You are on page 1of 16

TITLE NO.

68-S7
Seismic Resistance of Prestressed Concrete
Beam-Column Assemblies
By ROGER W. G. BLAKELEY
and ROBERT PARK
ACI JOURNAL I SEPTEMBER 1971
A series of four tests were conducted on full
size precast, prestressed concrete beam-column
assemblies under reversed cyclic loading of high
intensity. The test variables included the amount
of transverse confining steel for ductility and the
position of the plastic hinge in the members. It
was concluded that large post-elastic deformations
can be attained in prestressed concrete members
and that the energy dissipation at large deforma-
tions can be considerable, but only once the con-
crete in the compression zone has commenced to
crush and hence when structural damage has been
incurred.
Keywords: beams (supports J; bond (concrete to
reinforcement); columns (supports); connections;
cyclic loads; deformation; ductility; earthquake
resistant structures; energy dissipation; frames;
hinges (structural); joints (junctions); posttension-
ing; precast concrete; prestressed concrete; re-
search .
THE SATISFACTORY BEHAVIOR OF prestressed con-
crete under gravity loading has long been ac-
cepted but its application to primary seismic re-
sistant elements has not met with such ready
acceptance. This caution has been due to the lack
of information on the behavior of prestressed con-
crete structures under severe earthquakes and,
particularly, to the shortage of information on the
behavior of prestressed concrete members under
high intensity cyclic loading. A survey of the re-
search which has been conducted on the seismic
resistance of prestressed concrete has been pre-
viously published.
1
The three general regions of a moment-rotation
curve for a prestressed concrete member under
montonic loading are illustrated in an idealized
manner in Fig. 1. These correspond to an initial
stiffness for the uncracked concrete section, a re-
duced stiffness after cracking of the section, and
a further reduced stiffness once the prestressing
steel or the concrete begin to behave inelastically.
These three regions will be referred to as the
"elastic," "cracked elastic," and "inelastic" regions.
Also, the ratio of the rotation over a small incre-
677
Roger W. G. Blakeley gained his Bachelor of Engineering
degree at the University of Canterbury, New Zealand, in 1968.
Since graduating he has been engaged in research for his PhD
degree at the University of Canterbury on the strength and
ductility of prestressed concrete frames subiected to seismic
type loading. He received the N. Z. Portland Cement Associa-
tion postgraduate scholarship in 1968 and holds a study
award from the N. Z. Ministry of Works.
ACI member Robert Park is professor of civil engineering,
University of Canterbury, Christchurch, New Zealand. He gained
his BE and ME at the University of Canterbury and his PhD
degree at the University of Bristol, Engl,and. Professor Park has
held academic appointments at both of these universities. He
is the author of many papers on various aspects of reinforced
and prestressed concrete floors and frames. He received the
Guthrie Brown award of the Institution of Structural Engineers,
London, in 1969, and the Freyssinet award of the New Zealand
Institution of Engineers in 1970.
ment of length to the rotation over that length at
first cracking will be referred to as the "rotational
ductility factor," and the ratio of the perpendicular
displacement at the end of a member to the dis-
placement there at first cracking will be referred
to as the "displacement ductility factor."
This paper presents the results of a series of tests
which were conducted to allow the study of the
energy dissipation, ductility and structural re-
sponse of prestressed concrete frames.
Energy dissipation
The behavior of a prestressed concrete member
under cyclic loading due to earthquakes may be
likened to a bilinear elastic system with the
idealized moment-rotation hysteresis loop shown
in Fig. 2. The principal feature of this loop is the
large elastic recovery of prestressed concrete. On
the other hand, reinforced concrete under cyclic
loading may be likened to an elasto-plastic system
with the idealized moment-rotation loop shown in
Fig. 3. The area under the moment-rotation curve
represents the energy absorption. In the bilinear
elastic system the energy is largely released as
kinetic energy but in the elasto-plastic system it is
largely dissipated by plasticity. The area within
Inelastic
1
. . . ' , _____ ___
eA
c
<l.>
E

Rotation 8

I I
/E
I I
I '
(/uJ)
Fig. !-Idealized moment-rotation curve for prestressed
concrete
678
the hysteresis loop is a measure of the energy
dissipation characteristics of the systems. The
energy dissipation capacity of the members has an
important effect on the response of a structure to
earthquake excitation, yet little study has been
made of these characteristics for prestressed
concrete.
Ductility
Dynamic analyses of the elastic response of
structures, using earthquake acceleration records,
have shown that a structure is subjected to con-
siderably greater loads than are provided for by
the equivalent static load design coefficients rec-
ommended by codes. This means that structures
must be capable of developing large post-elastic
Moment
Rotation
Fig. 2-ldealized bilinear elastic system
Moment
Fig. 3-ldealized elasto-plastic system
ACI JOURNAL I SEPTEMBER 1971
deformations if they are to survive severe earth-
quakes. One object of the tests reported here was
to assess experimentally the available ductility
in members designed according to current pre-
stressed concrete codes, and to find if confinement
by special transverse reinforcement for ductility is
necessary in seismic resistant prestressed concrete
members.
Structural response
It is generally assumed that the response of a
prestressed concrete structure to a severe earth-
quake will be greater than that of a comparable
reinforced concrete structure. The reasons behind
this assumption are the lower percentage critical
viscous damping of prestressed concrete when
compared with reinforced concrete and the lower
energy dissipation characteristics. However, little
analytical study has been made of the response of
prestressed concrete structures and their conse-
quent ductility demands on individual members.
The results from the tests reported are in tended
for comparison with analytically obtained mo-
ment-rotation characteristics to facilitate a sub-
sequent dynamic analysis of the response of pre-
stressed concrete framed structures under actual
earthquake accelerations.
1- -L- ,4.. -L-
{a) Framed Structure
(C) /SOli!Jted Beam-Column
Assembly of Frame
J
'
-
(b)Structure under Lateri!JI
Ei!Jrthqui!Jke Load
---1
------- Tx
L
-2-
I

(d) rest Beam-Column Assembly
Rotated through Angle a
Fig. 4-Beam-column test specimen
ACI JOURNAL I SEPTEMBER 1971
EXPERIMENTAL PROGRAM
Selection of test specimen
The test specimen is the portion of a multistory
prestressed concrete frame shown shaded in Fig.
4a. The structure under lateral inertia loading due
to an earthquake is shown in Fig. 4b. Fig. 4c is an
isolated view of the external column-beam test as-
sembly under consideration. Points of contraflex-
ure are assumed to occur at the mid points of the
members. An imposed horizontal displacement
induces horizontal and vertical reactions. In Fig.
4d the entire beam-column assembly has been ro-
tated until the two column inflection points are on
the same vertical line giving a more convenient
orientation for testing purposes. The column axial
load P represents the loading due to the weight of
the building above the particular assembly and
the overturning moment on the structure. The
reversible load X applied at the end of the beam
and the reactive lateral loads H induced at the
ends of the column represent the shears applied by
the earthquake.
The behavior of joints such as the test speci-
mens is a critical factor affecting the behavior of
building frames under seismic load reversals. Of
particular interest is the behavior of the joint
assembly formed when precast members are post-
tensioned together with mortar joints at the criti-
cal sections.
Earthquake loading re,presentation
Static cyclic loading was used to represent high
intensity load reversals due to an earthquake. The
advantage of this method of testing over dynamic
loading lay in the ease of collection of test data.
The justification for its use may be found from
tests such as those carried out by Oladapo,
2
which
show little difference in the moment-rotation char-
-15
""-13
c:
. -ti -11
" 0
3 -9
-7
- 0
-5
.!:? :-
-3
15
5 10 _cycles--
Fig. 5-Earthquake loading sequence
679
.
acteristics between prestressed concrete members
tested dynamically (to failure in 1 sec) and those
tested statically. The effect of rate of loading is to
displace the moment-rotation curve upwards as
the loading rate increases. For example, Oladapo
found a 6 percent increase in ultimate moment
when the loading time was reduced from 40 min to
1 sec but little reduction in the ultimate curvature.
It is considered that the use of moment-rotation
curves from static cyclic loading tests would give
a slightly conservative estimate of the response of
a prestressed concrete frame to seismic ground
motions.
II
-c::::::::t---Steel plate(Typical)
f 1
!:::
c:
Cl>
(J
0

-1.
3
/s"' dia. stirrups
--
-UNIT 1-
, ..
dia. Ties
The cyclic loading sequence used in the tests is
similar to that followed by Hanson and Conner
3
for tests on reinforced concrete beam-column
joints except that greater deformations were en-
forced. This loading sequence is said to be repre-
sentative of the effect of two major earthquakes on
the structure. The extent of the imposed deflec-
tions was based on the rotational ductility factor at
the plastic hinge. For the first test (Unit 1) in
which the plastic hinge formed in the beam, the
maximum value taken for this factor in the load-
ing sequence was five. The basis for this figure was
that the only nonlinear dynamic analysis of a pre-
8-
3
1a'' dia. strands
3
/e" dia. tie
stirrup
0.276"dia cables
- '1.; dia. Spiral, ('pitch _ ___,

-
-
I -
680
-- 3-
3
ft; dia Stirrups
Fre ys sin et
anchorage

ELEVATION of ANCHOR BLOCK
(UNITS 1 & 2)
10'-('
I
3
18' dia. s rirrup s -1
af-
-UNIT 2-
SECTION B-8 (Typ:Units 1 & 2)
wires
SECTION C-C
strands
%"dia. tie
spirals
-8- 1;
8
" dia. strands
SECTION D-0
Fig. 6-Details of Units I and 2 (I ft = 12 in. = 30.48 em)
ACI JOURNAL I SEPTEMBER 1971
stressed concrete framed structure available,
which was carried out by Spencer,
4
showed this to
be the order of maximum beam rotational ductility
factor (same definition as in this paper) required.
In subsequent tests the maximum rotational duc-
tility factors for the test sequence were increased
to approximately 15. The loading sequence used
for Unit 2 is illustrated in Fig. 5. The loading se-
quences for the other units were approximately
simHar.
Design of test units
The four test units consisted of full-scale ele-
ments of dimensions as shown in Fig. 6 and 7. De-
tails of the material properties, design stresses and
strengths are given in the Appendix. The col-

'o'
I
-
I
0 1
"'' <lJ'
.!::'
c.
<l>l
.u:
(01
0
(/):

"'":i
!2:


- e::::=1- --Steel plate (Typical J
r 1
A A
,._
10'-1"
r+B
- - - - - - - - 1- ro,l.u
of!IH+IH-111-- ------- - -
- -1---1------ - - --I-
J "
/s stirrups
at tO"centres
-JJ!lJL2-
'
I
.j
i _1_
.f
T -,.
>,I"'
:\-
0)
4- dia. ties

"'' <o
. 1 +-
"' .
I <o
;f-

dia. spirals,
1" pitch .
Freyssinet
anchorage
umns were pretensioned. The beams were lightly
pretensioned for handling stresses, and post-ten-
sioned with cables passing through the column
into an exterior anchor block. Freyssinet anchor-
ages were used and the cables were grouted. Moist
pack mortar joints, 1 in. (2.54 em) thick, were
formed between the elements. The section sizes
were similar to those of an existing prestressed
concrete framed structure. In the prototype struc-
ture the prestressing force was concentric in the
joint region to give maximum flexural strength
under seismic reversals of load, and this configura-
tion was reproduced in the test specimens. The
post-tensioning cables in the beams of the test
specimens were not draped, thus ensuring that
prior to applying the earthquake loading in the
. t 0
<o
t -- 0
8- 'l8" dia. strands
--
3
1; dia. tie
I
" 3-12w 0.276 dia.
cables.
3
1
8
' dia, stirrup
0)
Lj _ .L-----'

-

stirrups


0
.c::
01
I , i Q.
"':::I ,..., ....
. "'
0
;
!
1 ELEVATION of ANCHOR BLOCK
D (UNITS 3 & 4 J
10'-1''
r+C
l.,.B '
dia. stirrups 13!/dia stirrupj
at S"centres 1 at 10"cl'ntres '

.!.I ":' f -
3
-8- Yadia.strands
l
I 5 " I
t-- 7 ?a-1
SECTION D-D
%'dia. tie
0.200"wires
,.-.,.._L--_ 8-
3
1/ dia. strands
SECTION E-E
Fig. 7-Details of Units 3 and 4 (I ft = 12 in.= 30.48 em)
ACI JOURNAL I SEPTEMBER 1971 681
tests the members were in direct stress only. This
same situation is achieved in common design prac-
tice in New Zealand where dead load plus one-
third of the live load are considered to be present
with seismic loading and hence if dead load plus
one-third of the live load is balanced by the pre-
stress the earthquake loading will act on a frame
which has direct stresses only.
Units 1 and 2 were designed to form plastic
hinges in the beam at the joint, and Units 3 and 4
were designed to form plastic hinges in the column
immediately above or below the beam connection.
The situation represented by Units 3 and 4 is
typical of many prestressed concrete framed struc-
tures. This arises from the design of long spans
requiring large cable eccentricity and hence a
large beam section size with a consequent high
ultimate moment capacity. The columns of the
frame, even when catering for the required seismic
design moment, may have a very much lower
ultimate moment capacity than the beams. Hence
under earthquake loading the plastic hinges would
form in the column rather than in the beam in
many prestressed frames. It is recognized that it is
desirable to have plastic hinges forming in the
beams rather than in the columns because in this
type of failure mechanism considerably greater
energy absorption is possible.G However this condi-
tion is not too critical for structures of three
storys or less.
The effect of varying the degree of confinement
on the ductility of the section was also studied.
The mild steel stirrups and ties in Units 1 and 3
satisfied the shear requirements in the commonly
used codes for prestressed concrete: ACI 318-63
6
(also ACI 318-71') and the British CP115.
8
Units
2 and 4 also contained special transverse steel for
confinement. For reinforced concrete, recommen-
dations for the amount of special transverse steel
required at beam-column joints for ductility are
given by Blume, Newmark, and Corning,n the
SEAOC Code
10
and ACI 318-717 but no such recom-
mendations are available for prestressed concrete.
However, since the column hooping recommended
by those authorities n,lo,< is to preserve the axial
load carrying capacity of the concrete after spall-
ing of the cover concrete, there is no reason
why the recommendations for reinforced concrete
should not be taken as a guide for prestressed con-
crete. The column regions adjacent to the joint
were provided with approximately one half of the
rectangular spiral steel required by both the SEA-
OC Code and ACI 318 .. 71, for special transverse re-
inforcement. Only one-half of the recommended
amount was placed since it is considered that the
recommendations are very conservative for col-
umns with small cross sections in which the ratio
of gross concrete area to area of concrete core is
high. In Units 2 and .4 the shear reinforcement
682
Fig. 8-Unit 4 under test
in the beams satisfied the SEAOC Code. The shear
capacity of the columns in those units was also
adequate but no attempt was made to analyze the
shear strength of the joint region because of the
complication of the arrangement of the prestress-
ing tendons across and down the depth of the joint
region, and the confining effect of the anchor
block.
When considering the behavior of packed mortar
joints under reversed loading of high intensity the
New Zealand Prestressed Concrete Insti tu te
11
rec-
ommends that "suitable binding or enclosure of
the joint itself, should be provided to prevent
loss of material." To study the effect of binding,
Units 2 and 4 which were detailed for ductility had
the mortar joint bound internally with a light wire
stirrup, whereas Units 1 and 3 had the joint moist
packed without binding. In all units the interfaces
of the precast members at the joints had been
roughened to a depth of approximately 1/16 in.
(1.6 mm) to help hold the mortar.
Test procedure
The test rig with a unit under test is shown in
Fig. 8. For Units 1 and 2 a 100 kips (45,400 kgf) ax-
ial load was applied to the column through a hy-
draulic jack. This column load was approximately
0.13 of the axial load capacity of the column and
was chosen to represent the loads carried by a
column in the lower storeys of a building frame.
For Units 3 and 4, where the column loading was
critical, axial loads were varied about this mean
value to simulate the effect of varying axial load
ACI JOURNAL I SEPTEMBER 1971
with lateral overturning moment. The beam end
was loaded by two 35 kips (15,900 kgf) mechanical
screw jacks for upward and downward loading.
The use of screw jacks enabled a controlled dis-
placement to be applied to the beam end. The time
taken to load each unit through the cycles of the
loading sequence and to collect the test data was
10 to 15 working days.
Instrumentation
Direct readings of hinge rotations were obtained
from dial gages and linear variable displacement
transformers attached to frames around the mem-
bers. Surface concrete strains for the whole unit
were measured using mechanical gages of small
gage length thus enabling the distribution of curv-
ature along the members to be calculated. Steel
strains were recorded with electrical resistance
strain gages attached to the stirrups, spirals, and
prestressing steel. Deflections were read with dial
gages and deflection rules.
TEST RESULTS
The performance of each test specimen is dis-
cussed in this section. The steel and concrete
strengths at the stage of testing are given in the
Appendix.

NEGATIVE MOMENT AND ROTATION
' '-'-'
'
CORRESPOND TO DOWNWARD LOAD
G:'
AT BEAM END.
n
.......
(Readings are for 12 inches beam
:>(
plastic hinge length)

';:: '7n
6
:L
'
Qn

.......
I
'
.-1n
I
.
A
Moment-rotation and displacement characteristics
Comparison of Units 1 and 2 - Both units de-
veloped a plastic hinge in the beam as expected.
Representative test results are illustrated in Fig.
9, 10, 11. The numbers on the curves correspond
to the peak of the cycle number given in Fig. 5.
Quantitative values are summarized in Table 1.
Fig. 9 is a moment-rotation curve for a 12 in.
(30.5 em) plastic hinge length on the beam adja-
cent to the column for Unit 2. It is apparent that
significant post-elastic rotation is available, and
the values of the rotational ductility factor reached
at the peaks of Cycle 15 were 31 and 56 for upward
and downward loading respectively. It was during
Cycles 4 and 15 that commencement of crushing
of the concrete in the compression zones of the
beam was observed, and the substantial area with-
in the hysteresis loop at this stage may be ob-
served in Fig. 9. The cycles before crushing com-
menced had considerably less energy dissipation
as is indicated by the smaller area within the
hysteresis loops. As is shown in Table 1, even
Unit 1 reached a high rotational ductility factor at
maximum moment. Unit 2, which was specially
detailed for ductility, had a slightly greater
maximum moment capacity than Unit 1. Table 1
Al'
v---
/
I
v
/
,
v/
I
--
5

v
m
-3()
"'"'
. z
:Y.J
-2(
p-.-
. 1 c-

10 po
pc 1(;
f'"\ 1 c:-
zo
,..,
[
"
J('
.....A...J !._; J'J _,l.J
BERI--1 f r; rqno

R"l'Yl C
I
]f4C
J __ I
--r---
VA
...
I

w
. , L.;
f !
----- ------
--! jj

. 'I "'J
--.......... Q
--
I
I
15
I
I
Fig. 9-Moment-rotation curve for Unit 2 (I in. = 2.54 em, I ft-kip = 138.25 m-kgf)
ACI JOURNAL I SEPTEMBER 1971
683
also shows that for these members the displace-
ment ductility factors were close to one third of
the rotational ductility factors.
After the 15 prescribed cycles had been com-
pleted the maximum allowable deflection at the
end of the beam was imposed in both directions.
Plots of the beam hinge moment against the beam
end deflection for the test are shown for Units
1 and 2 in Fig. 10 and 11, and include this final
failure cycle. Both units retained over 89 percent
of their maximum moment capacity up to the
maximum allowable deflection downward in the
rig of nearly 11 in. (27.9 em). On allowing the end
loading on the beam to return to zero, and im-
posing an upward deflection, both units attained
almost their previous maximum moment capacities
for upward loading. However, beyond this point
Unit 1 showed a gradual reduction of moment
capacity with increased deflection. The reason for
this behavior lay in a bond failure of the top
prestressing cable duct in the column (between
the duct tube and the column) which resulted in
the top cable pushing through the column and
causing a flexural failure of the anchor block. The
appearance of the unit at maximum upward de-
flection is shown in Fig. 12. This phenomenon did
not occur to such an extent in Unit 2, apparently
because the heavy rectangular steel spirals in the
'
cu
NEGATIVE MOMENT AND DISPLACEMENT
CORRESPOND TO DOWNWARD LOAD AT

1(-in
H
BEAM END.
"
---- -- --- ----'----- -----/---- -- ------ l2Q
LJ
L:
u
L
--1-------
v
<in
v/
Q
LJ
m
I
1t2
-lp -f-3 .
v

-17
/
l/
I
I/
--
1---
j
(
J
"'\
---

--
column prevented the splitting bond failure of the
top prestressing duct. In the case of Unit 2 a
moment close to maximum was maintained up to
an upward deflection of 12.5 in. (31.8 em) when
failure occurred suddenly, due to a buckling fail-
ure of the top prestressing cable in compression.
Fig. 10 and 11 show that very large post-elastic
deformations were available in these prestressed
concrete members even when they were not spe-
cially detailed for ductility. The heavily confined
Unit 2 had a higher moment capacity and retained
this moment capacity longer than did Unit 1.
However, the more gradual failure of Unit 1 at
very large deflections is preferable from a seismic
design viewpoint, although the deflections at
which failure occurred are greater than that which
could reasonably be expected under an earth-
quake.
Another feature of interest was that even
though the mortar joints were wide open over
one-half their depths during load reversals, little
mortar was lost, even for the case of Unit 1 which
had no binding reinforcement the mortar
itself.
Comparison of Units 3 and 4 - Plastic hinges
were formed in the columns as expected. Moment-
rotation curves for the plastic hinge in the column
------
15
(
/
---

----- -- -
i/
---------------
\

---- ----- ------- -
,
I

--- -- -----
I
--
'I
---------


.,
p
f3 'b
1/
lf'l
f-
'NJ r} T< PLRCU 11LNT, _NCHU
L- L- '
41 ..
-- 1---- ---
-t;\ ..
'1 zo
. ] b)
I
I
I
-----
Fig. I 0-Moment-displacement curve for Unit I (I in. 2.54 em, I ft-kip = 138.25 m-kgf)
684
ACI JOURNAL I SEPTEMBER 1971
above the beam are shown for Units 3 and 4 in Fig.
13 and 14. The reason for the elongated shape of
the curves is that plastic rotations occurred in the
column hinge above the beam for downward load-
ing at the beam end, and below the beam for
upward loading at the beam end. The moments
and ductilities for the two units are summarized
in Table 2.
For both units, Cycles 1 to 3 were in the elastic
range and Cycles 4 to 8 were in the cracked
elastic range. The rotational ductility factor at the
peak of Cycle 5 for downward loading was 5.0. Up
NEGATIVE MOMENT AND DISPLACEMENT
CORRESPOND TO DOWNWARD LOAD AT
B.EAM END.
-1 d
-1..,
-1 _: -P,
Fig. 11-Moment-displacement curve for Unit 2 (I in.
to this stage the hysteresis loops exhibited small
energy dissipation. It was only during Cycle 9,
when crushing of the concrete commenced, that
substantial energy dissipation became apparent.
In Unit 4 this was initiated by a sudden spalling
of the cover concrete, followed by a drop in mo-
ment resistance and an increase in rotation. The
sharp drop in moment capacity for Unit 4 during
Cycle 15 was due to a fracture of two prestressing
strands in tension.
The moment-rotation curves for Units 3 and 4
were very similar, despite the fact that the column
2.54 em, I ft-kip = 138.25 m-kgf)
TABLE I-MOMENT AND DUCTILITY CHARACTERISTICS OF UNITS I AND 2
At maximum beam moment
(e.g., at A and B of Fig. 9) Ratio of
maximum Beam moment
Beam Maximum beam Maximum at maximum
Beam hinge Beam Beam end displace- beam moment to beam end displace-
Direction curvature X 1()5, rotational displace- ment moment, cracking displace- ment,
Unit of beam radians per in. ductility ment, in. ductility ft-kips moment ment, ft-kips
No. loading (radians per em) factor (em) factor (m-kgf) at joint in. (em) (m-kgf)
Upward 91.7 28.1 2.19 12.3 183.9 2.62 15.25 40.3
1 (36.1) (5.56) (25,420) (38.7) (5,570)
Downward 111.4 33.0 2.13 11.0 169.0 2.41 10.71 149.5
(43.9) (5.41) (23,380) (27.2) (20,670)
Upward 132.8 30.8 2.48 9.9 195.2 2.78 12.50 126.7
2 (52.3) (6.30) (26,990) (31.8) (17,520)
Downward 242.5 56.2 3.83 19.5 183.1 2.61 10.93 170.0
(95.5) (9.73) (25,330) (27.8) (23,480)
Note: Beam rotatwns were measured over a 12 in. (30.5 em) gage length in beam adjacent to the column face. All beam moments in the
table are at the centre of the 12 in. (30.5 em) gage length
ACI JOURNAL I SEPTEMBER 1971
685
of Unit 4 contained heavy rectangular spirals and
Unit 3 contained only light ties. The main dif-
ference in behavior was that for Unit 4, spalling
occurred at a higher moment but lower hinge ro-
tation than for Unit 3. The rotational ductility
factors at spalling were 13.8 and 9.7 for Units 3 and
4 respectively. Table 2 lists the hinge rotational
Fig. 12-Unit I after failu-re
ductility factors at maximum moment points
and at maximum rotations. Both units showed
large ductilities available after spalling of the
cover concrete. Fig. 15 shows Unit 4 at maximum
downward deflection. Of particular interest are
the diagonal tension cracks across the column in
the joint region.
Strength of sections
The maximum moments measured during the
loading cycles are given in Tables 1 and 2 and may
be compared with the theoretical values calculated
for nominal concrete strengths and column loads
given in the Appendix.
Curvature distribution
The curvatures along the members of the units
were calculated from the concrete strains. The
strains measured in the concrete compression zone
were first used to find the neutral axis depth.
Then the curvature <f> at that section was calcu-
lated from:
where
eo = strain at the extreme compression fiber
kd =neutral axis depth
(1)
NEGATIVE MOMENT AND ROTATION CORRESPOND
TO DOWNWARD LOAD AT BEAM END
(Readings are for 8 inches plastic hinge length in
column above beam )
:,m J
---
5
k
-- - -- --t-!r::,t---'ll...lJ-II/--+-----
,------ ------ --- .
f-----1'------+-----+----ll-----+----------f--------- _______ ,__ ______ ------f------f-f- Rn J. --r----- __
s: /LJII
I AWl --LJJ
--. --- - ----- +----+------- f----+-----/---h'Mf--t'---fii--::::71T"r-+---l
f-----+----+-----+-----+----r:-/-+--- , j ''l --au
____ __ _ J _:-+-__ -_---j
-- -- --- ______ ----
!
r
--
------ ---- "r- -.... - ..
:
I
-
Fig. 13-Moment-rotation curve for Unit 3 (I in. = 2.54 em, I ft-kip = 138.25 m-kgfJ
686
ACI JOURNAL I SEPTEMBER 1971
Examples of the curvature distribution along
the members of Units 2 and 4 so calculated at
selected cycle peaks are shown in Fig. 16 and 17.
For Unit 2 the plastic hinge formed in the beam
and in Unit 4 plastic hinges formed in the column.
The curves show that the plastic curvature has
not been solely concentrated at the mortar joint
in Unit 2 but has extended over a good portion of
the beam. This curvature distribution also shows
that the gage length chosen of 12 in. (30.5 em) over
which to measure the beam hinge rotations does
include most of the plastic rotation which oc-
curred. The peaks of curvature in Fig. 16 and 17
correspond to the crack positions.
The area of the curvature diagram gives the ro-
tation which occurs along the member. This area
can be divided into a triangular region of elastic
curvature plus a region of plastic curvature near
the critical section as shown in Fig. 16 and 17.
The plastic rotation ()P can be conveniently ex-
nressed as:
(2)
where
cp,.,aa; = maximum curvature
cf>e =maximum elastic curvature calculated
assuming an uncracked section
Lp =equivalent plastic length.
The equivalent plastic hinge length, defined as
by Eq. (2), was calculated for each test unit from
the measured curvature distributions which were
plotted along the members at the peaks of major
load cycles. In these calculations the equivalent
plastic hinge length was found from the area of
plastic curvature divided by the difference be-
tween the measured maximum curvature and the
calculated maximum elastic curvature. The curva-
ture in the mortar joints was conservatively
ignored in these calculations. The mean value of
equivalent plastic hinge length for the beam of
Unit 1 so found was 14.1 in. (35.8 em) and that for
Unit 2 was 9.8 in. (24.9 em). The heavier stirruped
Unit 2 allowed greater local curvatures in the
joint region, resulting in lower equivalent plastic
hinge lengths. Since the beam was 21 in. (53.3 em)
deep a rough guide for equivalent plastic hinge
length could be taken as one half of the overall
depth of the member. For Units 3 and 4 in which
plastic hinges formed in the columns similar ef-
fects were observed. The mean value of equivalent
plastic hinge length for the Unit 3 column was
8.7 in. (22.1 em), whereas that for Unit 4 was 6.7 in.
(17.0 em). Again for these 16 in. (40.6 em) deep
members, one half of the overall depth provides a
NEGATIVE MOMENT AND ROTATION CORRESPOND
TO DOWNWARD LOAD AT BEAM END
(Readings are for 12 inches plastic hinge length in
column above beam)
1n0 If)
--f-------f-t:;-.I..Ui4-Jflll.l--+------l
! >l''f-
____ _
z
H
T.
-1' 0 1( c )}'JIJ
------r---
----r--
[17

g
.. j <..U
----1----L---- -- _ _._ __ ..J. _____ _J._ __ _..L ___ ___! ___ L_ __ _L_ __ _.l_ __ _
Fig. 14-Moment-rotation curve for Unit 4 (I in. = 2.54 em, I ft-kip = 138.25 m-kgf)
ACI JOURNAL I SEPTEMBER 1971
687
688
v
c
z
-(
m
II)
!::
z
::::>
LL
0
II)

I-
V)

w


-(
J:
(.)

:::::i
G
::::>
c
c
z
-(
1-
z
w

0

I
N
1--,-----i------+-----


<

" ...
"


II)
:6 ;; g
.s
Fig. IS-Unit 4 at maximum downward deflection
(Cycle 15)
rough approximation for the equivalent plastic
hinge length.
Transverse steel
Electrical resistance strain gage readings were
taken throughout the test sequence. An indication
of the stress levels in the stirrups and spirals of
Unit 4 at the peaks of selected cycles may be seen
in Fig. 18. For this unit the maximum stress re-
corded in a column spiral was 19,500 psi (1370
at the center of a leg in the compres-
sion zone; the maximum stress recorded at the
center of the first beam stirrup was 33,500 psi
(2360 Both stress levels were well below
the yield stress for the steel. Even the stress in
the ties of the lightly reinforced Unit 3 column
were below yield. However, strains greater than
yield were recorded on the beam stirrup closest to
the column of Unit 2.
DISCUSSION OF RESULTS
The following points arise from the experimen-
tal results reported:
Energy dissipation
The moment-rotation curves presented show
that prestressed concrete members have little
energy dissipation capacity while they remain in
ACI JOURNAL I SEPTEMBER 1971
the cracked elastic range. However, the nonlinear
response analysis of Spencer
4
indicated that the
maximum beam rotational ductility factor re-
quired was five for a framed structure under an
earthquake. If this figure is generally correct,
frame member rotations would not go beyond the
cracked elastic range, and there would be little or
no structural damage. If the rotational ductility
factor required is greater than five there is a large
reserve of energy dissipation available once the
concrete has commenced to crush. However, at
this stage the members have been structurally
damaged with consequent difficulty of repair back
to a fully prestressed condition. More dynamic
analyses need to be carried out on prestressed
concrete framed buildings to determine their
response to actual earthquake acceleration records
so that a better understanding of the ductility re-
quired of frame members under an earthquake
may be obtained.
Ductility
The tests showed that large post-elastic deforma-
tions were available in all units. For these units
there was no significant advantage in putting more
transverse steel in the members than was re-
quired as shear reinforcement by prestressed
concrete codes. The electrical resistance strain
gage measurements on the heavy rectangular
spirals in the compression zones of the column of
Unit 4 showed that this steel did not yield as is
commonly assumed, and in fact the stresses were
!:!!1lL (T) Cunature d15tnbut10n has on the tens1on !.ide of the bl'am
and on thi' comprtt5SJOn S1de of the column.
c
" 0.
35-


"'
:55-

Jso
(2}Peaks of h.lve been plotted at tens1on crack positions
BEAM & COLUMN SCALE : INCHES
0 5 10 15 20
Cycle N2 Curve Symbol Column above Beam Column below Beam Beam
!For /MOiflmum Momnr .Jnp fHI) /Ma1mum f.fomnt-klp ttl /1-tarlmum lrfomtlllt-IP 'IJ

29.0
75.5
70.9
29.0
75.5
71 .o
55.9
172.9
152.4
{
Lme of elastiC curvature based on uncracked section strffness IS denoted------
Calculated equ1va/ent pfast1c f'!lnge length on bear.- : 9.2 mches
Fig. 16-Curvature distribution along members of Unit 2 at peaks of upward loading (I in.
138.25 m-kgf)
2.54 em, I ft-kip
ACI JOURNAL I SEPTEMBER 1971
689
.
s 10

;;.;o


C:bo
BEAM & COLUMN SCALE:
"""''
"' "
10
Cycle N Curve Symbol Column above Beam Column blow Bum Bum
/For Curyatur! IMOJt.lolmntl IAifiOILood! I""QJI,I>fOmMII IA.rrlal LOOdJ (.AfO-;;;;;;;-.,,,
4
/lopfttl IHrpS} IN/pftl lk/ptJ (Jtlp(ff/}
1/NffRMEDIAffl 69.8 85.7 69.8 69.8 171.4
115.5 78.2 115.5 51.1 283.0
-0----0- 101.1 80.7 107.1 57.1 247.9
---e-------o- 116.1 77.7 116.1 50.4 284.6
{
Lme of e/asttc curvature based on uction strffness 1s denoted------
At Peak of Cycle 4
Calculated equrvatent plasl1c htnge length on,cotumn below Om: 7.2 rnct"s
Fig. 17-Curvature distribution along members of Unit 4 at peaks of upward loading (I in.
kgf, I ft-kip = 138.25 m-kgf)
2.54 em, I kip 453.59
690
Gauge
Location

c; ":-- --
87 82 B3 B4
r--
_ NOTE: __
11)Stresses at maximum
0
0
0
co
I
DOWNWARD LOAD AT
END
--Column
8
0
8
0
0 <o"
"
Spiral
0
8

t
I-.
UPWARD LOAD AT
lOIJ'D
Steel Stress psi.-
0
0
0
0 ca
-arn- ... ;
. -r---- ___.

4 , g '5 , ___
toad per cycle are
I ----.----
indicated.
12!The cycle numbers
are shown beside
the points.
I3!Positive stress is
tension.
14) Yield stress of steel:
Beam stirrups 49,800esi.
Column spirals 42,300psi.
..
01)
..
24,000
V)
Q;
..
v;
E
" ..
<ll

83 B4
Stirrup Gauge N2
Fig. IS-Stresses in stirrups and spirals of Unit 4 (I psi
,, 15
87 82 8'4
Stirrup Gauge N2
0.0703 kgf/cm
2
)
ACI JOURNAL I SEPTEMBER 1971
quite low. One rather undesirable effect of the
heavy retangular spirals was to precipitate the
spalling of the cover concrete. The only advantage
gained from the spirals in these tests was the pre-
vention of a bond failure of the prestressing duct
through the column at far advanced strains. How-
ever the applied column loads were relatively low.
The tests showed the advantage of spreading the
prestressing steel over the section. At high curva-
tures after crushing of the concrete had com-
menced the cables in the compression zone of the
member acted as compression reinforcement and
prevented a rapid deterioration of moment carry-
ing capacity.
Stiffness degradation
The moment-rotation curves show a marked re-
duction in stiffness between the initial and final
sets of elastic cycles of loading. This stiffness
degradation phenomenon would have a significant
effect on the response of a structure during an
earthquake.
Bond failure
The bond failure of the prestressing duct
through the column, when the top cable o ~ Unit 1
was in compression, is a point of concern when
considering extremely large rotations under cata-
strophic loading. However a more critical case
would be that of an interior beam-column assem-
bly, where the cable in the beam on one side of
the column is in compression and on the other
side of the column is in tension. The combined
compressive and tensile force in the cable may be
sufficient to break down the bond between the
tendons and the duct, or the duct and the column,
resulting in a reduction in the moment carrying
capacit of the beams when the tendons push
througYi. It is recommended that corrugated ducts
be used through the columns of interior and ex-
terior beam-column assemblies to minimize the
possibility of a bond failure between the duct and
the column as occurred in Unit 1.
Joint behavior
Some concern has been expressed about the be-
havior of mortar joints between precast members
when subjected to seismic load reversals. In these
tests the mortar held itself in place even without
internal binding, although the interfaces had
previously been roughened. The plastic curvature
extended well along the beam and did not concen-
trate only at the mortar. Apart from a stress con-
centration in the compression zone caused by a
recess of the mortar for architectural reasons, the
joint region behaved in a similar manner to that
expected of a monolithic joint, except that crack-
ing in the joint occurred earlier because of the
lack of tensile strength.
ACI JOURNAL I SEPTEMBER 1971
Future work
It is apparent that future work is necessary to
develop a theory which would allow the moment-
rotation hysteresis loops for prestressed concrete
members to be predicted accurately. Also, the de-
velopment of a reasonable idealization for the
loops would be of great assistance in the dynamic
analysis of prestressed concrete frames subject to
earthquake motions.
CONCLUSIONS
This series of tests showed that prestressed con-
crete framed structures can be capable of re-
sisting moderate earthquakes without structural
damage, and of withstanding severe earthquakes
although in this case structural damage may oc-
cur. The test results indicated that:
1. Energy dissipation is relatively small prior to
the commencement of crushing of the concrete,
but substantial once crushing has occurred. Evi-
dence is necessary from dynamic analyses to deter-
mine whether this reserve of energy dissipation is
required under severe earthquakes, and to obtain
the actual ductility demands on the members.
2. Large post-elastic deformations can be avail-
able in prestressed concrete members, even when
the transverse reinforcement satisfies only pre-
stressed concrete code requirements for shear.
Heavy column spirals to satisfy the requirements
of the SEAOC Code (and ACI 318-71) for rein-
forced concrete for ductility are felt to be unneces-
sary for the type of frame unit tested with rela-
tively low applied column loads.
3. Substantial stiffness degradation occurs when
prestressed concrete members are subjected to
high intensity cyclic loading.
4. It is recommended that corrugated metal
ducts be used for post-tensioning cables through
columns for both interior and exterior beam-col-
umn assemblies. This precaution is to minimize the
possibility of loss of moment capacity in the beams
due to bond failure between the ducts and the
column.
5. Mortar joints between precast post-tensioned
frame members at critical sections can behave
satisfactorily under seismic load reversals.
ACKNOWLEDCM ENTS
The experimental work reported in this paper was
carried out in the DepaTtment of Civil Engineering of
the University of Canterbury, New Zealand, by the
first named author during postgraduate studies super-
vised by the second author.
The encouragement of the New Zealand Prestressed
Concrete Institute, R. Shepherd, reader in civil en-
gineering, University of Canterbury, and the late I. L.
Holmes, consulting engineer, Christchurch, is much ap-
preciated.
691
The financial assistance of the University Grants
Committee, the New Zealand Portland Cement Asso-
ciation, the Ministry of Works, and Certified Concrete
Christchurch Ltd, is gratefully acknowledged.
REFERENCES
1. Blakeley, R. W. G.; Park, R.; and Shepherd, R.,
"Seismic Resistance of Prestressed Concrete," Bulletin,
New Zealand Society for Earthquake Engineering
(Wellington), V. 3, No. 1, Mar. 1970, pp. 3-23.
2. Oladapo, I. 0., "Rate of Loading Effect on Moment-
Curvature Relation in Prestressed Concrete Beams,"
ACI JouRNAL, Proceedings V. 61, No. 7, July 1964,
pp. 871-887.
3. Hanson, N. W., and Conner, H. W., "Seismic Re-
sistance of Reinforced Concrete Beam-Column Joints,"
Proceedings, ASCE, V. 93, ST5, Oct. 1967, pp. 533-560.
4. Spencer, R. A., "The Nonlinear Response of a
Multistory Prestressed Concrete Structures Subjected
to Earthquake Excitation," Proceedings, 4th World Con-
ference on Earthquake Engineering, Santiago, Chile,
Jan. 1969, V. II, pp. A4 139-154.
5. Park, R., "Ductility of Reinforced Concrete Frames
Under Seismic Loading," New Zealand Engineering
(Wellington), V. 23, No. 11, Nov. 1968, pp. 427-435.
6. ACI Committee 318, "Building Code Requirements
for Reinforced Concrete (ACI 318-63) ," American Con-
crete Institute, Detroit, 1963, 144 pp.
7. ACI Committee 318, "Proposed Revision of ACI
318-63: Building Code Requirements for Reinforced
Concrete," ACI JouRNAL, Proceedings V. 67, No. 2,
Feb. 1970, pp. 77-146.
8. "The Structural Use of Prestressed Concrete in
Buildings," CP 115: 1959, British Standards Institution,
London, 1959, 44 pp.
9. Blume, J. A.; Newmark, N. M.; and Corning, L. H.,
Design of Multistory Reinforced Concrete Buildings for
Earthquake Motions, Portland Cement Association, Sko-
kie, 1961, 318 pp.
10. "Recommended Lateral Force Requirements,"
Seismology Committee, Structural Engineers Associa-
tion of California, 1968, 100 pp.
11. "Seismic Design Recommendations for Prestressed
Concrete," New Zealand Prestressed Concrete Institute,
Sept. 1966, 9 pp.
12. Hognestad, E.; Hanson, N. W.; and McHenry, D.,
"Concrete Stress Distribution in Ultimate Strength
Design," ACI JouRNAL, Proceedings V. 52, No. 4, Dec.
1955, pp. 455-479.
APPENDIX
Material properties
Concrete-From graded aggregate with % in. ( 1.91
em) maximum size. Aggregate : cement : water = 4.57
: 1 : 0.43, by weight. Compressive strength .of 6 X 12 in.
(15 X 30 em) cylinders at time of testing units, fc', psi
(kgf/cm2):
692
Unit number I 1 I 2 I 3 I 4
Beam concrete 6630 (466) 6000 (422) 7640 (537) 7500 ( ~ 2 7 )
Column concrete 6570 (462) 8300 (584) 7640 (537) 8080 (568)
Mortar-From graded sand with 3/16 in. (0.48 em)
maximum size. Sand : cement : water = 2 : 1 : 0.35, by
weight. Compressive strength of 4 X 8 in. (10 X 20 em)
cylinders at time of testing units was 9130 to 11,200 psi
(642 to 787 kgf/cm2),
Grout-Cement : water : admixture = 1 : 0.40 : 0.01,
by weight.
Prestressing steel
Diameter of
prestressing steel
12/0.276 in. (7 mm)
cables
%in. (9.5 nun)
strand
0.2 in. (5.1 mm)
wire
Reinforcing steel
Diameter of
reinforcing steel
% in. (9.5 mm)
% in. (1.59 em)
Design stresses
0.2 percent
proof stress,
psi
(kgf/cm2 )
211,000
(14,800)
237,000
(16,700)
230,000
(16,200)
Ultimate tensile
strength,
psi
(kgf/cm2 )
Q42 000
(17,000)
271,000
(19,100)
258,000
(18,100)
Yield stress, psi (kgf/cm2)
49,800 (3500)
42,300 (2970)
Stress in prestressing steel at transfer = 149,300 psi
(10,500 kgf/cm2).
Compressive stress in concrete due to prestress at
transfer:
Beams: 1680 psi (118 kgf/cm2) for Units 1 and 2
1565 psi ( 110 kgf/cm2) for Units 3 and 4
Columns: 1035 psi (72.8 kgf/cm2) for all units
The,oretical maximum moment capacities
Calculated by a general analysis ensuring equilibrium
of forces, compatibility of strains and perfect bond
between steel and concrete using the compressive stress
block parameters and the maximum concretr strain
found by Hognestad, Hanson, and McHenry
1
2 and the
actual stress-strain curves for the steel:
Beam section of Units 1 and 2: 174.3 ft-kips (24,100
m-kgf) for fc' of 6000 psi ( 422 kgf/cm2).
Beam section of Units 3 and 4: 378.5 ft-kips (52,330
m-kgf) for fc' of 7500 psi (527 kgf/cm2),
Column section of all units: 105.6 ft-kips (14,600 m-
kgf) for axial load of 50 kips (22,700 kgf) and 120.7
ft-kips (16,690 m-kgf) for axial load of 125 kips
(56,700 kgf), for fc' of 8000 psi (562 kgf/cm
2
). (These
were approximately the axial loads at the critical sec-
tions of Units 3 and 4.)
This paper was received by the Institute Nov. 16, 1970.
ACI JOURNAL I SEPTEMBER 1971