Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title No. 112-S16

Behavior and Simplified Modeling of Mechanical


Reinforcing Bar Splices
by Zachary B. Haber, M. Saiid Saiidi, and David H. Sanders

Bridge seismic design codes do not allow mechanical reinforcing newer and more innovative bridge columns in earthquake-
bar splices in regions expected to undergo significant inelastic prone areas.
deformations during earthquakes, thus severely limiting precast Although previous studies have used mechanical splices
and innovative bridge column construction that uses such splices. in plastic hinge zones,8 there is little information as to the
The uniaxial behavior of two commercially available mechanical
deformation characteristics of mechanical splices and their
splices under different loading conditions was investigated
effects on local and global member behavior. Researchers
experimentally in this study with emphasis on deformation
response. Tests were performed with static, dynamic, and cyclic have studied the uniaxial behavior of mechanically spliced
loading. The performance of the splices was satisfactory under bar assemblies, but these studies have focused primarily on
all loading conditions in that bar fracture occurred outside the how strength is affected by fatigue loading,9 bar diameter,9
splice. Furthermore, the results revealed the effect of the relatively and blast loading rates.10 It was suggested by Haber et al.8
high stiffness of mechanical couplers. The responses of individual that the length of a splice is a critical factor that affects the
splices were used to interpret data from a series of cyclic tests on post-yielding flexural behavior of a member. That is, splices
half-scale bridge columns employing mechanical splices in plastic with smaller LSp/db ratios (<4) are less likely to change the
hinge zones. Lastly, a simple method was proposed and validated plastic hinging mechanism, where LSp is the length of the
for modeling these devices in reinforced concrete members. mechanical splice, and splices with larger ratios (>14) may
Keywords: accelerated bridge construction; acceptance criteria; coupler;
adversely affect hinge formation and behavior. The objective
ductility; mechanical splice; repair; seismic; shape-memory alloy. of this study was to evaluate the deformation characteristics
of two commercially available mechanical splices under
INTRODUCTION static, dynamic, and cyclic loading. The correlation between
Mechanical reinforcement splices have been used in cast- component- and system-level behaviors was addressed
in-place concrete construction when long, continuous bars by comparing uniaxial test results with a series of half-
or reinforcement cages are required. Unlike lap splices, scale bridge column test results conducted by the authors.
which can require lengths greater than 30 bar diameters (db), Lastly, experimental test data is used as a foundation for a
mechanical splices can be used to join bars at discrete locations. simple method to incorporate mechanical splices in member
Some of the mechanical reinforcing bar splices commercially deformation and capacity calculations.
available in the United States1 are shown in Fig. 1. Bridge
and building design codes use acceptance criteria such RESEARCH SIGNIFICANCE
as International Code Council (ICC) AC1332 and ASTM Previous studies have identified a number of applications
A1034/A1034M3 to quantify the ability of a splice to transfer for mechanical reinforcing bar splices in plastic hinge
load, withstand load reversals, and resist slip. Furthermore, regions. However, there is little understanding as to the local
some state departments of transportation (DOTs) have deformation response of these devices and how that response
developed their own acceptance criteria.4 After evaluation, can affect the global behavior of a ductile reinforced concrete
mechanical splices are given a performance classification member. This paper provides much-needed data and
compatible with the corresponding code provision of interest, insight into the local deformation behavior of mechanical
which is used to restrict placement in a structural member reinforcing bar splices through a series of uniaxial tests.
or limit stress/strain demands on spliced bars. In the United Using test data, a simple method for analytical modeling of
States, there is one significant difference between bridge mechanical splices in with concrete members is proposed
and building code requirements for mechanical splices. and validated with large-scale experimental test results.
ACI 318-025 allows Type 2 mechanical splices, which must
be able to develop the full tensile strength of the spliced bars EXPERIMENTAL INVESTIGATION
to be placed at any location within a member regardless of Specimen details
local inelastic demands. On the other hand, bridge design Two of the commercially available mechanical reinforcing
codes such as the AASHTO Bridge Design Specifications6 bar splices, shown in Fig. 1, were investigated in this study:
and Caltrans Seismic Design Criteria (SDC)7 prohibit
all mechanical splices from being placed in plastic hinge ACI Structural Journal, V. 112, No. 2, March-April 2015.
MS No. S-2013-319.R3, doi: 10.14359/51687455, received June 11, 2014, and
regions, which are subjected to high inelastic demands. Such reviewed under Institute publication policies. Copyright © 2015, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
provisions have prevented the use of mechanical splices in obtained from the copyright proprietors. Pertinent discussion including author’s
plastic hinges of bridge columns and have been a barrier to closure, if any, will be published ten months from this journal’s date if the discussion
is received within four months of the paper’s print publication.

ACI Structural Journal/March-April 2015 179


Fig. 1—Mechanical reinforcing bar splice commercially available in the United States.
through formation of compression struts in the grout which
transfer force to the sleeve.
HC specimens were constructed using No. 8 (D25)
Grade 60 ASTM A706 bars having an average measured
yield stress, ultimate stress, and percent elongation at rupture
of 67.9 ksi (468 MPa), 95.1 ksi (655 MPa), and 18.2%,
respectively. Specimens were prepared with two 16 in.
(406 mm) headed bar segments joined using the threaded
steel collars described previously. The two collars were
initially tightened by hand followed by a pipe wrench to
the manufacturer’s minimum specified torque of 150 lbf-ft
(203 N-m).
GC specimens were constructed using No. 8 (D25)
Grade 60 ASTM A615 bars having an average measured
yield stress, ultimate stress, and percent elongation at
rupture of 66.8 ksi (460 MPa), 111.3 ksi (767 MPa), and
15.8%, respectively. To construct GC specimens, reinforcing
bars were first placed into the tapered end of the sleeves and
the assembly was tied to a support frame. The prepackaged
high-strength cementitious grout was mixed according to
manufacturer specifications and the sleeves were filled
approximately three-quarters full. The grout was rodded
with a smooth 0.25 in. (6.5 mm) diameter rod to ensure good
consolidation, and the second reinforcing bar was inserted
into the sleeve. The average 28-day compressive strength of
Fig. 2—Uniaxial test setup and instrumentation plans. the grout was 15.7 ksi (108 MPa) according to ASTM C109/
C109M-02.
the upset headed coupler (HC) and grouted sleeve coupler
(GC). The HC splice consists of male and female threaded Test setup and loading protocols
steel collars that join bar segments with deformed heads, All specimens were tested in the Large-Scale Structures
which are created by heating the bar end and compressing Laboratory (LSSL) at the University of Nevada, Reno
the heated end with specially designed hydraulic ram. The (UNR), using a servo-hydraulic loading frame. The test setup
force transferring mechanism for the HC splice consists of was developed according to ASTM A1034/A1034M3 and
compression being transmitted directly through deformed Caltrans Test Method CT6704 (Fig. 2). Strain was measured
heads and tension through the threaded collars. The GC directly from the reinforcing bars using foil-backed resistive
splice has been commonly used in conventional and precast gauges (two on opposite sides per location), and over the
construction in East Asia11 and the United States.12 At the length of the splice using a digital extensometer. For GC test
precasting plant, reinforcing bars are inserted into the tapered specimens, a pair of strain gauges was also installed at the
end of the sleeve and the device is then cast within the concrete midheight of the sleeve. The extensometer gauge lengths
member. On site, the precast element is positioned such that over the coupler region LCR for HC and GC specimens were
reinforcing bar dowels protruding from the adjacent member 6 and 18 in. (152 and 457 mm), respectively. The clear length
enter the open sleeve ports. The connection is completed by between load frame grips, LClear, was selected as the minimum
pumping a proprietary high-strength (>14 ksi [96.5 MPa]) specified by CT670, which were 26.5 and 38 in. (673 and
cementitious grout into the sleeve. Force is transmitted 965 mm) for the HC and GC specimens, respectively.

180 ACI Structural Journal/March-April 2015


Table 1—Summary of test results
Yield stress, ksi Ultimate stress, ksi Strain at rupture, %
Bar Coupler region
ID Average Standard deviation Average Standard deviation Average Standard deviation Average Standard deviation
HC control bar 67.9 3.61 95.1 1.61 18.2 3.08 — —
HC-S 67.2 1.09 95.2 2.71 16.4 2.09 7.70 0.37
HC-D 71.9 0.42 98.1 0.31 15.5 1.09 8.46 0.27
HC-C1 68.5 — 93.3 — 16.9 — 7.80 —
HC-C2 67.7 — 94.6 — 12.6 — 8.48 —
GC control bar 66.8 3.69 111.3 1.61 15.8 0.44 — —
GC-S 66.3 0.66 108.6 1.08 15.9 1.19 5.61 0.35
GC-D 70.4 — 110.8 1.00 16.2 3.61 5.53 0.28
GC-C1 66.1 — 98.7 — 5.59 — 2.69 —
Note: 1 ksi = 6.89 MPa.

Five different loading protocols were used to evaluate load was reversed until the specimen reached a compression
the uniaxial behavior of the spliced bars. Three specimens stress of 20.7 MPa (10.5 kN). Both specimen types had long
per splice type were tested for each protocol with exception unsupported lengths. Thus, a low compression stress target
of the cyclic loading tests. The loading protocols and was selected to prevent buckling.
associated nomenclature were: monotonic static loading Elastic slip tests were conducted in accordance with
until failure (S), monotonic dynamic loading until failure Caltrans and AASHTO methods, which are used to determine
(D), slow cyclic loading until failure (C), single-cycle elastic the permanent relative deformation between the reinforcing
slip (SCS) loading, and multi-cyclic elastic slip (MCS) bar and mechanical splice. In single cycle slip tests (SCS),
loading. Specimens are identified by splice and loading samples were loaded to an initial stress of 3 ksi (20.7 MPa)
type, respectively. For example, a grouted splice specimen and the elongation over the sample measurement gauge
tested under monotonic dynamic loading would be identified length (DInitial) was measured. Samples were then stressed
as “GC-D.” to 30 ksi (207 MPa), held for 30 seconds, and subsequently
Loading was displacement-controlled for monotonic destressed to 3 ksi (20.7 MPa). Upon distressing, a final
static and dynamic tests. The loading rates for monotonic elongation measurement (DFinal) was recorded. The resulting
static tests were determined according to ASTM A370.13 slip, DSlip, is defined as the difference between final and initial
For HC-S specimens, pre- and post-yield displacement elongation measurements. After completing the single cycle
rates were 0.00625 and 0.05 in./s (0.159 and 1.27 mm/s, test, samples were subjects to three to five additional cycles
respectively. For GC-S specimens, pre- and post-yield to determine if slip increased with additional loading. This
displacement rates were 0.01875 and 0.15 in./s (0.476 and sequence is referred to as the multi-cycle slip test (MCS).
3.81 mm/s), respectively. The dynamic loading protocol The maximum permitted elastic slip for splices with No. 8
was selected to subject specimens to strain rates in the range (D25) bars according to Caltrans and AASHTO are 0.028
of those imposed by an earthquake event.14 A target rate and 0.01 in. (0.71 and 0.25 mm), respectively.
of 0.07 in./in./s was selected knowing that achieved rates
would be approximately 80 to 120% of the target.15 The EXPERIMENTAL RESULTS
corresponding displacement rates, which are based on LClear, A summary of test results is provided in Table 1, along with
for HC-D and GC-D specimens were 1.575 and 1.75 in./s the measured material properties for unspliced reinforcing
(40 and 44.5 mm/s), respectively. bars. On the day of testing, the average measured grout
The effect of tension-compression load reversals was strength for GC specimens was 18.5 ksi (128 MPa).
studied by applying cyclic loads. Although the widely used
ICC AC332 test criteria requires cyclic testing, spliced bars Monotonic static tests (S)
are only subjected to reversals up to five times the specified As would be expected, the average elongation over LCR
yield strain of the bar (5ey). Testing in this study subjected (otherwise referred to as the coupler region) was reduced
splices to load reversals beyond this level and was continued due to the presence of the threaded steel collars joining the
until failure. Cyclic tests were conducted in load control deformed heads. The average elongation at failure over
mode at rates of 1 kip/s (4.45 kN/s) during tensile loading and the coupler region was 7.70%, which was 53% less than
at 0.5 kip/s (2.22 kN/s) during compression loading, which that of the reinforcing bar. Figure 3 shows representative
correspond to stress rates of 1.27 and 0.635 ksi/s (8.72 and constitutive relationships for HC-S tests. The stress-strain
4.36 MPa) for tension and compression, respectively. Each curve for the coupler region exhibited a stiff initial slope
cycle of loading consisted of a single tensile and compression up to approximately 10 ksi (69 MPa), which subsequently
cycle. For each cycle, the peak tensile load was increased by softened and remained linear up to yielding of steel. Softening
increments of 0.2fy from 0.5fy to 1.1fy followed by increments occurs as the precompressive force on the deformed heads,
of 0.1fy thereafter. After the target tension was reached, the which is a result of the initial torque on the threaded collars,

ACI Structural Journal/March-April 2015 181


Table 2—Average measured strain rates during
dynamic tests (unit: strain/s)
HC-D GC-D
Standard Standard
Stress range Average deviation Average deviation
0 to yield 0.0175 0.0121 0.0781 0.0073
Yield to ultimate 0.0908 0.0178 0.0924 0.0041
Ultimate to failure 0.0633 0.0200 0.1060 0.0770

ductile reinforcing bar rupture, which occurred away from


and without damage to the splice.
In GC-S tests, the response of the spliced reinforcing
Fig. 3—Stress-strain curves from monotonic tests on bars was similar to that of the control bars (Fig. 4). The
HC device (static tests are solid lines; dynamic tests are deformation capacity over the coupler region was reduced
dashed lines). by 65% compared with strain measurements taken from
spliced reinforcing bars due to the presence of the grout-
filled cast-iron sleeve. Unlike the coupler region response in
HC-S specimens, the initial branch of the stress-strain curve
for GC-S specimens was similar to that of the reinforcing bar.
This indicates that the elastic stiffness of the GC assembly is
similar to that of mild steel bars. Similar to the HC-S tests, each
GC-S specimen failed by reinforcing bar rupture away from
and without damage to the splice. Strain measurements from
the midheight of the sleeves indicated that the average strain
in the sleeve at failure was 0.7%, which was three times the
average measured yield strain of steel. Strain measurements
from the sleeve also indicated that the sleeves undergo
nonlinear deformations once the stress in the reinforcing bars
reach approximately 70 ksi (482 MPa).
Fig. 4—Stress-strain curves from monotonic tests on Axially loaded reinforcing bars that are well-anchored in
GC device (static tests are solid lines; dynamic tests are cementitious materials undergo localized deformations from
dashed lines.) the anchoring material as a result of strain penetration,16
which is typically referred to as “bond-slip.” Although
not explicitly measured, Fig. 5 shows a grout cone failure
surface indicating that strain penetration occurs within the
grouted coupler assembly during loading. It was shown
by Haber et al.17 that bond slip can account for up to 40%
of the deformation over the coupler region (LCR). None of
the GC tests conducted in this study exhibited bar pullout
failure, but other studies have shown this can occur.10 Such a
failure mode would be caused by insufficient grout strength
or improper installation.

Monotonic dynamic tests (D)


Representative stress-strain curves from monotonic
dynamic loading are shown along with the static curves in
Fig. 3 and 4 for HC-D and GC-D tests, respectively. The
average measured strain-rates are listed in Table 2 according
to stress range. Previous tests on mild steel reinforcing bars
loaded at strain-rates similar to those shown in Table 2 have
exhibited increased yield and ultimate stresses by as much
Fig. 5—Evidence of strain penetration into grouted sleeve. as 30%18; Zadeh and Saiidi15 provide detailed discussion
regarding the behavior of axially loaded reinforcing bars
is overcome and the heads separate. Head separation is under high strain-rate loading. Thus, it is not unexpected
only evident up to yielding of steel. However, the heads that in both HC-D and GC-D tests that the yield and ultimate
continue to separate afterward, but the deformation over stresses were slightly larger than corresponding static tests.
the coupler region is controlled by nonlinear deformation Slight variations among the initial slopes of stress-strain
of the reinforcing bars. Each HC-S specimen failed due to curves can be observed between static and dynamic tests,

182 ACI Structural Journal/March-April 2015


Fig. 6—Cyclic test results: HC results (a) through (c); GC results (d) through (f).

which are expected due to differences in clamping forces at Cyclic loading tests
the grips. Previous studies have shown that dynamic loading Two HC-C specimens were tested, one with the
does not have a significant effect on the elastic modulus of manufacturer’s minimum specified torque applied to the
mild steel reinforcing bars.19 threaded collars, denoted as “HC-C1,” and a second with
In HC-D tests, the average yield and ultimate stresses collars hand-tight, denoted as “HC-C2.” The measured yield
increased 6% and 3%, respectively, compared with HC-S test and ultimate stresses of both specimens were within 2% of
results. In GC-D tests, the average yield and ultimate stresses those tested under monotonic static loading. Furthermore,
increased 6% and 2%, respectively, compared with GC-S the elongation over the coupler region was also comparable
test results. Similar to the static tests, both HC-D and GC-D with static tests. The cyclic stress-strain response for the
specimens exhibited reduced elongation over the coupler coupler region and the bar assembly of HC-C2 is shown in
region compared with measurements from the spliced Fig. 6(a) and (b), respectively, which indicate that the stress-
reinforcing bars. The average elongation at failure over the strain backbones for both the coupler region and reinforcing
coupler region was 45% and 66% lower in HC-D and GC-D bar are nearly identical to those from monotonic static tests.
specimens, respectively. All HC-D and GC-D specimens Both HC-C specimens failed due to ductile reinforcing bar
failed due to ductile reinforcing bar fracture away from and rupture away from and without damage to the splice.
without damage to the splices. This indicates that increased Once each peak stress level was reached, the load was
yield and ultimate stresses caused by dynamic loading were reversed to a target compressive stress of –3 ksi (–20.7 MPa).
sustained by both splice types without an adverse effect on During unloading, the slope of the stress-strain curves for the
the failure mode. Similar the GC-S tests, GC-D specimens coupler region and the reinforcing bar were approximately the
were inspected after testing and evidence of strain penetration same, indicating the reinforcing bars control the unloading
into the grouted sleeves was found in all specimens. Lastly, stiffness of the device. Once the load in the bar approached
dynamic loading did not affect the stress-strain curves in the zero, a distinct, instantaneous deformation occurred. It was
coupler regions of HC-D and GC-D specimens. hypothesized that separation of the deformed heads (otherwise
referred to as gap opening) within the steel collars occurred
Slip tests once precompression from the applied torque was overcome.
The maximum slip recorded for HC-SCS and GC-SCS Cyclic loading confirms this behavior and a relationship
samples were 0.007 and 0.0175 in. (0.178 and 0.044 mm), between peak stress and gap length can be established. The
respectively. The multi-cycle slip tests for each splice gap length, Dgap, was defined as the deformation during the
type did not indicate cumulative slippage with application transition between tensile and compressive force within the bar.
of three or more cycles. Both HC and GC splices passed There was an approximately linear relationship between the
single- and multi-cycles slip tests according to both Caltrans peak stress in the bar and the gap length between the deformed
and AASHTO maximum slip criteria, which are 0.028 and heads. The peak stress versus gap length plot (Fig. 6(c)) also
0.01 in. (0.71 and 0.25 mm), respectively. indicates elastic slip limits allowed by Caltrans and AASHTO.
It can be observed that these limits are significantly exceeded
even before yielding of the reinforcing bar. Further discussion
of this behavior is provided in subsequent sections.

ACI Structural Journal/March-April 2015 183


affect the characteristic behavior or failure modes of the
splice assembly. However, higher strain rates, such as those
expected in a blast, reduce the ultimate strength and ductility
of mechanically spliced bars.14
The SR plot indicates that the deformation response of
some splices could be as little as 25% (SR = 0.25) of that
of the reinforcing bar throughout the loading history, which
can be observed in test data from the GC device. The HC
splice did not exhibit as large of stiffness increase as the GC
splice. However, the gap-opening behavior is clearly visible
in Fig. 7, which allows the HC splice to deform significantly
compared to the reinforcing bars over a short part of the
loading history. Prior to a strain of 0.02, the majority of
deformation occurred within the coupler region, which is
indicated by a slope exceeding 1:1 (SR = 1) (Fig. 7).

CORRELATION BETWEEN UNIAXIAL AND


BRIDGE COLUMN TESTS
A series of half-scale reinforced concrete bridge column
Fig. 7—Relationships between strain in reinforcing bar and tests were conducted by the authors to investigate the
strain over splice. seismic performance of precast column-footing joints
with mechanical reinforcing bar splices.8 Two connection
One GC specimen (GC-C1) was tested under cyclic
configurations were studied: one employing two layers of HC
loading. The average yield stress was found to be
splices, referred to as “HCNP,” and the second using single-
comparable to static results, but the average ultimate stress
layer GC splices, referred to as “GCNP.” The reinforcement
and elongation of the reinforcing bar at rupture were 9% and
details and model geometry were designed assuming the
65% lower than static tests, respectively (Fig. 6(d) through
behavior would be similar to a conventional cast-in-place
(f)). The discrepancies between static and cyclic tests on
(CIP) benchmark column; this is otherwise referred to
the GC device were not caused by premature failure of
as “emulative” design. That is, a conventional column
the splice. In both cases failure was a result of reinforcing
model was designed for a target displacement ductility
bar rupture away from the splice. However, in GC-C1, the
capacity of µC = 7.0 (ultimate displacement/effective yield
reinforcing bar did not exhibit ductile behavior—that is,
displacement) according to Caltrans SDC, and the details
necking of the bar at the point of fracture did not occur rather
of the plastic hinge region were modified to incorporate the
the fracture surface was flat. This may have been caused by
mechanically spliced connections. The column models had
a low-cycle fatigue-type effect, but such failures typically
24 in. (610 mm) diameter cross sections, an aspect ratio of
require significantly larger strain reversals.20 There was also
4.5, and longitudinal and transverse reinforcement ratios
a slight difference between the static and cyclic stress-strain
of 1.9% and 1.0%, respectively. The ratio of axial load to
backbone curves for the coupler region. The slope of the
the product of the gross column cross-sectional area and
unloading curves for both the coupler region and reinforcing
the specified concrete compressive strength was 0.1. Along
bar were similar indicating the reinforcing bars control
with the precast models, a CIP benchmark model was also
the unloading stiffness. Unlike HC-C tests, there was not
constructed and tested for comparison. Each column was
a visible instantaneous deformation during the transition
tested in a cantilever configuration and subjected to slow
between tensile and compressive loads. The average stress-
cyclic loading at increasing drift levels. Figure 8(a) shows
strain history from the midheight of the cast-iron sleeve
the pertinent plastic hinge connection details of these three
indicates low strains and slight nonlinearity.
half-scale column models. A detailed discussion of these
tests can be found in Haber et al.8
Comparison and discussion
Figures 8(b) and (c) show the moment-rotation
It was observed that the presence of the mechanical splice
relationships for columns with HC and GC splices,
reduced the deformation capacity over the coupler regions for
respectively, measured near the base of the column (shown
both devices. Figure 7 presents representative relationships
in Fig. 8(a)). The data, which is only shown up to 5% due to
between the strain over LCR and the estimated strain over
localized bar buckling thereafter, captures the influence of
LSp in terms of the average reinforcing bar strain obtained
the mechanical splices on the hinge rotations. The effect of
from strain gauge measurements. The slope of the curves is
gap closure can be clearly observed in the moment-rotation
referred to as the splice-bar strain ratio (SR) and indicates
response of HCNP in the form of a slight pinch in the
the relative stiffness of the splice to the reinforcing bar. As
unloading branches. The pinch, however, is small and stable
mentioned previously, there was little difference between the
and does not lengthen as deformations increase. Except for
static and dynamic behavior of the coupler regions, which
the pinch, the loops are similar in shape and magnitude to
can be seen in Fig. 7. This indicates that strain rates similar
those of the CIP column. This is not surprising given the
to those expected from an earthquake would not adversely
relatively short length of the HC device. On the other hand,

184 ACI Structural Journal/March-April 2015


Fig. 8—Behavior of RC members with mechanical splices: (a) general column details; (b) moment-rotation relationships for
HCNP; (c) moment-rotation relations for GCNP; and (d) force-drift behavior.
the moment-rotation behavior of the GC column section a proven method to model these devices does not exist. Data
differs significantly from that of CIP. The maximum rotation provided in this paper indicate that the deformation response
in the GC section is approximately one-third the maximum of mechanical splices indeed is different from that of the
rotation of CIP, and very little plastic rotation is achieved. reinforcing bar. Furthermore, it was evident that the presence
This is consistent with observations from uniaxial tests, of the GC splices slightly increased the lateral load capacity
which indicated that the deformation capacity of the GC of GCNP and affected the local moment-rotation behavior.
splice could be as little as 25% of the reinforcing bar. Similar effects could occur using other devices with large
Although individual characteristics of the mechanical LSp:bd ratios. Thus, modeling techniques are required for
splices can be observed locally within each column, incorporating these devices in analysis procedures. Haber
the global force-displacement relationships are not as et al.21 described a method for modeling columns with GC
significantly affected in this case. However, the influence splices, but required detailed methods and specific material
of the mechanical splices could become more apparent and geometric properties of the mechanical splice. The
in the global force-deformation response with different following sections provide a simple method for modeling
column geometries and/or reinforcement details. As shown mechanical splices and validation with the previously
in Fig. 8(d), the hysteresis loops for three columns are described column test data.
comparable. GCNP had slightly higher peak loads after
2% drift due presence of GC splices. HCNP also exhibited SIMPLIFIED MODELING OF MECHANICAL
slightly higher peak load, but was a result of cementitious SPLICES IN BRIDGE COLUMNS
grout present within the hinge zone that was approximately Proposed method
twice as strong as the concrete for CIP. Furthermore, It is proposed that mechanical splices can be modeled in
the slight pinch observed at the local level can be seen in a simplified manner by defining an effective uniaxial stress-
global response, but is not significant. It should be noted strain relationship for spliced longitudinal reinforcement that
that although force-displacement relationships were similar can be employed in fiber section analysis. This model may
among the three columns, the presence of the GC splices in be used in most commercially available structural analysis
GCNP ultimately shifted plastic hinging to the footing and programs that have moment curvature, lumped-plasticity
above the couplers. This shifted hinge mechanism eventually frame element, or distributed plasticity frame element
caused longitudinal bars to rupture in the footing at 6% drift analysis capabilities. Distinct points on the effective stress-
due to strain concentrations and numerous load reversals. As strain curve can be calibrated to match test results in specific
a result, the drift capacity of GCNP was significantly less cases or may be based on the relative stiffness between the
than the drift at failure in CIP and HCNP, which was 10%. splice assembly and the reinforcing bar in a more generalized
Typically, designers do not account for the presence approach. The strain component of the effective stress-strain
of mechanical splices in design calculations for ductile relation is defined by strains occurring over LSp, which can
reinforced concrete members. This is primarily because be determined from measurements over LCR. This method is
splices are not typically placed in locations expected to advantageous because it does not require direct knowledge
undergo significant nonlinear deformations. Other reasons of material and geometric properties of mechanical splice
may include that designers assume splice behavior is assembly, which may be proprietary. That is, the intrinsic
approximately the same as the spliced reinforcing bar or that deformation characteristics of the device are captured using

ACI Structural Journal/March-April 2015 185


Fig. 9—Stress-strain model for reinforcing steel and
proposed splice model. Fig. 10—Analytical model for bridge column with grouted
sleeve column-footing connection.
standard evaluation and acceptance criteria tensile test data.
With respect to the GC device, these include influence of e*u e*sh Esh
the cast-iron sleeve on splice stiffness, deformation caused = = = SRI (2)
e u e sh Esh*
by unsupported reinforcing bar length, and deformation
resulting from bond slip within the sleeve.
It was observed in Fig. 7 that the GC device exhibited SRI ey
an approximately linear SR relationship, where the strain ≥ (3)
SRE e sh
over LSp was approximately 0.25 times that in the bar. Using
this information, an effective stress-strain relationship can
be created by scaling a constitutive stress-strain model for Validation with half-scale column test results
reinforcing steel to produce a similar SR relationship between The proposed model was validated using column model
the two stress-strain models. The effective stress-strain test results discussed previously for GCNP. Figure 10 shows
model can be applied in an individual nonlinear uniaxial fiber the general details of the half-scale test model along with
with the same cross-sectional area as the spliced reinforcing the key components of the analytical model. The analytical
bars. Figure 9 presents a typical stress-strain curve for mild model was developed in OpenSEES using force-based
steel reinforcing bars (solid line), which has an initial linear- distributed plasticity frame-elements with fiber sections at
elastic branch with slope Es, a yield plateau beginning at the each integration point. Because of the importance of bond-slip
yield strain ey, a nonlinear strain-hardening branch with a deformations, the method described by Haber et al.17 was used
slope Esh beginning at esh, and a plateau at the ultimate stress to incorporate the influence of the GC splice by including a
fu and corresponding strain eu. The stress-strain model for rotational spring at the column-footing interface. Two different
the splice, otherwise referred to as the “Proposed Model”, fiber-section assignments were used. Section “S1” was used in
is defined using the same constitutive model as the mild the frame element representing the shaft of the column above
reinforcing steel, but the characteristic parameters are the region with GC splices, and Section “S2” was used in the
scaled to achieve the desired SR response. The proposed region with the GC splices. For both sections, constitutive
curve representing the spliced region LSp is shown with a relationships for unconfined and confined concrete, and
dashed line and its parameters are identified with a “*”. The reinforcing bars were defined using available models from
parameters for the proposed curve are defined according to OpenSEES; namely “Concrete01,” “Concrete04,” and
Eq. (1) and (2), which correspond to the elastic and plastic “ReinforcingSteel,” respectively. Each constitutive model
portions of the stress-strain curve, respectively. Once tensile was calibrated using average measured materials properties.
testing has been performed, and the SR relationship has been The measured compressive strength of concrete on the day of
identified, the user can select appropriate elastic SRE and test was 4.7 ksi (32 MPa), and the average yield and ultimate
plastic SRI strain ratios for scaling the stress-strain curve for stress of longitudinal steel were 67 and 111 ksi (461 and
the splice model. Equation (3) must be used if the reinforcing 765 MPa), respectively. Confined concrete properties were
steel model includes a yield plateau to prevent calculation determined according to Mander’s model.21
errors as the stress state approaches yield. GC splices were defined in S2 as single fibers with the same
cross-sectional area as the No. 8 reinforcing bars. The actual
e*y diameter of GC device was approximately 2db – 2.5db, which
Es
= = SRE (1) corresponds to an area footprint of 4 to 6.25 times the bar
ey Es* area. Therefore, at each splice location, an equal-area segment
of confined concrete fibers were removed to account for the

186 ACI Structural Journal/March-April 2015


Fig. 11—Comparison of measured and calculated SR
relationships. Fig. 12—Comparison between measured and calculated
force-displacement response.
presence of the splice. The constitutive relationship for the GC
fibers was based on the previously described method of scaling displacement and lateral load, respectively. After 1.0% drift,
the reinforcing steel stress-strain properties. For this study, the measured data indicated initiation of strain hardening.
SRE and SRI were selected to be 1.0 and 0.26, respectively. There was very good correlation between the post-yielding
Strain data from the GCNP column test indicated that yielding parts of the calculated backbone curves and the measured
of longitudinal reinforcement first occurred at the column- curve. The calculated response including bond slip deviated
footing interface. In this model, the fibers at the column- slightly from the measured curved after 3.0% with the
footing interface employ effective properties, and therefore maximum difference between the measured and calculated
strains in the reinforcing bars cannot be calculated explicitly. lateral loads being 5.5% at 6% drift in the negative direction.
However, by setting SRE = 1.0, the first yield of longitudinal When bond slip was excluded, the maximum difference was
steel can be approximated. This is a reasonable selection 8.6%, which is still reasonable.
because GC-S tests indicated that the strain over the splice In general, the calculated force-displacement response
was approximately 75% of that from the reinforcing bar prior of the column showed good correlation with the measured
to yielding of steel. The ability to reasonably approximate the result using an effective constitutive relationship for GC
yield displacement of the column is critical for displacement- splices. In this study, a force-based distributed-plasticity
based designed methods, which require displacement ductility frame element model was used to validate the effective
calculations. Figure 11 shows a comparison between the material method for incorporating mechanical reinforcing
measured SR response and that calculated using the proposed bar splices within a ductile member. However, the proposed
method with SRE = 1.0 and SRI = 0.26. modeling method could also be used in moment curvature or
fiber section lumped plasticity analysis and can be applied to
Comparison with experimental results and other constitutive relationships for reinforcing steel using an
discussion approach similar to that described in this paper.
The frame element model shown in Fig. 12 was subjected
to monotonic lateral displacement twice, once in each SUMMARY AND CONCLUSIONS
direction because the column reinforcing pattern was not Two commercially available mechanical reinforcing bar
symmetric. Pushover analysis was conducted with and splices, namely an upset headed (HC) splice and a grout-filled
without the presence of bond slip due to strain penetration ductile cast-iron sleeve (GC) splice, where evaluated under
into the footing. The resulting pushover curves were uniaxial monotonic static and dynamic loading until failure,
compared with the measured hysteretic response. The slow reversed cyclic loading until failure, and a series of
calculated initial stiffness with bond-slip included was elastic slip tests. Analysis of results focused on characterizing
approximately the same as the measured stiffness. When the force-deformation characteristics of each device. Key
bond slip was excluded, the calculated initial stiffness was observations from uniaxial tests were then correlated with
slightly higher than the measured result, which is to be a series of half-scale bridge column models employing
expected. Nonetheless, the calculated and measured results mechanical splices in flexural plastic hinge zones. Local
were still comparable and the analytical model would be moment-rotation and global force-displacement relationships
adequate for design purposes. Circular markers identify the were presented for columns HCNP and GCNP, containing
measured and calculated first yield points of the longitudinal mechanical splices and for a conventional cast-in-place
steel. In the positive direction, there was very little column. Based on observations from uniaxial tests, a simple
difference between the measured and calculated results. In method for determining an effective stress-strain model was
the negative direction, the results were comparable with a proposed for mechanical splices. As an example, effective
6% and 17% difference in the measured and calculated yield uniaxial properties were established for the GC splice

ACI Structural Journal/March-April 2015 187


using the measured coupler behavior. Effective properties ACKNOWLEDGMENTS
were implemented in a nonlinear frame-element model for The research presented in this document was funded by the California
Department of Transportation (Caltrans) under contracts No. 65A0372
GCNP, and pushover analysis results were compared with the and 65A0425. The support and advice of M. Mahan, R. Bromenschenkel,
measured force-displacement response. Based on the results M. Keever, and T. Ostrom of Caltrans are appreciated. The authors would like
of this study, the following conclusions can be made: to thank Headed Reinforcement Corp. (HRC), Splice Sleeve Japan, and Splice
Sleeve North America for donating splices and bars used in this study. Special
Mechanical splices can significantly reduce the thanks are expressed to R. Nelson and C. Lyttle for their help with testing.
deformation capacity of spliced reinforcing bars by as much
as 75%, which can have a noticeable effect on the local REFERENCES
moment-rotation behavior of a ductile reinforced concrete 1. ACI Committee 439, “Types of Mechanical Splices for Reinforcing
member depending on the size and stiffness of the splice Bars (ACI 439.3R-07),” American Concrete Institute, Farmington Hills,
MI, 2007, 20 pp.
relative to the reinforcing bar. 2. ICC-ES AC133, “Acceptance Criteria for Mechanical Connector
The characteristic stress-strain behavior for HC and GC Systems for Steel Reinforcing Bars,” International Code Council Evalua-
splices is not significantly affected by strain rates expected tion Service, Whittier, CA, 2010, 9 pp.
3. ASTM A1034/A1034M-10, “Standard Test Methods for Testing
during earthquakes. These devices were able to sustain the Mechanical Splices for Steel Reinforcing Bars,” ASTM International, West
higher ultimate stress demands associated with the strain Conshohocken, PA, 2010, 5 pp.
rate effect on mild steel reinforcing bars without adverse 4. CT670, “Method of Tests for Mechanical and Welded Reinforcing
Steel Splices,” California Department of Transportation, Division of Engi-
effect on failure modes. neering Services, Sacramento, CA, 2011, 11 pp.
The gap formation between the heads of the HC splice, 5. ACI Committee 318, “Building Code Requirements for Structural
which significantly exceeded the elastic slip limits for Concrete (ACI 318-02) and Commentary,” American Concrete Institute,
Farmington Hills, MI, 2002, 443 pp.
AASHTO and Caltrans bridge design codes, does not have 6. American Assiciation of State Highway and Transportation Officials
an adverse effect on the force-displacement behavior of (AASHTO), “AASHTO Guide Specifications for LRFD Seismic Bridge
members containing these splices in plastic hinge zones. Design,” second edition, Washington, DC, 2011, 301 pp.
7. California Department of Transportation, “Seismic Design Criteria
The procedure proposed for determining the effective (SDC) Version 1.7,” Division of Engineering Services, Sacramento, CA,
stress-strain properties for a mechanical splice can be 2013, 180 pp.
employed using test results from standard test methods 8. Haber, Z. B.; Saiidi, M.; and Sanders, D. H., “Seismic Performance of
Precast Columns with Mechanically Spliced Column-Footing Connection,”
and acceptance criteria. Thus, proprietary geometric ACI Structural Journal, Vol. 111, No. 3, May-June 2014, pp. 639-650.
specifications and material properties for a mechanical 9. Paulson, C., and Hanson, J. M., “Fatigue Behavior of Welded and
splice are not required. Mechanical Splices in Reinforcing Steel,” NCHRP Report 10-35, Dec.
1991, 158 pp.
The calculated member response using the proposed 10. Rowell, S. P.; Grey, C. E.; Woodson, S. C.; and Hager, K. P., “High
model showed very good correlation with the measured Strain-Rate Testing of Mechanical Couplers,” Report ERDC TR-09-8, U.S.
test results with and without incorporation of bond-slip Army Corps of Engineers, Washington, DC, Sept. 2009, 74 pp.
11. Aida, H.; Tanimura, Y.; Tadokoro, T.; and Takimoto, K., “Cyclic
deformation. The calculate column load and displacement at Loading Experiment of Precast Columns of Railway Rigid-Frame Viaduct
first yield of longitudinal bars also were in close agreement Installed with NMB Splice Sleeves,” Proceedings of the Japan Concrete
with the measured data. Institute, V. 27, No. 2, 2005, pp. 613-618.
12. Culmo, M. P., “Connection Details for Prefabricated Bridge Elements
The proposed effective strain-strain model for mechanical and Systems,” Report FHWA-IF-09-010, Federal Highway Administration,
splices can be easily implemented in available software Washington, DC, Mar. 2009, 568 pp.
packages. 13. ASTM A370-03a, “Standard Test Methods and Definitions for
Mechanical Testing of Steel Products,” ASTM International, West Consho-
hocken, PA, 2003, 49 pp.
AUTHOR BIOS 14. Motaref, S.; Saiidi, M. S.; and Sanders, D. H., “Seismic Response
ACI member Zachary B. Haber is a Bridge Research Engineer with of Precast Bridge Columns with Energy Dissipating Joints,” Report No.
Professional Service Industries (PSI) at the Federal Highway Administration CCEER-11-01, Center for Civil Engineering Earthquake Research, Depart-
Turner-Fairbank Highway Research Center in McLean, VA. He received ment of Civil Engineering. University of Nevada, Reno, Reno, NV, 2011.
his BS and MS in civil engineering from the University of Central Florida, 15. Sadrossadat-Zadeh, M., and Saiid Saiidi, M., “Effect of Strain Rate
Orlando, FL, and received his PhD in civil engineering from the University on Stress-Strain Properties and Yield Propagation in Steel Reinforcing
of Nevada, Reno, Reno, NV. His research interests include large-scale Bars,” Report No. CCEER-07-02, Center for Civil Engineering Earthquake
testing, advanced materials in civil engineering, and bridge engineering. Research, Department of Civil Engineering. University of Nevada, Reno,
Reno, NV, 2007.
M. Saiid Saiidi, FACI, is a Professor of civil and environmental engineering 16. Otani, S., and Sozen, M. A., “Behavior of Multistory Reinforced
and the Co-Director of USDOT University Transportation Center on Concrete Frames during Earthquakes,” Structural Research Series No. 392,
Accelerated Bridge Construction-Seismic at the University of Nevada, Reno. University of Illinois, Urbana, IL, 1972, 551 pp.
He is the Founding and former Chair and a current member of ACI Committee 17. Haber, Z. B.; Saiidi, M.; Ou, Y. C.; and Sanders, D. H., “A Method for
341, Earthquake-Resistant Concrete Bridges, and a member of Joint Calculating the Seismic Response of Bridge Columns with Grouted Sleeve
ACI-ASCE Committee 352, Joints and Connections in Monolithic Concrete Column-Footing Connections,” Proceedings, Seventh National Seismic
Structures. He is also a member of ACI Subcommittee 318-D, Subcommittee Conference on Bridges & Highways, Oakland, CA., May 20-22, 2013.
on Flexure and Axial Loads (Structural Concrete Building Code). 18. Malvar, L. J, “Review of Static and Dynamic Properties of Steel
Reinforcing Bars” ACI Materials Journal, V. 95, No. 5, Sept.-Oct.1998,
David H. Sanders, FACI, is a Professor at the University of Nevada, Reno. He pp. 609-614.
received his BS from Iowa State University, Ames, IA, and his MS and PhD from 19. Fu, H. C.; Erki, M. A.; and Seckin, M., “Review of Effects of Loading
the University of Texas at Austin, Austin, TX. He a member of the ACI Board Rate on Reinforced Concrete,” Journal of Structural Engineering, ASCE,
of Direction and former Chair of the ACI Technical Activities Committee; ACI V. 117, No. 12, Dec. 1991, pp. 3660-3679.
Committee 341, Earthquake Resistant Concrete Bridges; and Joint ACI-ASCE 20. Mander, J. B.; Panthaki, F. D.; Kasalanati, A. “Low-Cycle Fatigue
Committee 445, Shear and Torsion. His research interests include concrete Behavior of Reinforcing Steel,” Journal of Materials in Civil Engineering,
structures with an emphasis on seismic performance of bridges. ASCE, V. 6, No. 4, 1994, pp. 453-468.
21. Mander, J. B.; Priestley, M. J. N.; and Park, R., “Theoretical Stress-
Strain Model for Confined Concrete,” Journal of Structural Engineering,
ASCE, V. 114, No. 8, 1988, pp. 1808-1826.

188 ACI Structural Journal/March-April 2015

You might also like