure 2.1.
Moment-curvature analysis was used to determine flexural strength according to the strain
compatibility method, Mn(s.c.). The steel tube was modeled as a bilinear material with a strain
hardening ratio of 15%. The concrete constitutive relationship was the Kent, Scott and Park
model (Mazzoni, McKenna and Fenves, 2005). The effects of confinement on the concrete
compressive strength and ductility were calculated according to Mander, et al. (1988).
7
Table 2.1 Comparison of predicted to experimental flexural strengths
Test
D/t
Fy
(ksi)
34
92
59
41.0
73.0
54
fc
(ksi)
P/P0
Mexp
(k-in)
Mn(p.s.)/Mexp
Mn(s.c.)/Mexp
5.4
5.9
5.1
5.1
0.22
0.60
0.44
0.59
5239.9
3929.9
8213.9
3151
0.96
0.97
1.04
0.92
0.76
0.45
0.71
0.56
13.3
13.3
13.3
13.3
0
0.29
0.61
0.17
185.87
262.88
269.96
266.42
0.82
1.17
1.22
0.97
0.78
0.98
0.74
0.87
4.2
3.8
5.1
1.71
0.64
0.21
99.13
151.35
138.96
1.29
1.08
0.98
0.55
0.62
1.04
13.0
10.4
11.0
0.27
0.08
0.56
435.48
330.15
434.49
0.96
0.99
0.96
0.76
0.87
0.56
5.7
12.8
5.2
12.3
0.7
0.53
0.73
0.54
1531.2
1787.9
1349.7
1944.2
0.69
0.86
1.09
1.04
0.38
0.54
0.61
0.58
14.5
15.1
5.8
10.2
0.43
0.26
0.69
0.45
4800.87
5461.17
3991
4530.91
0.94
0.98
0.77
0.85
0.63
0.75
0.44
0.56
4.0
15.1
9
0
1.71
0.49
0.69
1.29
0.98
0.14
0.38
1.04
0.67
0.17
The range in the ratio of predicted to experimental strengths, Mn/Mexp, is wide for both
methods. However, the mean of the plastic stress distribution method is 0.98, while the mean of
the strain compatibility method is 0.69, indicating an underprediction of CFT flexural strengths.
At high axial load ratios, both the accuracy of both methods decreases. The range of results in the
plastic stress distribution method become larger, while strain compatibility method predictions
become more conservative with increasing axial load ratio, as shown in Figure 2.3. No other
trends in accuracy were observed over the experimental parameters considered.
8
1.5
Mn(s.c.)
Mn(p.s.)
Mn/Mexp
1.25
1
0.75
0.5
0.25
0
0.2
0.4
0.6
0.8
The low predictions made by the strain compatibility approach can be attributed to the strain
limit of 0.003 in/in imposed on the concrete, which may be inappropriate and excessively low for
concrete fill in a CFT column. The strain limit of 0.003 in/in for reinforced concrete sections
limits spalling of concrete cover. CFT members do not have cover, and their flexural strength is
often controlled by buckling of the steel tube. The compression strain limit may also explain the
effect of axial load ratio on the accuracy of the strain compatibility method. At high axial loads
the triaxial stress state, and thus confinement on the concrete fill, increases to improve the
compression strength and strain capacity of the concrete. With increasing axial load, the 0.003
in/in strain limit becomes less suitable for a CFT column.
Due to slenderness effects and other design considerations, CFT columns will rarely be
subjected to high axial load ratios in actual construction. Maximum CFT moment capacity occurs
at approximately 0.3 axial load ratio. This limit is indicated by the dashed line in Figure 2.3. In
this range of low axial load ratios, the accuracy of both methods is improved. For the studies
considered with P/P0 less than 0.3, the mean Mexp/Mn(s.c.) increases to 0.85, and the mean
Mexp/Mn(p.s.) remains the same. The percent of unconservative predictions made by the plastic
stress distribution method, with Mexp/Mn(p.s.) greater than one, decreased from 32% for all
specimens to 12% for specimens with P/P0 less than 0.3.
9
2.2 Survey of Column Base Connections
In moment resisting frames and bridges, column bases are subjected to large moment and
rotation demands during extreme seismic events. Therefore, a column base connection must
provide adequate strength to transfer the column forces and sustain the inelastic deformation
demands in the column due to seismic loading. A survey of previous research on column base
connections for steel and CFT columns indicates that a variety of connection types have been
used and tested for adequacy to meet these seismic demands. These connections can be
categorized into the following types identified in Figure 2.4: (a) exposed base plate connection,
(b) embedded connection, (c) embedded structural steel connection and (d) semi-embedded
connection. The following sections describe the load resisting mechanisms and recent research on
each connection type, respectively.
(b) embedded
(d) semi-embedded
Figure 2.4 Types of column base connections for steel and CFT sections
10
P
V
Recent major earthquakes have led to detailed evaluations of the behavior of existing column
base designs. For example, damage and unsatisfactory performance was observed in several base
plate connections after the Northridge Earthquake of 1994 (Gruavilardell, et al., 2004). Extensive
damage to steel base plates due to the Kobe Earthquake of 1995 led Japanese researchers to revise
design methods for exposed column base connections, and have motivated research of
connections which retain strength after the column base has yielded to improve seismic
performance (Kadoya, Kawaguchi and Morino, 2005). Current research has focused on
developing column base connections with improved strength, stiffness and ductility (Hitaka, Suita
and Kato, 2003).
11
the behavior of the steel or CFT column base can be considered fixed (Gravilardell, et al., 2004;
Hitaka, Suita and Kato, 2003). With these large embedment lengths, the full moment strength of
the column is resisted by bearing of the column face, and the failure mechanism for such
connections is plastification of the column at the foundation level. The large embedment length
requires a very deep foundation. The base plate contributes to the response of the connection to
axial forces by providing bearing beneath the column base in compression and increasing
anchorage of the column with the foundation. Although bearing is the predominant mechanism of
force transfer, some consideration of the reinforcement in the concrete foundation is necessary to
prevent excessive damage and strength deterioration due to tension stresses in the foundation.
Based on a survey of previous research, Grauvilardell, et al. (2004) concluded that columns
embedded 2D or more are expected to have larger rotational stiffness values than shallowly
embedded columns, and show ductile behavior and good energy dissipation.
P
M
le
Grauvilardell, et al. (2004) report that Japanese researchers have investigated shallow
embedded connections (le < 2D) for both steel and CFT columns and have shown that embedment
length, base plate and anchor rod dimensions all have significant effects on the behavior of these
column bases. In shallow embedded connections, bearing of the column face against the column
foundation is not adequate to develop the full moment strength of the column. Generally,
plastification of the column has been found only when large, thick base plates and numerous
anchor rods are used (Grauvilardell, et al., 2004). Heavy cracking in the concrete foundation often
accompanies column plastification. Damage in the concrete foundation often controls the
deterioration of lateral strength in shallow embedded connections. Horizontal reinforcement helps
12
to reduce cracking at the surface of the foundation; however, vertical reinforcement has not been
found to prevent the development of foundation cracks. The research surveyed by Grauvilardell,
et al. (2004) shows that connections with shallow embedded columns have had less strength and
stiffness, exhibited greater foundation damage and yielded more pinched hysteresis loops in
cyclic tests compared to deeply embedded columns.
Hsu and Lin (2003) performed experimental tests on full-scale embedded CFT column base
connections to investigate the effects of embedment length and base plate details on rigidity
(fixity) of the column base and the ability of the connection to develop the full strength of the
column. The columns were square CFT with D/t ratio of 50 and Fy of 47 ksi. Embedment length
was varied from 0.5D to 1.5D. The concrete foundations were cast in two lifts, with a
construction joint at the depth of embedment. The concrete compressive strengths were 3.5 and
5.1 ksi for the first and second lift, respectively. The end plate and anchor bolt configuration was
the same in each specimen; however, some of the specimens were stiffened at the column base as
shown in Figure 2.7. The stiffeners were intended to move the failure surface from the embedded
region of the column to the foundation surface, reducing the force demands placed on the
foundation and ensuring that the moment capacity of the CFT columns could be fully developed.
The specimens were subjected to constant axial load equal to 15% of P0 and cyclic lateral
loading.
stiffeners
The test results showed that the failure mode and the stiffness of the connection were
governed primarily by the embedment length. Specimens with 0.5D embedment failed due to
excessive cracking and uplift in the foundation without developing the full CFT moment capacity,
and achieved base rigidities equal to 7080% of a fixed end condition. A large portion of the
column base moment in the shallow embedded connections was resisted by the anchor bolts.
13
Columns embedded 1.0D and 1.5D were able to develop the full strength of the CFT with limited
damage to the foundation; although, specimens embedded 1.5D did not show significant increase
in strength or stiffness compared to specimens with 1.0D. The flexural stiffness of the connection
increased with embedment; specimens embedded 1.0D and 1.5D had stiffness ranging from 85
97% of a fixed end condition. The presence of the stiffeners did not influence the failure mode of
the connections, but did reduce the demands on the anchor rods and increased base rigidity. The
stiffeners increased base rigidity 15% in specimens embedded 0.5D; the effect of the stiffeners on
base rigidity decreased with increased embedment.
Marson and Bruneau (2004) performed experimental tests on a series of 300 mm (11.8 in.)
diameter columns with the proposed connection. The foundation details were as shown in Figure
2.8 and remained constant for all specimens; the D/t ratio of the columns was varied from 34 to
64. The average measured concrete compressive strength was 5.5 ksi, the yield strengths of the
14
steel tubes and foundation components were 64 ksi and 42 ksi, respectively. The specimens were
subjected to constant axial load and cyclic horizontal loading. The proposed column base
connection was able to develop the full composite capacity of each test specimen. Failure was
controlled by buckling of the steel tube above the foundation, and damage to the concrete
foundation was not observed in any of the tests. All columns exhibited high ductility and were
able to reach drifts of 7% without significant loss of strength.
produced full hysteretic curves with large energy dissipation. However, strain data taken during
testing indicated that the steel sections used in the connection were subjected to insignificant
stresses; the authors suggest that the concrete foundation likely provided most of the resistance to
the column forces.
15
strut in the concrete and vertical foundation reinforcement prevent column pull-out (Morino et al.,
2003).
M
V
acting forces
(a) bearing
Morino et al. (2003) tested semi-embedded connections with embedment length equal to
0.1D, 0.5D, 1D and 2D. The columns were square tubes with D/t ratio of 22. Tubes with two
different yield strengths were used: 55 kis and 45 ksi. The average measured compressive
strength of the concrete foundation was 4.6 ksi. Reinforcement in the foundation included
longitudinal bars in the direction of loading, stirrups and hoop steel and longitudinal bars around
the column and anchor bolts. The specimens were subjected to monotonic and cyclic horizontal
loads.
Four ultimate states were identified during testing: 1) anchor bolts yielding, 2) anchor bolts
yielding and cracking in foundation, 3) both anchor bolts and column yielding and concrete
cracking, and 4) only column yielding. The type of loading had little effect on the strength or
failure mode of the specimens. For short embedment lengths (0.1D and 0.5D) the steel tube
remained elastic and yielding in anchor bolts controlled ultimate strength. The foundations of
specimens with 0.5D and 1D embedment were heavily cracked, which led to strength
deterioration during cyclic loading. Severe damage in the foundation controlled the maximum
strength of specimens with 1D embedment, although the anchor rods and column reached
yielding. At 2D embedment, no foundation cracking was observed and the maximum strength of
the specimen was determined by column yielding and subsequent local buckling. In the shallowly
embedded connections, resistance was provided mainly by the anchor rods, while the embedded
portion of the column provided additional resistance to the deeper embedded connections.
16
Based on the experimental results, the authors concluded that the maximum strength of the
specimens with longer embedment can be predicted by the sum of the strengths of mechanisms
(a) and (b) shown in Figure 2.10. The sum of all three mechanisms is necessary to estimate the
strength of specimens with shallower embedment.
17
CFT
tube anchored to lower
foundation
upper lift
foundation
lower lift
18
3.2 Description of Analytical Model
An numerical model of the connection described above was developed to investigate the
effects of varying connection parameters on the response of the connection to aid in the
development of an experimental test matrix. A series of finite element analyses were performed
to determine the effects of footing depth, df, embedment length, le, annular ring dimensions, bf
and tf, steel-concrete friction, and footing reinforcement on the behavior of the CFT-footing
connection. Table 3.1 lists each of the analyses performed in this study.
Table 3.1 Analyses performed for parametric study
df
le
1.2
1.5
1.8
1.8
1.8
1.8
0.6
0.6
0.6
0.9
1.2
1.5
Coefficient of
Friction
~0 0.25 0.5 ~1
0.2Dxt
0.2Dxt
0.3Dxt
0.3Dxt
Vertical
Reinforcement
Figure 3.2 describes the modeled connection, including column and footing dimensions and
footing reinforcement. Member sizes and footing reinforcement representative of typical
construction were selected. A 20-in. diameter by 0.25-in. thick tube was selected for the column,
a model representative of a full-scale column in a medium- to high-rise building; or a one half- to
one third-scale bridge pier. The diameter-to-thickness, D/t, ratio of 80 for this configuration is
intended to take advantage of the high yield strength of vanadium alloyed steel. For a full-scale
building column, a 12 ft. column height would be typical. An inflection point exists at
approximately mid-height of a building column subjected to lateral displacement, and only the
portion of the column below the inflection point is considered in the analyses. The width, bf, and
thickness, tf, of the annular ring were varied.
The footing was 76 in. in the direction of loading by 68 in wide. These dimensions were
selected to minimize the effects footing size and boundary conditions had on the local behavior of
the connection. Although a construction joint may occur in the footing in actual construction, the
numerical model considers this concrete continuous. The column embedment, le, was varied, and
was measured from the bottom of the column flange to the surface of the footing. Footing depth,
df, varied. The flexural reinforcement in the footing was scaled from typical reinforcement found
in Washington and California state bridge pier footings, and remained constant among each of the
19
analyses. It consists of #6 bars spaced at 4 in. top and bottom in the direction of loading, and
#4 bars spaced at 9 in. top and bottom in the direction perpendicular to loading,. Both California
and Washington employ significant flexural reinforcement in concrete footing construction, but
California bridge pier footings also include significant shear (vertical) reinforcement while
Washington does not. Vertical reinforcement was included in one analysis to determine if the
additional reinforcement improves connection behavior and strength of the footing.
D = 20 in.
#6 bars
at 4 in.
#4 bars
at 9 in.
t = in.
CFT and
annular ring
6 ft.
68 in.
le
df
tf
bf
(a) cross-section view
76 in.
(b) plan view
The MSC.Marc finite element analysis program was used to perform each analysis. The
connection model and typical mesh are shown in Figure 3.3. Eight-node brick elements were used
to model the steel tube, concrete fill, and concrete in the footing. Typical elements in the
connection region had linear dimensions approximately 10% of D. To reduce computation time,
the element sizes were gradually increased with distance from the connection region, as shown in
Figure 3.3. Rebar elements, which define a plane of distributed reinforcement with uniaxial
stiffness, were used to represent reinforcement planes with stiffness equivalent to the flexural
reinforcement described above. The vertical reinforcement was modeled discretely as a series of
two-node, uniaxial truss elements. Deflections in the shear reinforcement elements were
controlled by the displacement of the surrounding footing elements.
20
z
y
connection region
flexural reinforcement layers
Linear elastic material models defined the behavior of the materials in the CFT columnfooting connection model.
reinforcement was 29,000 ksi. The footing and column fill were modeled as concrete with 5 ksi
compressive strength, fc, and 4030 ksi (57000fc (psi)) modulus of elasticity, Ec. The Poissons
ratio for steel and concrete were 0.33 and 0.2, respectively.
As the column is subjected to horizontal load and the connection deforms, the embedded
portion of the steel tube will come into come into contact with, and move relative to, the column
fill and concrete footing. These contact surfaces were modeled using the contact option in
MSC.Marc. This option controls the geometric constrains of the contact and calculates the shear
force transferred between elements from the normal force between them. The contact bodies
considered in this model were the column fill, steel tube and annular ring, and concrete footing.
The bodies were defined as initially touching; separation between bodies was allowed and
penetration, or overlap of bodies, was prevented. Bond and sliding behavior were modeled as
Coulomb friction behavior. The coefficient of friction, , between steel and concrete was varied
as a parameter of the study, the values are shown in Table 3.1. Contact causes geometric
nonlinearities between adjacent elements and was the only nonlinearity considered in the
preliminary parametric study.
Symmetry of the connection permitted modeling only one half of the column and footing, as
shown in Figure 3.4. To enforce the symmetry condition, displacement in the global y-direction
was prevented at each of the nodes on the symmetry plane. The footing was continuously
21
supported at the base and side to prevent rigid body motion in the global z- and x-directions,
respectively. In actual conditions, the footing would be supported at the base and sides by elastic
soil; however, accurately defining such conditions was beyond the scope of this study. Rotation
was allowed at the top of the column to ensure the absence of bending moment at the location of
the inflection point.
P
()
z
symmetry plane
y
Constant 258 kip axial load, P, was applied to the column, equal to approximately 0.1P0. This
was applied as a distributed load to the top face of the concrete fill. Horizontal load was applied
as a linearly increasing displacement, (), at the top of the modeled column along the plane of
symmetry. In each analysis, the top of the column was displaced to 4 inches, 5.6% drift, in 0.05
inch increments. The horizontal load, H, required to reach each displacement was determined
from the reaction forces on the system. The applied loads and material definitions used in the
analyses are summarized in Table 3.2.
Table 3.2 Preliminary analytical model definition
P/P0
max / Lcol
Increment / Lcol
E (ksi)
Poissons Ratio
0.1
0.55
0.0007
Steel Concrete
29,000
4030
0.33
0.2
22
3.3 Results of Parametric Study
The behavior of the connection was evaluated using several criteria. To identify the yield
condition of the steel tube, the Von Mises stresses, v at the base of the tube and 2 inches above
the surface of the footing were investigated. Maximum and minimum principle stresses, 1 and
3, were examined in locations of the concrete footing likely to sustain damage, including
locations beneath the annular ring and at the tube bearing surface. Anchorage provided by the
annular ring may induce a cone-shaped failure plane, extending from the column base at a plane
approximately 45 through the footing, as indicated in Figure 3.5(b). Principle stresses at the 45
plane were also evaluated to assess the stresses in the footing in the region of the assumed failure
plane. Figure 3.5 shows the locations on the steel tube and the concrete footing where the stresses
were evaluated. The results of the analysis are presented in terms of Fy for the tube, fc for
concrete in tension, and fc for concrete in compression. Further, the numerical results are
evaluated based on the drift level at which the connection components reach these strength limits.
foundation
bearing surface
assumed
45 failure plane
23
concrete footing uplifted slightly and the annular ring deformed, as shown in the detail of
Figure 3.6.
uplift
separation
downward
deflection
Figure 3.7 plots v at each location over increasing column drift, indicating that the base of
the tube reached Fy at approximately 1.1% drift. Above the footing, the tube reached Fy at 1.25%
drift. Slight nonlinearities in the Von Mises stress vs. column drift plot occur due to the nonlinear
behavior of the steel-concrete contact. At 1.5% drift, the calculated tube stress had exceeded Fy
over the entire embedded length of the tube and up to approximately 1D above the footing
surface. To evaluate the stress state in the connection at full plastification of the tube, the results
of all analyses were compared at this drift level. The calculated values of v in the steel were
largest at the base of the tube, and decreased along the column height, as shown in Figure 3.8. On
the compression side of the tube, the stress decreased slightly where the tube bears against the
concrete footing; at this location, friction effects allow the tube to transfer some compressive
stress into the footing. At 1.5% drift v was1.36Fy at the base of the tube, near the annular ring on
the tension side. On the compression side, this stress was 1.25Fy. Two inches above the footing,
v was 1.19Fy and 1.2Fy on the tension and compression sides, respectively.
24
1.5
base (tens.)
1.25
base (comp.)
2" above foundation (tens.)
0.75
0.5
0.25
0
0.0%
0.3%
0.6%
0.9%
1.2%
1.5%
Column Drift
Figure 3.8 Von Mises stress distribution (ksi) in baseline tube at 1.5% drift (Note: 100 ksi = 1.4Fy)
The calculated maximum and minimum principles stress distributions in the concrete footing
at 1.5% drift are shown in Figure 3.9 and Figure 3.10. Stress concentrations occurred where the
tube bears against the concrete on the compression face, and near the annular ring. Tension
stresses also occurred below the tension side of the annular ring. At 1.5% drift, the largest 1 of
171fc occurred at this location, where high drift levels caused the column fill to separate from
the footing concrete. At the 45 plane, 1 was 14.5fc. Below the compression flange and at the
bearing face of the concrete, the 1 values were 41.4fc and 109fc, respectively. Design
25
provisions published by the American Concrete Institute (ACI) suggest a nominal cracking
stress of 6fc for concrete in tension. The footing in the baseline study reaches this limit at
approximately 0.2% drift at the bearing face of the concrete, and at approximately 0.7% drift in
the 45 plane, well before the tube reaches its yield stress.
Figure 3.9 Maximum principle stress distribution (ksi) in baseline footing at 1.5% drift
(Note: 13 ksi = 184fc)
Figure 3.10 shows that at 1.5% drift, the maximum 3 was 7.9fc, located near the tension
flange where the upward motion of the tube caused significant bearing stress. However, stress at
this location was highly concentrated, and was neglected as a measure of global connection
performance in the parametric study. The 3 values below the compression flange and at the
bearing surface of the footing were 2fc and 4fc respectively. Due to confinement effects, 3
values exceeding fc may not indicate crushing in the footing.
26
Figure 3.10, Minimum principle stress (ksi) distribution in baseline footing at 1.5% drift
(Note: 40 ksi = 8fc)
27
Table 3.3 Stress values at 1.5% drift for varying footing depths
Von Mises stress in tube (v/Fy)
df
at base
above footing
tens.
comp.
tens.
comp.
1.2
1.5
1.8
1.36
1.35
1.35
1.26
1.25
1.23
1.19
1.18
1.18
1.19
1.19
1.18
45 plane
f 'c
14.46
13.85
13.64
below flange
f 'c
-0.08
-0.07
-0.07
f 'c
41.44
40.04
39.13
bearing surface
f 'c
f 'c
-1.97
-1.94
-1.92
108.9
109.1
108.8
f 'c
-4.01
-4.04
-4.06
le
at base
above footing
tens.
comp.
tens.
comp.
0.6
0.9
1.2
1.5
1.63
1.21
0.87
0.63
1.44
1.15
0.75
0.53
1.15
1.10
1.07
1.05
1.19
1.14
1.11
1.10
f 'c
12.55
7.42
5.04
2.84
below flange
3
f 'c
-0.07
-0.09
-0.12
-0.10
f 'c
40.0
32.9
26.5
49.2
f 'c
-2.11
-1.39
-0.95
-0.28
bearing surface
f 'c
112.5
125.4
125.4
128.7
f 'c
-3.92
-4.2
-4.32
-4.35
28
3.3.4 Effects of Annular Ring Dimensions
As the primary component in the mechanical anchorage of the connection, it is necessary to
investigate the effects of the annular ring dimensions on the connection behavior. Five analyses
were performed on models with a footing that was 1.2D deep and with 0.6D embedment. The
combination of ring width and thickness was varied. In four analyses the rings were 0.2D wide by
t thick, 0.2D by 2t, 0.3D by t, and 0.3D by 2t. In one analysis there was no annular ring. The
stress values at 1.5% drift for connections with varying annular ring dimensions are listed in
Table 3.5.
Excluding the annular ring from the connection significantly reduced the calculated
connection stresses and the global load-drift response. As shown in Figure 3.11, the stiffness of
the connection without an annular ring was approximately 90% of the stiffness for connections
with a ring. The anchorage provided by the annular ring increased the demands on the tube and
increased the forces developed in the footing. The presence of an annular ring approximately
doubled v in the tube at the base and above the footing, and increased 1 at the 45 plane by
400%. Compression stress calculated at the location beneath the annular ring increased 150%
with the presence of an annular ring.
150
120
no flange
4x0.25
4x0.5
6x0.25
6x0.5
90
60
30
0
0.0%
0.5%
1.0%
1.5%
Column Drift
Figure 3.11 Load-drift response for models with varying annular ring dimensions
The stress values in Table 3.5 indicate that the dimensions of the annular ring have a small
effect on the various response measures. The rings with an increased tf have increased tensile
stress demands at the tube base. The 1 measured at the 45 plane increased with increases in
either bf or tf, indicating increased anchorage and transfer of column forces into the footing.
29
Table 3.5 Stress values at 1.5% drift for varying annular ring dimensions
Von Mises stress in tube (v/Fy)
at base
above footing
bf by tf
tens.
comp.
tens.
comp.
none
0.2Dt
0.2D2t
0.3Dt
0.3D2t
0.79
1.64
1.76
1.62
1.75
1.09
1.47
1.58
148
1.57
0.52
1.16
1.18
1.14
1.17
0.88
1.20
1.23
1.19
1.21
45 plane
below flange
f 'c
f 'c
3.08
13.34
14.56
14.84
16.40
-0.32
-0.08
-0.04
-0.10
-0.07
f 'c
f 'c
-1.39
-2.16
-2.12
-2.57
-2.44
109.3
112.9
113.1
115.5
116.7
f 'c
31.4
42.0
39.2
39.3
36.9
bearing surface
3
f 'c
-4.25
-3.90
-3.93
-4.05
-4.0
~0
0.25
0.5
~1
above footing
tens.
comp.
tens.
comp.
1.36
1.36
1.36
1.35
1.27
1.47
1.24
1.20
1.20
1.16
1.20
1.20
1.18
1.20
1.25
1.22
f 'c
13.76
13.30
14.47
14.24
below flange
3
f 'c
-0.08
-0.07
-0.08
-0.08
f 'c
43.5
42.03
41.04
40.3
f 'c
-2.1
-2.1
-1.95
-1.92
bearing surface
f 'c
11.9
112.9
106.6
101.2
f 'c
-3.97
-3.89
-4.05
-4.11
30
3.3.6 Effects of Vertical Reinforcement
To investigate the potential benefits of vertical reinforcement in the footing, an analytical
model with shear ties was compared to an identical analysis with no vertical reinforcement. The
vertical reinforcement consisted of No. 3 ties located at each intersection of flexural reinforcing
bars. The results of the analyses are shown in Table 3.7. The Von Mises stresses in the tube base
were reduced by 17% with the presence of vertical reinforcement and vertical reinforcement
caused a 7% increase in the predicted 1 at the 45 plane. This may imply that the presence of
shear reinforcement will act to distribute forces in the concrete footing away from the tube. The
total load-drift response of the connection with shear reinforcement shows an increase of
approximately 2% in connection stiffness compared to the connection without shear
reinforcement. Inspection of the analytical model showed that the vertical elements representing
shear reinforcement carried stress equal to 3% of the yield stress of normal reinforcing steel.
Bond along the length of one of these reinforcing bars would lead to an average stress of 3.5fc
in the surrounding concrete.
Table 3.7 Stress values at 1.5% drift with presence of vertical reinforcement
Von Mises stress in tube (v/Fy)
at base
above footing
Vert.
Rfmt.
tens.
comp.
tens.
comp.
none
included
1.64
1.36
1.47
1.26
1.16
1.19
1.20
1.18
f 'c
13.31
14.34
below flange
3
f 'c
-0.08
-0.08
f 'c
42.03
41.38
f 'c
-2.16
-1.97
bearing surface
f 'c
112.9
109.0
f 'c
-3.89
-4.01
the annular ring wass necessary to develop anchorage of the column in the footing;
however, the width and thickness of the ring did not appear to have significant effect on
connection response.
in the presence of the annular ring, and with linear elastic material behavior, changes in
friction did not have an effect on the stresses calculated in the steel tube or footing.
31
Vertical reinforcement reduced the stress demands in the embedded portion of the steel
tube and increased stress measured in the footing.
32
The concrete in the footing and column fill was modeled as a Drucker-Prager material (which
MSC.Marc refers to as a Mohr-Coulomb material), accounting for the increased compression
strength of confined concrete. The MSC.Marc implementation for this material requires definition
of the initial yield stress, , and the factor . The failure condition for a linear Drucker-Prager
material can be described by the yield function in Equation 4.1 (MSC.Software Corporation,
2004).
f = J1 +
J2
J1 = ii
J2 =
1
2
=0
(4.1)
(4.2)
ijd ijd
(4.3)
ijd = ij J1 ij
(4.4)
Where J1 is the first invariant of stress, , and J2 is the second invariant of the deviatoric stress,
d. The initial yield stress, , can be found by solving Equation 4.1 for the case of a uniaxially
loaded plane stress element with compression strength fc, resulting in the following expression:
= f 'c 1 3
(4.5)
The factor is calculated from the internal angle of friction, f, by Equation 4.6. For this model,
33
3
1 3
= sin f =
sin f
9 + 3sin 2 f
(4.6)
A smeared cracking model was used to define the tensile behavior of the concrete footing and
column fill. The MSC.Marc cracking implementation requires that the softening modulus, critical
cracking stress, crushing strain, and shear retention factor be defined. The softening modulus,
Esoft, was set at the maximum value allowing convergence of the model up to 5.5% drift, 10500
ksi. Accurate modeling of concrete cracking would require that the softening modulus be defined
in terms of the fracture energy and the characteristic element length, a parameter dependent on
mesh size; however, this level of analytical detail was not within the scope of this study. The
critical cracking stress, cr, was set at 6fc, The crushing strain, cu, was zero; hydrostatic
compression does not cause crushing. The shear retention factor, Cshear, was 0.05. The material
definitions are summarized in Table 4.1.
The steel tube, annular ring and reinforcement were modeled as linear-elastic, perfectly
plastic materials, with the yield plateau at the design yield stress for each component: 70 ksi for
the tube, 50 ksi for the annular ring, and 60 ksi for the reinforcing steel.
Table 4.1 Nonlinear material parameters
Es (ksi)
Fy,tube (ksi)
Fy,ring (ksi)
Fy,reinforcement (ksi)
fc (ksi)
Ec (ksi)
(ksi)
Esoft (ksi)
cr (ksi)
cu
Cshear
29, 000
70
50
60
5
4,030
37
0.19
3.35
10,500
0.42
0
0.05
The elastic baseline analysis described in Section 3.3 was repeated with the nonlinear
inelastic material definitions described above. The mesh shown in Figure 3.3 was refined in
connection region, as shown in Figure 4.1, to reduce the effects of stress averaging in locations
where stress values changed rapidly. The same boundary conditions and loading defined for the
elastic analyses were used.
34
The deformations in the nonlinear connection model are shown in Figure 4.2. Uplift and
separation occurred on the tension side of the connection, and downward deflections in the tube
and column fill occur on the compression side. In the inelastic analysis, deformations in each of
the connection components were larger and more concentrated than in the elastic model, as
shown in the detail of Figure 4.2. Outward deflections on the compression side of the tube above
the footing were large. The annular ring was significantly warped on both the tension and
compression sides of the connection and footing uplift near the tube increased relative to the
elastic analysis. In addition, elements in the concrete fill near the footing surface indicate regions
of concentrated deformation and bending.
uplift
separation
downward
deflection
35
On the tension face of the tube, highly concentrated stresses occurred at the tube-annular
ring joint, and decreased with distance from the tube base. The Von Mises stress, v, in the
normal strength steel (Fy = 50 ksi) annular ring first reached yield at 0.7% drift, and yielding
progressed until the all of the elements in the annular ring yielded at approximately 2% drift. The
v distribution in the tube base and annular ring are shown at increasing drift levels in Figure 4.3.
In the tube, v first reached the yield stress (70 ksi) in tension at 1.25% drift; however, this stress
was highly localized, and yielding progresses into the tube at 3.5% drift. Compression stresses in
the steel tube were highest at the bearing face of the column, where contact with the footing
occurred; and just above the footing surface due to combined bearing force and axial force caused
by column bending. Compression yielding first occurred at 2% drift, in the embedded portion of
the tube.
36
37
At low drifts, the maximum principle stress, 1, distribution in the analytical model
provides information about the progression of forces and damage in the footing. The 1
distribution in the footing is shown in Figure 4.4, and indicates that localized stresses occurred on
the tension face of the tube at the joint with the annular ring, and at the bearing face on the
compression side of the connection. The distribution of 1 confirmed the presence of an inclined
failure plane extending from the base of the tube. Stress values reached the critical stress limit of
6fc in this region at approximately 0.3% drift. With increasing drift, the failure plane spread
away from the column and the angle of the plane became increasingly shallow. The numerical
solution implies that 1 exceeds the critical stress limit in some regions of localized stress;
however, this is due to interpolation of the stress between solution points of regions where the
stress state changes rapidly. The progression of 1 over increasing drift shows that as the critical
stress limit was reached, the force and damage in the connection was redistributed into other
regions of the footing. At higher drift levels, the concrete elements became highly nonlinear,
making the 1 distribution less clearly reflective of total connection behavior.
38
39
At higher drifts, displacement-based measures of behavior provide qualitative insight into
the connection performance and the damage state in the concrete footing and column fill.
Cracking strains represent the plastic tension strain in concrete elements. The progression of
cracking strain, shown in Figure 4.5, indicates locations where cracking was likely to occur in the
footing and column fill. Cracks began to form in the footing at the tip of the annular ring at very
low drift, approximately 0.2%. With increased drift, these cracks spread from the annular ring
into the footing, and the initial cracks continued to widen. By 1.4% drift (Figure 4.5(a)) cracks
formed on the surface of the footing in the plane parallel to loading, first on the compression face
of the connection, and then on the tension side. As with the cracks in the footing, these cracking
regions spread with increased drift, and existing cracks experienced increased strain. In the
column fill, cracking occurred near the surface of the footing and at the base of the column, on
the tension side of the column. The cracking strains first occurred at approximately 0.4% drift,
and continued to grow with increased drift. The location of the cracks in the column fill indicates
that loading caused the column to pull out of the footing, but the footing and annular ring
provided anchorage, causing significant bending in the column at the surface of the footing.
40
41
The minimum principle stress, 3, distribution in the footing and column fill is shown in
Figure 4.6 for increasing drift levels. These figures show that compression stresses in the
connection were highly concentrated at the compression face of the connection and near the
annular ring. At approximately 0.2% drift, the bearing surface of the footing reached the concrete
compressive strength, fc; at 0.3% drift, the compression side of the column fill reached this stress
as well. The effects of confinement in the column fill and the connection region allowed
compressive stresses to exceed the compression strength of the concrete. With increased drift
demand, these stresses became larger within the same region but did not spread significantly,
indicating the horizontal loading caused localized crushing in the connection near the column and
annular ring.
42
43
Each of the stress distributions in Figure 4.3Figure 4.6 indicate that the connection
response became nonlinear at low drifts. Damage in the connection was concentrated around the
embedded portion of the tube; although, maximum principle stresses exceeding the critical stress
were widely distributed through the footing. The horizontal load-column drift response of the
inelastic model is plotted in Figure 4.7. This plot shows that after an initial elastic response,
gradual softening began at approximately 0.5% drift. (Note that the horizontal load was that
acting on the analytical model, i.e., one half of the full connection.) At this drift level, neither the
tube nor annular ring had yielded; although, localized principle stresses in the footing and column
fill had exceeded the fc limit in compression and cracking had begun in the connection region of
the footing and in the column fill.
90
80
70
60
50
40
30
20
10
0
0%
1%
2%
3%
4%
5%
Specimen Drift
6%
44
A 20-in. diameter, high-strength vanadium alloy steel tube was selected to model a full-scale
building column or half- to third-scale bridge pier. The tube was 0.25 in thick, resulting in a
diameter to thickness, D/t ratio, of 80. This exceeds the D/t limit of 62 (Equation 2.1) in AISC
design provisions for 70 ksi yield stress. The column met all other AISC design limits. The
design compressive strength of the concrete fill was 10 ksi, resulting in a nominal P0 of 4070
kips. The flexural strength of the CFT column was predicted using the plastic stress distribution
method described in Section 2.1.1 In the presence of axial load equal to 0.1P0, the predicted
flexural strength of the column was 11,065 k-in.
A normal-strength steel (50 ksi yield stress) annular ring 6 in. wide by 0.25 in. thick was
welded to the base of each steel tube using a complete joint penetration weld. The rings projected
4 in. on the outside of the tubes and 2 3/4 in. on the inside of the tubes.
The geometry and reinforcement common to all specimens is shown in Figure 5.1. The
footing was 68 in by 76 in. in plan and 24 in. deep. The dimensions were selected to prevent
footing size from influencing the connection behavior or failure mode, and to ensure that the
potential 45 failure plane did not exceed the footing dimensions or was not influenced by
specimen anchors. The columns in Specimens I, II and IV were embedded 12 in., 0.6D; an
increased embedment of 18 in., 0.9D, was used for Specimen III. Flexural reinforcement was
selected to ensure the footing had capacity to resist the predicted column flexural strength. For all
45
specimens, the flexural reinforcement in the direction of loading, l,l, consisted of No. 6 bars
spaced at 4 in, top and bottom. In Specimens I and II, flexural reinforcement in the direction
perpendicular to loading, l,p, consisted of No. 4 bars spaced at 9 in. top and bottom.
In
Specimens III and IV, flexural reinforcement perpendicular to loading was increased to match the
reinforcement in the direction of loading, No. 6 bars spaced at 4 in. To ensure development of the
longitudinal reinforcement, the bars were bent as shown in Figure 5.1(a). Approximately 2 in. of
clear cover were allowed on all sides of the footing, and between the CFT and longitudinal
reinforcement.
D = 20 in.
in.
point of loading
6 ft.
#6 @ 4 in (all spec.)
top and bottom
68 in.
approx. 2 in.
clear spacing
24 in.
#3 ties
(Spec. II, III & IV)
le
6
(a) cross-section
76 in.
(b) plan view
The vertical reinforcement in Specimens II, III and IV was No. 3 bars detailed with standard
seismic hooks, as shown in Figure 5.1(a).
represented the range of reinforcement found in construction in all seismic zones. In Specimen I,
shear reinforcement, v was not included, as is typical in most U.S. construction. The shear
reinforcement in Specimens II, III and IV represents the maximum level of reinforcement used
for bridge footings in high seismic zones.
46
5.2 Specimen Construction and Material Properties
Specimens I and II were constructed using the two-lift method described in Section 3.1. The
footing for Specimen III was constructed in a single lift. A modified construction procedure was
used for Specimen IV. The steel tube and annular ring were prepared and placed prior to
construction of the footing. Anchor bolts placed through the annular ring were intended only to
provide stability during construction. A 30 in. diameter corrugated steel pipe was placed around
the tube to create a void in the connection region, as shown in Figure 5.2(a), and the footing was
cast in a single lift. Finally, the tube was filled with concrete, and the void was filled with highstrength, fiber reinforced grout (Figure 5.2(b)). The measured compressive strength of the
reinforced grout was 8.75 ksi. The modifications were intended to simplify construction of the
connection by eliminating the need for multiple construction trades working simultaneously on
the project, while preserving the strength and stiffness benefits from embedment. In actual
practice, erection of the column could take place after construction of the reinforced concrete
footing is completed.
void created by
corrugate steel pipe
high-strength,
fiber-reinforced
grout
The design yield stress of the steel tubes was 70 ksi. The spirally welded tubes were formed
using a double submerged arc weld process. The weld designation for Specimens I and II was
LA70/860, and for Specimens III and IV was LA75/865.
High-strength, self-consolidating
concrete with 10 ksi compressive strength was specified to ease construction. Low-shrinkage
properties of the specified mix improved bond condition within the CFT. The measured material
properties and experimental parameters for each specimen are summarized in Table 5.1. Concrete
compressive strength was measured on the day of testing. The strength and ultimate strength of
the steel tube yield were measured in the longitudinal direction of the tube.
47
Table 5.1 Experimental variables and measured material properties
Specimen
le
(in.)
I
II
III
IV
12
12
18
12
le
D
0.6
0.6
0.9
0.6
l,p
Construction
Procedure
fc
(ksi)
Fy
(ksi)
Fu
(ksi)
0
0.3%
0.3%
0.3%
0.1%
0.1%
0.5%
0.5%
Embedded
Embedded
Embedded
Grouted
11.0
11.0
10.3
10.3
76.3
87.4
Specimens I and II were tested to assess the adequacy of shallowly embedded column bases
for low and moderate seismic zones, and to evaluate the influence of shear reinforcement on the
strength and behavior of the connection. The increased embedment in Specimen III was expected
to provide strength and stiffness necessary for column bases in high-seismic zones, with shear
reinforcement typical of bridge footings in such regions. In addition to evaluating the alternative
construction procedure, Specimen IV was selected to investigate methods for reducing footing
damage. The high-strength, fiber reinforced grout was expected to sustain local damage in the
connection region.
5.3 Test Setup
The CFT-footing connection specimens were tested using the self-reacting test rig shown in
Figure 5.3. The test rig was constructed below a 2400-kip Baldwin Universal Testing Machine
which applied axial load to the specimens. The rig consists of two L-shaped steel frames, posttensioned at the horizontal legs to a reinforced concrete anchor block. The legs were attached to
the laboratory floor to prevent rigid body movements in the rig. A steel beam spans the vertical
legs of the steel frames to support a 220-kip actuator used to apply horizontal load to the
specimens. The cantilever specimens were placed on top of the concrete block and anchored by
high-strength threaded rods. Simple support conditions at the top of the column were guaranteed
by a swivel on the horizontal actuator and a spherical bearing attached to the column. During
lateral loading, the top of the CFT column moved along a track attached to the load head of the
Baldwin test machine, preventing out-of-plane drifts. To minimize friction losses, a greased
Teflon-stainless steel sliding surface was placed in the track. In the presence of high axial
pressure, such contact surfaces have been shown to have a coefficient of friction approaching
zero (Campbell and Kong, 1988).
48
load head of Baldwin test machine
swivel
W1490
spherical bearing
W2494
HSS663/8
6 ft.
high-strength
anchor rods
concrete anchor
block
W2494
Figure 5.3 Experimental Test Rig
5.4 Loading
The Baldwin Test Machine applied a vertical load of 410 kips to each specimen
(approximately 10% of P0). The axial load was applied before horizontal loading began and was
held constant for the duration of each test. Cyclic horizontal loads were controlled by the internal
linear variable displacement transducer, LVDT, on the horizontal actuator.
The cyclic horizontal drift history, shown in Figure 5.4, was based on ATC-24 protocol (ATC
1992). The nominal yield displacement for the total connection, y, was estimated using the
nonlinear analyses described in Chapter 4:. First, the predicted column yield force was
determined, and the total specimen drift corresponding to 75% of that yield force was identified.
Then, as shown in Figure 5.5, the secant to the specimen drift-75% column yield force point was
calculated and extrapolated to the column yield force. This intersection point was taken as y, 0.9
in. (1.25% specimen drift). Elastic deflections in the test frame caused significant differences
between the applied drift history and the measured displacement at the top of the CFT-footing
specimen.
49
10
8
6
4
2
0
-2
-4
-6
-8
-10
0
10
12
14
16
18
20
22
24
26
Cycle Number
200
160
y
40
connection behavior
calculated secant
5.5 Instrumentation
Throughout testing, the overall load and displacement in the specimen, local column
displacement and rotation, and steel tube strain distributions were monitored. The instrumentation
is shown in Figure 5.6. Horizontal displacement of the specimen was measured at three locations
on the column using horizontal potentiometers attached to an independent reference column (pot.
1, 4 and 11).
Prior to casting, horizontal threaded rods were attached to the steel tube 1 in., 7 in. and 19 in.
above the footing as shown in Figure 5.6(a). Vertical potentiometers (pot. 710) extended
between adjacent rods on both sides of the column to measure segment rotation. Similarly,
50
vertical potentiometers (pot. 5 and 6) were placed between the top horizontal rods and the
footing to measure column base rotation relative to the footing. The mounting for these
instruments is shown in Figure 5.7.
Uniaxial high-elongation strain gauges were placed on the tube both along the height of the
column and into the footing as shown in Figure 5.6(b) to measure longitudinal tube strains during
testing. Due to longer embedment length, only Specimen III had strain gages placed 16 in. below
the footing surface. At the base of the column, 3 in. above the footing surface, four additional
strain gages spaced equally around the column circumference measured hoop stress in the tube.
pot. 1
52 in.
+30 in.
pot. 4
pot. 5
pot. 7
12 in.
pot. 8
pot. 9
6 in.
pot. 10
pot. 6
+21 in.
19 in.
+12 in.
1 in.
+3 in.
-2 in.
-4 in.
-10 in.
-16 in.
pot. 11
(a) external instrumentation
pot. 8
pot. 6
pot. 10
51
Additional instrumentation, shown in Figure 5.8, was used to monitor displacement of the
specimen and anchor block. Horizontal potentiometers measured specimen and anchor block slip.
Rotation at the base of the specimen and anchor were measured by pairs of vertical
potentiometers. These measurements were used to calculate corrected horizontal drifts of the
column.
point of loading
horizontal pots:
rig & spec. slip
anchor block
Figure 5.8 Instrumentation of CFT-footing test apparatus
52
For each specimen, the sequence of damage that occurred in the foundation and column, as
observed during testing, is described at low drift ratios (02%), at moderate to high drifts (2%
6%) and at the final state (after 6%). The drift ratio is defined as displacement at the point of
loading divided by column length, /Lcol. These observations are followed by a comparison of the
behavior of all specimens to evaluate the relative performance, and all of the cyclic load-drift
curves are presented together for comparison.
6.1.1 Specimen I
At approximately 0.4% drift, a crack formed around the column-footing interface. At 0.7%
drift compression strain exceeding the yield strain was measured in the tube 6 in., 0.3D, below the
footing surface. At 0.7% drift bisecting cracks formed in the footing surface, extending from the
column both parallel and perpendicular to the direction of loading. Figure 6.1(a) shows the
interface crack and bisecting cracks at 0.75% drift. The bisecting cracks propagated through the
footing depth, as shown in Figure 6.1(b), and the crack widths increased with subsequent cycles
at 0.7% and 1% drift. At 1.6% drift cracks formed in the footing surface extending diagonally
from the column. At the peaks in drift, a gap opened on the tension side of the column-footing
interface.
53
interface crack
bisecting crack
At each cycle of drift at 2% and larger, damage in the footing increased. The bisecting and
diagonal cracks in the footing continued to widen, and these existing cracks branched and new
cracks formed in various directions across the footing surface. Damage in the footing at drifts
greater than 2% is shown in Figure 6.4. The maximum horizontal load, 130.5 kips, was reached at
2.4% drift, and the lateral resistance of the connection degraded with each increasing drift level.
At this drift, cracks oriented diagonally through the footing depth (Figure 6.4(e)) formed on all
four sides of the footing. At 3.6% drift, the interface gap on the tension side was approximately
0.6 in. (Figure 6.4(a)) and the concrete at the compression side of the column-footing interface
had crushed and begun to break away (Figure 6.4(b)). This localized damage at the interface
allowed rigid body rotations of the column within the footing. The bisecting cracks in the
direction of loading were approximately 0.2 in. wide and uplift in the surface of the footing was
approximately 0.375 in. (Figure 6.4(c)). Pull-out failure was defined as 0.5 in. or more of uplift in
the footing surface due to cracks forming diagonally through the footing depth. This state was
reached during cycles between 3.6% and 5% drift.
54
Loading was terminated at 8% drift, and the damage in the specimen at this drift is shown in
Figure 6.4(e). Uplift in the surface of the footing exceeded 0.9 in. and large pieces of the footing
had broken away from the footing surface and along the cracks oriented diagonally through the
footing depth. Tension yielding of the tube above the footing was not detected during the test.
The steel tube was removed from the footing after testing, and showed that the tube base and
annular ring experienced permanent deformation and warping, as shown in figure xxx. Observed
damage states are identified in the cyclic load-drift plot in Figure 6.5.
55
150
pull-out
100
50
0
-50
-100
-150
-200
-10% -8% -6% -4% -2%
0%
2%
4%
6%
8% 10%
56
6.1.2 Specimen II
The design of Specimen II was identical to that of Specimen I, except for the presence of
vertical reinforcement. Prior to testing, hairline cracks in the footing surface were noted at the
column-footing interface and radiating from the column in the four bisecting directions.
Variations in ambient temperature may have caused shrinkage and temperature cracking to occur
in Specimen II which was not observed in other specimens.
The sequence of damage and subsequent strength deterioration observed in Specimen II was
similar to that in Specimen I. At 0.5% drift an interface crack formed. At 0.7% drift horizontal
bisecting cracks formed on the footing surface and into the footing depth parallel to the direction
of loading (Figure 6.6(a)). Bisecting cracks formed perpendicular to the direction of loading at
0.9% drift. Diagonal cracking initiated at 1.4% drift. Diagonal and bisecting cracks at this drift
are shown in Figure 6.6(b).
At each drift level 2% and larger, the number of cracks and the width of existing cracks in the
footing increased. At 2% drift the bisecting crack parallel to loading was approximately 0.12 in.
wide and the column-footing interface gap was approximately 0.2 in. The peak horizontal load,
134.5 kips, was reached at 3% drift and corresponded to measurement of tension yield strain in
the tube just above the footing. The cracking damage in the footing at 3% drift is shown in Figure
6.9(a). Thereafter, the horizontal strength deteriorated, as shown in the cyclic load-drift response
57
in Figure 6.10. At 4.2% drift, the major bisecting cracks had widened to 0.2 in. and uplift at the
footing surface was 0.25 in.
At 7% drift, the concrete near the column-footing interface had crushed and began to break
away, as shown in Figure 6.8(a). At this drift, uplift at the footing surface was 0.6 in, which was
considered pull-out failure. Footing uplift increased with drift exceeding 7%, and is shown in
Figure 6.8(b) at the final drift level, 9.5%. Visual inspection of the specimen after testing found
that the tube had buckled within the embedded region of the column, just above the annular ring,
as shown in Figure 6.9.
58
tube yield
150
pull-out
100
50
0
-50
-100
-150
-200
-10% -8% -6% -4% -2%
0%
2%
4%
6%
8% 10%
Specimen III had a deeper embedment length than other specimens and had shear
reinforcement. At small deformations the sequence of damage in the footing in Specimen III was
similar to that of the first two specimens; however, at drift ratios larger than 2%, the response of
Specimen III was controlled by the behavior of the CFT column. At 0.5% drift, a crack formed at
the column-footing interface. At 0.9% drift bisecting perpendicular to the direction of loading
formed at the footing surface. At 1.2% drift tension and compression strains exceeding the yield
strain were measured in the tube just above the footing. The horizontal load was 134.3 kips. At
this drift level, the perpendicular bisecting cracks had propagated into the footing depth.
59
Additional perpendicular cracks formed at the footing surface at 1.8% drift. Cracks in the
footing at this drift are shown in Figure 6.11.
Figure 6.11 Specimen III damage at low drift ratios (1.8% drift)
(cracks highlighted)
The maximum horizontal load, 165.1 kips, was reached at 2.4% drift. Bisecting cracks
parallel to the direction of loading formed in the footing surface and through the footing depth.
At 4.3% drift local buckling of the tube was visible, as shown in Figure 6.12(a), and corresponded
to a slight decrease in horizontal strength, as shown by the cyclic load-drift response in Figure
6.14. Figure 6.12(b) shows that the footing had sustained limited cracking and damage at this
drift.
Figure 6.12 Specimen III damage at moderate and high drift ratios
60
At 6% drift, local buckling led to initiation of ductile tearing at the highly strained buckle.
The tearing initiated at holes drilled in the tube for instrumentation, as shown in Figure 6.13(a).
Crushed concrete spilled from inside the tube at the location of the tear. At 7.2% drift, tearing had
spread around nearly the entire circumference of the steel tube at the local buckled region (Figure
6.13(b)), and led to termination of the test. The observed damage states are identified in Figure
6.14.
150
tube buckling
tube yield
100
50
0
-50
-100
-150
-200
-10% -8% -6% -4% -2%
0%
2%
4%
6%
8%
10%
61
6.1.4 Specimen IV
Although the connection detail for Specimen IV differed from the other specimens, the
sequence of behaviors was similar. Damage in the concrete footing was preceded by similar
behavior in the grouted region of the connection. At 0.5% drift a crack formed at the columngrout interface. At 0.7% drift, cracks initiated in the grout in the direction of loading, and were
followed by bisecting cracks both parallel and perpendicular to the direction of loading in the
surface of the concrete footing at 1% drift. At increasing drifts, these bisecting cracks widened
and propagated along the footing surface and into the footing depth. Diagonal cracks formed in
the grout at 1.4% drift and in the footing at 1.9% drift. The damage to the grout and footing
surface at 1.9% drift is shown in Figure 6.15.
The peak horizontal load, 138.8 kips, occurred at 2.8% drift and corresponds to measurement
of tension yield strain in the tube just above the footing. Separation between the grouted
connection and the surrounding concrete was also noted at this drift level, as shown in Figure
6.16(a). An uplift of approximately 0.2 in. of the grout relative to the surface of the footing was
measured. With increased drift cycles the horizontal strength deteriorated, as shown in the cyclic
load-drift response in Figure 6.18. The number and width of cracks in the grout and in the footing
surface increased, and damage at the grout-footing interface increased (Figure 6.16(b)). The
width of cracks through the footing depth remained small, as shown in Figure 6.16(c).
62
At 6.6% drift the surface of the footing exhibited uplift similar to that observed in Specimens
I and II, as shown in Figure 6.17(a). A 0.2-in. horizontal gap and 0.3-in. of uplift were measured
between the grout and the footing surface. Testing was terminated at 7.8% drift, with significant
cracking and uplift in the footing.
63
grout uplift
tube yield
150
100
50
0
-50
-100
-150
-200
-10% -8% -6% -4% -2%
0%
2%
4%
6%
8% 10%
Evaluation of the impact of vertical reinforcement was made by comparing the load-drift
response and observed behavior in Specimens I and II, which were similar. The cyclic load-drift
response for Specimen I is plotted in Figure 6.19(a). This figure shows that the response was
highly asymmetric, the peak positive load was approximately 30% greater than the peak negative
load. The horizontal load capacity decreased significantly after the initial cycle at a given drift
level and the lateral resistance degraded immediately after the peak strength was reached. The
cyclic response is notably pinched, indicating large amounts of rigid body deformation in the
specimen and decreased energy dissipation. The hysteretic response of Specimen II was similar to
64
that of Specimen I, with similar quantities of horizontal load resistance, deformation capacity,
ductility and energy dissipation capacity. Figure 6.19(b) indicates that the response of Specimen
II was much more symmetric than Specimen I, the difference in the positive and negative peak
loads was approximately 7%.
Damage states in the footing were the same in both Specimens I and II, and occurred at
approximately equal loads and drift levels. However, the presence of shear reinforcement in
Specimen II appeared to reduce the severity of the footing damage compared to Specimen I. At
drifts greater than 4%, the crack widths and uplift in the footing surface measured in Specimen II
were smaller than in Specimen I. Large pieces of concrete broke away from the footing in
Specimen I, which was not observed in Specimen II. The presence of vertical reinforcement had
little effect on the load capacity of the connection. The compressive strength of the concrete
forming a compression strut along the cracks oriented diagonally through the footing depth may
control the strength of these specimens with shallow embedment.
The effect of embedment length was assessed by comparing the response and behavior of
Specimens II and III. The cyclic load-drift response for Specimen III is shown in Figure 6.19(c).
The specimen had higher load capacity (approximately 25%) and improved drift capacity relative
to Specimen II. Specimen III experienced less loss of strength during load cycles at the same drift
level. The hysteretic loops are full, with very little pinching, indicating improved energy
dissipation capacity. The response indicates the specimen had large ductility and experienced
very little loss of lateral strength after the tube yielded, and even after the tube buckled.
The increased embedment depth in Specimen III improved the observed behavior compared
to Specimen II. The maximum horizontal load capacity in Specimen III was controlled by local
buckling, and ultimately ductile tearing, in the steel tube rather than damage in the footing.
Specimen III sustained limited damage to the footing; the number and width of cracks was
significantly reduced relative to Specimen II, and no uplift in the footing surface was observed.
The effects of the grouted construction procedure were evaluated by comparing the response
and behavior of Specimen IV to Specimen II. The embedment and shear reinforcement in
Specimen IV was identical to that of Specimen II; however, Specimen IV did not exhibit a
significant improvement in horizontal load capacity. Specimen IV did have improved
deformation capacity and reduced strength degradation, as shown in the cyclic load-drift plot in
Figure 6.19(d). Compared to Specimen II, damage in the concrete footing in Specimen IV was
delayed and less severe. Pull-out occurred at a higher drift level and well after tension yielding
was measured in the tube.
65
200
100
150
Hmax = 130.5 k
2.4% drift
150
200
50
0
-50
-100
-150
0
-50
-100
2%
4%
6%
8% 10%
(b) Specimen II
-200
-10% -8% -6% -4% -2% 0% 2%
Hmax = 165.1 k
2.5% drift
200
tube
buckling
tube
yielding
150
100
50
200
150
0%
100
-150
(a) Specimen I
-200
-10% -8% -6% -4% -2%
50
0
-50
-100
-150
0%
2%
4%
100
4%
6%
8% 10%
Hmax = 138.8 k
2.8% drift
50
0
-50
-100
-150
-200
-10% -8% -6% -4% -2%
tube yielding
Hmax = 134.5 k
2.9% drift
6%
8% 10%
(d) Specimen IV
-200
-10% -8% -6% -4% -2% 0% 2%
4%
6%
8% 10%
66
6.2 Measurement Corrections and Adjustments
In order to determine the response and evaluate the performance of the test specimens,
corrections to and assessment of the measured data were made. In the following section, a
correction method is presented which removes deflection of the test rig from the measured
specimen drift. The measured effective horizontal load acting on the specimen due to axial load
effects is calculated. In addition, measurements of tube segment elongation and rotation of the
column are assessed for accuracy.
Horizontal potentiometers measured column drift at three locations along the column height
(Figure 5.6(a)) throughout each test. These measurements were made relative to an independent
reference column attached to the laboratory floor, and therefore included rotation and rigid body
slip of the specimens and test rig. To obtain the specimen drifts, these specimen movements and
frame deformations must be removed.
Pairs of potentiometers placed at the base of the specimen were used to calculate the rigid
body rotation of the test specimens, spec,uplift, and the contribution it made to column drift,
spec,uplift, as illustrated in Figure 6.20. Equations 6.1 and 6.2 provide expressions to calculate the
rotation and deformation due to uplift at the base of the specimen.
spec ,uplift =
(6.1)
(6.2)
Where north,spec and south,spec are the potentiometer readings on the North and South side of the
specimen, respectively, bfooting is the distance between the potentiometers, and lspec is the distance
to the column drift measurement. Positive rotation and drift correlate to positive (Southward)
horizontal load. Rigid body rotation at the base of the anchor block, rig,uplift, and the subsequent
contribution to column drift, rig,uplift, are calculated in the same manner from potentiometers
readings at the base of the anchor block, north,rig and south,rig, respectively. The total corrected
specimen drift, , is calculated in Equation 6.3.
= raw spec ,uplift rig ,uplift spec , slip rig , slip
(6.3)
Where raw is the raw measured drift, and spec,slip and rig,slip are the horizontal slips measured in
the specimen and rig, respectively.
67
spec,uplift
spec,slip
load
load
spec
lspec
spec-slip
north
anchor block
south
anchor block
bfooting
(a) specimen uplift
The flexural response of the specimens was affected by the applied horizontal load as well as
the eccentricity in the applied load that occurred due to the column drift. To evaluate the load
drift response of the specimens including the effects of axial load eccentricity, the total effective
horizontal force, Heff, was calculated as the sum of the applied actuator force, H, and the P-
effects in Equation .
H eff = H +
P
Lcol
6.4
Figure xxx shows the cyclic effective load-drift response for each specimen. At low drifts the
measured and effective horizontal loads were similar but the difference in the effective and
measured load increased with drift. The effective load, Heff, may be a better measure of flexural
demands; however, the horizontal load, H, used in the remaining calculation in this document is
the measured load.
68
100
Heff,max = 139.6 k
2.4% drift
50
0
-50
-100
-150
(a) Specimen I
-200
-10% -8% -6% -4% -2%
150
100
0%
2%
4%
6%
8% 10%
150
50
0
-50
-100
-150
(b) Specimen II
-200
-10% -8% -6% -4% -2% 0% 2% 4%
200
Hmax = 173.1 k
2.5% drift
50
0
-50
-100
-150
-200
-10% -8% -6% -4% -2%
0%
2%
4%
Hmax = 145.7 k
2.9% drift
100
200
150
200
200
6%
8% 10%
150
100
6%
8% 10%
50
0
-50
-100
-150
-200
-10% -8% -6% -4% -2%
(d) Specimen IV
0%
2%
4%
6%
8% 10%
69
6.2.3 Assessment of Vertical Potentiometer Measurements
Pairs of potentiometers were mounted vertically to the CFT column specimens to measure
segmental elongations at the tube base and to calculate column flexural rotations, as shown in
Figure 5.6(a) and in Figure 6.22. Problems with the measurements made by these instruments
were noted during and after testing. During testing, column rotation caused the mounting of the
lowest potentiometers (closest to the footing) to collide with the surface of the concrete, which
moved the instruments out of position. After testing, measurements from each of the
potentiometers were assessed. The vertical potentiometer measurements of tube elongation,
lmeas, were compared to the theoretical upper bound on elongation, ltheory, and to elongations
calculated from measurements of longitudinal strain in the tube, lstrain.
12 in.
pot. 7
pot. 8
6 in.
pot. 9
pot. 10
specimen foundation
Figure 6.22 Placement of potentiometers measuring segmental tube elongation
The theoretical elongation between two points, x1 and x2, is derived from the equation for
deflection of a cantilever, (x), in Equations 6.56.9 (Gere, 2001) and is expressed in terms of the
experimental column drift at the point of loading, , in Equation 6.9.
( x) =
HL3col 3 x 2
x3
2 + 3
3EI 2 Lcol 2 Lcol
(6.5)
HL3col
=
3EI
( x) =
d
dx
(6.6)
( x)
(6.7)
ltheory = ( ( x2 ) ( x1 ) ) r
ltheory =
3
2 L3col
(( x L
2
col
(6.8)
(6.9)
70
Where x is measured from the footing surface, H is the applied load, Lcol is the column length,
EI is the elastic stiffness of the column, (x) is the column rotation at x, and r is the radius of
curvature, taken as D/2.
As a second measure of instrument accuracy, lmeas was compared to elongation calculated
from tube strains, meas, measured at approximately the middle of the potentiometer gage length,
lgage. Elongation is calculated as a function of strain in Equation 6.10.
x2
(6.10)
Figure 6.23 compares the measured elongation history of one potentiometer with the
corresponding calculated elongations. During early cycles of loading, the measured and
calculated elongations show good correlation. However, after only seven cycles of loading, the
potentiometer reading jumps, which implies a large permanent increase in tube elongation
occurred that is not supported by either of the calculated values.
1.5%
1.0%
0.5%
0.0%
-0.5%
-1.0%
lmeas
ltheory
lstrain
-1.5%
The tube elongation comparison was performed for each of the four potentiometers shown in
Figure 6.22, for each test specimen. Discrepancies between the measured and calculated
elongation were noted in at least one potentiometer in each test. In the upper potentiometers this
problem may be caused by slip or loosening of the instrument mounting. Due to these suspected
errors, the vertical potentiometer readings were not used in the data analysis, and an alternative
method was used to calculate the flexural response of the columns.
71
6.3 Measured Response
Displacement readings and strain data collected during testing provide information about the
specimen response that can not be gained from visual observations. The following sections
present the tube strain and column curvature distributions over a portion of the column length, the
flexural response of the CFT column and the specimen, the contribution of the footing and
connection to specimen response, and the energy dissipated during testing.
72
36
South
36
North
30
South
24
18
Spec I
Spec II
Spec III
Spec IV
12
6
0
-0.0005
0.0005
24
-0.001
0.001
18
12
6
0
-0.0015
-0.001
-0.0005
-6
-12
0.25% drift
-18
36
North
30
24
18
12
6
0
0
North
30
South
24
18
12
6
0
-0.003
-0.002
-6
-0.001
0.001
-6
-12
1% drift
-18
-12
-18
Tube Strain
Tube Strain
36
North
30
South
24
18
12
6
0
0.001
0.002
0.003
0.004
-6
1.5% drift
0.0015
Tube Strain
36
0.75% drift
0.001
0.5% drift
-18
Tube Strain
0.0005
-6
-12
South
North
30
-12
-18
Tube Strain
0.002
0.003
73
36
North
36
North
South
30
24
18
Spec I
Spec II
Spec III
Spec IV
12
6
0
-0.0005
0.0005
24
-0.001
0.001
18
12
6
0
-0.0015
-0.001
-0.0005
-6
-6
-12
-12
-0.25% drift
-18
30
South
South
24
12
6
0
0
0.001
0.002
0.003
-0.5% drift
30
North
18
-0.001
-6
18
12
6
0
-0.003
-0.002
-0.001
0.001
-18
-18
Tube Strain
36
30
South
24
0.003
-12
-0.75% drift
Tube Strain
18
12
6
0
-0.004 -0.003 -0.002 -0.001
0.002
-6
-12
North
0.0015
36
24
-0.002
0.001
Tube Strain
36
-0.003
0.0005
-18
Tube Strain
North
South
30
0.001
0.002
0.003
0.004
-6
-12
-18
-1.5% drift
Tube Strain
-1% drift
74
Table 6.1 Maximum drift level recorded by embedded strain gages
Gage
South -2
North -2
South -6
North -6
South -10
North -10
South -16
North -16
I
1%
1.6%
1%
1%
1%
1.6%
Specimen
II
III
1.6%
2.9%
1.6%
1.4%
1.6%
1.4%
2.5%
NF
0.7%
1.6%
1.6%
1.6%
IV
4.2%
4.2%
4.2%
3%
3%
Peak tension strains in Specimen IV occurred above the footing surface and strains below the
footing were much lower than other specimens. Typically, the maximum compression strain
measured in each specimen occurred above the surface of the footing. To approximately 1% drift,
the compression strain distributions in Specimens I, II and IV were similar above and below the
footing surface. At 1.5% drift, Specimen III sustained larger compression and tension strains
above the footing surface, and therefore developed larger moments in the column. On the tension
face of the column, the maximum strain in Specimens I, II and III was measured below the
footing surface.
It is postulated that column forces are resisted by the footing through two mechanisms shown
in Figure 6.26: (a) bearing of the concrete foundation forms a moment couple and resists
horizontal loads, and (b) anchorage from the annular ring and shear stress transfer. Either
mechanism can support the decrease in measured longitudinal tube strains over the embedded
length.
P
H
acting forces
(a) bearing
75
6.3.2 Column Curvature Distribution
The measured tube strains were used to calculate curvature distributions in the column above
the footing surface. Column curvature at location i, i, was calculated from longitudinal strains
using Equation 6.11.
i =
north ,i south ,i
D
(6.11)
Where north,i and south,i are strains measured on the North and South faces of the tube,
respectively. The curvature distribution over the lower third of the column height at drift levels
from 0.5% to 4% is plotted for each specimen in Figure 6.27 and Figure 6.28. Curvatures 3 in.
above the footing surface in Specimen III are not plotted after 2% drift due to failure of the strain
gages at this location.
76
36
Spec I
Spec II
Spec III
Spec IV
30
0.5% drift
36
24
18
12
1% drift
30
24
18
12
0
0
0.00003
0.00006
0.00009
0.00012
0.00005
0.00015
0.0002
0.00025
36
1.5% drift
30
36
24
18
12
2% drift
30
24
18
12
0
0
0.0001
0.0002
0.0003
0.0004
0.0001
0.0002
0.0003
0.0004
0.0005
0.0006
0.0007
36
36
3% drift
30
0.0001
24
18
12
30
4% drift
24
18
12
0
0
0.0001
0.0002
0.0003
0.0004
0.0005
0.0006
0.0007
0.0001
0.0002
0.0003
0.0004
0.0005
0.0006
0.0007
77
30
24
18
12
-1% drift
30
24
18
12
0
-0.00009
-0.00006
-0.00003
-0.00025
-0.0002
-0.00015
-0.0001
-0.00005
-1.5% drift
30
24
18
12
36
36
-2% drift
30
24
18
12
0
-0.0003
-0.0002
-0.0001
-3% drift
30
24
18
12
36
-0.0004
36
-4% drift
30
24
18
12
0
-0.0007 -0.0006 -0.0005 -0.0004 -0.0003 -0.0002 -0.0001
-0.00012
-0.5% drift
36
36
Spec I
Spec II
Spec III
Spec IV
0
-0.0007 -0.0006 -0.0005 -0.0004 -0.0003 -0.0002 -0.0001
At low drifts, the curvature distribution was similar in all specimens. However, with increases
in drift demand, the curvature, and therefore the moment developed, at the tube base in Specimen
III became significantly larger than the other specimens. The curvature distribution in this
specimen increased most just above the footing surface and became nonlinear at 1.5% drift,
78
shortly after tube yielding occurred. Column curvature in Specimens I and II remained
approximately unchanged after 2% drift, although the columns Specimens III and IV continued to
sustain increasing demands.
load
lel
4
3
2 x
1
lgage,4
lgage,3
lgage,2
lgage,1
Lcol Lcol
col =
( x)dxdx
0
(6.12)
(6.13)
The gage length for a pair of strain gages, lgage,i, was defined as the average distance between
adjacent gages, and xi is the distance from the footing surface to the location of strain gage pair i.
79
Strain measurements 30 in. above the footing indicated that the tubes remained elastic at
this location and above. For this top portion of the column, the component of column drift, col,el,
was determined using the equation for elastic deflection of a cantilever:
col ,el
Hlel3
=
3EI eff
(6.14)
where H is the horizontal load, lel is the length of the elastic portion of the column, and EIeff is
calculated as defined in Equation 2.2. Finally, flexural column drift was calculated as:
(6.15)
Figure 6.30 plots the positive moment-column drift ratio envelopes for each test specimen.
Experimental moment at the column base, Mexp, includes P- effects and is defined as:
M exp = HLcol + P
(6.16)
where P is the measured axial load. The moment-drift ratio responses are very similar, indicating
that differences in total specimen behavior can be largely attributed to the connection response.
The moment-drift ratio envelopes for these columns exhibit a short elastic range initially and
gradual softening. Tube yielding, in Specimens II, III and IV, occurred at approximately the same
moment and column drift ratio, 0.65%. Specimens I and II were slightly stiffer than Specimens III
and IV, which may be due to small differences in the concrete compressive strength at testing.
14000
12000
10000
tube yielding
8000
6000
Spec I
Spec II
Spec III
Spec IV
4000
2000
0
The positive moment-drift ratio envelopes of total specimen response are plotted in Figure
6.31. The initial moment-drift ratio responses are very similar to approximately 1% drift,
80
indicating that the variations in shear reinforcement and embedment length had little effect on
the initial stiffness of the specimen. The detail of the response to 1% drift shows that the grouted
connection in Specimen IV led to slightly decreased initial stiffness. The increased embedment in
Specimen III increased the specimen stiffness after 1% drift, while the other specimens exhibited
softening in this range.
Although the tube yield moment was approximately the same in all specimens that exhibited
yielding, this occurred at a much lower drift in Specimen III than other specimens. The large
yield drift in Specimens II and IV can be attributed to large deflections caused by damage in the
concrete footing. Both Specimens III and IV sustained large inelastic drifts without significant
reductions in resistance, while Specimens I and II lost strength rapidly after the maximum
moment was reached. Specimen I had no vertical reinforcement in the footing, and the rate of
strength deterioration is the largest in this specimen.
14000
10000
12000
8000
10000
6000
8000
tube yielding
6000
4000
Spec I
Spec II
Spec III
Spec IV
4000
2000
2000
0
0%
2%
4%
6%
8%
10%
0
0.0%
0.5%
1.0%
The moment-drift envelopes of the experimental column and specimen response indicate
significant softening and degradation of moment resistance occurred in all of the test specimens.
These changes are due to damage occurring the concrete footing and in the CFT column. To
quantify this change in resistance, the flexural stiffness values of specimens, EIspecimen, were
calculated from the measured horizontal load and drift using the expression in Equation 6.17. The
flexural stiffness of columns, EIcolumn, were calculated by substituting col for in Equation 6.17.
The degradation in the stiffness values with increasing drift demand are shown in Figure 6.32. On
the vertical axes, the calculated stiffness values are normalized by EIeff, defined in Section 2.1.1.
81
EI specimen =
HL3col
3
(6.17)
1.4
Specimen I
Specimen II
Specimen III
Specimen IV
EIcolumn / EIeff
1.2
1
0.8
0.6
0.4
0.2
0
0%
2%
4%
6%
8%
10%
EIspecimen / EIeff
0.8
0.6
0.4
0.2
0
0%
2%
4%
6%
8%
10%
The change in total specimen stiffness was similar among all specimens, indicating that the
experimental variables had little effect on the flexural specimen stiffness. Specimens III and IV
exhibited significant reduction in the column stiffness values over increasing drift, due to the
development of larger moments and inelastic deformations in those columns.
82
6.3.4 Connection Contribution to Specimen Drift
The total CFT-footing specimen response has two components, the flexural behavior of the
CFT column, and the degradation of the connection and footing. This relationship is expressed in
terms of specimen drift in Equation 6.18.
= col + cnxn
(6.18)
Where col is as defined in Equation 6.15, and cnxn is the component of total drift occurring due
to deformation and flexibility in the footing and connection. For a perfectly fixed base, tot and
col components would be equal. The difference between specimen and column stiffness values
implies that a perfectly fixed column base was not achieved in the four test specimens; this is
supported by the ratios of total drift due to connection flexibility plotted in Figure 6.33. Initially,
approximately 50% of the specimen drift in each of the specimens occurs in the connection. With
increased drift demand and subsequent damage to the connection and footing, the connection
contribution to total drift in Specimens I, II and IV increases. The connection drift in Specimen
III does not exceed 50%, and begins to decrease at 2% drift, shortly after initial tube yielding.
Specimen III response is terminated 4% drift due to local buckling of the steel tube. Although the
increased embedment provided sufficient strength to prevent excessive connection degradation,
the initial response was not affected.
100%
80%
60%
40%
Spec I
Spec II
Spec III
Spec IV
20%
0%
0%
2%
4%
6%
8%
10%
83
enclosed by the cyclic load-drift response (Figure 6.19).The cumulative energy dissipated by
the specimen, ESPEC, at time i was calculated using the linear time-stepping method in Equation
6.19. The cumulative energy dissipation, normalized by the product of the peak load and peak
drift, Hmaxmax, in each specimen is plotted in Figure 6.34.
ESPEC ( i ) =
1 i
( H ( j 1 ) + H ( j ) )( ( j ) ( j 1 ) )
2 j =1
(6.19)
7
Spec I
Spec II
Spec III
Spec IV
6
5
4
3
2
1
0
0
0.2
0.4
0.6
0.8
Figure 6.34 shows that the incremental and total energy dissipated by Specimens I, II and IV
was similar over the majority of the test time; although, the incremental energy dissipation in
Specimen IV increased slightly in the last few cycles of testing. The normalized total energy
dissipated by Specimen III was approximately 50% larger than the other specimens. During initial
loading cycles, incremental energy dissipated by Specimen III was similar to the other specimens.
Large inelastic deformations occurred in the column due to local buckling of the steel tube and
led to fuller hysteretic curves (Figure 6.19) and significant increase in the incremental energy
dissipated by Specimen III in the late load cycles.
To assess the energy dissipation of the CFT-footing components, the energy dissipated by
flexure in the column, ECOL, and by the connection, ECNXN, was calculated by substituting in
Equation 6.19 with col and cnxn, respectively. In Figure 6.34, the energy dissipated by each
specimen is presented as the sum of the energies dissipated by the column and by the connection.
84
tube yield
(a) Specimen I
tube yield
(b) Specimen II
tube yield
(d) Specimen IV
The components of energy dissipation plotted in Figure 6.35 indicate that in Specimens I,
II and IV the connection and concrete footing was the primary source of energy dissipation. The
column component of energy dissipated by Specimen IV increases slightly with loading cycles.
Specimen III developed more significant moment in the column and exhibited larger column
drifts than other specimens, and so the component of specimen energy dissipation occurring in
the column was much larger. This comparison is terminated at 0.45% of total loading time in
Specimen III due to failure of the strain gages at the location of the local buckle on the tube.
85
The experimental results were compared to current AISC design expressions for concretefilled tube construction, which were described in Section 2.1.1. The material strengths of the
experimental specimens met the AISC provisions for maximum nominal steel yield stress, Fy, and
concrete compressive strength, fc. However, the diameter to thickness, D/t, ratio of the steel tubes
was 80, which exceeded the AISC limit of 62, as calculated by Equation 2.1.
The maximum experimental column base moment for each specimen was compared to the
nominal CFT moment capacity calculated using the plastic stress distribution method, Mn(p.s.), and
the strain compatibility method, Mn(s.c.) (Section 2.1.1). The nominal strengths were calculated
using measured material strengths for each specimen (Table 5.1). The results of this comparison
are summarized in Table 7.1. All of the specimens were able to resist loads greater than the
nominal strength calculated using the strain compatibility method; however, this method was
shown to consistently underpredict experimental CFT moment strengths in Section 2.1.2. Only
Specimen III developed strength adequate to exceed the nominal capacity calculated using the
plastic strain distribution method. However, only Specimen III was controlled by CFT column
behavior, the strength of Specimens I, II and IV were dominated by the reinforced concrete
footing.
Table 7.1 Comparison of experimental connection strength to design column strength
Mn(p.s.) (k-in)
Mn(s.c.) (k-in)
Mexp. (k-in)
Mexp./Mn(p.s.)
Mexp./ Mn(s.c.)
11917
9978
Specimen I Specimen II
10048
10511
0.84
0.88
1.01
1.05
11802
9882
Specimen III Specimen IV
12844
11230
1.09
0.95
1.30
1.14
86
Flexural stiffness values, EIi,col, EIi,spec, and EIsec were calculated to approximate the
experimental column and specimen response. The initial column stiffness, EIi,col, and the initial
specimen stiffness, EIi,spec, are linear approximations to experimental data from all four
specimens. The secant column stiffness, EIsec, is the slope of the line extending from the origin to
the average experimental moment and column drift ratio at which tube yielding occurred. The
calculated stiffness values are compared to the experimental response in Figure 7.1.
1.2
Mexp / Mn(p.s.)
1
tube yielding
0.8
0.6
EIsec = 0.74EIeff
0.4
0.2
Spec I
Spec II
Spec III
Spec IV
EIi = 0.99EIeff
0
0.00%
0.25%
0.50%
0.75%
1.00%
1.25%
1.50%
Mexp / Mn(p.s.)
1
0.8
0.6
Spec I
Spec II
Spec III
Spec IV
0.4
EIi = 0.488EIeff
0.2
0
0%
2%
4%
6%
8%
10%
The EIsec value is a reasonable approximation of column response prior to yielding. The EIsec
value provides an estimate of the yield point but may underestimate the experimental moment by
as much as 25% for a give column drift ratio.
87
The initial column and specimen stiffness for each specimen are compared to the AISC
design stiffness, EIeff, in Table 7.2. This comparison shows that the AISC equation for design
stiffness closely predicted the experimental column stiffness; however, the CFT-footing
specimens achieved approximately 50% of the column stiffness value because the full base fixity
was not achieved in any of the specimens. Despite having adequate strength to develop the full
column capacity, the increased embedment of Specimen III does not show improved initial
stiffness or fixity compared to the other specimens. This is consistent with similarities in the
experimental response among the specimens for elastic load levels. The initial stiffness of
Specimen IV is slightly lower than that measured for the other specimens, which may be caused
by the presence of grout in the connection region, which had a lower compressive strength than
the concrete used for all of the footings.
Table 7.2 Comparison of experimental stiffness to design stiffness
EIeff (k-in2)
EIi,spec (k-in2)
EIi,spec./ EIeff
EIi,col (k-in2)
EIi,col./ EIeff
51618312
Specimen I Specimen II
24110928
21835968
0.47
0.42
54,870,432 51,746,112
1.06
1.00
50671368
Specimen III Specimen IV
21099744
18675120
0.42
0.37
48,761,112
48,783,960
0.96
0.96
In the following sections, the experimental response of the specimens is interpreted in terms
of theoretical expressions and predictions. The effective column length and footing shear stress
demand are calculated for each specimen. Moment-curvature analysis can be used as an effective
design tool, and analytical predictions are compared to the moment-curvature response of the
experimental columns. Finally, the nonlinear finite element model described in Chapter 4: is
verified against the experimental results.
88
1
3EI eff 3
Leff =
(7.1)
The effective length of each specimen is plotted over increasing drift levels in Figure 7.2.
Initially, the Leff of all specimens was similar and increased approximately linearly with drift.
Specimens I and II have larger Leff compared to the other specimens, due to damage to the
concrete footing and the subsequent reduction in lateral resistance after 3% drift.
4.5
1.75
Spec I
Spec II
Spec III
Spec IV
4
3.5
1.5
3
2.5
1.25
2
1.5
1
0.0% 0.5% 1.0% 1.5% 2.0%
1
0%
2%
4%
6%
8%
10%
89
D
c
t
Ts
le
bf
When the tube is fully yielded in tension, the maximum force in the tube is Ts. This force is
transferred into the footing through shear friction along the embedment length and anchorage at
the annular ring, and leads to the principle stress state if the concrete at the failure plane shown in
Figure 7.3. Equilibrium of vertical forces is expressed in Equations 7.27.4.
Ts = c cos Acone
Ts = d
D
tFy
2
(7.2)
(7.3)
le
D
d
l2
D
d
Acone = + b f + cot
d = le + b f le + e cot
2
2
sin
2
sin
0
(7.4)
Where Ac is the surface area of the failure plane and d is an infinitesimal angle in the plane of the
column cross-section. The angle of inclination of the failure plane, , ranges from 30 to 45
(MacGregor, 1997). For the conservative value of 45 the stress demand in the footing, c, to
develop plastification of the tube is defined by Equation 7.5.
D
tFy
2
c =
le2
D
b
l
+
+
f e
2
2
(7.5)
To determine the capacity of the footing at the failure plane, the stress demands placed on the
footing to develop the maximum experimental moments, c*, were calculated for each specimen.
Because the c calculation assumes full plastification of the tube, which did not occur in
Specimens I, II and IV, c* was calculated for a tube with the same geometry, but with design
90
strength Mn(p.s.) (see Section 2.1.1) equal to the maximum experimental moment Mexp. The
experimental stress demands for each specimen are listed in Table 7.3. The c* values are the
maximum demands that could be placed on the footing at the maximum experimental moments.
Because other resisting mechanisms may contribute to the capacity of the connection, the c*
demands are an upper bound estimate of the pull-out failure mechanism. Because specimens I, II
and IV exhibited pull-out failure, the capacity of the pull-out mechanism is not greater than the
maximum demands in these specimens. The upper bound estimate of the capacity of the pull-out
mechanism is 5.9fc.
Table 7.3 Experimental footing stress demands
Mexp (k-in)
c* / fc
I
10048
5.9
Specimen
II
III
10511 12844
6.3
5.1
IV
11230
7.3
91
12000
Moment (k-in)
10000
8000
6000
analysis
Spec I
Spec II
Spec III
Spec IV
4000
2000
0
0
Curvature (rad/in)
Moment-curvature analysis was used to determine nominal CFT moment capacity calculated
using the plastic strain compatibility method, Mn(s.c.). The strain compatibility method
recommends that concrete compressive strain be limited to 0.003 in./in., which is not
representative of the failure modes occurring during experimental testing. To determine a more
appropriate limit state for CFT flexural capacity, the response of Specimen III, which reached full
plastic capacity, was compared to the response predicted by moment-curvature analysis.
The maximum experimental moment capacity of Specimen III was controlled by local
buckling of the steel tube. The measured curvature and tube strains at tension yielding in the tube,
and at the drift ratio prior to local buckling, are compared to values calculated using the moment
curvature analysis in Table 7.4. The moment curvature analysis conservatively calculated the
compression strain in the tube just before buckling; however, moment curvature analysis does not
correlate well with local response parameters.
Table 7.4 Experimental and moment-curvature response (Specimen III)
Mexp (k-in)
Measured
Analysis
Meas./Anal.
Tension Yielding
9,638.5
1.2%
s+
s
(in./in.)
(in./in.)
(rad./in)
0.000173 0.001473 -0.001996
0.000337 0.003723 -0.00293
0.51
0.40
0.68
Prior to Buckling
11,973
2.5%
s+
s
(in./in.)
(in./in.)
(rad./in)
0.000927 0.007812 -0.01073
0.001565 0.021704 -0.00911
0.59
0.36
1.18
92
7.2.4 Comparison to Nonlinear Finite Element Analyses
The results of the nonlinear finite element baseline analysis (Section 4.2) adequately
predicted the progression of damage states in the experimental CFT-footing connections. The
analysis indicated that yielding and permanent deformation would occur in the tube base and
annular ring, which was verified by post-test investigation of Specimens I and II. The nonlinear
analysis predicted high compression stresses at the bearing surface of the connection and crushing
occurred at the column-footing interface at low and moderate drifts in all of the experimental
specimens. The progression of cracking in the experimental footing is supported by the spread of
high maximum principle stresses through the footing elements in the analytical solution, and the
wide bisecting crack in the footing surface was predicted by the concentrated cracking strains
parallel to the direction of loading in the finite element model.
An analytical model which can accurately predict the nonlinear response of the experimental
connections is a valuable tool in assessing and developing design recommendations for an
adequate column-footing connection. To further assess the accuracy of the nonlinear finite
element model defined in Section 4.1, analyses were performed on models representing
experimental Specimens I, II and III. The concrete material and smeared cracking definitions
were adjusted to the measured strength, and the applied axial load was increased to maintain an
axial load ratio, P/P0, of 10%. The lateral load history was the displacement controlled,
monotonic history used in the baseline analysis. The analytical boundary conditions were
modified to more closely represent the test conditions. The rigid support preventing lateral motion
n the x-direction was removed and substituted by prestressed truss elements placed vertically in
the footing, representing the high-strength anchor rods shown in Figure 5.3. In models of
Specimens II and III, vertical reinforcement elements were included.
The nonlinear finite element analysis results are compared to the measured specimen
responses in Figure 7.5. The initial stiffness and onset of softening in the test specimen responses,
to approximately 2% drift, were well matched by the analytical results. The lateral load resistance
of the analytical models increased even at high drifts, and do not predict the significant reductions
in resistance exhibited by Specimens I and II. This overstrength in the analytic models may be
due to the cracking modulus defined for the concrete foundation and column fill elements. The
cracking modulus in the models was selected to ensure convergence of the finite element solution
even at high drifts; and may prevent failure of the numerical concrete foundation.
93
200
150
100
50
0
-50
Experiment
FEA
-100
-150
Specimen I
-200
-10% -8% -6% -4% -2%
200
0%
2%
4%
6%
8% 10%
150
100
50
0
-50
-100
-150
Specimen II
-200
-10% -8% -6% -4% -2%
200
Experiment
FEA
0%
2%
4%
6%
8%
10%
150
100
50
0
-50
-100
Experiment
Specimen
FEAIII
-150
-200
-10% -8% -6% -4% -2%
0%
2%
4%
6%
8%
10%
94
7.3 Assessment of Performance
In order to assess and compare the experimental performance of the connection specimens
and make recommendations for the use and design of the proposed connection, the observed
experimental behavior and failure modes are related to measured and calculated response
quantities. The following section describes a series of performance (damage) states that occurred
during experimental testing including the progression of damage in the concrete footing, in the
steel tube and in the grouted connection of Specimen IV. The measured and calculated responses
of the specimens, specimen and column drift, experimental moment, specimen stiffness, and
maximum tube strain and curvature, at each damage state are tabulated and compared.
95
(d) pull-out
96
Interface cracking occurred at 0.40.5% drift, in the first two drift levels, for all
specimens.
Bisecting cracks formed at 0.91% drift in shallow embedded specimens, and at 1.2%
drift in the deeper embedded column.
Damage states identifying crack initiation occurred at similar drifts in shallow embedded
columns; however, the drift ratio corresponding to pull-out varied from 3.67% drift.
In specimens that exhibited pull-out, the specimen stiffness was 0.060.12 times the
design effective stiffness, EIeff, at that damage state.
Column drift was between 0.640.7% drift at tube yielding in specimens that yielded in
tension above the footing surface.
97
Concrete
Footing
Performance State
Interface Cracking
Bisecting Cracking
Diagonal Cracking
Pull-out
M exp
Mexp
(k-in)
M max
5,406
9,036
9,748
10,009
0.54
0.90
0.97
1.0
0.4%
1%
1.6%
3.6%
EI spec
EI eff
0.57
0.41
0.26
0.12
col
max
(rad./in.)
max+
(in./in.)
max(in./in.)
0.24%
0.48%
0.54%
0.56%
0.0000864
0.0001731
0.000196
0.000214
0.000893
0.001755
0.001842
0.001741
-0.00084
-0.0017
-0.00208
-0.00254
col
max
(rad./in.)
max+
(in./in.)
max(in./in.)
M yield
Concrete
Footing
EI spec
Interface Cracking
Bisecting Cracking
Diagonal Cracking
Pull-out
4,872
7,183
8,821
8,013
0.46
0.68
0.84
0.76
0.5%
0.9%
1.4%
7%
0.49
0.37
0.29
0.04
0.22%
0.36%
0.47%
0.46%
0.000075
0.000128
0.00017
0.000207
0.000715
0.00105
0.00124
N.M.
-0.00079
-0.00150
-0.00217
-0.00416
Steel
Tube
M exp
Mexp
(k-in)
Tension Yielding
Local Buckling
Ductile Tearing
10,494
X
X
1.0
X
X
2.9%
X
X
0.15
X
X
0.64%
X
X
0.000288
X
X
0.00255
X
X
-0.00431
X
X
Performance State
EI eff
97
98
M yield
Concrete
Footing
Interface Cracking
Bisecting Cracking
Diagonal Cracking
Pull-out
4836
10,042
X
X
0.48
1.0
X
X
0.5%
1.2%
X
X
Steel
Tube
M exp
Mexp
(k-in)
Tension Yielding
Local Buckling
Ductile Tearing
10,042
12,844
12,803
1.0
1.28
1.27
1.2%
4%
6%
Performance State
EI spec
col
max
(rad./in.)
max+
(in./in.)
max(in./in.)
0.49
0.37
X
X
0.25%
0.64%
X
X
0.0000963
0.000267
X
X
0.000677
0.00241
X
X
-0.0012
-0.0029
X
X
0.37
0.16
0.1
0.64%
N.M
N.M.
0.000267
0.000803
N.M.
0.00241
0.00217
N.M.
-0.0029
-0.0139
N.M.
col
max
(rad./in.)
max+
(in./in.)
max(in./in.)
EI eff
M yield
Concrete
Footing
EI spec
Interface Cracking
Bisecting Cracking
Diagonal Cracking
Pull-out
4,212
6,500
9,002
10,211
0.40
0.62
0.87
0.98
0.5%
1%
1.8%
6.6%
0.43
0.32
0.56
0.06
0.21%
0.37%
0.56%
0.95%
0.0000783
0.000144
0.000237
0.000679
0.000568
0.00119
0.00192
0.000417
-0.000143
-0.000161
-0.000137
-0.000109
Steel
Tube
M exp
Mexp
(k-in)
Tension Yielding
Local Buckling
Ductile Tearing
10,400
X
X
1.0
X
X
2.8%
X
X
0.17
X
X
0.71%
X
X
0.000356
X
X
0.00243
X
X
-0.000028
X
X
Bisecting Cracking
Diagonal Cracking
Uplift at Interface
5,487
8,219
10,814
0.53
0.79
1.03
0.7%
1.4%
4%
0.36
0.27
0.11
0.29%
0.49%
0.87%
0.000114
0.000202
0.000559
0.00924
0.00171
0.00309
-0.00016
-0.000145
-0.000097
Grout
Performance State
EI eff
98
99
Chapter 8: Conclusions
A research program was undertaken investigating the behavior of an embedded CFT columnreinforced concrete footing connection through finite element studies and experimental testing.
The following sections summarize the research program, present the major findings of the
research and discuss topics for future research phases.
Previous research on column base connections for both steel and CFT columns was surveyed.
These studies indicated that embedded column base connections can provide the strength and
stiffness required to develop the full flexural strength of the CFT column and can exhibit
adequate ductility for seismic loading. Based on this conclusion, an embedded CFT-footing
connection was proposed and investigated.
To determine the effect of various design parameters on the connection behavior, and to aid
in the development of an experimental test matrix, a series of finite element analysis were
performed on a model of the proposed connection detail. The preliminary parametric study used
linear-elastic material models to determine the effects of footing depth, embedment length,
annular ring dimensions, vertical footing reinforcement, and steel-concrete friction on the
connection behavior. The numerical results indicated that embedment length and the presence of
the annular ring have significant effects on the stresses calculated in the steel tube and concrete
footing and the development of column forces.
The preliminary analytical model was modified to included nonlinear, inelastic material
properties. The results of the nonlinear finite element model provided qualitative predictions of
the experimental behavior, such as locations of concentrated stress in the footing and column, and
the presence of a cone-shape failure plane extending from the base of the embedded column.
Four full-scale CFT-footing connection specimens were designed and tested to assess the
effects of embedment length, vertical footing reinforcement and construction procedure on the
nonlinear response of the connection. The influence of the seismic load and deformation demands
on the connection components, the CFT column and the reinforced concrete footing, were
evaluated. One specimen was constructed using an alternative grouted connection procedure
intended to improve the construction efficiency and reduce damage sustained in the footing.
100
8.2 Research Results and Conclusions
Connections with shallow embedment, 0.6D, had flexural resistance near or equal to the
yield strength of the CFT column. However, these specimens failed due to cracking and
damage in the footing and exhibited significant deteriorations in strength.
The presence of vertical reinforcement did not significantly increase the flexural strength
or stiffness of the connection, but damage was reduced at drifts exceeding 2% and pullout failure was delayed
The connection with embedment equal to 0.9D developed the flexural strength of the
CFT column. Failure of the connection was controlled by local buckling, and subsequent
tearing of steel tube near the column base, and occurred without considerable cracking or
damage in the footing.
The deeply embedded connection exhibited large deformation capacity, minimal strength
degradation after initial tube yielding, and improved energy dissipation capacity
compared to other specimens.
Column forces are developed in the footing by multiple resisting mechanisms, including:
column bearing against the concrete foundation forms a moment couple over the
embedment length, and anchorage from the annular ring and shear stress transfer are
resisted by the footing resistance to pull-out.
The connections exhibited initial stiffness equal to approximately 50% of the stiffness
expected for an ideal, fixed base condition calculated with the AISC column stiffness
equation. The experimental parameters had limited effect on the initial specimen
stiffness.
The experimental CFT columns exhibited strength and stiffness consistent with design
expressions published by AISC.
The specimen drifts occurring due to damage in the footing were initially approximately
50% of the total drift, and increased to 90% of total drift with large inelastic deformation
101
demands. The contribution to specimen drift occurring due to footing damage did not
exceed 50% in the deeper embedded connection.
The shallow embedded columns provided adequate strength but limited deformation
capacity and energy dissipation, and may only be appropriate for low- or non-seismic
demands. Longer embedment provided the strength load and deformation capacities
required for column bases subjected to high seismic demands.
Embedded column base connections are intended to provide fully fixed end conditions;
however this was not achieved in the specimens tested. Further research is required to determine
the embedment length necessary to achieve this ideal end condition and the effect that the column
base stiffness has on the full structural response to seismic demands. The analytical and
experimental studies identified typical behaviors and failure modes, and possible resistance
mechanisms, for embedded CFTfooting connections with varying embedment length, vertical
reinforcement and construction procedure. To accurately quantify the effects of these parameters
on the connection strength and stiffness, further analytical or experimental studies considering a
larger number of embedment lengths and vertical reinforcement ratios and other design
parameters are necessary. With a sufficient volume of experimental or numerical research results,
the contribution and capacity of the potential resistance mechanisms can be determined and used
to develop adequate design recommendations for embedded column base connections for CFT
members.
102
Bibliography
American Concrete Institute (ACI) (2003)
American Institute of Steel Construction (AISC) (2005) Specification for Structural Steel
Buildings, AISC, Chicago, Illinois.
Applied Technology Council (ATC) ATC-24 Guidelines for Testing Steel Components, Applied
Technology Council, Redwood City CA, 1992.
Azizinamini, A. and Schneider, S. (2004) Moment Connections to Circular Concrete-Filled Steel
Tube Columns. Journal of Structural Engineering, Volume 130, Issue 2, pp. 213-222,
February 2004.
Campbell, T.I., and Kong, W.L. (1988) Laboratory Study of Friction in TFE Sliding Surfaces for
Bridge Bearings, Report MAT-88-04, Ministry of Transportation, Downsview, Ontario,
Canada, May 1988.
Gere, J.M. (2001) Mechanics of Materials, Fifth Edition, Brooks/Cole, Pacific Grove, California.
Hajjar, J. F. (2002) "Composite Steel and Concrete Structural Systems for Seismic
Engineering,"Journal of Constructional Steel Research, Vol. 58, Nos. 5-8, May-August, pp.
703-723.
Hitaka, T., Suita, K., and Kato, M. (2003) CFT Column Base Design and Practice in Japan,
Proceedings of the International Workshop on Steel and Concrete Composite Construction
(IWSCCC-2003), Report No. NCREE-0.-0.26, National Center for Research in Earthquake
Engineering, Taipei, Taiwan, October 8-8, 2003, National Center for Research in Earthquake
Engineering, Taipei, Taiwan, pp. 291-290.
Hsu, H. and Lin, H. (2003) Performance of Concrete-Filled Tube Base Connections Under
Repeated Loading, Proceedings of the International Workshop on Steel and Concrete
Composite Construction (IWSCCC-2003), Report No. NCREE-0.-0.26, National Center for
Research in Earthquake Engineering, Taipei, Taiwan, October 8-8, 2003, National Center for
Research in Earthquake Engineering, Taipei, Taiwan, pp. 291-299.
Kadoya, H., Kawaguchi, J., and Morino, S. (2005) Experimental Study on Strength and Stiffness
of Bare Type CFT Column Base with Central Reinforcing Bars, Composite Construction in
Steel and Concrete V, Ed. Roberto Leon and Jrg Lange, United Engineering Foundation,
Inc., July 2004.
MacGregor, J.G. (1997). Reinforced Concrete: Mechanics and Design, Prentice Hall, Upper
Saddle River, New Jersey.
Mander, J.B., Preistley, M.J.N., and Park, R. (1988). Theoretical Stress-Strain Model for
Confined Concrete, ASCE Journal of Structural Engineering, Vol. 57, No. 8, pp.1 804-1826.
103
Marson, J. and Bruneau, M. (2004) Cyclic Testing of Concrete-Filled Circular Steel Bridge
Piers Having Encased Fixed-Base Detail, ASCE Journal of Bridge Engineering, Vol. 9, No.
1, pp. 14-23.
Mazzoni, S., McKenna, F., and Fenves, G.L. (2005) Open System for Earthquake Engineering
Simulation User Manual; Version 1.6.0, Pacific Earthquake Engineering Research Center,
Uinversity of California, Berkeley. (http://opensees.berkeley.edu)
Morino, S., Kawaguchi, J., Tsuji, A., and Kadoya, H. (2003) Strength and Stiffness of CFT
Semi-Embedded Type Column Base, Proceedings of ASSCCA 2003, Sydney, Australia, A.
A. Balkema, Sydney, Australia.
MSC.Software Corporation (2004) Volume A: Theory and User Information, Santa Ana,
California.
104
Ac
As
Cshear
D/t
EIcol
EIeff
EIi,col
EIi,spec
EIsec
EIspec
ECNXN
ECOL
ESPEC
Ec
Es
Esoft
Fu
Fy
Heff
J1
J2
Ki
Ksec
Lcol
Leff
Mexp
105
Mn(p.s.)
= Nominal flexural strength of composite member calculated using the plastic stress
distribution method, as defined by AISC (2005)
Mn(s.c.)
= axial load
P0
Ts
bf
bfooting
df
= foundation depth
fc
le
lel
lgage
lspec
= tube thickness
tf
cnxn
col
col,base
col,el
north,spec
south,spec
north,rig
south,rig
raw
rig,slip
rig,uplift
spec,uplift
spec,uplift
lmeas
= specimen drift at location of loading due to anchor block slip and uplift,
repsectively
= specimen drift at location of loading due to specimen slip and uplift, respectively
= anchor block base uplift at North and South face of anchor block, respectively
= measured specimen drift at location of horizontal load
106
lstrain
ltheory
()
cu
north
south
rig
spec,uplift
= coefficient of friction
l,l
l,p
c*
cr
= time
= column curvature
= column rotation
107
These
methods were then applied to the results of the experimental tests conducted at the University of
Washington (UW). Experimental variables in the UW study were embedment depth, foundation
reinforcement, and construction procedure.
The experimental investigation at UW included four CFT-footing connection specimens: C1
had shallow column embedment and no shear (vertical) reinforcement in the foundation; C2 had
shallow embedment, but included shear reinforcement; C3 had the same shear reinforcement as
C2, but with deeper embedment; C4 had shallow embedment, but the construction sequence was
altered such that the embedded portion of the column was surrounded by grout rather than
concrete.
Under the direction of Dr. Nakashima and his research group, the following evaluation of the
UW experimental tests was performed. For each test specimen, a hysteretic model approximating
the experimental moment-rotation relationship of the CFT column base connection was defined
using the hysteretic material model defined in the OpenSEES finite element analysis framework.
These models combine the nonlinear behavior of the column and connection into a single
hysteresis, and include the strength degradation and post-peak behavior unique to each
experimental connection. The hysteretic models were used to define column base hinges
108
incorporated into discrete analytical building frame models. The frame model is based on a
typical 3-story, steel frame structure, designed for construction near Seattle, WA. The steel
columns are replaced by equivalent strength CFT columns. The columns were assumed to remain
elastic, with the exception of the hysteretic hinges at the column base; plasticity is concentrated at
the beam ends. One frame model was defined for each column base model, to determine the
effect of the connection hystereses on frame behavior. Incremental dynamic analyses were carried
out on each frame using a suite of eighteen artificial ground motion records, at increasing
intensity levels. Each ground motion was scaled such that the ground motion spectrum matched
the design spectrum for the frame; corresponding to intensity level one. The intensity level was
increased incrementally to ten.
At each intensity level, the probability of reaching column base yielding, column base
deterioration, and exceeding the maximum allowable story drift was evaluated for each
connection model. Figures 1-3 present the results of this evaluation. The results show that at all
intensity levels, connections C1 and C2 have similar probability of reaching yielding or failure.
At most intensity levels, connection C3 is least likely to reach column base yielding. Connection
C4 has the lowest probability of reaching column base deterioration at all intensity levels. The
results also indicate that at moderate intensity levels, connections C3 and C4 have slightly lower
probability of exceeding the allowable story drift limit.
18
Mom
16
14
Rot
12
C1
C2
C3
C4
10
8
6
4
2
0
1
109
18
Mom
16
14
Rot
12
10
8
6
4
2
0
1
18
16
14
12
10
8
6
4
2
0
1