Wei Zheng, Ph.D., P.E.

Senior Structural Engineer

Flad & Associates
Madison, Wisconsin
Michael G. Oliva, Ph.D.
Professor
Department of Civil and
Environmental Engineering
University of WisconsinMadison
Madison, Wisconsin
The in-plane exibility of untopped precast concrete double-tee
diaphragms is often ignored in current design practice. The performance
of parking structures during earthquakes and subsequent research
studies have shown, however, that in-plane exibility of diaphragms
may contribute to the development of large displacements, a concern
for stability. The objective of this paper is to provide designers with
a practical approach to judge the linear elastic in-plane exibility of
discretely connected untopped double-tee diaphragms. Based on the
results of detailed nite element model analyses of commonly used
untopped diaphragms, a simplied rational approximation is used to
establish equivalent beam models. Methods of dening linear elastic
stiffness parameters of the equivalent beam are derived and can be
directly used in manual calculations or computer analysis. Compared
to a complex nite element analysis, the proposed equivalent beam
model can predict the linear elastic in-plane diaphragm deformation
under wind or seismic load with reasonable accuracy for design.
A deection calculation example is provided in Appendix B to illustrate
the proposed approach.
A Practical Method to Estimate
Elastic Deformation of Precast
Pretopped Double-Tee Diaphragms
2 PCI JOURNAL
D
ouble tees are commonly
used to form oor and roof
diaphragms for parking struc-
tures, commercial and industrial build-
ings, and other structures. Two types
of diaphragms are used. In some parts
of the United States, designers prefer
to use discretely-spaced mechanical
ange-to-ange connectors to join ad-
jacent pretopped double tees and cre-
ate untopped double-tee diaphragms.
In other locations and particularly in
high seismic zones, however, 2 to 4 in.
(51 to 102 mm) thick reinforced cast-
in-place concrete slabs (topping) may
be overlaid on double tees and across
joints, or they may be combined with
mechanical connectors to create con-
nections between double tees and form
topped double-tee diaphragms.
Both double-tee oor and roof sys-
tems were developed primarily for car-
rying vertical gravity loads, but they
are also key elements in transferring
March-April 2005 3
lateral wind or earthquake forces to
lateral load-resisting systems, such as
shear walls. In parking structures, the
double-tee diaphragms are divided by
ramps into one-bay diaphragm seg-
ments. Even though the span-to-width
ratio of individual double-tee dia-
phragm segments often range from 3 to
5, the in-plane exibility of double-tee
diaphragms is often neglected in de-
sign practice.
As a result of structural failures that
occurred during the 1994 Northridge
earthquake, the response of parking
structures to seismic events has been
the subject of investigation. Observa-
tions after the earthquake indicated that
the damage to some parking structures
may have been caused by gravity load-
resisting systems failing due to large
lateral displacements (drifts) of the oor
at locations away from shear walls.
1-3

Shear walls were expected to resist
lateral loads. Separate column and in-
verted tee framing, intended to only
resist the gravity loads, made up the
gravity load system. Large lateral
displacement of the columns, unre-
strained by stiff diaphragms, may have
led to instability under vertical load.
Subsequent analytical studies
4,5
on
topped double-tee diaphragms have in-
dicated that the exural deformation of
individual diaphragm segments caused
lateral displacements several times
larger than the supporting shear wall
story drifts. The structural system used
to carry vertical (gravity) loads to the
ground is often not designed to be able
to sustain such large drifts.
In a recent analytical investiga-
tion,
6
the inuence of elastic in-plane
exibility of untopped (or pretopped)
double-tee diaphragms on the seismic
behavior of parking structures was also
addressed. That study revealed that:
1. The dynamic response of
parking structures with an
untopped diaphragm may be
substantially different from the
dynamic response based on a
rigid diaphragm assumption,
due to more participation of
diaphragm-driven vibration
modes;
2. The diaphragm shear forces
developed at stories varied with
the in-plane exibility and might
be larger than specied values
estimated from building code
methods; and
3. Predictions of elastic diaphragm
deformation show lateral
displacement demands on the
gravity loadbearing system
much larger than would be
expected considering only story
drift at supporting shear walls.
Two studies
5,6
have suggested using
a design procedure that recognizes that
a parking structure oor is composed
of exible diaphragm segments.
Some precast concrete designers
have argued that double-tee diaphragms
should be designed to remain elastic
during seismic events. This argument
is very convincing when the concept
of inelasticity is considered. If yielding
is allowed to develop in a precast dia-
phragm, it is likely to be concentrated
within one or a few joints adjacent to
shear walls since all joints are normal-
ly provided with the same connections
and the same strength.
When the deformability of existing
mechanical connectors is examined,
6
it becomes obvious that the demand
developed in the yielding joints would
exceed the limited connector capaci-
ties. Thus, maintaining elastic behav-
ior in mechanically connected precast
diaphragms seems imperative. For
this design approach, the diaphragm
shears obtained from model building
code estimation procedures need to be
increased by an overstrength or magni-
cation factor since they are the design
yield values of the lateral load resisting
system, not the expected peak values.
This approach for an elastic design
load is presented in the Fifth Edition of
the PCI Design Handbook.
7
Still, the
elastic in-plane exibility of double-
tee diaphragms should remain a major
concern in design because of excessive
story drifts that may be imposed on
gravity load-resisting systems causing
instability.
The International Building Code
(IBC 2003)
8
and its predecessor, the
Uniform Building Code (UBC 1997),
9

clearly stipulate that diaphragms shall
not be considered rigid for the purpose
of distributing story shear and torsional
moment when the maximum lateral de-
formation of the diaphragm is more than
two times the average story drift of the
associated story. The average story drift
is based on displacement of the load-
resisting system, such as shear walls.
Both codes also stipulate that the in-
plane deection of the diaphragm shall
not exceed the permissible deection of
the attached elements, such as columns.
The P- effects on such elements shall
be considered. Since there is no infor-
mation or analytical method suggested
in the codes for determining the in-
plane exibility of untopped double-tee
diaphragms, these code provisions are
often not satisfactorily followed.
This paper provides a simple method
that designers can use to check defor-
mations in untopped precast concrete
double-tee diaphragms. Existing infor-
mation on diaphragm components and
analysis is reviewed, and then the de-
velopment of a detailed nite element
model (FEM) for predicting diaphragm
behavior is briey explained. The de-
velopment of a simple method, based
on the FEM, and using beam modeling
for the diaphragm, is shown with ex-
ample calculations of deformation in a
prototype structure.
OBJECTIVES AND SCOPE
This paper presents a simple method
for predicting elastic in-plane displace-
ments of untopped double-tee dia-
phragms, considering both exural and
shear deformation at joints between
Table 1. Comparison of predicted diaphragm deformations.
Accurate FEM
analysis*
FEM with
uniform connector
stiffness

Uniform
connector stiffness
and planar
behavior in joints

Diaphragm in-plane
deection (in.)
0.224 0.296 0.239
* Correct FEM analysis with different axial stiffness under tension and compression.
Approximate FEM analysis with uniform connector stiffness (tension value) in connectors.
Model with uniform connector stiffness (tension value) and rigid ange edges.
4 PCI JOURNAL
pretopped precast members as well as
within members.
First, behavioral simplications
are proposed based on analytical re-
sults from detailed planar nite ele-
ment analysis of typical untopped
diaphragms. Then, with those assump-
tions, the elastic exural deformation
and shear deformation in the pretopped
double tees and their ange-to-ange
connections are smeared into deforma-
tions of an equivalent beam.
The elastic stiffness parameters of
the equivalent beam are derived and
can be easily used in manual calcu-
lations or computer analysis. Shear
deformation at the connection of the
diaphragm to the remaining lateral
load system is also considered and
smeared into the shear deformation of
the equivalent beam.
The proposed equivalent beam
model is intended to provide an eas-
ily calculated and reasonable estimate
of the elastic in-plane deformation of
an untopped diaphragm. The method
is suitable for the design of untopped
double-tee diaphragms to resist wind
and seismic loads.
BACKGROUND
The impact of in-plane exibility in
eld-topped double-tee diaphragms on
seismic performance of parking struc-
tures has been analytically investigated
with complex nite element model-
ing at Lehigh University.
5
The model,
however, was based on exural defor-
mation of topped diaphragms only and
did not include shear deformations of
double tees or shear sliding in the joints
between adjacent mechanically con-
nected double tees.
In follow-up studies,
10,11
static analy-
ses of detailed nite element models
were conducted for both topped and
pretopped mechanically connected
double-tee diaphragms. The analytical
model of the diaphragm, including wire
and chord steel, was similar to that ad-
opted in the previous research, but me-
chanical connectors were added using
nonlinear springs possessing both ten-
sion (axial) and shear resistance. For
topped diaphragms, the shear-friction
effect due to the steel reinforcement
crossing the cracked concrete topping
section at the joints was considered and
determined by using empirical data.
In the pretopped diaphragm model,
contact friction occurring at joints in
regions of compression was included
by using friction-capable gap elements.
The investigation proposed a practical
method to calculate elastic in-plane de-
ection based on an equivalent elastic
modulus, determined by calibrating the
data with nite element analysis. The
equivalent elastic modulus was ob-
tained for three types of connectors.
For other connectors with different
properties, the equivalent elastic mod-
ulus needs to be determined either by
further calibrating with a nite element
model or by an interpolating method.
For typical dry mechanical connec-
tion systems preferred in pretopped
diaphragms, assuming contact friction
at joints in the region of compression
may unconservatively estimate a high
diaphragm stiffness.
Nakaki
12
proposed another rational
method for calculating the deformation
of diaphragms. The diaphragm elastic
exural stiffness was derived based
on the cracked section property of a
monolithic diaphragm, which is similar
to a cracked beam model. For topped
and untopped diaphragms, with dis-
crete cracks along precast panel joints,
the elastic tension strength of the re-
inforcement in cast-in-place topping
or the discrete connectors at joints is
converted into an equivalent web rein-
forcement in a monolithic diaphragm.
The elastic modulus of the equiva-
lent web reinforcement is proportioned
so that the tension deformation of top-
ping reinforcement, or discrete connec-
tors at joints, can be simulated. A simi-
lar treatment, however, is not included
for chord steel reinforcing. Therefore,
the tension deformation of chord steel
near the joints in a discrete cracked dia-
phragm is not explicitly included in the
Nakaki model.
For shear deformation, the Nakaki
method does not make a distinction be-
tween discretely connected diaphragms
and monolithic diaphragms. It simply
assumes that the concrete shear mod-
ulus, G, is equal to 0.4E and uses the
cracked monolithic diaphragm proper-
ties to determine the diaphragm shear
deformation, regardless of the shear
stiffness of discrete connectors across
the joints in an untopped diaphragm.
When the method is applied to an
untopped diaphragm, the stiffness con-
tribution of the double-tee ange is
omitted in the deection calculation.
Thus, for untopped diaphragms, this
method does not recognize any bene-
cial effect on reducing the diaphragm
deection by using thicker double-tee
anges or mechanical connectors with
higher shear stiffness. The Nakaki
method may not be appropriate for es-
timating the deformation of untopped
diaphragms, especially for common
diaphragms with a span-to-width ratio
of 3, in which the shear deformation
can be a signicant source of the total
diaphragm deformation.
In a recent analytical investigation
on the impact of elastic behavior in
untopped double-tee diaphragms,
6
a
detailed two-dimensional nite ele-
ment diaphragm model with discrete
mechanical connections between
double tees was developed. Planar -
nite elements were used to model the
individual double-tee members. The
Fig. 1. Segment
extracted from
a double-tee
diaphragm
(plan view).
March-April 2005 5
discrete mechanical connectors joining
the double-tee anges were represent-
ed by truss elements (or springs) for
axial and shear behavior. The connec-
tor properties were determined from
experimental data.
The nite element model was used
to predict elastic in-plane deformation
of a set of untopped double-tee dia-
phragms. It included both shear defor-
mation in the double-tee members and
shear sliding at joints between adjacent
members.
Several experimental tests have
been conducted on typical mechani-
cal connectors which join double-tee
panels.
13-16
The main focus of most of
these experimental studies was on the
connector strength, although a few in-
vestigators also examined stiffness.
17

One such study that addressed stiff-
ness in detail was a test program re-
cently conducted at the University of
Wisconsin-Madison.
6,16
DIAPHRAGM FINITE
ELEMENT MODEL
The proposed simple beam model
is based on well-dened mechanical
connector behavior measured in ex-
perimental testing and on experience
derived from complex nite element
(FEM) analyses of a series of untopped
diaphragms with discrete connections.
The detailed FEM analysis was car-
ried out using planar nite elements to
model double-tee members and special
joint elements with the measured dis-
crete connector properties.
Tests have shown that the compres-
sion stiffness of a mechanical connec-
tor may be ten times the tension stiff-
ness.
6
This high compression stiffness
and initial joint gap prevent the con-
crete anges from ever coming directly
into contact with one another during
linear elastic response. This means that
only the connector compression and
shear stiffnesses needed to be modeled
at the joints rather than adding contact
elements having a friction-shear com-
ponent as included in the Farrow and
Fleischman model.
10,11
Using discrete elements to model
the mechanical joint connectors posed
a particular challenge. In a real ange,
the forces transferred through connec-
tors are distributed into the ange con-
crete gradually through the connectors
anchor bars and create a complex three-
dimensional stress eld. Since the con-
nector test results actually included
local deformations that occurred in the
double-tee ange, the connector ele-
ments in the model were joined to a
series of nodes in the region around the
connector location to avoid the concen-
trated load and local deformation prob-
lem that might occur in the FEM.
When modeling different axial com-
pression and axial tension stiffnesses
of chords and mechanical connectors,
analytical studies using the detailed
nite element model showed that the
neutral axis of the diaphragm, with the
diaphragm in exure, was eccentric
from the diaphragm center by about
20 percent of the diaphragm depth. An
assumption that the compression stiff-
ness of connectors and chords at a joint
is the same as the tension stiffness is
adopted here to simplify calculations.
It places the neutral axis at mid-depth
and can overestimate rotation in the
joints between double tees, resulting in
an overestimate of deection.
If this incorrect assumption is used,
calculated in-plane diaphragm deec-
tions are 20 to 30 percent larger than
they should be. This error is compensat-
ed for, however, when the model with
equal tension and compression connec-
tor stiffness is also assumed to have a
rigid ange edge (i.e., plane section
assumption along joints). The error in
predicted diaphragm deformation was
reduced to 7 to 15 percent. The exure
results from these three model varia-
tions for one diaphragm conguration
are listed in Table 1.
Results from the nite element mod-
eling indicated that a series of model-
ing assumptions might be acceptable
for a simplied beam model:
1. The axial compression stiffness
of the connectors and chord steel
at joints could be assumed equal
to their tension stiffness;
2. The axial force distribution to
connectors at a joint, due to
exure of the diaphragm, could
be assumed to vary linearly
(plane section assumption);
3. The shear force transferred
across the joint could be
assumed to be uniformly
distributed to the connectors
at the joint because of the high
ange stiffness; and
4. The total deection of the
diaphragm could be assumed
as the sum of deections of
a monolithic diaphragm with
section equal to the double-
tee anges plus additional
deformation caused by shear and
rotation in the joints.
The assumptions reached on the
basis of the nite element results ap-
pear to relate well to the physical be-
havior except for the rst assumption.
Considering the high transverse stiff-
Fig. 2 Comparison of
in-plane deection
between equivalent
beam and direct
calculation. Note: 1
in. = 25.4 mm.
Fig. 3. Assumed
stress in embedded
steel reinforcing bar.
6 PCI JOURNAL
ness that a precast concrete double tee
possesses and the low stiffness of the
mechanical connectors, the second, or
rigid edge assumption appears logical.
A similar argument can be made for
the third assumption; namely, the tee
is nearly rigid compared to the shear
stiffness of mechanical connectors.
The fourth assumption uses superposi-
tion as commonly accepted for linear
elastic behavior.
The one assumption that may be
questionable is the rst. The neutral
axis of the joint will not be located
at mid-depth under in-plane exure,
as was observed in the accurate FEM
model, since the connectors have a
higher compression axial stiffness than
tension. When considering forces in-
side the uncracked double-tee ange,
however, the neutral axis would be as-
sumed to be at mid-depth because of ac-
tual uniform stiffness across the depth
(during linear elastic condition). There
is a discontinuity between double-tee
members and the joints. This may be
considered a disturbed region mak-
ing selection of a simple approximate
analysis method difcult.
The use of a detailed FEM analy-
sis, with appropriate connector axial
stiffnesses, provided a set of initial as-
sumptions for a simplied beam model.
A comparison with the detailed FEM
analysis is used again later to judge the
error and acceptability of the proposed
beam model.
EQUIVALENT BEAM MODEL
Consider that a segment consisting of
a double-tee member and one joint con-
nection at the right edge of the member
is extracted from a complete double-
tee diaphragm system (see Fig. 1). The
connectors that would be on the left
side of the member would be included
as part of an adjacent segment. Based
on elementary beam theory,
18
with an
in-plane bending moment M assumed
constant over the segment, the exural
rotation,
1,
occurring in the double-tee
panel is:

1
=
Mb
E
c
I
(1)
where
b = width of double-tee ange
(see Fig. 1)
E
c
= elastic modulus for concrete
of double tee
I = in-plane transverse moment
of inertia of double-tee ange
section (webs ignored)
The exural rotation occurring at
the connection joint can be determined
using a plane section assumption at the
ange edge. Suppose a rotation
2
oc-
curs at the joint between two double
tees. Then, from equilibrium of forces
in the axial direction of the connectors
across the joint:

i
K
ai
(y r
i
)
2
= 0 (2)
where
K
ai
= axial stiffness of ith
connector or chord element
r
i
= distance from ith connector or
chord to top end of double tee
y = distance from neutral axis to
top end of double tee
Since
2
0, Eq. (2) can be rear-
ranged as:

i
K
ai
(y r
i
) = 0 (3)
Assuming that the compression stiff-
ness of connector and chord, respec-
tively, is the same as their tension stiff-
ness, the neutral axis will be located at
the center of the section; thus, y = d/2,
where d is the depth of the diaphragm
(see Fig. 1).
From equilibrium in exure:

i
K
ai
(y r
i
)
2

2
= M (4)
Rearranging Eq. (4) results in the
following equation:

2
=
M

i
K
ai
(y r
i
)
2
(5)
Thus, the in-plane exural stiffness
of a joint between double tees, denoted
as

, can be expressed as:

=
i
K
ai
(y r
i
)
2
(6)
The parameter

given in this form

can be used in general for any dia-
phragm. Precasters often place connec-
tors closer together near the mid-region
of a double tee. Close spacing near the
midspan will not be effective in resist-
ing exure as witnessed by the squared
term in the stiffness factor.
If connector spacing is uniform and
the same type of connector is used at
all locations, the calculation of

in
Eq. (6) can be further simplied by
assuming the total axial stiffness of
the connectors to be spread uniformly
along the connection joint. The K
ai

terms in Eq. (6) are replaced with K
t

and K
c
for the ange connectors and
chord connectors, respectively. An ad-
ditional axial stiffness is present when
a chord exists, in excess of the normal
ange connector stiffness K
t
. This is
K
c
K
t
when no connector is placed at
the chord location.
Eq. (6) becomes:

= (K
c
K
t
)
d
2
2
+ nK
t

d
2
12
(7)
where
= 1 when a mechanical
connector is not placed at
the diaphragm edge (normal
case), or 0 when a connector
is placed at the chord location
K
c
= tension stiffness of chord
connection
K
t
= tension stiffness of each
ange connector
n = number of ange connections
(including chord connections)
Therefore, the total exural rotation
over the segment, including a double
tee and a connection joint, is:
=
1
+
2
=
Mb
E
c
I
+
M

(8)
The rst term,
1
, represents the
deformation in the double-tee anges
due to exure and might be neglected
when an axially soft connector is used
and the ange is relatively rigid. Then,
the ange would be assumed as a rigid
body and all of the exural deforma-
tion of the diaphragm would arise from
joint deformation. The entire equation
will be continued here.
Under a shear force, V, over the seg-
ment as shown in Fig. 1(b), the shear
distribution among the connectors is
assumed to be uniform. The in-plane
shear deection occurring at the joint
can be expressed as:

1
=
V

(9)
where
K
vi
= shear stiffness of ith
connector

= total shear stiffness of

connectors at a joint,
i
K
vi
If all the connectors are identical,
then
v
= (n 2)(K
v
) when no me-
March-April 2005 7
chanical connectors are placed at the
diaphragm edges, or
v
= nK
v
when
connectors are placed at the chord lo-
cations. Closer spacing of connectors
near the midspan of double tees will
simply add to the shear stiffness.
Note that the shear stiffness of chord
connectors is ignored here (conserva-
tive) since little information is avail-
able to accurately model their stiffness
even though the strength contribution
can be estimated from shear-friction.
A shear term for the chords, similar to
that in Eq. (7), could be added if the
shear stiffness of chords was known.
The shear deformation occurring in
the double-tee panel can be expressed
as:

2
=
1.2Vb
AG
c
(10)
where
A = cross-sectional area of
a double-tee ange in
longitudinal section
G
c
= shear modulus of concrete =
E
c
/[2(1 + )]
= Poissons ratio for concrete
(usually taken as 0.17 to 0.3)
Thus, the total shear deformation
in the segment including a double-tee
panel and a connection joint is:
=
1
+
2
=
1.2Vb
AG
c
+
V

(11)
where

1
= shear deformation of ange

2
= shear deformation in joint
= total shear deformation of
ange and joint
Consider that a segment with a length
b was extracted from an equivalent
beam model of the diaphragm. Based
on elementary beam theory, the ex-
ural rotation over the segment of the
equivalent beam model is:
=
Mb
EI
(12)
where
E = elastic modulus of equivalent
beam
I = moment of inertia of
equivalent beam, in which
the moment is considered
constant over the length of
the segment
The shear deformation over the
same segment of the equivalent beam
model is:
=
1.2Vb
AG
(13)
where
A = cross-sectional area of
equivalent beam
G = E/[2(1 + )] = shear
modulus of equivalent beam
= Poissons ratio of material in
equivalent beam, in which
the shear is considered
constant over the length of
the segment
To reach deformation equivalence
between the equivalent beam model
and the diaphragm, the segment of the
equivalent beam model and the seg-
ment of the double-tee diaphragm must
have the same deformation, i.e., =
and = . Thus:

Mb
EI
=
Mb
E
c
I
+
M

(14)

1.2Vb
GA
=
1.2Vb
G
c
A
+
V

(15)
Rearranging Eqs. (14) and (15) and
taking the thickness and Poissons ratio
of the equivalent beam model as those
of the actual double-tee ange yields
Eqs. (16) and (17):
d = d
1 +
G
c
A
1.2

b
1 +
E
c
I

b
(16)
where d is the depth of the equivalent
beam, and:
E = E
c
d
d
1 +
G
c
A
1.2

b
(17)
These parameters dene the equiva-
lent beam model of the double-tee dia-
phragm.
The shear deection from the equiv-
alent beam is smaller than that from
an accurate calculation as shown in
Fig. 2. This is attributed to the missing
deformation in the left joint where the
diaphragm is attached to a support (at
location 0 in Fig. 2). Since the equiv-
alent beam properties were calculated
for the segment in Fig. 1, the deforma-
tion on the left end of the diaphragm
is missing. To reach the correct deec-
tion at midspan, the shear sliding at the
end support should be smeared into the
shear deformation occurring within the
equivalent beam model.
A modication of the shear modu-
lus of the equivalent beam model is
made and a new equivalent depth d
and modulus E are given in Eqs. (18)
and (19):
d = d
1 + k
G
c
A
1.2

b
1 +
E
c
I

b
(18)
E = E
d
d
1 + k
G
c
A
1.2

b
(19)
where
k = a joint modication factor
k = (m + 2)/m with an even
number of double tees
k = (m + 2)
2
/m
2
with an odd
number of double tees
m = number of double tees
in diaphragm span being
considered
The in-plane deection of untopped
diaphragms can be estimated using the
equivalent beam either by manual cal-
culation or structural analysis software.
For instance, the midspan deection of
a one-span simply supported untopped
diaphragm can be calculated using the
deection formula from elementary
beam theory.
18

The properties of the section, such as
A and I, are calculated assuming that
the beam section is equal to the lon-
gitudinal cross section of the double-
tee ange with the equivalent depth d
rather than the actual depth. For more
complicated support conditions, struc-
tural analysis software can be used
with the E and d.
METHOD TO ESTIMATE
TENSION STIFFNESS OF
CHORD STEEL
Diaphragm chord members can actu-
ally be created by casting a raised sec-
tion of concrete, a pour strip or curb, on
top of the double-tee diaphragm along
two longitudinal edges. Chord steel re-
inforcement is embedded in this pour
strip. The deformation behavior of
chord steel members at joints has not
been specically tested in full-scale
diaphragm segments. The behavior,
however, might be reliably predicted
8 PCI JOURNAL
from the engineering properties of the
chord steel.
The following proposed method
can be used to estimate a conservative
chord steel linear elastic tension stiff-
ness in the absence of test data.
The force in the chord steel at the
joints will be developed through bond
with the ange concrete adjacent to
joints. Thus, the tensile stress in the
chord steel will vary from a lower
value within the double-tee anges
to a higher value or its yield stress at
the joints. The development length for
the embedded reinforcement to de-
velop its yield strength is dened in
ACI 318-02.
19
The stress distribution
in embedded reinforcement, at high
strains, has been determined by pull-
out tests.
20,21
Based on these pull-out
test results, the stress distribution in
the chord steel on both sides of a joint
is simplied as a parabolic curve (see
Fig. 3).
When the steel is at a stress substan-
tially lower than the yield level, and
the concrete surrounding the bar is at
a strain below the cracking strain, the
stress variation in the reinforcing bar
could be signicantly different. It is
likely that the variation would be the
inverse of the parabola shown in Fig.
3, high where it enters the concrete but
rapidly dropping off. An average be-
tween those two parabolic distributions
might be a linear variation, as indicated
by the dotted line in Fig. 3. A parabolic
assumption will provide a low (conser-
vative) chord stiffness estimate.
The linear elastic deformation of the
chord steel model between double-tee
anges is controlled by the assumed
stress distribution along the chord steel
as it becomes anchored in a ange. The
deformation can be determined by in-
tegrating the chord steel strain over the
development length on both sides of
the joint and taking the chord deforma-
tion as:

s
= 2

ld
0
dx
= 2

ld
0

f
E
s
dx
= 2
1
E
s

2
3
f
max
l
d

=
4f
max
l
d
3E
s
(20)
where
f
max
= T/A
s
A
s
= cross-sectional area of chord
steel
E
s
= elastic modulus of chord steel
l
d
= development length of chord
steel from ACI
T = tension in chord steel
Then, the tension stiffness of chord
steel at the joint between double tees,
K
s
, can be taken as:
K
s
=
T

s
=
A
s
E
s
4
3
l
d
=
3
4

A
s
E
s
l
d
(21)
or, if a linear stress is assumed along
the anchorage length:
K
s
=
A
s
E
l
d
(22)
where K
s
is the chord stiffness.
EXAMPLES AND
DISCUSSION
To illustrate and evaluate the equiva-
lent beam method, an example is given
for a one-span untopped diaphragm
with plan dimension of 120 60 ft (36.6
18.3 m). The diaphragm is attached
to one-story shear walls at both ends,
as shown in Fig. 4. In this example, the
double tees are connected to adjacent
panels by nine discrete mechanical
connectors spaced at 6 ft (1.8 m).
Four types of mechanical connec-
tors

are considered, ranging from light-
duty type connectors such as a hairpin
or bent wing connector, to heavy-duty
type connectors such as a structural-
tee connector. The characteristic shear
behavior of the bent wing connector
is shown in Fig. 5. This connector is
similar to a widely marketed commer-
cial connector that has been rigorously
tested with test results published.
22
The
behavior of the other connectors is de-
scribed elsewhere.
6,16
Chords are cast in pour-strips along
both longitudinal edges of the dia-
phragm with three No. 6 (19 mm) em-
bedded steel reinforcing bars. The
thickness of the double-tee ange is
4 in. (102 mm) if it is a oor and 2 in.
(51 mm) if it is a roof. Concrete strength
is taken as 5000 psi (34.5 MPa). Five
diaphragm congurations, four as
oors, and one as a roof (see Table 2),
are examined.
Drag beams are assumed to transfer
shear forces to the walls at either end of
the diaphragm and deformations in the
drag beams should be included in nor-
mal calculations, but are excluded here
to focus on the diaphragm alone. Re-
gardless of the diaphragm type, the total
horizontal lateral (in-plane) load on the
diaphragm is assumed to be 123.5 kips
(549 kN) from an IBC 2003
8
calcula-
tion (0.156 times the seismic weight of
the diaphragm) uniformly distributed
over the diaphragm area. The concrete
elastic modulus E
c
= 4070 ksi (28063
MPa) and Poissons ratio is taken as
= 0.3. The steel modulus is 29,000 ksi
(200000 MPa).
Table 2. Parameters for different diaphragm congurations in example.
Analysis
case
Type of
connector
Thickness
of panel, t
(in.)
Chord
steel
Stiffness of
connector
in shear, K
v

(kips/in.)*
Stiffness of
connector in
tension, K
t

(kips/in.)*
Tension yield
strength of
connector
(kips)
Stiffness of
connectors in
compression
(kips/in.)
Case 1 Bent wing 4 3 No. 6 300 16 3.5 3000
Case 2 Structural tee 4 3 No. 6 600 160 11.9 3000
Case 3 Bent wing 2 3 No. 6 300 16 3.5 1500
Case 4 Hairpin 4 3 No. 6 500 70 7.8 3000
Case 5 Stud-to-plate 4 3 No. 6 350 150 7.4 3000
* Connector tension stiffness and yield is based on test results presented in Reference 6.
Connector compression stiffness is based on test results presented in Reference 6.
Note: 1 in. = 25.4 mm; 1 kip = 4.45 kN; No. 6 bar = 19 mm diameter.
March-April 2005 9
Calculation of the midspan dia-
phragm deection for one of the cases
of this example is illustrated in a step-
by-step procedure in Appendix B using
the proposed equivalent beam model.
The stiffness parameters of the equiva-
lent beams are shown in Table 3 and
are based on measured connector prop-
erties.
6
In Table 4, the calculated mid-
span deections are tabulated and com-
pared with deections from the Nakaki
method
12
and more detailed planar -
nite element model analysis.
The equivalent beam deection pre-
dictions are very close to the accurate
nite element model predictions. If
the double tees were considered rigid
[the rst term in Eq. (8) being zero]
the displacements would be predicted
as 7 to 10 percent smaller. The maxi-
mum error between the beam model
and FEM, nearly 19 percent, occurs in
Case 2 where the axially stiff structural
tee type connector is used.
Case 5, with the axially stiff stud-
to-plate connector, has a similar error
(17 percent). Fortunately, these connec-
tors are not common for practical use
because the high axial stiffness causes
resistance to volume change in the
structure and is likely to result in long-
term ange deterioration problems.
The error ranges from 7 to 14 percent
on the conservative (overestimate) side
in most of the other cases. The only un-
derestimate of deection occurs in Case
3 due to the plane section assumption
at the joint. The thinner 2 in. (51 mm)
ange of Case 3 develops more local
ange deformation in the FEM analy-
sis due to the discrete connector forces
on the thin concrete.
The plane section assumption ignores
this deformation and overestimates the
diaphragm stiffness. The model devel-
oped by Farrow and Fleischman
10,11
pre-
dicted a lower deformation, 0.115 in.
(3 mm), for the Case 1 diaphragm.
23

The lower prediction occurs because
the Farrow and Fleischman model in-
cluded a shear-friction component in
the joint compression region.
For dry mechanical connections
in untopped diaphragms, it may be
unconservative to assume shear-fric-
tion can develop between the anges
because measured connector com-
pression stiffness is very high and the
anges are unlikely to come in con-
tact. Since untopped parking structures
use pretopped [4 in. (102 mm)] double
tees, the equivalent beam model ap-
pears to be acceptable, and conserva-
tive, for parking structure displacement
estimates.
In the Nakaki method,
12
the shear
stiffness of connectors at a joint is not
included in the deection calculation.
The method does not reect the bene-
cial effect on reducing the diaphragm
deformation from using connectors
with a high shear stiffness. The ange
thickness of the double tees is actually
omitted in the deection calculation.
Thus, the calculated midspan deec-
tion of a diaphragm with 4 in. (102 mm)
thick anges is the same as that of one
with 2 in. (51 mm) thick anges. Over-
all, the deection obtained from the
Nakaki method is larger than that from
a more detailed nite element analysis
or the equivalent beam model.
SUMMARY AND
CONCLUSIONS
A simplied deection analysis was
performed to provide precast designers
with a practical approach for calculat-
ing linear elastic in-plane exibility and
lateral deformation of untopped precast
double-tee oor diaphragms that use
discrete mechanical ange connectors.
Complex nite element modeling
and mechanical connector properties
derived from tests served as the basis to
develop the simplied analysis meth-
od. Simplifying assumptions, justied
through the accurate FEM model, led to
the development of an equivalent beam
model for practical applications. The
Fig. 5. Shear test results and schematic of the bent wing connector. Face plate at edge
of tee ange is white. Note: 1 in. = 25.4 mm; 1 kip = 1.45 kN.
Fig. 4. Example of one-span diaphragm layout. Note: 1 ft = 0.305 m.
10 PCI JOURNAL
equivalent model was shown, through
example calculations and compari-
son with the FEM analysis, to provide
deection prediction accuracy within
14 percent of that of a much more com-
plicated and accurate analysis.
Using the proposed equivalent beam
model, the exibility of linear elastic
pretopped double-tee diaphragms can be
modeled with reasonable accuracy, and
deections can be calculated from con-
ventional beam equations. To achieve a
satisfactory analytical model and reect
the exibility of untopped double-tee
diaphragms, both the shear and tension
stiffnesses of mechanical connectors are
required. The model is appropriate for
pretopped diaphragms with:
1. Closely spaced ange
connectors [less than 10 ft
(3.1 m)].
2. Connectors that have axial
tension stiffness between 16
and 160 kips/in. (2.8 and
28 kN/mm).
3. Similar connectors used
throughout the diaphragm.
4. Chords formed from special
mechanical connectors or a
reinforced pour strip.
The equivalent beam model can be
directly established by considering
the diaphragm to be a horizontal beam
with a width equal to the double-tee
ange thickness and span equal to the
diaphragm span, but setting the depth,
d, and the material elastic modulus, E,
to equivalent values as given in Eqs.
(18) and (19).
The proposed model provides a sim-
plied but reasonably accurate and
practical method for predicting the
linear elastic deection of double-tee
diaphragms resisting lateral in-plane
loading. There is certainly the oppor-
tunity to further improve and verify
the proposed method through tests on
chords, full diaphragm testing, and
building eld studies. At present, pre-
cast concrete diaphragm design should
be based on maintaining elastic behav-
ior because of the low deformability of
mechanical ange connectors.
Further research is desirable, through
testing and analysis, to understand the
behavior of precast diaphragms when
the linear elastic capacity threshold is
exceeded. The likelihood of concentra-
tion of inelasticity in a few joints and
the accompanying deformation demand
during an extreme event level earth-
quake motion should be identied.
ACKNOWLEDGMENTS
The research reported in this paper
was partially funded by the National
Science Foundation (NSF) under Grant
No. CMS-9412906. The following PCI
Producer Members also participated in
funding portions of the study through
the PCI Research and Development
Program: Atlanta Structural Concrete,
Blakeslee Prestress, Concrete Technol-
ogy, Ferreri Concrete Structures, Me-
tromont Prestress Corporation, Rocky
Mountain Prestress, Spancrete Indus-
tries, Spancrete Midwest Inc., Tindall
Corporation, and Wells Concrete.
The opinions, ndings and conclu-
sions expressed in this paper are those
of the authors and do not necessarily
reect the views of the sponsoring or-
ganizations.
The technical input provided by an
industry advisory panel is also grate-
fully acknowledged. Special recogni-
tion is due the PCI JOURNAL review-
ers who provided invaluable comments
and recommendations to the authors.
REFERENCES
1. EERI, Northridge Earthquake of
January 17, 1994 Reconnaissance
Report, Volume 2, William T. Holmes
and Peter Sommers (Technical Editors),
Earthquake Spectra, Earthquake
Engineering Research Institute,
Supplement C to Volume 11, January
1996.
2. NIST, 1994 Northridge Earthquake:
Performance of Structures, Lifelines,
and Fire Protection System, NIST
Special Publication, National Institute
of Standards and Technology,
Gaithersburg, MD, 1994.
3. Iverson, J. K., and Hawkins, N. M.,
Performance of Precast/Prestressed
Concrete Building Structures During
the Northridge Earthquake, PCI
JOURNAL, V. 39, No. 2, March-April
1994, pp. 38-55.
4. Wood, S. L., and Stanton, J. F.,
Performance of Precast Parking
Garages in the Northridge Earthquake:
Lesson Learned, Proceedings of
Structures Congress XIV, Volume 2,
American Society of Civil Engineers,
New York, NY, 1996, pp. 1221-1227.
5. Fleischman, R. B., Sause, R., Rhodes,
A. B., and Pessiki, S., Seismic
Behavior of Precast Parking Structure
Diaphragms, PCI JOURNAL, V. 43,
No. 1, January-February 1998, pp.
38-53.
6. Zheng, W., Analytical Method for
Assessment of Shear Capacity Demand
for Untopped Precast Double-Tee
Diaphragms Joined by Mechanical
Connectors, Ph.D. Dissertation,
University of Wisconsin-Madison,
Madison, WI, 2001.
7. PCI Design Handbook: Precast and
Prestressed Concrete, Fifth Edition,
Table 4. Maximum in-plane deection at midspan of diaphragm (in.).
Analysis
model
Equivalent
beam model
Nakaki
method
Planar nite
element model
Case 1 0.239 0.471 0.224
Case 2 0.184 0.229 0.154
Case 3 0.254 0.471 0.273
Case 4 0.203 0.338 0.178
Case 5 0.216 0.238 0.184
Note: 1 in. = 25.4 mm.
Table 3. Parameters of equivalent beam model.
Analysis
model
Thickness
of section,
t (in.)
Depth
of section,
d (in.)
Elastic
modulus,
E (ksi)
Shear
modulus,
G (ksi)
Case 1 4 683.33 295.34 113.59
Case 2 4 524.84 719.50 276.73
Case 3 2 685.52 550.85 211.87
Case 4 4 547.17 587.74 226.05
Case 5 4 661.92 351.66 135.25
Note: 1 in. = 25.4 mm; 1 ksi = 6.89 MPa.
March-April 2005 11
Precast/Prestressed Concrete Institute,
Chicago, IL, 1999.
8. ICC, International Building Code,
International Code Council, Falls
Church, VA, 2003.
9. ICBO, 1997 Uniform Building Code,
International Conference of Building
Ofcials, Whittier, CA, 1997.
10. Farrow, K. T., and Fleischman, R. B.,
Effect of Dimension and Detail on the
Capacity of Precast Parking Structure
Diaphragms, PCI JOURNAL, V.
48, No. 5, September-October 2003,
pp. 46-61.
11. Fleischman, R. B., and Farrow, K. T.,
Seismic Design Recommendations
for Precast Concrete Diaphragms in
Long Floor Span Construction, PCI
JOURNAL, V. 48, No. 6, November-
December 2003, pp. 46-62.
12. Nakaki, S. D., Design Guidelines for
Precast and Cast-in-Place Concrete
Diaphragms, The 1998 NEHRP
Professional Fellowship Report,
Earthquake Engineering Research
Institute, Berkeley, CA, 1998.
13. Venuti, W. J., Diaphragm Shear
Connectors Between Flanges of
Prestressed Concrete T-Beam, PCI
JOURNAL, V. 15, No. 1, February
1970, pp. 67-79.
14. Venuti, W. J., and Nazarian, D.,
Diaphragm Shear Connectors Between
Flanges of Prestressed Concrete T-
Beam, Report No. ST-0007-68, San
Jose State College, San Jose, CA, 1968.
15. Kallros, M. K., and Spencer R. A.,
An Experimental Investigation of the
Behavior of Connections in Thin Precast
Concrete Panels Under Earthquake
Loading, MS Thesis, University of
British Columbia, Vancouver, British
Columbia, Canada, April 1987.
16. Pincheira, J. A., Oliva, M. G., and
Kusumo-Rahardjo, F. I., Tests on
Double-Tee Connectors Subjected to
Monotonic and Cyclic Loading, PCI
JOURNAL, V. 43, No. 3, May-June
1998, pp. 50-67.
17. Aswad, A., Selected Precast
Connections: Low Cycle Behavior
and Strength, Second U.S. National
Conference in Earthquake Engineering,
Stanford, CA, August 1979.
18. Timoshenko, S. P., and Goodier, J. N.,
Theory of Elasticity, Third Edition,
McGraw-Hill, New York, NY, 1970.
19. ACI Committee 318, Building Code
Requirements for Structural Concrete
(ACI 318-02) and Commentary
(ACI 318R-02), American Concrete
Institute, Farmington Hills, MI, 2002.
20. ACI Committee 408, Bond Stress
The State of Art, ACI Journal,
Proceedings, V. 63, No. 11, November
1966, pp. 1161-1190.
21. Alsiwat, J., and Saatcioglu, M.
Reinforced Anchorage Slip Under
Monotonic Loading, Journal of the
Structural Division, American Society
of Civil Engineers, V. 118, No. 9,
September 1992.
22. Oliva, M. G., Testing of the JVI
Flange Connector for Precast Concrete
Double-Tee Systems, Structures and
Materials Test Lab Report, University
of Wisconsin-Madison, Madison, WI,
June 2000 [www.jvi-inc.com].
23. Farrow, K. T., Private Communication,
May 10, 2004.
APPENDIX A NOTATION
= 1 when mechanical connector
is not placed at diaphragm
edge (normal case)
= 0 when connector is placed at
chord location
A = cross-sectional area of
a double-tee ange in
longitudinal section
A = cross-sectional area of
equivalent beam
A
s
= cross-sectional area of chord
steel
b = width of double-tee ange (see
Fig. 1)
d = depth of equivalent beam
d = depth of diaphragm (see Fig. 1)
E = elastic modulus of equivalent
beam
E
c
= elastic modulus of concrete in
double tee
E
s
= elastic modulus of chord steel
f
max
= chord steel stress at joint, T/A
s
G = shear modulus of equivalent
beam
G
c
= shear modulus of concrete
I = in-plane transverse moment
of inertia of double-tee ange
section (webs ignored)
I = moment of inertia of
equivalent beam
k = joint modication factor
K
ai
= axial stiffness of ith connector
or chord steel
K
c
= tension stiffness of chord
connection
K
s
= tension stiffness of chord steel
at joint between double tees
K
t
= tension stiffness of each ange
connector
K
vi
= shear stiffness of ith connector
l
d
= development length of
embedded steel
L = span of diaphragm
M = in-plane moment between
adjacent double tees
m = number of double tees in
diaphragm span being
considered
n = number of ange connections
(including chord connections)
r
i
= distance from ith connector or
chord to top end of double tee
(see Fig. 1)
T = tension in chord steel
t = thickness of equivalent beam
t = thickness of double-tee ange
V = in-plane shear force between
two double tees
w = uniform lateral load on
diaphragm per unit length
x = distance from support of
equivalent beam (see Fig. 2)
y = distance from neutral axis to
top end of double tee (see
Fig. 1)
= shear deformation of
equivalent beam section
= total shear deformation of
ange and joint

1
= shear deformation of ange

2
= shear deformation in joint
= strain in chord steel at joint
= Poissons ratio for material of
equivalent beam
= Poissons ratio for material of
actual concrete

= in-plane exural stiffness of

joint between double tees

= total shear stiffness of

mechanical connectors in
joint =
i
K
i
= exural rotation in equivalent
beam

1
= exural in-plane rotation of
single double tee

2
= exural in-plane rotation
within joint between double
tees
= total exural rotation of joint
and double tee
12 PCI JOURNAL
The midspan deection of the single-span untopped dia-
phragm shown in Fig. 5 is calculated below using the equiva-
lent beam model. The diaphragm conguration is chosen as
Case 1, with the bent wing connector in a 4 in. ange, from
Table 1.
Starting data:
Connector shear stiffness K
vi
= 300 kips/in.
Connector tension stiffness K
t
= 16 kips/in.
Chord steel area, 3 No. 6 bars
A
s
= 3(0.44)
= 1.32 sq in.
Number of connections,
including two chords
n = 11
Number of double tees in span m = 12
Panel ange thickness t = 4 in.
Panel width
b = 10(12)
= 120 in.
Panel length
d = 60(12)
= 720 in.
Diaphragm span
L = 120(12)
= 1440 in.
Total load on diaphragm V = 123.5 kips
Concrete strength f
c
= 5000 psi
Reinforcement strength f
y
= 60,000 psi
Elastic modulus of concrete E
c
= 4070 ksi
Elastic modulus of steel E
s
= 29,000 ksi
Poissons ratio of concrete = 0.3
Calculations:
1. Calculate the development length for a No. 6 reinforcing
bar in accordance with ACI 318-02:
No. 6 rebar diameter, d
e
= 0.75 in.
A 1 in. thick cover over chord steel is assumed;
therefore, cover dimension, c = 1 + d
e
/ 2 = 1 + 0.75/2 =
1.375 in.
No transverse reinforcement is used; therefore,
transverse reinforcement index, K
tr
= 0
(c + K
tr
) / d
e
= (0.375 + 0) / 0.75 = 1.833 < 2.5
Reinforcement location factor, = 1
Coating factor, = 1
= 1 < 1.7
Concrete aggregate factor, = 1
= 1
No. 6 rebar development length:
L
d
= d
e

3
40

f
y
f
c

c + K
tr
d
e

= 0.75
3
40

60,000
5000

(1)(1)(1)(1)
1.375 + 0
0.75
= 26.03 in.
Conservatively, take L
d
= 27 in.
2. Calculate chord steel tension stiffness at joints in
accordance with Eq. (21):
K
c
=
3
4

A
s
E
s
L
d
=
3
4

1.32(29,000)
27
= 1063.33 kips/in.
3. Calculate shear stiffness at joints:

V
= (n 2)K
V
= (11 2)(300) = 2700 kips/in.
4. Calculate rotation stiffness at joints in accordance with
Eq. (7):
No connector at chord steel location; therefore, = 1.

= (K
c
K
t
)
d
2
2
+ nK
t

d
2
12
= 1063.33 (1)(16)
720
2
2
+ 11(16)
720
2
12
= 2.791 10
8
kips-in.
5. Calculate double-tee panel properties:
Cross-sectional area, A = td = 4(720) = 2880 sq in.
Cross-sectional moment of inertia:
I = td
3
/12 = 4(720)
3
/12 = 1.244 10
8
in.
4
6. Calculate concrete shear modulus:
G
c
= E
c
/2(1 + ) = 29,000/2(1 + 0.3) = 1565.39 ksi
APPENDIX B EQUIVALENT BEAM EXAMPLE
March-April 2005 13
7. Calculate shear modulus modication factor:
k = (m + 2)/m = (12 + 2)/12 = 1.167
Section depth [Eq. (18)]:
d = d
1 +
kG
c
A
1.2

b
1 +
E
c
I

r
b

= 720
1 +
1.167(1565.39)(2880)
1.2

(2700)(120)
1 +
4070(1.244 10
8
)
(2.791 10
8
)(120)
= 683.51 in.
Elastic modulus [Eq. (19)]:
E = E
c

d
d
1 +
kG
c
A
1.2

b
=
720
683.51
1 +
1.167(1563.39)(2880)
1.2(2700)(120)
= 295.11 ksi
Section thickness, t = t = 4 in.
Poissons ratio, = = 0.3
Equivalent cross-sectional area:
A = dt = 683.51(4) = 2734.03 sq in.
Equivalent section modulus:
I = td
3
/12 = 4(683.51)
3
/12 = 1.064 10
8
in.
4
Equivalent shear modulus:
G = E/[2(1 + )] = 295.11/[2(1 + 0.3)] = 113.50 ksi
8. Calculate diaphragm midspan deection:
Diaphragm span, L = 12(10)(12) = 1440 in.
Load on diaphragm, w = 123.5/L = 0.086 kips/in.
Deection due to shear:

s
= 1.2wL
2
/8GA
= 1.2(0.086)(1440)
2
/8(113.50)(2734.03)
= 0.086 in.
Deection due to bending:

b
= 5wL
4
/384EI
= 5(0.086)(1440)
4
/384(295.11)(1.064 10
8
)
= 0.153 in.
Total deection:
=
s
+
b

= 0.086 + 0.153
= 0.239 in.