Engineering Structures 86 (2015) 58–71

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

A coupled connector element for nonlinear static pushover analysis

of precast concrete diaphragms
Ge Wan a, Dichuan Zhang b, Robert B. Fleischman c,⇑, Clay J. Naito d
a
SDL Structural Engineers, 550 Maryland Way, Suite 250, Brentwood, TN 37027, United States
b
Nazarbayev University, 53 Kabanbay Batyr Ave, Astana 010000, Kazakhstan
c
University of Arizona, 1209 E 2nd Street, Tucson, AZ 85721, United States
d
Lehigh University, 117 ATLSS Drive, Bethlehem, PA 18015, United States

a r t i c l e i n f o a b s t r a c t

Article history: This paper describes the formulation of a diaphragm connector element developed for use in two-dimen-
Received 6 March 2014 sional finite element (2D-FE) modeling of precast concrete diaphragms. The connector elements, com-
Revised 6 October 2014 posed of assemblages of standard element types readily available in most FE software package
Accepted 15 December 2014
libraries, are nonlinear, coupled for shear–tension interaction, enable friction mechanisms, and possess
Available online 10 January 2015
descending branch behavior. Element construction is based on data from full-scale tests of common pre-
cast diaphragm connectors. The 2D-FE models have been employed in nonlinear static ‘‘pushover’’ anal-
Keywords:
ysis of isolated floor diaphragms to determine diaphragm stiffness, strength, deformation capacity, and
Precast concrete
Floor diaphragms
limit state sequence. The use of discrete elements to model the precast diaphragm connectors permits
Nonlinear analysis the direct evaluation of local force and deformation demands acting on these details. Further, the coupled
Finite elements formulation is adaptable to complex force histories and deformation patterns in the floor diaphragm,
thereby permitting the element to respond in realistic fashion. The models, verified for accuracy using
large scale testing, are providing crucial information on capacity and limit states for calibrating perfor-
mance-based design factors for a new seismic design methodology for precast concrete diaphragms.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Since the collapse of several precast concrete structures due to

Floor diaphragms are often treated as rigid and sufficiently
strong in the analysis and design of building structures under seis-
mic excitation. This assumption cannot always be made for precast
concrete construction in which the floor system is composed of
series of individual precast floor units. A critical feature of these
diaphragms is the nature of the force transfer across the joints
between the floor units. The units are joined by connectors alone
or in concert with a thin reinforced topping slab. Thus, unlike the
distributed force transfer in monolithic floor slabs that serve as
diaphragms in reinforced concrete structures for instance [1],
forces in precast diaphragms are transferred at the discrete con-
nector locations. The floor joints serve as critical sections in the
precast diaphragm, and thus precast diaphragm behavior is highly
dependent on the characteristics of the connectors [2], including
impacting the diaphragm global properties [3] and local failure
modes [4,5].
⇑ Corresponding author.
E-mail address: rfleisch@email.arizona.edu (R.B. Fleischman).
failures of the floor systems in the 1994 Northridge earthquake [6],
steady progress has been made on understanding the behavior of
precast floor diaphragms for the purposes of improving their seis-
mic design. These advances in knowledge have been driven largely
through analytical and experimental research, culminating in a
project [7] developing a new precast diaphragm seismic design
methodology [8], currently in the codification process [9,10].
Diaphragm analytical models used in research, like those used
in the design practice, originally involved monolithic models,
including Bernoulli beam formulations [11], fiber element and
smeared crack models [12], and elastic plane stress finite elements
[13–15]. However, recognizing the limitations associated with
modeling precast diaphragms using monolithic models, precast
diaphragm finite element (FE) models with discrete representa-
tions of the connectors have been developed. Early versions of
these ‘‘discrete’’ diaphragm models included those capturing dia-
phragm flexure response only [12], and later flexure and shear
response as uncoupled degrees of freedom (DOF) [4,16].
For the current research [7], the discrete diaphragm models are
being used in extensive parameter studies to calibrate design

http://dx.doi.org/10.1016/j.engstruct.2014.12.029
0141-0296/Ó 2014 Elsevier Ltd. All rights reserved.
 G. Wan et al. / Engineering Structures 86 (2015) 58–71
Nomenclature
A area of the link element
Cu concrete peak compression force within effective
area
E concrete elastic modulus
0
fc concrete cylinder compressive strength
k, k2L, k3L initial, secondary, softening stiffness of link
k1,2 stiffness in the link element
ts vs
k2 , k2 secondary tension, shear stiffness of uncoupled
spring
kc contact stiffness
Kic, Kuc theoretical initial, secant compression stiffness of
concrete
cs cs
ki , ku initial, secant stiffness of compression spring
ts vs
ki , ki initial tension, shear stiffness of uncoupled spring
L length of the link element
n1,2 force in one of link element
ny yield strength of link element
T tension force
TY connector tension yield strength
t thickness of the plane stress element
ty tension yield strength of uncoupled spring
factors for the new diaphragm design methodology [8]. For this
work, a general ‘‘connector element’’ that can respond accurately
to complex force histories in realistic fashion is a key feature of
the discrete diaphragm model. This paper describes the formula-
tion of such a connector element for two-dimensional finite ele-
ment (2D-FE) representations of the precast concrete floor
system. The elements are nonlinear, coupled for shear and tension
interaction, enable friction mechanisms, and possess descending
branch behavior. The connector elements are composed of assem-
blages of standard elements readily available in most FE software
package libraries, and are feasible for large DOF models.
The formulation is derived generically in this paper. Element
construction is then demonstrated for a number of common pre-
cast diaphragm connectors. The element construction depends on
data from full-scale tests performed on an assortment of common
precast connectors [17,18], tested as part of the overall research
effort [7], and supplemented with data from existing tests of pre-
cast connectors [19–21]. The resulting element response is com-
pared to physical tests and a comparable user-defined element.
The 2D-FE discrete models have been used extensively in the
research project for nonlinear static analyses of isolated dia-
phragms under increasing monotonic in-plane inertial forces to
determine the precast diaphragm stiffness and strength character-
istics, deformation capacity, and limit states sequence [22]. Such
analyses are similar to so-called ‘‘pushover’’ analysis [23] used to
evaluate the capacity of the lateral force resisting system (LFRS)
e.g. shear walls, moment frames, and this same terminology is
adopted here. The importance of using the coupled element in
the 2D-FE discrete model is demonstrated in pushover analyses
for a number of common diaphragm layouts at the end of this
paper.
Determining diaphragm capacity through pushover analyses is a
key initial step in developing the performance-based seismic design
methodology. A complementary step, determining anticipated dia-
phragm seismic demands, is accomplished via three-dimensional
(3D) nonlinear dynamic time history analysis [24]. The extension
of this connector element for use in the 3D-FE model, and the ver-
ification of these models using large scale testing of joints and
structures is presented in [25–27] respectively.
59
V shear force
VY connector shear yield strength
V CY connector shear yield strength under compres-
sion
vy shear yield strength of uncoupled spring
D T, D V tension, shear deformation
DTOT Total deformation
DTY, DVY tension, shear yielding deformation
DTYf, DTYs tension full, soften yielding deformation
DuT,red, DuV,red reduced ultimate tension, shear deformation
DVYf, DVYs shear full, soften yielding deformation
jt, jv tension, shear soften stiffness ratio
DuT, DuV connector ultimate tension, shear deformation
d1,2 deformation in one of link
du ultimate deformation of link
dut, duv ultimate tension, shear deformation of uncoupled
spring
l full yielding-to-soften yielding ductility ratio
ls coefficient of friction for contact element
h orientation angle of line element
2. Precast diaphragms
2.1. Precast diaphragm design
Diaphragms are reinforced to carry the in-plane shear, flexure
and anchorage forces associated with seismic loading. This action
is required in order to provide a complete load path for the inertial
forces to reach the primary elements of the LFRS, i.e. shear walls or
moment frames [1]. In precast structures, reinforcing of the dia-
phragm involves placing connectors between the precast floor
units. Transfer is made solely by these connectors in an untopped
diaphragm [28,29]. In a topped (composite) diaphragm, a thin cast-
in-place topping also participates, but since the diaphragm cracks
along the joints, the force transfer mechanism is similar.
Fig. 1(a) shows a simple diaphragm schematic with typical floor
diaphragm dimensions and connector layouts, indicating the con-
nectors intended to transfer: (b) in-plane shear force between the
units; (c) chord forces associated with in-plane flexure; and (d)
collector/anchorage forces to walls and frames. Typical diaphragm
connector layouts and connector dimensions are shown in the figure.
2.2. Precast diaphragm connectors
Element construction for the 2D-FE connector models is dem-
onstrated in this paper for a set of commonly-used flange-to-flange
connectors [see Fig. 1(b)–(d)]: (b) a JVI Vector1 for shear reinforce-
ment; (c) a dry chord connector for flexure reinforcement; and (d)
angled bar-plate connectors, often used one-sided [as shown in
Fig. 1(d)] to connect precast floor units to beams and walls. It is
noted that a wide variety of connectors are used in practice to pro-
vide precast diaphragm reinforcement, including standard industry
hardware and proprietary connections [3]. These particular connec-
tors are chosen as they: (1) have traditionally found widespread use;
(2) have been extensively tested [17–21]; (3) together provide the
complete set of reinforcement required for an untopped precast floor
diaphragm; and, (4) transfer force similarly to most connections, i.e.
1
Proprietary precast flange-to-flange connector, JVI, Inc. Lincolnwood, IL 60712.
60 G. Wan et al. / Engineering Structures 86 (2015) 58–71

Fdia
(a) (b) 6.4 102mm (c) 8 216mm (d) 8 216mm

Anchorage Beam
[see Fig. (d)] Shear 1~2.5m or wall
102mm
connectors 152mm
[see Fig.(b)]
432mm #5 317mm
Chord Shear
Collector connectors wall
10~30m [see Fig. (c)] U
P
FO
R Precast Face plates
units 216x51x9.6mm #3

5~15m 1.2~4.6m Slug PL 85x25.4x9.5mm Slug PL 216x38x9.5mm

Fig. 1. Precast diaphragm: (a) plan (mod. from [22]); (b) JVI Vector; (c) dry chord; and (d) angled bar-plate.

(a) (2) Compression (4) Shear with compression (b)

Fdia
Fdia

N.A.
V N

MV M
Collector/
Confinement
(5) Shear (1) Tension (3) Shear with tension

Fig. 2. Precast diaphragm schematic: (a) internal force condition and (b) combined forces on precast units.

through anchorage mechanisms created by bars extending into the
precast unit, and using field-welds to complete the connection
between panels.
2.3. Diaphragm connector force demands
Consider Fig. 2(a), a schematic of a simple precast diaphragm
under lateral loading in the same direction as shown in Fig. 1(a).
Note that due to a significantly higher connector compressive stiff-
ness relative to the tension stiffness (due to contact), the neutral
axis is not at the centroid [22]. Indicated on the schematic is the
different instantaneous internal force conditions to which connec-
tors may be subjected at various discrete transfer points within the
diaphragm. Fig. 2(a) implies that in addition to individual force
components, shear and axial force combinations can occur. Like-
wise, for more realistic configurations, joints can be subjected to
some combination of moment, axial and shear forces under hori-
zontal earthquake loads, leading to combined shear and axial
actions on individual connectors [see Fig. 2(b)]. These force combi-
nations can occur in critical regions of the diaphragm under
earthquake loading [26], particularly when collector actions [‘‘N’’
in Fig. 2(b)] are also present. Thus, two combined-force conditions
must be considered in testing and modeling of the connector: (1)
response under combined tension and shear; and, (2) shear
response in the presence of axial compression.
The typical manner in which a connector transfers shear and ten-
sion is illustrated in Fig. 3 for a generic steel strap plate with two 45°
anchorage legs embedded in a precast unit. Indicated are contours
of principal stresses from a numerical analysis of a continuum FE
model. Grey indicates crushing of concrete and yield in steel. The
plots underscore the distinctly different load transfer mechanisms
that occur under each force component [30]. The mechanisms
involved are complex and vary significantly based on loading,
including local crushing of the concrete at the return of the faceplate
and flexural, tensile, and compressive response of the strap plate.
The connector illustrated is relatively simple; most connection
details possess greater geometric complexity. Continuum modeling
of connectors does not lend itself for models of an entire diaphragm
or structure. Consequently it is advantageous to utilize simplified
numerical models to capture connector performance.

Fig. 3. Precast diaphragm connector anchorage and local stress mechanisms: (a) shear and (b) tension.
 G. Wan et al. / Engineering Structures 86 (2015) 58–71
Shear only test
(a) 80
Tension only test
70
60
Force (kN)
50
40
30 Tension
Shear
20 Tension-shear test
10
0 0
0 0.2 0.4 0.6
Deformation (cm)
Fig. 4. Connector response (modified from [19,17]): (a) tension/shear and (b) tension/compression.
Consider next connector response under combined tension and
shear. As seen in the monotonic connector tests results [19] of
Fig. 4(a), precast connectors exhibit strength and deformation
capacity reductions under combined tension and shear loading,
relative to a connector subjected to only tension or only shear load.
Finally consider the shear response under axial compression [17]
shown in Fig. 4(b). As seen, precast connectors exhibit shear stiff-
ness and strength increases when axial compression is present.
Note that this can be shear–friction [31] in a topped-precast dia-
phragm, or simply direct friction behind the face plate [refer to
Fig. 1(c)] in an untopped diaphragm [28].
3. 2D-FE discrete model
3.1. Description of discrete model
Fig. 5 shows an example of the 2D-FE discrete model, in this
case for the top level diaphragm of a precast parking structure. In
the model, precast units are modeled as elastic plane stress ele-
ments while the connectors between precast units are modeled
as assemblages of nonlinear springs, link elements and contact ele-
ments. Thus, the intent of the modeling is that the nonlinear
response due to yielding, slip, and geometric change (e.g. catenary
action or buckling) in the steel connectors and reinforcing bars, and
degrading action caused by weld tearing or local damage to the
surrounding concrete (cracking, crushing, and spalling), all be cap-
tured in the analysis within the connector elements.
Key advantages of the discrete diaphragm model include the
ability to: (1) map individual connector demand directly to global
diaphragm response; and (2) simulate localized mechanisms that
can lead to global non-ductile failure modes. Further, the 2D-FE
discrete model has the ability to provide a realistic representation
of the precast floor global boundary conditions (e.g. confinement
perpendicular to the joint) and the local deformation patterns
(e.g. concentrated deformations, joint opening not conforming to
(a)
(c)
(b)
11m
55.6m
3.7m
91.4m
Fig. 5. 2D-FE discrete model: (a) parking structure (courtesy of Wm. Blanchard Co.); (b) typical diaphragm layout; (c) full model; and (d) close-up of panels and joints.
61
(b) 120 Shear under 44.5kN
100 compression
Shear (kN)
80
60
40
20 Pure shear
0.8 1 0 0.2 0.4 0.6 0.8 1
Sliding (cm)
Bernoulli assumptions, etc.) as the precast diaphragm responds
to seismic loading. This feature is essential in the modeling of pre-
cast diaphragms in realistic structures subjected to a ground
motion of uncertain attack angle. In such cases, complexity/irregu-
larity in both floor system and LFRS layout (frames or walls) will
place the diaphragm connectors under simultaneous and varying
non-proportional combinations of tension/compression and shear
conditions that are amplified in the presence of bi-directional
earthquake components [26].
3.2. Rationale for modeling approach
The calibration of design factors for the new diaphragm design
methodology involves extensive parameter studies [8]. These stud-
ies require versatile parameter modification and agile handling of
spatial geometry and complex boundary conditions in the creation
of large DOF models; robust computational engines and reasonable
run times, and effective post-processing. The process was thus
greatly facilitated by the incorporation of the discrete diaphragm
models within commercial FE software packages. These packages
possess powerful pre-processors with graphical user interfaces
(GUIs) for complex model troubleshooting and results visualiza-
tion; powerful solvers suitable for models with complex geometry,
large DOFs and high nonlinearity; and efficient post-processing of
large data sets. This approach, the most viable option at the pro-
ject’s outset, is still considered so by the research team now for
the reasons given, despite advances in open source computing,
e.g. [33], producing more available research tools.
The discrete force transfer required for the diaphragm connec-
tor element does not lend itself to continuum elements (e.g. with
a multi-dimensional principal stress yield criterion [34]), and
instead can be achieved using a custom element (e.g. user pro-
grammable features [34] or a multi-DOF spring elements with
user-defined material properties [33]), or at the macro level using
the standard element library available in commercial FE packages.
Macro modeling approaches have been shown to be effective in the
(d)
Diaphragm
connectors
Precast units: elastic
plane stress element
Shear
Boundary conditions connector
representing LFRS
Chord connector
62 G. Wan et al. / Engineering Structures 86 (2015) 58–71
(a)
(b) Force
Fig. 6. Connector tests: (a) setup (after [3]) and (b) qualification backbone (modified from [32]).
past, including diagonal truss elements to model the flexural–shear
coupling response of RC shear walls [35] and 3D diagonal trusses to
model the in-plane and out-of-plane coupling response of masonry
infill walls [36]. The macro model approach was chosen for this
research primarily to facilitate timely results for the applied
research program, where the model can be developed on top of
the existing FE program without significant computer coding or
compiling effort. This approach also provides versatility, since the
model is easily incorporated into the FE programs using their stan-
dard material and element library, and thus should be operable on
different programs (e.g. ANSYS, ABAQUS, ADINA,2 etc.) and plat-
forms. The first of these software packages is used for the material
presented in this paper. The incorporation of the discrete model into
user defined elements is also discussed later in the paper.
3.3. Discrete connector element
An effective connector element must possess an accurate repre-
sentation of connector stiffness, strength and deformation capac-
ity, both for shear and axial response. However, since connectors
in precast diaphragms are often subjected to these shear and axial
force actions simultaneous and non-proportionally [26], including
at critical diaphragm regions, and exhibit modified characteristics
in these conditions, modeling both the tension–shear coupling
effect and the friction mechanism associated with compression
in the precast diaphragm connector element becomes essential.
Because the FE model has the ability to provide a realistic rep-
resentation of floor boundary conditions and local deformation
patterns that develop under seismic load, a general ‘‘connector ele-
ment’’ that can respond accurately to complex force histories in
realistic fashion, rather than being prescribed a priori, becomes a
key feature of the discrete diaphragm model with general flex-
ure–shear–axial coupling capabilities. These elements must also
be feasible for insertion into precast diaphragms models with doz-
ens to hundreds of these connectors.
As such, the 2D-FE discrete modeling technique has the ability
to more accurately capture the local demands on individual precast
connectors or key diaphragm joints in an earthquake, provided the
connector element model can replicate the key behaviors exhibited
by the precast diaphragm reinforcement. Three steps are needed to
accomplish this goal: (1) An experimental program that establishes
the needed characteristics of a given connector; (2) The creation of
an element that can distinguish and respond to the conditions
2
ANSYS (v11), Canonsburg, PA; ABAQUS 6.10. Dassault Systems. Waltham, MA;
ADINA (v9), Watertown, MA.
Tension behavior
Vy Tu
K2T
Ty
Type 1
Type 2
KiT K2V
K3T
KiV
Shear behavior
(Type 3)
ΔuV ΔuT ΔuT Deformation
described above; and (3) The derivation of algorithms that convert
the test data into model input. These steps are described in this
paper.
4. Connector testing program
The connector element models were developed in conjunction
with a comprehensive testing program on common precast dia-
phragm connectors [17,18], and also make use of results from pre-
vious testing [19–21]. The test setup for the former tests is shown
in Fig. 6(a) and fully described in [3,17,18]. As seen in Fig. 6(a), the
tests are performed on isolated connectors anchored in concrete
representing half-spaces of the surrounding precast panels, in a
test frame that can independently control shear, moment, and axial
forces or displacements. The purpose of these tests was to: (1)
qualify precast diaphragm connectors for use in the design meth-
odology; and (2) determine characteristics for construction of pre-
cast diaphragm FE models. A distinction is drawn in the test
program between those tests required for connector qualification
in the new precast diaphragm seismic design methodology [8]:
Connection Qualification (CQ) tests, and supplemental tests per-
formed to calibrate the models during the connector element for-
mulation: Model Calibration (MC) tests.
4.1. Connector Qualification (CQ) Tests
The testing program produced a set of qualification protocols
for precast connectors [32]. Fig. 6(b) shows the connector qualifica-
tion backbone for tension and shear response, based on FEMA-356
curve typology [23]. It is noted that for precast connectors, shear
response is typically associated with a FEMA-356 Type 3 curve
(force-controlled non-ductile behavior), while tension response is
typically associated with a FEMA-356 Type 1 or 2 curve (deforma-
tion-controlled ductile behavior) [23]. The curves are constructed
from the cyclic test envelope and express the key characteristics
of the connector in tension and shear: initial stiffness (KiT, KiV),
yield strength (TY, VY), secondary stiffness (K2T, K2V), deformation
capacity (DuT, DuV), and, for Type 1, peak strength (Tu) and soften-
ing stiffness (K3T). Dashed lines represent response beyond the
range of expected demand of properly designed precast
diaphragms.
The upper part of Table 1 lists the tests used to populate values
for the qualification backbone curves: (1) a cyclic tension test with
zero applied shear force; and (2) a cyclic shear test with zero ten-
sion deformation. These tests, performed to qualify connectors
within the design methodology framework [8], also produce data
 G. Wan et al. / Engineering Structures 86 (2015) 58–71 63

Table 1
Characteristics measured in isolated connector tests.

Test series Load protocol Characteristic

Initial stiffness Yield strength Secondary stiffness Deformation capacity
Connector Qualification (CQ) Tension load only KiT TY, Tua K2T, K3Ta DuT
Shear load only KiV VY K2V DuV
Model Calibration (MC) Shear w/tension – V VT VT
Y , TY
– –
Shear w/compression – V CY – –

a
Type 1 connector.

that serve as the basic information to create connector element
models. The parameters that describe the uncoupled tension char-
acteristic {KiT TY Tu K2T K3T DuT} are obtained directly from the first
test. The four parameters that describe the uncoupled shear char-
acteristic {KiV VY K2V DuV} are obtained directly from the second
test. In the past, this set fully described response [4]. In this paper,
these parameters are considered as the basic set of characteristic
values to build a more advanced model.
4.2. Model Calibration (MC) tests
A set of supplemental tests was performed in the research pro-
gram (see Table 1, lower portion). These tests, external to the qual-
ification procedure, were used to calibrate the connector elements
for the 2D-FE discrete model. The tests subjected isolated connec-
tors to: (1) shear in the presence of constant axial compression;
and, (2) combined tension and shear loading. The former is used
to determine an effective coefficient of friction; the latter to cali-
brate tension–shear coupling within the connector element.
Two important ratios are identified for model calibration tests
under combined shear and tension: (1) the applied tension-to-
shear force ratio a, a = T/V; and (2) the applied tension-to-shear
deformation ratio k, (k = DT/DV). These values do not typically coin-
cide in response due to differences in tension and shear stiffness.
Thus, one of these ratios is enforced in testing, either a for force
control [19], or k for deformation control [17,18], and the other
ratio measured through instrumentation.
5. Model formulation: response to individual load components
The precast connector element for the 2D-FE discrete model is a
nonlinear shear–tension coupled friction-enabled element. It is
composed of assemblages of standard elements readily available
in most commercial FE software package libraries [34]. Fig. 7(a)
shows the precast diaphragm connector element (extracted from
a 2D-FE discrete model, e.g. Fig. 5; discretization within the 2D-
FE model is detailed in Section 7.3). The configuration shown,
selected after careful evaluation of several possible configurations,
was found to be the most capable of reproducing observed behav-
ior in tests.
The individual components of the connector element are iden-
tified in Fig. 7(a) for Type 2 and Type 3 connectors. The properties
for each component are derived to match the characteristic values
obtained in the testing. The connector element is constructed from
an assemblage of three components, each being a different type of
finite element [34]: (1) inelastic link elements, inclined in order to
capture the coupled tension and shear behavior; (2) inelastic ten-
sion and shear springs in parallel with the link elements to par-
tially uncouple the response as observed in the testing; and, (3) a
contact element with friction capabilities to mobilize the effect
of applied compression on shear resistance and an inelastic com-
pression spring in series to capture compression limit states (soft-
ening/crushing) of the surrounding concrete. The spring and
contact element are zero-length elements taking direct stiffness
input; the link element has finite length and thus is provided with
an equivalent axial rigidity (see Section 7.3).
The characteristic backbones are shown in: Fig. 7(b) for those
components producing tension and shear response; and Fig. 7(c)
for compression. The derivation of the component properties to
produce appropriate connector behavior follows using small defor-
mation assumption. The derivation uses the following notation:
small letters indicate the response of individual components of
the connector element; capital letters indicate the global degrees
of freedom (DOFs) of the overall connector element; and, bold cap-
ital letters indicate the experimental data derived from tests on
isolated connectors (see Table 1).
5.1. Coupled link formulation
The tension–shear coupled behavior is modeled through the
inclined link components. It will be instructive to consider first a

Link element
Force Compression
Compression Link element Contact
spring vy element
Contact k2L kc
element cu
ny Compression
k2ts spring
ty Inelastic

k2vs
compression
Tension spring kits d eff
UP
FOR

spring
0.5cu
θ k Tension Topped
diaphragm
d eff
Shear kics
UP
FOR

ki vs spring ku cs UP
FO
R

shear spring spring UP

Contact
element
Untopped
diaphragm
FOR

δuv δu δut
Panel
b
Center

Deformation Deformation
(a) (b) (c)
Fig. 7. Connector element: (a) components: (b) shear and tension; and (c) compression.
64 G. Wan et al. / Engineering Structures 86 (2015) 58–71
pair of angled links acting alone, with geometry as shown in
Fig. 8(a). The relationships are derived here for single displacement
components (DV, DT) as would be applied in the CQ tests listed in
Table 1.
The relationship between the connector global stiffness,
strength and deformation capacity characteristics {KiT KiV TY VY
DuT DuV} and the properties of the individual links (axial stiffness
k, strength ny, and deformation capacity du), are obtained through
contragredient laws [see Fig. 8(b) and (c)]:
2 5.2. Partially-coupled formulation
K iT ¼ 2k sin h T Y ¼ 2ny sin h DuT ¼ du = sin h
K iV ¼ 2k cos2 h V Y ¼ 2ny cos h DuV ¼ du = cos h
Connector symmetry dictates that links 1, 2 have the same nominal
properties, e.g. k1 = k2 = k; etc., and thus the internal forces in the
derivation are identical in magnitude and with same and opposing
senses for shear and tension respectively (see insets in Fig. 8).
Parameters h and k, needed to establish the elastic properties of
the coupled link element, can be obtained by combining Eqs. (1a)
and (2a), and setting the connector characteristics to the corre-
sponding values measured in the CQ tests (see Table 1):
h ¼ tan1 ½ðKiT =KiV Þ1=2 
2
k ¼ KiT =2 sin h
In this way, a single value of k in the links will produce the desired
connector shear and tension stiffness via the parameter h, an
approach that is adequate for elastic models. However an over-con-
strained solution exists for nonlinear connector models regarding
strength (Eqs. (1b) and (2b)) and ductility (Eqs. (1c) and (2c)), since
there is no certainty that the selected h will also fulfill:
h ¼ tan1 ½ðTY =VY Þ
h ¼ tan1 ½ðDuV =DuT Þ
Many precast connectors resemble the two link assemblage shown
in Fig. 8(a) [e.g. refer to Fig. 1(b) and (d)], and thus it might be first
thought that the strength and deformation capacity of the connec-
tor may well be approximated by Eqs. (4a) and (4b). However, the
precast connector tension and shear response can be markedly dif-
ferent because the presence of the surrounding concrete creates
boundary conditions due to bearing, local concrete crushing, deb-
onding, etc., that produce dissimilar gage lengths for tension and
shear (refer to Fig. 3), and these conditions can differ in the elastic,
yielded and near ultimate state [17]. Thus, for a h determined using
Eq. (3a), parameters ny and d can typically only accurately specify an
n1
link1
θ Inelastic θ n2
link: k , n y
ΔV
n2= -n1= n= k Δ V cosθ
θ link2 θ
(a)
Fig. 8. Coupled link kinematics: (a) model geometry; (b) shear, and (c) tension transformation.
element strength and deformation capacity for one of the force
components (e.g., tension), and will produce inaccurate measures
for the other (e.g. shear). Alternately, h can be determined using
Eq. (4a) instead of Eq. (3a) thereby enforcing accurate connector
strength at the expense of accurate initial stiffness. These limita-
tions render a directly proportional coupled connector element
not practical for diaphragm nonlinear pushover analyses. A par-
tially-coupled formulation is instead adopted, described next.
ð1a; b; cÞ
The partially coupled formation is achieved through the addi-
ð2a; b; cÞ
tion of uncoupled inelastic shear and tension springs to the cou-
pled link elements, as shown in Fig. 7(a). In this approach, three
parameters are available for matching the CQ test results: (1) the
angle h; (2) a tension strength ratio for the uncoupled tension
spring (xt = ty/TY); and (3) a shear strength ratio for the uncoupled
shear spring (xv = vy/VY).
In the research applications [22,24,26], the uncoupled inelastic
springs employed a multi-linear characteristic to improve accuracy
along the entire backbone curve. For simplicity, the derivation that
follows is based on uncoupled springs with a bilinear response
ð3aÞ characteristic, as the formulations follow the same basic approach.
The derivation uses yield deformation capacity to determine link
ð3bÞ orientation (as opposed to matching stiffness or strength), consis-
tent with the element derivations in the research application, as it
was found straightforward and produced sufficiently accurate
results (see Section 8):
The link orientation angle h is determined by aligning the con-
nector tension and shear yield deformation measured in the CQ
tests (DyT = TY/KiT, DyV = VY/KiV):
ð4aÞ h ¼ tan1 ½ðVY Ki;T =TY Ki;V Þ ð5Þ
The strength parameters xt and xv are selected to match the
ð4bÞ strength reduction under tension–shear coupling load observed
in MC tests, or an assumed level of coupling in the absence of such
tests.
Once the strength ratio parameters are established, the input
properties of the link element (yield strength ny and elastic stiff-
ness k) can be determined by modifying Eqs. (1) and (2):
ny ¼ ð1  xt ÞTY =2 sin h ¼ ð1  xv ÞVY =2 cos h ð6aÞ
2
k ¼ ð1  xt ÞKiT =2 sin h ¼ ð1  xv ÞKiV =2 cos2 h ð6bÞ
It is important to recognize that in order to produce a unique
solution for ny and k in Eq. (6), the ratio of xt and xv has to meet:
V n1
θ T
n2
δ 1= -Δ V cosθ δ 1= Δ T sinθ
δ 2= Δ V cosθ
δ 2= Δ T sinθ
ΔT
n2= n1= n= k Δ T sinθ
θ
(b) (c)
 G. Wan et al. / Engineering Structures 86 (2015) 58–71 65

ð1  xv Þ=ð1  xt Þ ¼ TY cos h=VY sin h ð7Þ Eqs. (10)–(12) express the equilibrium, constitutive and kine-
matic transformation relationships respectively, under small
Likewise, the input properties for the uncoupled springs are also
deformation for the individual link elements with respect to the
straightforwardly established, directly from the definition of the
global connector DOFs. Note the forces in the link elements are
strength ratio parameters:
no longer equal and are designated n1 and n2.
ts
t y ¼ xt TY ki ¼ xt KiT ð8a; bÞ
ðn1 þ n2 Þ cos h ¼ V ð10aÞ
vs
v y ¼ xv V Y ki ¼ xv KiV ð8c; dÞ
ðn1 þ n2 Þ sin h ¼ T ð10bÞ
Eqs. 9(a–f) indicate how post-yield properties are determined.
For typical connector response [refer to Fig. 6(b)], post-yield n1 ¼ k1 d1 ¼ kd1 ð11aÞ
tension response exhibits strain hardening behavior while the
post-yield shear response typical involves strength degradation n2 ¼ k2 d2 ¼ kd2 ð11bÞ
(negative stiffness). As such, the secondary stiffness and deforma-
tion capacity of the link elements were selected to mimic the
d1 ¼ DT sin h  DV cos h ð12aÞ
tension response (Eqs. (9a) and (9b)), while the descending branch
behavior was captured in the uncoupled shear spring (Eq. (9d)):
d2 ¼ DT sin h þ DV cos h ð12bÞ
2
kjL ¼ ð1  xjt ÞKjT =2 sin h j ¼ 2; or 3 ðfor Type I onlyÞ ð9aÞ
The derivation of response under combined loads focuses on
strength reduction. Considering first the force interaction, solving
du ¼ DuT = sin h ð9bÞ Eqs. (10a) and (10b) for n1 and n2 provides:
ts
k2 ¼ x2t K2T ð9cÞ n1 ¼ 1=2 T= sin h  1=2V= cos h ð13aÞ

K 2v s ¼ K2V  2k2L cos2 h ð9dÞ n2 ¼ 1=2 T= sin h þ 1=2V= cos h ð13bÞ

Eq. (13) indicates that for positive values of both force compo-
dut ¼ DuT ð9eÞ
nents (V > 0, T > 0), link 2 will yield first. Accordingly, substituting
the tension-to-shear force ratio a into Eq. (13b) and rearranging
duv ¼ DuV ð9fÞ
provides the relationship between global and local force:
ts
nu ¼ ½Tu  t y  ðT Y =K iT þ T u =K 2T Þk2 =2 sin h ð9gÞ V ¼ 2n2 ½ðsin h cos hÞ=ða cos h þ sin hÞ ð14Þ

where x2t is the equivalent of xt in the post-yield range, which will
not necessarily be the same and can be approximated as zero; and
k3L and nu are softening stiffness and peak strength needed only for
Type 1 connectors. Note that for the condition shown in Fig. 6, K2V is
negative while K2T is positive. Thus, Eq. (9a) produces a positive k2L,
and hence Eq. (9d) will always produce negative stiffness for this
case.
6. Model calibration: response to general load trajectories
6.1. Strength reduction
Strength reduction under combined loading is used to calibrate
the xt and xv parameters. Consider first the mechanics of inclined
links subjected to a trajectory of shear and tension [see Fig. 9(a)].
Introducing the constant Ca = 1 (a cos h + sin h), substituting
n2 = ny (n1 < ny), based on the yielding of link 2, and finally normal-
izing by the direct (unreduced) yield loads (Eqs. (1b) and (2b)) pro-
vides the reduced tension and shear strength ratios:
V Y;red =V Y ¼ C a sin h ð15aÞ
T Y;red =T Y ¼ aC a cos h ð15bÞ
The resulting strength reduction is shown in Fig. 10(a) for dif-
ferent connector element link angles. Note that this plot indicates
that a steeper angle can be used for connectors whose response
exhibits a higher reduction in tension strength due to the presence
of shear load and a shallower angle for those whose response
exhibits a higher reduction in shear strength due to the presence
of tension load.

θ Link2
δ 1< δ y δ 1= δ y
V (Δ V)
T/V=α
(ΔT /ΔV = λ)

T (Δ T)
Link1
V

n1 δ 2= δ y δ 2> δ y
θ T

n2

(a) (b) (c)

Fig. 9. Combined tension and shear: (a) layout; (b) yield of link 1; and (c) yield of both links.
66 G. Wan et al. / Engineering Structures 86 (2015) 58–71

1 ω t =1.0 ω v =1.0
TY,red /TY , θ =30º,45º,60º

0.8 ω v =0.75

Strength reduction
ω t =0.75
0.6 ω v =0.5
ω t =0.5
0.4 ω v =0.25

ω t =0.25 VY,red /V Y
0.2 ω v =0
TY,red /TY
V Y,red /VY ,θ =30º,45º,60º
0 ω t =0
0 1 2 3 4 5 0 1 2 3 4 5
α =T/V α =T/V
(a) (b)
Fig. 10. Strength reduction: (a) for h with xv = xt = 0 and (b) for xv and xt with h = 45°.

The relative strength reduction in shear and tension for a given The same process used to create Eq. (17) can be repeated for
orientation angle h, which is fixed in Eq. (15) for a given a [as link 1 by instead using Eqs. (11a) and (12a), thereby providing
shown in Fig. 10(a)], can be further adjusted by the introduction the normalized ‘‘fully’’ yielded condition:
of the uncoupled shear and axial springs with properties xv and
DVYf =DVY ¼ cos h=ðk sin h  cos hÞ ð18aÞ
xt. For this case, Eqs. (15a) and (15b) become:
V Y;red =V Y ¼ ð1  xv ÞC a sin h þ xv ð16aÞ DTYf =DTY ¼ k sin h=ðk sin h  cos hÞ ð18bÞ
The normalized softening and fully-yielded deformations of Eqs.
T Y;red =T Y ¼ ð1  xt ÞaC a cos h þ xt ð16bÞ
(17) and (18) are shown in Fig. 11(a) for h = 45°. It is seen that the
Fig. 10(b) shows the Eq. (16) relationships for h = 45° with xv shear softening deformation decreases and tension softening
and xt varying from 0 (fully coupled) to 1 (fully uncoupled). deformation increases with k, and both are less than the unreduced
yield deformations. The total resultant softening deformation
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
6.2. Post yield behavior
Dy;TOT ¼ D2VYs þ D2TYs is seen to be reduced slightly for combined
In the post-yield regime, where response is deformation con- trajectories.
trolled, it is more useful to define the coupled response in terms The region between the softening and fully-yielded deformation
of the displacement ratio k. Note that for any trajectory other than is characterized by a secondary (reduced) stiffness, even for link
outside of the bounding cases of pure shear (k = 0) or tension elements provided with elastic-perfectly plastic material response.
(k = 1) where the link elements yield simultaneously, one link will Relationships are written in incremental form. Consider a positive
yield prior to the other, thus a distinction is made between a ‘‘soft- increment of deformation components (@ DV > 0, oDT > 0), with a
ening’’ state [one link yielded, see Fig. 9(b)] and the fully yielded perfectly plastic link 2 yielding first, producing the state (k1 = k;
state [both links yielded, see Fig. 9(c)]. For the fully coupled case k2 = 0):
shown in Fig. 9(a) with positive deformation component values
ð@ n1 þ @ n2 Þ cos h ¼ @V ð19aÞ
(DV > 0, DT > 0), Eqs. (11) and (12) indicate that link 2 will yield
first. Inserting Eq. (11b) into Eq. (12b), setting n2 = nY, and substi-
ð@ n1 þ @ n2 Þ sin h ¼ @T ð19bÞ
tuting the trajectory parameter k, provides the following ‘‘soften-
ing’’ deformations, here normalized by the unreduced yield
@ n1 ¼ k1 @d1 ¼ k@d1 ð20aÞ
deformations (DVY = VY/KiV and DTY = TY/KiT):
DVYs =DVY ¼ cosh=ðk sin h þ coshÞ ð17aÞ @ n2 ¼ k2 @d2 ¼ 0 ð20bÞ

DTYs =DTY ¼ k sin h=ðk sin h þ cos hÞ ð17bÞ @d1 ¼ @ DT sin h  @ DV cos h ð21aÞ

2 Shear softening 5
Normalized yield deformation

Tension softening μ
Deformation or stiffness ratio

Total softening 4 κv
Shear fully-yielded
1.5 Tension full yield κt
3
λ =tanθ
1 2

1 λ =tanθ
0.5 θ =45 º
0
0 2 4 6 8 10
-1
0 θ =45 º
0 2 4 6 8 10
-2
λ=ΔT /ΔV λ=ΔT /ΔV
(a) (b)
Fig. 11. Staged yielding condition: (a) normalized yield deformation and (b) softening stiffness.
 G. Wan et al. / Engineering Structures 86 (2015) 58–71
@d2 ¼ @ DT sin h þ @ DV cos h ð21bÞ
Substituting the values in Eq. (20) into Eq. (19), and in turn
combining with Eq. (21) provides the secondary stiffness of the
connector element:
K 2V ¼ @V=@ DV ¼ kðk sin h cos h  cos2 hÞ ð22aÞ
2
K 2T ¼ @T=@ DT ¼ kðsin h  sin h cos h=kÞ ð22bÞ
Normalizing these values by the original elastic stiffness
provided by the links (Eqs. (1a) and (2a)) provides the softening
stiffness ratios jv, jt, as plotted in Fig. 11(b):
jv ¼ K 2V =K iV ¼ ðk tan h  1Þ=2 ð23aÞ
jt ¼ K 2T =K iT ¼ ð1  cot hkÞ=2 ð23bÞ
The ratio between the fully-yielded and softening deformations
(l = DVYs/DVYf) is also shown in Fig. 11(b). It is noted that a defor-
mation trajectory exists (k = tan h) in which a ‘‘fully’’ yielded condi-
tion is not reached (k = 1.0 in Fig. 11), but this coincides with the
trajectory in which the secondary softening slope possesses a zero
value, thus this trajectory does not create a performance issue for
the connector element within the pushover analysis.
6.3. Response under compression
The response to axial force perpendicular to the joint is highly
asymmetric. In tension the elements of the connector are free to
deform, while in compression these elements bear directly on
the surrounding concrete. This characteristic must be included
for the model to provide an accurate neutral axis, thereby impact-
ing both local (opening) deformation demand as well as global dia-
phragm response.
The asymmetric response is accomplished in the model through
the use of contact pseudo-elements. These elements use a penalty
function formulation to limit incompatible penetration of the pre-
cast units [34]. The contact elements are linear and the surround-
ing precast panels in the 2D-FE discrete model are represented
with elastic plane stress elements (refer to Fig. 5). Thus, an inelastic
compression spring is introduced in series with the contact ele-
ment [refer to Fig. 7(a)] to capture the softening and eventual lim-
iting compressive strength due to concrete damage, thereby
preventing artificially high compression forces from developing
in the floor system. The properties for these elements are deter-
mined by aligning the characteristics of the contact element, com-
pression spring and the elastic plane stress element to MC test data
for an equivalent strut of concrete [refer to Fig. 7(c)]. The equiva-
lent strut has a gage length representing two half precast panels,
thus equal to one panel width b, and an effective area equal to
the product of the precast panel flange thickness (t) and an effec-
tive width (deff), taken as the connector spacing for a topped dia-
phragm, and the width of face plate for an untopped diaphragm
[see inset, Fig. 7(c)]. The initial (KiC) and secant (KsC) compressive
stiffness from MC test data can be used to determine the corre-
cs cs
sponding stiffness of the compression spring [ki and ku , refer to
Fig. 7(c)] using the following equations, based on springs in series:
1
K cs
i ¼ ð1=K iC  b=tEdeff  1=kc Þ ð24aÞ
K cs
1
u ¼ ð1=K sC  b=tEdeff  1=kc Þ ð24bÞ
where kc is contact element stiffness, selected as the minimum
acceptable higher order of magnitude of surrounding elastic ele-
ments for penalty function convergence [34], e.g. kc = 20 tEdeff/b.
cs
In the absence of test data, KiC  tEdeff/b, thus ki can be assigned
cs cs
a high value (ki ¼ 100 tEdeff =b) and ku can be calculated in Eq.
67
0
(25b) using a KsC based on secant stiffness at peak stress f c for a
concrete cylinder test.
Compression across the joint mobilizes friction mechanisms
between the precast units. An effective coefficient of friction (ls)
is therefore assigned to the contact element. The parameter ls is
calibrated using the results of MC shear tests in the presence of
axial compression with the CQ test [17]:
ls ¼ ðV CY  V Y Þ=C ð25Þ
where V CY ,
VY are the peak shear strengths obtained from the MC,
CQ shear tests and C is the constant axial compression force applied
in the MC shear test (see Table 1). In the absence of the MC test, the
accepted value of ls = 1.0, 0.6 for topped, untopped systems in
accordance with code procedures [28].
7. Connector element construction, verification and
demonstration
7.1. Element construction
The procedure to construct the connector element model is
demonstrated here for the three connectors shown in Fig. 1. Table 2
shows the connector properties obtained in the CQ and MC tests
[18–20]. The construction of the diaphragm connector element
uses the formulation from Section 5, as calibrated using the proce-
dures of Section 6. The resulting model parameters (h, xt, xv and
x2t) and connector model properties are given in Tables 3 and 4
respectively.
The step-by-step connector element construction procedure is
summarized as follows: (1) Calculate the connector element link
orientation angle h using Eq. (5). Example (JVI Vector): h = tan1
[(81  67)/(21  395)] = 33°. (2) Create the appropriate strength
reduction charts based on the selected link orientation angle calcu-
lated in Step 1 (see Fig. 12). (3) Read a trial uncoupled tension (xt)
and shear (xv) strength factor from the strength reduction charts
(Fig. 12) using MC test data (or average of multiple tests) or
assumed behavior in the absence of test data. Example (Dry Chord):
In Fig. 12(b) interpolate between lines xv = 0.25 and 0.5 for the tri-
angle; and between lines xt = 0.75 and 1.0 for the square, produc-
ing a trial set for (xv, xt) of [0.45, 0.83].
(4) Adjust the strength factors to meet Eq. (7). Example (Dry
Chord): Insert trial xt = 0.83 into Eq. (7) to calculate xv = 0.55.
The final selected set for xv, xt are 0.55 and 0.83 (see Table 3).
(5) Calculate the model properties for the link [ny (nu), k, k2L
vs vs
(k3L), du], the uncoupled shear spring (vy, ki , k2 , duv), and the
ts ts
uncoupled tension springs (ty, ki , k2 , dut) using Eqs. (6), (8)
and (9). (6) Determine the compression assemblage properties
cs cs
(ki , ku ls) using Eqs. (24) and (25). The equivalent concrete
strut geometry (b, deff in Table 4) is measured based on Fig. 7(c)
and concrete strut properties are based on the nominal compres-
0
sive strength f c . The coefficient of friction is calculated using
data from Table 2 (see Table 4). Example (JVI Vector):
ls ¼ ðV CY  V Y Þ=C ¼ ð105  81Þ=44:5 ¼ 0:55.
7.2. Comparison to tests
Fig. 13 show the comparison of the connector element model
response obtained from the construction of the previous section
with the connector tests results for the CQ tests and the MC tests
under tension–shear proportional load with k = 2.0 for the JVI Vec-
tor and 2#5 chord connector, and a = 1.0 for #3 angled bar-plates.
As seen, the model produce responses that closely match the CQ
tests results, and quite reasonably match the MC tests results
under combined loading, including similar strength reduction,
stiffness reduction, and descending branch behavior. This result
68 G. Wan et al. / Engineering Structures 86 (2015) 58–71

Table 2
Connector characteristics measured in qualification and model calibration tests.

Characteristic JVI Vector 2#5 Chord connector #3 Angled bar-plate

Tension Shear Tension Shear Tension Shear
KiT or Kiv (kN/cm) 67 395 917 309 298 665
TY or VY (kN) 21 81 186 67 45 76
K2T or K2V (kN/cm) 9 67 71 22 32 131
K3T (kN/cm) – – 40 – – –
Tu (kN) – – 219 – – –
DuT or DuV (cm) 1.26 1.24 1.53 1.87 0.76 0.64
T VT VT
Y or V Y (kN)
18 56 173 44 27 38
T VT VT
Y /TY or V Y /VY
0.86 0.69 0.93 0.66 0.6 0.5
V CY (kN) 44.5 105 – – – –

Table 3 including instantaneous stiffness for: tension (Kt), shear (Kv), ten-
Model parameters for partially-coupled formulation. sion–shear coupling (Ktv, Kvt). An event-based formulation is used
Model parameters h (°) xt xv x2t to modify the instantaneous stiffness based on the total displace-
JVI Vector 33 0 0.6 0
ment vector and a set of displacement-based criteria.
2#5 Chord connector 47 0.83 0.55 0     
K t ð DT ; DV Þ K tv ðDT ; DV Þ dDT T
#3 Angled bar-plate 37 0 0.2 0 ¼ ð26Þ
K v t ðDT ; DV Þ K v ðDT ; DV Þ dDV V

Table 4 7.3. Demonstration of 2D-FE discrete model

Connector model properties.

Elements JVI 2#5 Chord #3 Angled The importance of the coupled precast connector element fea-
Vector connector bar-plate tures is illustrated here in the application of the 2D-FE discrete
Link ny (kN) 20 22 38 model to a precast diaphragm with two common LFRS layouts:
k (kN/cm) 112 148 414 (1) an interior wall structure; and (2) exterior walls with an eccen-
k2L (kN/cm) 9 67 44
k3L (kN/cm) – 38 –
tric longitudinal wall [see plan in Fig. 14(a)]. The diaphragm design
nu (kN) – 44 – is based on a 4-story precast structure in Berkeley CA (SDC E) with
du (cm) 1.11 1.11 0.46 a floor mass of 5.5 kPa. For simplicity, the transverse walls in the
Tension ts
ki (kN/cm) – 760 – 2D-FE discrete model are represented via pin-roller boundary con-
spring ty (kN) – 155 – ditions while springs are used to model the 6.1 m longitudinal wall
ts
k2 (kN/cm) – 0 – out-of-plane (cracked section) stiffness, calculated for a first story
dut (cm) – 1.53 – floor-to-floor height of 3.66 m [see Fig. 14(b)]. The diaphragm flex-
Shear vs
ki (kN/cm) 238 170 135 ural reinforcement is the dry chord [Fig. 1(c)]. For shear reinforce-
spring vy (kN) 50 37 15 ment, layout 1 uses the JVI Vector [Fig. 1(b)] while layout 2 uses
vs
k2 (kN/cm) 79 25 188 the stiffer and less ductile angled bar-plate connection [two-sided
duv (cm) 1.24 1.67 0.64 version of Fig. 1(d)]. Fig. 13(c) shows the connector discretization:
Elastic plane E (MPa) 27,789 27,789 27,789 The shear (Y direction) and axial (X direction) springs are placed
stress element t (cm) 10 10 10 between node 1 and node 2; The (X direction) contact element
deff (cm) 15 25 15
and compression spring are between nodes 2,5 and 1,5, respec-
b (cm) 305 305 305
tively. The inclined link elements are connected from nodes 1 to
Contact kc (kN/cm) 28,234 47,056 28,234
3 and 1 to 4. The link elements transfer forces between four plane
l 0.55 0.6 0.6
cs
stress elements through nodes 1 and 2. Linear constraint equations
Compression ki (kN/cm) 141,168 235,280 141,168
are provided to couple the DOFs of nodes 3 and 4 to node 2, appro-
spring cs
ku (kN/cm) 2511 4185 2511
priate for small deformation analyses. The connector elements are
constructed as shown in Section 8 to produce the response shown
in Fig. 13.
implies that the connector models can be used in the 2D-FE dis- The 2D-FE discrete diaphragm model in Fig. 14 is analyzed in a
crete model for ‘‘pushover’’ analyses to capture the softened pushover analyses by subjecting the isolated floor diaphragm to an
behavior under complex and non-predetermined loadings where increasing body force, an approach appropriate for diaphragms
the trajectory ratios vary with location and loading state. Note cyc- that do not have large transfer conditions [1]. The pushover analy-
lic tension–shear test results for the JVI were unavailable, and are sis is performed dynamically [34], rather than statically, to capture
instead based on the monotonic tests [top plot of Fig. 13(b)] under- the diaphragm descending branch behavior when non-ductile limit
scoring the importance of using a cyclic backbone characteristics to states are reached. Each analysis is performed twice, once each for:
construct the connector elements. (1) the connector elements developed in this paper, and (2) a sim-
The bottom plots of Fig. 13 also show a comparison with a user- ple connector model with uncoupled axial and shear springs [4].
defined material created to reproduce the coupled diaphragm con- Fig. 15 shows the pushover results comparison for layout 1.
nector response. Though not considered fully viable for the studies Fig. 15(a) shows the global diaphragm force vs. mid-span deflec-
performed, developments in open source structural analysis soft- tion curves. As seen, the simple connector model predicts higher
ware [33] permit this comparison. A link element with two trans- strength and significantly more ductility than the coupled model.
lational springs for the in-plane DOFs (axial and shear) and a de- Turning to the diaphragm joint level, Fig. 15(b) shows the flexural
activated rotational spring. A user-defined material model is response of the mid-span and support joint, while Fig. 15(c) shows
assigned for the springs with the properties shown in Eq. (26), the shear response of the support joint. It is seen that the simple
 G. Wan et al. / Engineering Structures 86 (2015) 58–71 69

1 ω t=1.0 ω v =1.0 ω t=1.0 ω v =1.0 ω t=1.0 ω v =1.0

Strength reduction

0.8 ω v =0.75 ω v =0.75 ω v =0.75

ω t=0.75 ω t=0.75 ω t=0.75
0.6 ω v =0.5
ω v =0.5 ω v =0.5
ω t=0.5 ω t=0.5 Tension ω t=0.5
0.4 ω v =0.25
Shear ω v =0.25
ω v =0.25
0.2 ω t=0.25 ω t=0.25 VY,red /VY ω v =0 ω t=0.25
TY,red /TY θ =47 0 θ =37 0
ω v =0
ω t=0 θ =33 0 ω v =0 ω t=0
0 ω t=0
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
λ = ΔT /Δ V λ= Δ T /Δ V α =T/V
(a) JVI Vector (b) 2#5 chord (c) #3 Angled bar-plate
Fig. 12. Strength reduction chart with test data mapping.

30
(a) (b)
80
25
70
Tension (kN)

20 60 Monotonic

JVI Vector
Shear (kN)
50
15
40
10 30 Test

20 Cyclic Est.
5
10
0 0
0 0.5 1 1.5 0 0.5 1

250 70
60
200
50
Tension (kN)

2#5 Chord
Shear (kN)

150 40
Black: Model
100 30
Grey: Test
Solid: Tension or shear 20
50 Dashed: Tension-shear
10
0 0
0 0.5 1 1.5 0 0.5 1 1.5 2

70 80
70
#3 Angled bar-plate

60
50 60
Tension (kN)

Shear (kN)

50
40
40
30 User-defined User-defined
30 material
20 material
20
10 10
0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
Opening (cm) Sliding (cm)

Fig. 13. Connector response comparison between model and test: (a) tension and (b) shear.

model completely misses the diaphragm failure mode (defined as the support joint, indicating that plane section Bernoulli beam con-
the onset of loss of diaphragm load carrying capacity), predicting cepts will not produce a realistic prediction of deformation
a ductile flexural mechanism at mid-span [Fig. 15(b)], while the demands along the diaphragm joint. Fig. 15(e) and (f) show the
coupled model produces a non-ductile shear failure at the support local response of individual shear connectors in the support joint
[Fig. 15(c)]. The reason for the discrepancy is that the tension asso- compression and tension zones respectively, indicating the signif-
ciated with diaphragm flexure at the support is sufficient to lower icant effect of axial force, both tensile and compressive on shear
the overall shear strength of the joint to make it the critical failure response.
mode. Note that this combined force also lowers the flexural Fig. 16 show the results for layout 2, the diaphragm with
strength of the support joint, leading to the simple model underes- orthogonal walls. As before, the simple connector model overesti-
timating the rotational ductility demand for the support joint as mates diaphragm strength and deformation capacity [see Fig
well. This plastic deformation and the shear failure protect the 16(a)]. The combination of shear and tension at the joint flanking
midspan, so the simple connector model actually predicts a flexure the longitudinal wall (that serves doubly as a collector region for
failure that cannot occur. Fig. 15(d) shows the deformed shape of earthquakes with bi-directional components) is sufficient to lower
70 G. Wan et al. / Engineering Structures 86 (2015) 58–71

(a)
Longitudinal Transverse 18.29m
Column
shear wall shear walls
3m (c)
Precast
panels

73.15 m 3 Link
Layout 1 Springs for wall
Layout 2 Constraints element
out-of-plane stiffness 2 (L=25.4cm, A=6.5cm 2)
(b)
5 1
Transverse (direction of body force)
Y 4
Longitudinal Plane stress
X element
(t=10cm, E=27.8GPa)

Fig. 14. Connector model implementation: (a) structural plan; (b) 2D-FE model and (c) close-up.

7000 100
(a) 20000 (b) (e)
Moment (kN-m)

Shear

Force (kN)
6000 15000 Flexural failure 50

10000
5000 0
Midspan
5000
Supports
Compression
Force (kN)

4000 0 -50
0 0.5 1 0 0.1 0.2 0.3
Rotation (rad) x 10-3 Deformation (cm)
80
3000 2500
(c) (f) Shear
2000 60
Shear (kN)

Force (kN)

2000
1500
Shear failure 40 Tension
1000
1000 (d)
Coupled Model 20
500
Refer to (d) Simple Model
0 0 0
0 2 4 6 8 0 0.2 0.4 0 0.2 0.4 0.6 0.8
Deflection (cm) Sliding (cm) Deformation (cm)

Fig. 15. Layout 1 results: (a) global response; (b) flexural response; (c) shear response at support; (d) deformed shape at support; shear connector response at (e) compression
zone and (f) tension zone.

7000 12000
(a) (b) (d)
Moment (kN-m)

Flexural 10000
failure 8000
6000
6000
4000 Flexural
5000 failure
Flexural/
shear 2000
failure 0 Coupled Model
Force (kN)

4000 0 0.2 0.4 0.6 0.8 1

Simple Model
Rotation (rad) -3
x 10 (e)
3000 1,400
1,200
(c) 60
Flexural
Force (kN)
Shear (kN)

2000 1,000
failure
800 40
Refer 600 Shear failure
1000 to (d)
400 20 Shear Tension
200
failure
Shear
0 0 0
0 2 4 6 0 0.2 0.4 0.6 0 0.2 0.4 0.6 0.8
Deflection (cm) Sliding (cm) Deformation (cm)

Fig. 16. Layout 2 results: (a) global response; Joint flanking longitudinal wall: (b) flexure response; (c) shear response; (d) deformed shape; and (e) shear connector response.
 G. Wan et al. / Engineering Structures 86 (2015) 58–71
the joint flexural strength [Fig. 16(b)] and cause significantly more
shear ductility demand [Fig. 16(c) and (e)]. These combined flex-
ural and shear deformations are shown in Fig. 16(d). In this case
the difference in global response is due more to different local
characteristics to similar deformation patterns than due to differ-
ent deformation patterns, as occurred for layout 1. This outcome
is due to the larger relative contribution and non-ductile character-
istic of the shear connectors in flexure.
The inaccurate predictions of behavior of the uncoupled dia-
phragm connector models in terms of strength, ductility, local
deformation demands and failure mode are demonstrated here
for relatively simple diaphragm layouts. This result underscores
the importance of using the coupled connector element in analyses
assessing the reliability of the new diaphragm design methodology
[8], in which models of realistic diaphragms with complicated floor
layouts are evaluated under complex force inputs.
8. Summary and conclusions
Discrete connector elements have been developed for use in
two-dimensional finite element models for nonlinear static mono-
tonic ‘‘pushover’’ analysis. These connector elements were con-
structed on the basis of cyclic backbone curves from full scale
tests of isolated connector elements under individual loading com-
ponents, i.e. tension/compression and reversing shear. Tests of iso-
lated connectors under combined loading indicate modified
behavior: Shear and tension together produces tension–shear cou-
pled response (strength and stiffness reduction); shear in the pres-
ence of compression produces higher stiffness and strength. A
partially-coupled element formulation was shown to be effective
in capturing the tension–shear interactions. Likewise, contact
pseudo-elements were able to reproduce the compression effects.
A procedure is presented that permits construction of the connec-
tor element from test data of Connector Qualification (CQ) tests
under tension or shear only loadings, and calibrated on the basis
of a small set of Model Calibration (MC) tests under combined
loadings. The connector elements produce response that closely
matches the individual component loading, and that reasonably
match the response obtained in the connector tests under com-
bined loading. The connector elements successfully reproduce
key behaviors exhibited in the tests including tension–shear cou-
pling behavior, high stiffness contact, and friction under compres-
sion. This result implies that the connector models can be used
effectively in the 2D-FE discrete model for ‘‘pushover’’ analyses,
where complex loading on the connector element will translate
into softened behavior relative to the response under simple indi-
vidual force components. These analyses have been successfully
performed in support of developing the capacity of precast dia-
phragms, a key step in developing a seismic design methodology
for precast concrete floor diaphragms.
Acknowledgements
This research was supported by the Precast/Prestressed Con-
crete Institute (PCI), the Charles Pankow Foundation, and the
National Science Foundation (NSF) under Grant CMS-0324522
and SGER Supplement CMMI-0623952. The authors are grateful
for this support. Any opinions, findings, and conclusions or recom-
mendations expressed in this material are those of the author(s)
and do not necessarily reflect the views of the National Science
Foundation.
References
[1] Moehle JP, Hooper JD, Kelly DJ, Meyer TR. Seismic design of cast-in-place
concrete diaphragms, chords, and collections. NEHRP seismic design technical
brief no. 3; 2010.
71
[2] Biondini F, Ferrara L, Toniolo G. Capacity design criteria for connections in
precast structures. In: 14th World conference on earthquake engineering.
Beijing, China; October 12–17, 2008.
[3] Ren R, Naito C. Precast concrete diaphragm connector performance database.
ASCE J Struct Eng 2013;139(1):15–27.
[4] Farrow KT, Fleischman RB. Effect of dimension and detail on the capacity of
precast concrete parking structure diaphragms. PCI J 2003;48(5):46–61.
[5] Wood SL, Stanton JF, Hawkins NM. New seismic design provisions for
diaphragms in precast concrete parking structures. PCI J 2000;45(1):50–65.
[6] Iverson JK, Hawkins NM. Performance of precast/prestressed concrete building
structures during northridge earthquake. PCI J 1994;39(2):38–55.
[7] Fleischman RB, Restrepo JI, Naito CJ, Sause R, Zhang D, Schoettler M. Integrated
analytical and experimental research to develop a new seismic design
methodology for precast concrete diaphragms. ASCE J Struct Eng
2013;139(7):1192–204. Special Issue.
[8] DSDM Task Group. Development and design of untopped precast concrete
diaphragm system for high seismic zones. Charles Pankow Foundation.
<http://www.pankowfoundation.org/grants.cfm?grantid=1070>; 2014.
[9] Building Seismic Safety Council, Committee TS-4. Seismic design methodology
for precast concrete floor diaphragms. Part III, NEHRP Recommended seismic
provisions federal emergency management agency. Washington (DC); 2009.
[10] Building Seismic Safety Council, Committee IT6. Diaphragm Design Force
Level. Appendix for PART 3 2014 NEHRP.
[11] Nakaki SD. Design guidelines for precast and cast-in-place concrete
diaphragms. Technical Report, EERI Professional Fellowship, Earthquake
Engineering Research Institute; 2000.
[12] Fleischman RB, Sause R, Pessiki S, Rhodes AB. Seismic behavior of precast
parking structure diaphragms. PCI J 1998;43(1):38–53. January–February.
[13] Tena-Colunga A, Abrams DP. Seismic behavior of structures with flexible
diaphragms. ASCE J Struct Eng 1996;122(4):439–45.
[14] Ju SH, Lin MC. Comparison of building analyses assuming rigid or flexible
floors. ASCE J Struct Eng 1999;125(1):25–31.
[15] Lee HJ, Kuchma DA. Seismic response of parking structures with precast
concrete diaphragms. PCI J 2008;53(2):71–94.
[16] Zhang W, Oliva MG. Analytical method to determine the elastic in-plane
behavior of pretopped, precast concrete double-tee diaphragms with discrete
connections. PCI J 2007:106–23. September/October.
[17] Naito C, Peter W, Cao L. Development of a seismic design methodology for
precast diaphragms – phase 1 summary report. ATLSS Report No. 06-03. ATLSS
Center, Lehigh University, PA; 2006.
[18] Naito C, Jones C, Cullen T, Ren R. Development of a seismic design
methodology for precast diaphragms – phase 1b summary report. ATLSS
Report No.07-04. ATLSS Center, Lehigh University, PA; 2007.
[19] Pincheira JA, Oliva MG, Kusumo-rahardjo FI. Tests on double tee flange
connectors subjected to monotonic and cyclic loading. PCI J 1998;43(3):82–96.
[20] Oliva MG. Testing of the JVI flange connector for precast concrete double-tee
system. Structures and materials test laboratory. Madison (WI): College of
Engineering, University of Wisconsim; 2000.
[21] Shaikh AF, Feile EP. Load testing of a precast concrete double-tee flange
connector. PCI J 2004:85–94. May–June.
[22] Fleischman RB, Wan G. Appropriate overstrength of shear reinforcement in
precast concrete diaphragms. ASCE J Struct Eng 2007;133(11):1616–26.
[23] Federal Emergency Management Agency. Prestandard and commentary for the
seismic rehabilitation of buildings. FEMA 356, Washington (DC); 2000.
[24] Zhang D, Fleischman RB. Establishment of performance-based seismic design
factors for precast concrete floor diaphragms. Earthquake Eng Struct Dynam
2014. submitted for publication.
[25] Zhang D, Fleischman RB, Naito CJ, Wan G. Development of diaphragm
connector elements for three-dimensional nonlinear dynamic analysis of
precast concrete structures. PCI J 2014. in preparation.
[26] Zhang D, Fleischman RB, Naito C, Ren R. Experimental evaluation of pretopped
precast diaphragm critical flexure joint under seismic demands. ASCE J Struct
Eng 2011;137(10):1063–74.
[27] Schoettler MJ, Belleri A, Zhang D, Restrepo J, Fleischman RB. Preliminary
results of the shake-table testing for development of a diaphragm seismic
design methodology. PCI J 2009;54(1):100–24.
[28] ACI 318-05. Building code requirements for structural concrete (ACI 318-05)
and commentary (ACI 318-05). ACI committee 318; 2005.
[29] Precast/Prestressed Concrete Institution. PCI design handbook: precast and
prestressed concrete. 6th ed., Chicago (IL); 2004.
[30] Cao L, Naito C. Design of Precast Diaphragm Chord Connections for In-Plane
Tension Demands. ASCE J Struct Eng 2007;133(11):1627–35.
[31] Mattock AH, Hawkins NM. Research on shear transfer in reinforced concrete.
PCI J 1972;17(2):55–75.
[32] Naito C, Ren R. An evaluation method for precast concrete diaphragm
connectors based on structural testing. J Precast/Prestressed Concr Inst
2013;58(2):106–18.
[33] OpenSees Version 2.4.3. OpenSees Comand Language Manual. (Rev 5621). CA:
University of California at Berkeley; 2013.
[34] ANSYS version 11. Elements (00853) and theory reference (00855). SAS Inc.;
2007.
[35] Panagiotou M, Restrepo JI, Schoettler M, Kim G. Nonlinear cyclic truss model
for reinforced concrete walls. ACI Struct J 2012;109(2).
[36] Kadysiewski S, Mosalam KM. Modeling of unreinforced masonry infill walls
considering in-plane and out-of-plane interaction. Pacific Earthquake
Engineering Research Center Report 2008-102; 2009.