Design Considerations for the Connection

of Precast Concrete Structures

3
S.H. Rizkalla', M.L. Lau2 , and K.A. Soudki

SUMMARY
The structural integrity of precast concrete buildings depends mainly on the connections between
the precast structural components. Large precast concrete panels are used extensively in North America
for low, medium, and highrise construction due to their ease and speed of assembly, and high quality
control.
This paper presents the results of twenty-two prototype specimens tested to investigate the
behaviour and the capacity of typical connections used for precast loadbearing shear wall panels subjected
to shear loading.
The different configurations include dry-packed for plain surface panels, two sizes of dry-packed
multiple shear keys, dry-packed plain surface and continuity bars with and without mechanical shear
connectors, and post-tensioned connections. Shear walls supporting hollow-core floor slabs were also
included. Different levels of load nomlal to the connection were used to simulate the effects of gravity
and permanent loading conditions.
Test results were used to refine and calibrate rational mathematical models developed to predict
the maximum shear capacity and the nominal shear strength of the connection. Various failure
mechanisms are presented and discussed.
The findings of this investigation were implemented in the design of three 32-storey apartment
buildings in Winnipeg, Manitoba, and provided significant savings in construction time and cost.

Professor, Civil Engineering Department, University of Manitoba, Winnipeg, Manitoba, R3T 2N2
Engineering Manager, Con-Force Structures, Winnipeg, Manitoba, R2W 3R4
Doctoral Candidate, Civil Engineering Department, University of Manitoba, Winnipeg, Manitoba,
R3T 2N2

Rizkalla et al., "Design Considerations for ... " 1

INTRODUCTION
Behaviour of the connections typically used for precast concrete load-bearing shear wall panels
greatly influences the structural integrity of the entire structure. Current design sources (ACI. 1989; PCI.
1985; CSA. 1984; CPCI. 1987) provide very limited infomlation to describe tile behaviour and design
of such connections. The information available in the literature (Birkeland and Birkeland. 1966; Mast.
R.F. 1968; Hofbeck and Abrallim. 1969; Mattock and Hawkins. 1972; Mattock. 1974). which mainly
describes the shear friction concept and the dowel action mechanism of initially cracked concrete. is not
directly applicable to the behaviour of such connections.
This paper discusses the results of tllree phases of an experimental program undertaken to
exanline the behaviour of nine connection configurations as given in Table I. The first phase (Foerster
ct al .• 1989) included four types of connections: a dry-packed plain surface connection. a dry-packed
connection with continuity reinforcement. and dry-packed connections with two types of mechanical shear
connectors. in addition to the continuity bars. The second phase (Serrette et al .• 1989) included two
different types of dry-packed multiple shear key configurations. The third phase (Hutchinson et al .• 1991)
included vertical post-tensioning tendons to replace the continuity bars. Testing included shear walls
supporting hollow core slabs. All the connections were subjected to a monotonic shear load up to failure.
Three different levels of load normal to the connection were used to simulate the effects of gravity and
permanent loading conditions.

RESEARCH SIGNIFICANCE
The primary objective of this paper is to investigate the limit states behaviour of nine connection
configurations used for precast concrete load-bearing shear wall panels. Twenty-two prototype precast
concrete shear wall panel specimens were used to test nine different connections currently used by the
construction industry. The study identified tlle contribution of each component used for tllese types of
connections. The results of tlle experimental program were used to refine and calibrate proposed rational
mathematical models introduced to predict tlle strength of such connections before and after cracking.

EXPERIMENTAL PROGRAM

Test Specimens
A total of twenty-two specimens were tested. The nine connection configurations tested in this
study are as follows:
Type I: Dry pack only.
Type II: Dry pack and 25M continuity bars.
Type III: Dry pack. 25M continuity bars. and shear connectors Type A.
Type IV: Dry pack. 25M continuity bars. and shear connectors Type B.
Type V: Large-size dry-packed multiple shear keys.
Type VI: Small-size dry-packed multiple shear keys.
Type VII: Dry pack and hollow core slab.
Type VIII: Dry pack. hollow core slab and post-tensioning.
Type IX: Dry pack and post-tensioning.
The configuration of tile post-tensioned connections supporting hollow core slabs and tlle dimensions of
the precast panels used for typical specimens are shown in Figure I. Details of tlle otller six connections
are shown in Figure 2.
Each specimen consisted of two precast panels. as shown in Figure 1. All parameters. such as
steel details and concrete material. were kept constant. Prior to testing. specially-designed temporary steel
brackets were used for specimens with dry pack only. The dry pack mix consists of 2 parts concrete
sand. I part nonnal portland cement. and approximately 0.5 parts water. The mix was placed and
compacted into tlle 20 mm wide joint space between the two panels and tlle multiple shear keys. The
mixture was of a dry consistency. which allows tlle tamping and compaction of the material.
For phase III specimens. tlle hollow-core slab rests on "Korolatll" bearing pads on the bOllom
panel. The cores of tlle hollow-core slab and tlle gap between tlle ends of two slabs are filled witll a

Rizkalla et al., "Design Considerations for .. , " 2

 flow able concrete fill. The gap between the hollow-core and the top panel is dry-packed. Seven days
after drypacking, the strands were post-tensioned and the ducts were filled using an expansive grout.

Testing Apparatus and Procedure

Each specimen was positioned vertically into the testing machine, as shown in Figure 3. The
joint centerline of the specimen was aligned with the vertical centerline of the top and bottom heads of
the testing machine. To study the effect of the gravity load, Ulree levels of preload applied nornlal to the
joint were used. The testing machine was used to apply the vertical shear load in increments of 100 kN.
At each increment, readings of all instrumentation were recorded followed by marking of the cracks. The
experiment was terminated after extensive deformation and a significant drop in the load-carrying capacity
was observed.

TEST RESULTS AND DISCUSSION

Selected typical failure patterns of Ule various connections tested in Phases I, II and III are shown
in Figures 4a, 4b, 4c, and 4d. In general, Ule predominant mode of failure was due to slip along the
drypack to panel interface. Cracking of the hollow core at Ule bottom of the connection was observed
in Phase III, as shown in Figure 4d. Typical load-slip behaviour of the connections tested in Phase I
under a preload level of 4 MPa is shown in Figure 5. The behaviour of the specimens with multiple
shear keys, tested in Phase II under the sanle preload level is shown in Figure 6. The load-slip behaviour
of the post-tensioned connections, with and without Ule hollow core slab, subjected to preload levels of
8 MPa is shown in Figure 7.
The test results of Phase I indicate that the cracking strength of the connection depends mainly
on Ule bond strength at the dry pack concrete interface. The residual shear, which represents Ule nominal
shear strength of the connection, is related to Ule level of the load normal to the joint, the dowel action
of the continuity bars, and the resistance of Ule mechanical connectors. Thus, the nominal shear strength,
V"' of these connections can be predicted using the following proposed model:

Vn = I.l on Ae + Ab f/"3 + Vw ( I)

where
~L = friction coefficient factor, 0.7 is proposed
0" = compressive stresses nonnalto the connection, psi
Ae = cross-sectional area of the connection, in2
fy = yield strength of Ule continuity bars, psi
Ab = area of the continuity bars, in2
Vw = shear strength of the weld, Ib
The predicted nominal shear resistance based on the proposed model was found to be in good
agreement with the measured values of Phase I, as shown in Figure 8.
The test results of Phase II indicate that using multiple shear keys will enhance Ule maximum
shear capacity by as much as 60 percent in comparison to the plain surface connections at the same level
of preload. The behaviour of these connections suggests that the shear capacity depends mainly on the
strength of the weaker material wiUlin the vicinity of the connection and the level of the load nomlal to
the connection. The difference in multiple shear key configurations used in this study had no measurable
effect on the ultimate shear load capacity of the connection. It should be noted that the strengUI of Ule
dry pack used in Phase II is considerably lower than the dry pack strengUI used in Phase I.
Based on the observed behaviour after cracking, the maximum shear capacity, VR!' of the multiple
shear key connection is mainly governed by the compressive strengUI of the struts between the diagonal
cracks and the shear friction resistance along the slip surface, as illustrated in Figure 9. Therefore, Ule
predicted maximum shear capacity, Vm' can be expressed in ternlS of these two components as follows:

Vm = Vme + Vmr (2)

where Vme is the shear resistance of the strut mechanism and VillI is the shear friction resistance along the
slip surface. In this analysis, Ule shear wall panels are assumed to act as rigid bodies connected by n-I

Rizkalla et aI., "Design Considerations for ... " 3

struts, where n is the number of shear keys. For the three keys in Figure 9(a), the struts arc shown
schematically in Figure 9(b). The compressive strength of the cracked drypack. fel• could be taken as
0.67 fa' for these types of connections, where fa' is the compressive strength of the drypack (Serrelte et
aI., 1989). TIlUS, the shear resistance of the strut mechanism, Vme , may be estimated as:

(3)

where At. is the average cross-sectional area of the diagonal portion of the strut and a is the inclination
of tbe diagonal portion of the strut to the horizontal.
Representing the connection by a rectangular strip, the distribution of forces at the connection,
including the shear friction resistance, V one ' provided by slip along Ille drypack-panel interface and along
the diagonal cracks, is shown in Figure 9(c). The shear friction resistance, V me , may be evaluated as:

(n-l)fo2Ac,cosa
(4)
Vm! = Il( on - ) At
Ac
where II is Ille friction coefficient and on is the compressive stress normal to the connection. Therefore,
the maximum shear capacity after cracking, V... according to Eq. (2), can be estimated as:

Vru = (n-1) fo2 A.. sina + II (on - (5)

The predicted maximum shear capacities according to Eq. (6) were in good agreement WiUl the measured
values of Phase II, using a value of 0.6 for the friction coefficient.
The nominal shear resistance. V'ik' of the multiple shear key connection mainly depends on the
level of load nonnal to the connection, and the bearing stresses and shear friction along the slip surfaces.
As discussed earlier, the configuration of the shear keys considered in this investigation was found to
have an insignificant effect on Ule shear capacity. Using a linear regression analysis, the following model
was developed to predict the nominal shear resistance of the multiple shear key connections in temlS of
the bearing and shear resistances:

(7)

where ACk is the cross-sectional area for Ule pOition of U1e connection covered by the shear keys, and Ac
is the cross-sectional area for Ule entire length of the connection. Ack is equal to Ac if the shear keys
cover the entire length of the connection.
The predicted nominal shear resistance, using Eq. (6), was in excellent agreement with the
measured values of Phase II, as shown in Figure 10. .
The test results in Phase III indicate that Ule increase in the maximum shear capacity and nominal
shear strength is directly proportional to the increase of the load normal 10 the connection. At the low
stress level, 4 MPa, nonnalto the connection, Ule presence of hollow-core slab had no or little effect on
the behaviour or the capacity of the post-tensioned connection. However, at higher levels of stress normal
to the connection, the presence of hollow-core slab significantly reduced Ule maximum and nominal shear
capacities of the connection.
The maximum shear capacity could be predicted as the lesser of that determined by U1e friction
resistance model, Ve, Eq. (7), and the cracking capacity of the hollow-core, Vh, Eq. (8b).

(7a)

(7b)

Rizkalla et aL, "Design Considerations for ... " 4

The shear friction resistance. Vr. can be predicted using the shear friction model which is related to the
area of the concrete interface. Ac. and the coefficient of friction. ~. of the dry pack to panel interface.
A coefficient of friction of 0.7 is proposed based on the test results of Phase I investigation.
The cracking capacity of the connection. Vh' based on the capacity of the hollow-core slab with
concrete fill. can be predicted using the areas of the hollow-core and concrete fill in contact with the
drypack. Acl • and Aa. respectively. The magnified tensile capacities of the concrete of the hollow-core
and the concrete fill. Fu and F12 • respectively. are based on the tensile strength of each material. ftl and
f12• The compressive normal stress. crill' is the poltion of the total nonnal stress. including the effect of
prestressing acting on the hollow core slab portion of the connection. cr,ll is the nomlal stresses acting
on the concrete fill portion of the connection (Hutchinson et aI.. 1991).
The nominal shear strength of horizontal connections with hollow-core could be predicted as the
lesser of that determined by the shear friction model. Vr• Eq. (7a). and the ultimate capacity of the
hollow-core. VlUI' Eq. (8):

(8)

with

Where ACIl is the summation of the area of the webs at mid-height of the hollow-core beneath the contact
surface area Acl ' For the given geometry of the hollow-core tested in this investigation. ACIl == A,/4.
In this model. the contribution from the centre concrete fill remains unchanged while the
contribution from the hollow-core slabs are modified to reflect the loss of bond. The area of the hollow-
core in contact with the dry pack. Ad' is reduced by a factor of four. However. the normal stress. cr'ul'
is increased, also by a factor of four. due to the reduction of the contact area at ultimate.
The comparison between the measured and predicted nominal shear strength suggests that the
proposed model provides a conservative lower bound for the nominal shear capacity of horizontal
connections with hollow-core slab as shown in Figure 11.

CONCLUSION
Twenty-two specimens with nine different connection configurations were tested under monotonic
shear loading conditions to investigate the various limit states behaviour of horizontal connections. The
effect of connection configuration. load normal to connection. presence of hollow-core. and post
tensioning were investigated.
Based on the results of this study. Ule following conclusions could be drawn:
1. An increase in Ule level of load nonnal to the connection increases the maximum shear capacity
of this connection.
2. The presence of shear keys in the horizontal connection enhances the shear capacity in
comparison to the plain surface connection.
3. The difference in the shear key configurations considered in this study had insignificant effect on
tile behaviour or capacity of the connection.
4. The shear Cllpacity depends mainly on the strengUI of the weaker material witilin the vicinity of
the cOImection.
5. The maximum shear capacity of the connection with hollow-core slab is governed by the lower
magnitude of the shear friction capacity and the hollow-core shear capacity.
6. TIle nominal shear strength of the connection with hollow-core slab is based on complete loss of
bond. The predicted values provide a conservative lower bound.
7. Replacing the continuity bars in the shear wall panels WiUI vertical post-tensioning for connecting
the panels enhances the shear capacity. in addition to its economical advantages.

Rizkalla et aI., "Design Considerations for ... " 5

ACKNOWLEDGEMENTS
This experimental program was conducted at the University of Manitoba. Winnipeg. Canada. with
financial assistance from the Natural Sciences and Engineering Research Council of Canada (NSERC)
and Con-Force Structures Ltd .• Winnipeg. Manitoba.

REFERENCES

1. American Concrete Institute (ACI). 1989. Building Code Requirements for Reinforced Concrete.
Detroit. Michigan.

2. Birkeland. P.W. and Birkeland. H.W. 1966. Connections in Precast Concrete Construction. ACI
Joumal. March. pp. 345-367.

3. Canadian Prestressed Concrete Institute (CPCI). 1987. "Metric Design Manual". Ottawa. Ontario.

4. Canadian Standards Association (CSA). 1984. Design of Concrete Structures for Buildings
(CAN-A23.3-M84). Rexdale. Ontario.

5. Foerster. H.R .• Rizkalla. S.H.. and Heuvel. 1.S.. 1989. Behaviour and Design of Shear
Connections for Load-Bearing Shear Wall Panels. PCI Joumal, Vol. 34, No. I. lanuary/February.
pp. 100-119.

6. Hofbeck. J.A. and Abrahim. 1.0. 1969. Shear Transfer in Reinforced Concrete. PCI JOllmal.
February. pp. 119-128.

7. Hutchinson. R.L.. Rizkalla, S.H .• Lau. M.L. .. and Heuvel. T.S. 1991. Horizontal Post-Tensioned
Connections for Precast Concrete Load-Bearing Shear Wall Panels. Accepted for publication in
the PCI Joumal.

8. Mast. R.F. 1968. Auxiliary Reinforcement in Concrete Connections. Journal of Structural

Division. ASCE. June. pp. 1485-1504.

9. Mattock. A.H. 1974. Shear Transfer in Concrete Having Reinforcement at an Angle to the Shear
Plant. ACI Special Publication SP42. Volume 1. pp. 17-42.

10. Mattock, A.H. and Hawkins, N.M. 1972. Shear Transfer in Reinforced Concrete -- Recent
Research. PCI Joumal. March, pp. 55-75.

11. Prestressed Concrete Institute (PCI). 1985. PCI Design Handbook. Chicago, Illinois.

12. Serrette, R .• Rizkalla. S.H., Auiogbe. E.• and Heuvel. 1.S. 1989. Multiple Shear Key
Connections for Precast Load-Bearing Shear Wall Panels. PCI Joumal, Vol. 34. No.2.
March/April, pp. 104-120.

Rizkalla et aI., "Design Considerations for ... " 6

 Table I. Overall test parameters.

Dry Pack Load nonnal Type of Specimen

Configuration to connection connection mark
MPa

Phase I
Plain surface 2 SPit
II SPI2
III SPI3
IV SPI4

4 I SP21
II SP22
III SP23
IV SP24
Phase II
Multiple shear 2 V lLK2
key VI ISK2
4 V 2LK4
VI 2SK4

Plain surface 2 INK4

Phase III
Dry pack and 4 VII lHD4
hollow core slab 8 VII 2HD8
8 VII 3HD8
Dry pack. hollow 4 VIII 4HP4
core slab and 8 VllI 5HP8
post -tensioning 8 VIII 6HP8
Dry pack and 4 IX 7PD8
post -tensioning 8 IX 8PD8
8 IX 9PD8

Rizkalla et al., "Design Considerations for ... " 7

 20mm wide
I,Jroufed jont
spoce Continuity bars
(2!5M renforcinq welded
to ~ 11 ~"IO a191e iron )

Type I Type II

====~~~~=Shea type A
connector
(S-sectionwelded
===:=::=~~~~=
heaconnector

b
j]~~~ with
to coonec1ion
ongla iron
plate) -
type B
(15M reinforcinq
;::= .---- CI'Id plate with
~ comection plate)
~~= ~~~
Type m Type :&

Type ~ Type '5lI

Figure 2. Details of the nine connections considered in this study.

Rizkalla et al., "Design Considerations for ... " 9

 Figure 3. Test sel-up

Rizkalla et al., "Design Considerations for ... " 10

Rizkalla et al., "Design Considerations for ... " 11
 3LK4E:>
5\DE A

(c)

(d)

Figure 4. Typical failure of Ule connection (cont'd)

RizkaJla et al., "Design Considerations for ... " 12

 1.6 • SP 21

. 1.4
o SP
+ SP
22
23
.
"0
c:
0
::J
0
1.2
ASP 24

~ 1.0 -
z
-'"
• 0.6
a
8J 0.6
a::
;5 0.4
:r
II)
0.2

0.0
-2 0 2 6 10 14 16 22
SLIP (mm)
Figure 5. Load-displacement of the connections tested in
Phase I under preload of 4 MPa
1.2
'"c:
"0
<> ZLK4
a 1.0 D. ISK4
'"a::J • INK4
.£:.
f- 0.8 -
Z
-'" 0.6
0
«
0
...J
0.4
0::
«
w 0.2 -
J:
en
0.0
-2 a 2 6 10 14
SLIP (mm)
Figure 6. Load-displacement of the connections tested in
Phase II under preload of 4 MPa

1200
,-..
~ 1000
'-/

Cl 800
«
0
....J 600
0::
«w 400
A 8PD8
::x: 200 .5HP8
(/)

0 I
0 2 3 5
SLIP (mm)
Figure 7. Effect of the presence of hollow-core slab at high
stress level nomlal to the connection

Rizkalla et aL, "Design Considerations for ... " 13

 1600~-------------------------------'

SP24 23
~
z.x
1200 • •
oc::(
g 800
o
W
I-
U
aw
Ii 400
•" PROPOSED MODEL

o
a 400 800 1200 1600
MEASURED LOAD ( kN )

Figure 8. Relationship between the measured and predicted

nominal shear strength of the connection for
Plmsc I

ac(
o

.g
t • Ponel Thickn."

(01

P SHEAR fRICTION

STRUT
FORCES
\
\~ STRU~j
fORCE~
(ZP' ~

J\ \p
....
w~g-l
z _
ZO..J
.t :j
,,"

(bl (el

Figure 9. Force mechanism for multiple shear connections

Rizkalla et al., "Design Considerations for ... " 14

 ~ 1100'-.---.---.---.----.---,---,---,---,---,---.",

~
'-'
NOMINAL
t'IOOO
·8
oj

~ unconservative
c..> 800

3LK4B
lSK • • 2LK4
3SK48

1~~~ conservative

ZOO~-.---II._-,r--ri--_r-_r-__._-_r--_,_-_4
200 400 SOO BOO 1000 1200
Experimental Results (KN)

Figure 10. Relationship between measured and predicted

nominal shear strengUl of the connection for
Phase II

~IIOO 9PDB B 6
NOMINAL SHEAR
~
'-'
STRENGTH •
1:
:f1bII
~

.......
Q)
unconservative
(J)

~ 800 conservative
Q)
.c:
(J)
7PD
-uQ)
....tJ 5HPB 6HP8
.....
-uII> •
3HDS. .2HD6 •
... 4HP4

• •
ll. HD4

400
400 BOO 11.00
Experimental Results (kN)

Figure 11. Relation between measured and predicted nominal

shear strengtlJ of ilie connection for Phase III

Rizkalla et aJ., "Design Considerations for ... " 15