Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 94-S47

Seismic Assessment of the Santa Monica


Viaduct Bent Details

by M. J. N. Priestley, F. Seible, G. A. MacRae, and Y. H. Chai

The paper describes tests carried out on large scale reinforced


concrete test units with details modeling those encountered in
sections of Los Angeles’ I-10 Santa Monica Viaduct. Lap-splices of
column longitudinal reinforcement, and joint details including
inadequate development of column reinforcement in the cap beam,
and a total absence of special joint reinforcement would be consid-
ered unacceptable in new designs. The purpose of the tests was to
obtain verification of analytical predictions that, despite the poor
detailing the bents, could be expected to survive the design level
earthquake without collapse. Test results confirmed that the lap
splices should perform adequately under seismic attack, and the
column/cap beam connections would maintain gravity load
capacity after joint degradation.

INTRODUCTION
The Santa Monica Viaduct forms a segment of the Inter-
state I-10 freeway immediately adjacent to downtown Los Fig. 1—Typical structural configuration of Santa Monica
Angeles, and contains more than 2300 columns in multi- Bents.
column bents with circular columns extending below ground
level as pile shafts with the same diameter and reinforcing
details as the columns. Figure 1(a) shows a typical four- of 0.6g, or 50 percent higher than used for modern building
column bent, though 3-column bents are also common. The design in the adjacent downtown Los Angeles.
viaduct, built in the 1960s and reputedly the most heavily The cost of a full retrofit of the Santa Monica Viaduct, and
trafficked section of freeway in the world, contains a number other similar elevated sections of the 1-10 freeway was
of details which would be considered unacceptable in expected to exceed $100 million. Consequently, any reduction
modern design, and hence was an early candidate for seismic in retrofit effort that would not jeopardize the safety of the
review in the Phase II retrofit program of the California viaduct was essential. A peer review3 of the seismic assessment
Department of Transportation (Caltrans).1 Potential prob- requested by Caltrans concluded that, despite the inadequate
lems include a lap-splice of all column longitudinal rein- lap-splice details, splice failure was unlikely, and that though
forcement at ground level, where moments were expected to column/cap-beam failure was to be expected in the design
be close to the maximum developed in the in-ground plastic level earthquake, this should not result in bridge failure,
hinge (Fig. 1[b]), inadequate development of large-diameter since the lateral displacement capacity of typical bridge
column longitudinal reinforcement in the cap beam, and a bents was greater than predicted demand, even with the
complete lack of transverse reinforcement in the column/ effective moment capacity at the top of the columns reduced
cap-beam joint region, leading to a potential for joint failure to zero as a consequence of joint strength degradation. A
under seismic attack. concern existed, however, that following joint failure, the
columns could punch through the superstructure, as had
On the other hand, this section of the I-10 freeway, unlike
occurred with the Struve Slough bridge in the 1989 Loma
other sections further to the west which collapsed in the 1994 Prieta earthquake.4
Northridge earthquake,2 contained surprisingly good trans-
verse reinforcement details, with #4 (D12.7) spirals at 3.5 in.
(89 mm) centers over the full column height. As a conse-
quence, calculations for shear strength and flexural ductility ACI Structural Journal, V. 94, No. 5, September-October 1997.
Received January 22, 1996, and reviewed under Institute publication policies.
capacity of plastic hinges indicated that flexural or shear Copyright © 1997, American Concrete Institute. All rights reserved, including the
failure would not be expected under the maximum credible making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion will be published in the July-August 1998 ACI Structural
earthquake, corresponding to a site peak ground acceleration Journal if received by March 1, 1998.

ACI Structural Journal/September-October 1997 513


M. J. N. Priestley, FACI, is a professor of structural engineering at the University of
California, San Diego. He is a member of ACI Committee 531, Masonry Structures
Research. He is the recipient of the 1997 Arthur R. Anderson Award and the 1997
Wason Medal for Most Meritorious Paper.

F. Seible is a professor of structural engineering at the University of California, San


Diego. He is a member of joint ACI-ASCE Committees 334, Concrete Shell Design
and Construction, and 343, Concrete Bridge Design; and ACI Committees 341,
Earthquake Resistant Concrete Bridges, and 437, Strength Evaluation of Existing
Concrete Structures.

G. A. MacRae is an assistant professor of Civil Engineering, at the University of


Washington, Seattle. He completed his doctoral training at the University of Canter-
bury, NZ, and has since conducted research work both at the Public Works Research
Institute, Tsukuba, Japan, as well as at the University of California, San Diego. His
research interests include the design and retrofit of reinforced concrete structures sub-
ject to seismic loading.
Fig. 2—Failure model for lap-splice involving circumferential
Y. H. Chai is an assistant professor in the Department of Civil and Environmental and radial cracks.
Engineering at the University of California, Davis. His research interests are in exper-
imental testing, characterization of structural damage under earthquake loads, and
analysis and design of concrete and masonry structures.
vertical and radial crack surface must form between the
lapped bars, and also between the column bar and the starter
bar in the adjacent pair of bars. Also, since all bars are
To verify the conclusions reached by the peer review connected to the core concrete, relative slip of the bars can
committee, a program of testing was initiated at the Univer- only occur if a circumferential crack surface forms, sepa-
sity of California, San Diego, on large scale structural rating the bars from the core. These surfaces are shown in
elements representing the critical sections of the Santa Fig. 2. Assuming 45 deg compression struts forming in the
Monica Viaduct bents. This paper summarizes results from concrete around the bars, the longitudinal tension force in a
these tests. column bar can be directly equated to the tension strength of
the sum of the potential failure surfaces surrounding each
RESEARCH SIGNIFICANCE bar. This implies a certain plasticity of tension strength.
The paper describes tests and presents results that showed Thus, the maximum force that can be sustained by a lapped
that existing methods for evaluating seismic performance of bar will be
sub-standard details of the Santa Monica Viaduct were
unnecessarily conservative. As a consequence of the Tb = pls ft (1)
research described herein, the level of retrofit required for
the viaduct was greatly reduced, with savings estimated at
more than $50 million. where

p = πD′/2n + 2(db + c) (2)


LAP-SPLICE TESTS

Theoretical considerations is the perimeter of the crack surface (see Fig. 2), ft is the
Columns of the Santa Monica Viaduct are either 3 ft (914 direct tension strength of the concrete, taken as ft = 4 f c′
mm) or 4 ft (1219 mm) diameter, with reinforcement varying psi (0.33 f c′ MPa), n is the number of bars lapped at the
between 12 #11 (D34.8) and 16 #18 (D57.2) bars of grade 40 section (assumed to be uniformly distributed around the
(fy = 275 MPa nominal) strength. Typical lap-splice details core diameter D′), c is the cover to the main bars, and ls is
consist of all bars being lapped at ground level with a lap the lap splice length parallel to the column axis.
length equal to 20 bar diameters. Under seismic loading, The approach could be expected to be a little conservative,
initial plastic hinging is expected to form at the top of the since it assumes that the tension force over the length of the
columns, followed by a second set of hinges forming lap-splice is constant instead of reducing as the moment
approximately three column diameters below ground. As a reduces with distance from the critical section. However,
consequence, column moments at the base of the ground even if the analysis indicates that the yield strength can be
level lap splice are expected to reach about 70 percent of the achieved (i.e; Tb > Ab fy), lap-splice failure may still occur at
moment of the in-ground hinge (see Fig. 1). At this level of comparatively low ductility. This is because under cyclic
moment, longitudinal tension reinforcement with maximum response the concrete adjacent to the lap-splice will be
distance from the neutral axis will be at or very close to yield. subjected to longitudinal compression strain under the
Based on ACI 318-89 equations, required lap length would reversed direction of seismic displacement to that placing the
be about twice the 20db provided, though lap splicing is not spliced bars in tension. As the peak displacements increase,
permitted for #14 or #18 bars. Consequently it could be inter- and the compression strain around the bars reaches about εc
preted that the longitudinal reinforcement would only be = 0.002, longitudinal microcracking will develop, reducing
able to sustain a stress of about 0.5fy, and splice failure the tension strength of the concrete. Thus, as the response
would occur before the in-ground plastic hinge could form. direction reverses again, placing the spliced bars in tension,
The lap splices were reviewed using a fracture analysis splice failure will eventually occur as the value of Tb found
procedure developed by Priestley.5 from Eq. (1) will decrease as ft decreases.
With reference to Fig. 2, it is noted that for starter bars Transverse hoop reinforcement will potentially apply a
from the pile shaft to slide relative to the lapped column bars, clamping pressure to the lap splices. However, since the
which is a necessary condition for a splice failure, an essentially clamping pressure will be developed only as a result of radial

514 ACI Structural Journal/September-October 1997


Table 1—Lap-splice columns material strength
Concrete compression strength fc′ Longitudinal rebar Transverse rebar
Lower section* Upper section† Size fy l Size fyh
Unit psi MPa psi MPa # mm ksi MPa # mm ksi MPa
SM1 4063 28.0 4136 28.5 11 35.8 44.2 307 4 12.7 54.5 376
SM2 4930 34.0 4160 28.7 18 57.2 41.8 286 4 12.7 54.5 376
*
Below lap-splice

Lap-splice and higher

dilation, the crack pattern suggested in Fig. 2 will already


have developed before significant clamping action is
achieved. Therefore it is unlikely that the clamping action will
be additive to the strength found in accordance with the predic-
tion of Eq. (1), and the splice strength after formation of
cracking will be that of the clamping action alone, given by

Tb = πp′ls fl (3)

where

fl = 0.5ρs fsh (4)

is the lateral confining pressure developed by the hoop


reinforcement of volumetric ratio ρs at a tension stress fsh,
p′ is a reduced perimeter length over which the lateral
pressure acts, given by

p′ = πD′/2n + 2db (5)

and μ = 1.4 is the coefficient of friction on the naturally


forming crack surfaces.
Tests6 indicate that it may not be possible to utilize the full
yield strength of the hoop or spiral reinforcement to provide
effective clamping, since the radial dilation may be so large
that the crack surfaces shown in Fig. 2 open to the extent that
the effective coefficient of friction reduces to μ < 1.4. A
stress corresponding to a hoop strain of εh = 0.0015 appears
to be an effective upper limit.7
Unless the bar strength Tb predicted by either Eq. (1) or (3)
exceeds the maximum expected under seismic response,
splice failure is predicted, with degradation of capacity Fig. 3—Lap-splice test units (1 in. = 25.4 mm).
occurring under cyclic reversals of response.

Column lap-splice test units plastic hinge at the base of the column, rather than lap-splice
Two full-size test units modeling the region between the failure might be considered to indicate adequate strength of
column point of contraflexure and the location of maximum the lap-splice.
moment of the in-ground hinge, identified as region 1 in Fig. 1, However, it was desired to investigate the actual strength
were constructed and tested under simulated seismic of the lap-splice to ensure that an adequate reserve of
loading. Both columns represented 3 ft (914 mm) diameter strength existed to cope with the development of higher
columns, one (unit SM 1) reinforced with 16 #11(35.8 mm) moments at the lap-splice as a consequence of higher than
longitudinal bars, and the other (unit SM2) with 12 #18 (57.2 expected longitudinal reinforcement yield strength, strain
mm) longitudinal bars. Both columns were tested as simple hardening of reinforcement, and the possibility of the in-
vertical cantilevers, and contained lap-splices of 20db ground plastic hinge forming closer to the ground surface
starting 48 in. (1219 mm) above the base of the column. than expected. As a consequence, the lower 48 in. (1219
Transverse reinforcement consisted of #4 (12.7 mm) spirals mm) of column below the lap-splice was strengthened by the
at 3.5 in.(88.9 mm) pitch. Test unit dimensions and rein- inclusion of an inner ring of reinforcement (16 #8 [25.4 mm]
forcement details are shown in Fig. 3. With the column for unit SM1; 16 #11 [35.8 mm] for unit SM2) of Grade 60
proportions as shown, the moment at the base of the lap was reinforcement, as shown in Fig. 3. This ensured that the base
75 percent of the moment at the base of the column, and thus of the lap-splice, rather than the base of the column, was the
was a little larger than the predicted ratio of 70 percent critical section for flexure. Note that it would have been
occurring in the prototype situation. Thus, development of a possible to achieve the same result by eliminating the lower

ACI Structural Journal/September-October 1997 515


natural strength gain with age, and partly to conservative
batch design common in the 1960s and still prevalent today
when low strength concrete is specified.
Actual material strengths for two units are summarized in
Table 1. Concrete strengths for concrete below and above the
construction joint at the base of the lap splice obtained at
time of testing the columns were lower than expected. Since
the lap-splice strength is believed to be dependent on the
concrete tension strength, the low concrete compression
strength should ensure that the results from the tests are
conservative. Based on these material strengths, and the
section dimensions (see Fig. 3), the maximum strengths
predicted by Eq. (1) are Tb = 76.4 kips (340 kN) and 145.5
kips (647 kN) for the #11 and #18 bars of SM1 and SM2,
respectively. These translate to maximum stress levels of
49.0 ksi (338 MPa) and 36.4 ksi (251 MPa), respectively.
Consequently, it is predicted that the splice of SM 1 should
be able to achieve the yield strength of the #11 bars, but that
the #18 bars of SM2 should not quite be able to reach yield.

Instrumentation
The columns were extensively instrumented with electric
resistance strain gauges, and deflection and curvature
measuring devices. Full details of instrumentation are avail-
able in References 8 and 9.

Results of lap-splice tests


Crack patterns for the two columns are shown in Fig. 5 for
Fig. 4—Column test setup (1 in. = 25.4 mm; 1 kip = 4.45 kN). the condition corresponding to maximum expected proto-
type response (design level) and for a displacement ductility
48 in. (1219 mm) of column and locating the lap-splice factor of μΔ= 3. At levels of response lower than the design
immediately adjacent to the footing. This option was rejected level of SM 1 only flexural cracks developed. At the design
at the design stage because of concern that the stiffness of the level, a number of vertical splitting cracks developed at the
footing might act to confine the lower portion of the lap- base of the lap-splice, but did not propagate under cycling.
splice, resulting in unrealistic enhanced performance. Figure 5(a) shows conditions after three complete cycles to
The test units were tested as vertical cantilevers, as shown the design level, with the extent of the lap-splice being
in Fig. 4, and were subjected to an axial compression force defined by the two straight circumferential lines. As the
of 400 kips (1779 kN) which was typical of dead load levels column was cycled to higher force and displacement levels,
in the bents modeled by the test units. Lateral loading was the extent of vertical splitting increased to cover the full
initially applied under force-control until first yield of height of the splice by the third cycle to μΔ = 2. Strength
tension reinforcement was obtained, and then under degradation did not occur, however, until after the first cycle
displacement control to displacement ductility factors of μΔ= to μΔ = 3. By this stage extensive crushing of cover concrete
0.84, 1, 1.25, 1.5, 2 and 3. At all except the lowest level of had developed, centered about 10 in. (250 mm) below the
force-controlled testing three complete cycles of imposed base of the lap-splice. Figure 5(b) shows the condition of the
displacements were applied. lap-splice and lower region after the first cycle to μΔ = 3.
The yield displacement, and hence the reference for Under additional cycles to μΔ = 3, the lap-splice began to
calculating displacement ductility, was based on a bilinear fail, with extensive spalling of concrete in the lap splice
representation of force-displacement response, and was region.
found by multiplying the average displacement measured at Initial response of SM2, reinforced with 12 #18 (57.2 mm)
the column top at first yield by the ratio of ideal flexural bars was also essentially flexural in the early stages of
capacity to yield moment. Ideal moment capacity was based response. However, vertical splitting of cover concrete in the
on measured material strengths, and an extreme fiber lap-splice region initiated earlier, at 73 percent of the design
compression strain of 0.005. Both yield and ideal moment response level. This splitting continued to develop into a
capacity related to conditions at the base of the lap splice. fairly extensive pattern of vertical cracks after three cycles to
the design level of 68 percent of ultimate strength, as shown
Material properties in Fig. 5(c). However, there was no significant strength
Main column longitudinal reinforcement and transverse degradation at this level of response. Based on a plane-
spirals were grade 40 (nominal fy = 275 MPa) of a special section moment-curvature analysis, peak reinforcement
order of steel rolled for Caltrans for structural testing tension stress at the base of the lap-splice was 38 ksi (262 MPa),
research purposes. Specified concrete strength for the Santa 5 percent higher than the capacity predicted by Eq. (1).
Monica Viaduct was fc′ = 3250 psi (22.4 MPa). However, Taking the effect of tension shift into account, the extreme
measurements of strength from cores and by impact hammer tension rebar would be expected to be at yield at this stage.
measurements indicated a minimum strength in situ of 5000 psi This was confirmed by strain gauge readings.9 As lateral
(34.5 MPa). The strength increase can be attributed in part to force and displacement levels were increased, the extent of

516 ACI Structural Journal/September-October 1997


Fig. 5—Damage to lap splice units at different stages of testing.

vertical splitting increased both within and below the lap- since the level of shear in both columns was sufficiently high
splice region. Crushing of cover concrete initiated some 10 to cause shear inclination to the flexural cracks, the critical
in. (250 mm) below the construction joint at μΔ = 1.5, as was section for moment corresponding to a given tension stress
the case for SM 1. However, for SM2, the first cycle to μΔ = at the base of the lap-splice was depressed somewhat below
1.5 corresponded to the maximum strength of the unit. the lap-splice. The observed inclination corresponded to a
Further cycles to μΔ = 1.5 and to higher displacements tension shift of approximately 10 in. (250 mm), as noted
resulted in strength degradation and spalling of cover above in the location of initial concrete cover spalling.
concrete in the lap-splice region. After three cycles to μΔ =3 Because the sections at the base of the lap-splice and the
this spalling was extensive, as is evident in Fig. 5(d). base of the column are almost equally critical for flexure, the
Lateral force-displacement response for SM1 and SM2 is expected curvature distribution is as shown by the solid line
shown in Fig. 6(a) and 6(b) respectively. Included in these in Fig. 7(a). This is simplified to the constant curvature
figures are the theoretical force-displacement envelopes, distribution over the bottom 48 in. (1219 mm) shown by the
based on flexural response, taking account of tension shift; dashed line. Note that strain-penetration of reinforcement

ACI Structural Journal/September-October 1997 517


Fig. 6—(a) SM1 (16 #11); and (b) SM2 (16 #11).

Fig. 8—SM3 column/cap beam unit dimensions (1 in. =


25.4 mm).

If My and φy are the moment and curvature at first yield of


the longitudinal reinforcement, the corresponding “first-
yield” displacement is

1 2
Δ y = φ y [ 144 ⁄ 3 + ( 48 + l sp ) ( 168 + 0.5l sp ) ] [in.] (7)

Fig. 7—Curvature model for deflection prediction of lap- The total displacement at higher stages of response is
splice columns. predicted from the moment-curvature response (M-φ) as

M
Δ = Δ y ------- + φ p ( 144L p1 + 192L p2 ) [in.] (8)
into the foundation results in an increase to the effective column My
length of lsp = 0.15fyl dbl ,7 where fyl is the longitudinal rebar
yield stress in ksi. By the same reasoning, two potential regions where
of plasticity can develop—one at the base of the lap splice, and
the other at the column base (see Fig. 8[b]). M
φ p = φ – ------- φ y
Using normal assumptions for effective plastic hinge length,7 My

lp1 = 0.08L1 + 0.15fyl dbl ≥ 0.3fyl dbl (6a) is the plastic curvature.
The force-displacement envelopes of Fig. 6 were based on
lp2 = 0.08L2 + 0.15fyl dbl ≥ 0.3fyl dbl (6b) moment-curvature analyses using measured material properties

518 ACI Structural Journal/September-October 1997


and a stress-strain model for confined concrete developed by
Mander, Priestley and Park.10
The force-displacement response of SM1 (Fig. 6[a]),
shows dependable response under cycling to μΔ ≤ 2 with
very little degradation of strength on successive cycle to
each displacement level. The envelope of response agrees
very well with that predicted by the analysis described
above. However, at μΔ = 3, the strength loss on successive
cycles is considerable, as a consequence of the lap-splice
failure with residual strength after five cycles being approx-
imately 50 percent of the initial maximum. Back-calculation
based on the force-displacement model outlined above indi-
cates that the peak compression strain at the level of the
longitudinal reinforcement would be approximately 0.004 at
μΔ = 3, indicating that the assumed strain of 0.002 at onset of
tension strength degradation may be too low.
Response of SM2 shows much earlier degradation, with
stable response only being obtained up to μΔ = 1. Up to the
first cycle to μΔ = 1.5 the envelope of response agrees well
with the prediction of Eq. (8). Strength on successive cycles
to μΔ = l.5, and to higher displacements shows moderately
rapid capacity-degradation as the lap-splice failure spread Fig. 9—Column/cap beam test unit and bending moments.
around the circumference of the column. However, behavior
exceeded that predicted using the model of Fig. 2, which cap-beam in transverse flexural response. The width of slab
implied a peak strength less than yield strength. modeled agreed with Caltrans design assumptions12 but was
Both columns thus performed better than predicted. There expected to be conservatively low.
appear to be two main reasons for this. First, the analytical Model cap-beam longitudinal reinforcement at the column
model assumes that the tension force to be transferred is face consisted of 7 #14 (43.0 mm) bars in the bottom of the
constant along the length of the lap-splice, whereas the cap-beam as tested (i.e., in the top, in the prototype configu-
moment gradient results in a reduction in tension force with ration). One of the #10 (31.7 mm) bars was lapped with a
height above the base of the splice. This is also apparent in l6db lap at the column centerline, outside the column cage
the development of splitting cracks, which initiated at the (see Fig. 8[b]). Shear reinforcement did not continue through
base of the splice in both column tests. the column-cap-beam joint region.
Second, it is assumed that all tension reinforcement is The column clear height of 108 in. (2743 mm) modeled
equally stressed, whereas the bar force to be transferred the expected prototype distance from the superstructure
reduces with circumferential distance from the critical bar, soffit to the point of contraflexure in the column. Longitu-
particularly for SM2, where splice failure before first yield dinal reinforcement was developed in the cap-beam a length
was predicted. Since sliding of one bar relative to its lapped of l6db, which was significantly less than the development
partner is not feasible without sliding of adjacent pairs, some length of 52.9 in. (1344 mm) required by ACI 318-89,13 for
averaging must occur in the circumferential direction. It expected material strengths of f'c= 5 ksi (34.5 MPa) and fy =
appears, based on these considerations and the test results, 44 ksi (303 MPa). More importantly, this short development
that the approach described by Fig. 2 and Eq. (1) is suitably length meant that the column bars were terminated well short
conservative for seismic assessment of lap-splices in of the cap-beam top reinforcement (bottom reinforcement as
existing circular columns. tested) with the consequence that the ends of the bars could
not benefit from being clamped in the flexural compression
Column/cap-beam test unit (SM3) zone of the cap beam. Column transverse spiral reinforce-
A 3/4-full-size model, designated Unit SM3, of the most ment was not continued into the joint region. Conditions for
critical column/cap-beam joints was constructed repre- joint force transfer were thus very poor.
senting the portion of bent identified as region 2 in Fig. 1.11
A 48 in. (1219 mm) diameter column reinforced with 16 #18 Test setup
(57.2 mm) bars was chosen for the prototype since this The test setup is shown in Fig. 9 and 10. Vertical support
potentially developed the highest joint shear stress of column was provided by two reaction blocks and one tie-down on
types in the Santa Monica Viaduct. A 3/4-scale model was either side of the column centerline. Vertical load representing
chosen to keep column axial and lateral loads within capacities column dead load was applied by high strength steel rods
of laboratory equipment, and because at this scale the key reacting against a strong back system at the column top load
prototype reinforcement sizes (#18 (57.2 mm) longitudinal, #6 stub, and anchored to the strong floor. This load reacted
(19.05 mm) and #4 (12.7 mm) transverse) are exactly modeled against the inner pair of vertical reaction blocks, providing
by #14 (43.0 mm), #4 (12.7 mm) and #3 (9.53 mm) bars. the correct gravity load moment/shear ratio in the cap beam
Test unit details are summarized in Fig. 8. For conve- adjacent to the column. The dead load moment diagram is
nience the test unit was constructed and tested inverted. As shown in Fig. 9(b) by profile D. Under lateral force, applied
is apparent in Fig. 8(b), the 36 in. (914 mm) deep cap-beam by an actuator reacting against a lateral load frame, the cap
included a section of superstructure 30 in. (762 mm) wide on beam tended to lift off the reaction block on one side of the
either side to provide some representation of the effective test unit, eventually developing a tension force in the tie-
flange width of soffit and deck slab expected to act with the down rods, causing positive moment bending (in prototype

ACI Structural Journal/September-October 1997 519


Fig. 11—Potential punching shear failure after joint
degradation.

Fig. 10—Column/cap beam test setup.

configuration) in the cap beam. On the other side of the cap


beam, the increased reaction was shared between the two
vertical reaction blocks, producing the cap beam moment
profile in Fig. 9(b), marked E + D. In the event of a positive
Fig. 12—Applied lateral force or displacement test sequence.
moment hinge forming in the cap beam, the double vertical
support system would enable the negative bending moment
to continue to increase on the opposite side of the column, involving variation of the column axial load between low
thus modeling the redistribution capabilities of the hypo- and high limits would represent a worst-case condition.
static cap beam in the prototype. The test pattern of applied forces and displacements is
The distribution of cap beam moments could be improved summarized in Fig. 12. Initial stages of testing were carried
by testing under a reduced dead load. As a consequence, an out under force control with later stages under displacement
axial load of PE = 130 kips (578 kN) or 54 percent of the control. Generally, three cycles were imposed at each level,
scaled prototype value PD was used for the basic test pattern. though an additional cycle was carried out at μΔ = 1.5, 3, 4.5,
At selected stages of the test program additional cycles were and 6 with axial load increased to the full dead load value of
carried out under increased axial load (PD) to investigate PD = 240 kips (1068 kN).
influence on joint behavior. At the end of the test program,
axial load was increased to 1.5PD to investigate resistance Instrumentation
against a punching shear failure of the column through the Instrumentation consisted of linear potentiometers to
cap beam. This mechanism is shown in Fig. 11. Under the measure column lateral and cap beam vertical displacements,
combined effects of high joint shear stress and poor and also to monitor column curvatures adjacent to the cap
anchorage of longitudinal column reinforcement, a pattern of beam. Electric resistance strain gauges were placed on the
intersecting diagonal and vertical cracks could occur. These reinforcing steel to enable stress distribution to be investi-
would be most likely to develop with low column axial gated. Of particular interest were the stresses developed in the
loads, since column axial compression reduces the joint prin- column longitudinal reinforcement within the joint region.
cipal tension stress. If the joint region deteriorated suffi- Full details of instrumentation are included in reference.11
ciently, it could be expected that the full column axial force
would be transmitted through the upper portion of the cap Material strengths
beam, developing a potential for punching shear failure on Based on concrete cores and impact hammer measure-
the cracks identified in Fig. 11. This would be critical under ments on the Santa Monica Viaduct, a target concrete
maximum dead load. It was thus felt that the test pattern, strength of fc′ = 5000 psi (34.5 MPa) was selected. This was

520 ACI Structural Journal/September-October 1997


Fig. 13—SM3 cap/column joint damage patterns.

expected to be on the conservatively low side for 30-year old first diagonal cracking was observed in the joint region,
concrete with specified 28-day compression strength of though a full pattern of diagonal cracking did not develop
3250 psi (22.8 MPa). Concrete was placed in three different until H = 120 kips (534 kN). Flexural cracks in the cap-beam
batches, forming the deck slab, cap beam and soffit slab, and extended across the full width of the deck slab indicating the
column respectively. Compression strengths at time of latter was fully participating with the cap-beam in resisting
testing were 5130, 4750, and 4040 psi (35.4, 32.8, and 27.8 flexural tension.
MPa) for the three batches, respectively. During cycling to the nominal “first yield” load of H =
Yield strengths of critical reinforcement were #14 (D35.8)
140 kips (623 kN), joint cracks widened significantly, and
bars: 42.5 ksi (293 MPa); #10 (D32.3) bars: fy = 44.0 ksi (303
new flexural cracks formed in the column. On cycling to the
MPa); #3 bars: fy = 65.2 ksi (450 MPa). The nominally grade
40 #3 bars were thus substantially overstrength. However, nominal experiment yield displacement (i.e., μΔ =1) damage
since failure occurred in the joint region it is not felt that this continued to concentrate in the joint region, with joint crack
overstrength contributed in any way to unrealistically widths as large as 0.05 in. (1.3 mm) being measured for both
enhanced performance. directions of response. Joint shear cracks were deflected
horizontally along the interface between the cap-beam and
Results of column/cap-beam tests deck slab, tending to separate the slab from the cap-beam.
Visual observations—Initial stages of testing produced After three cycles to μΔ=1.5, this separation had a maximum
only flexural cracks in the column and cap-beam. During width of 0.25 in. (6.4 mm). Figure 13(a) shows the crack
response to peak lateral force of H = ±80 kips (356 kN) the pattern at this stage of response.

ACI Structural Journal/September-October 1997 521


decreased dramatically, as is apparent from Fig. 14. At first
yield (H = 140 kips [623 kN]) the measured displacement
was 94 percent higher than the predicted value of 0.435 in.
(11.05 mm). The discrepancy between predicted and
measured force-displacement response continued to increase
as the joint shear failure developed.

Joint stress levels


The horizontal joint shear force may be approximately
calculated as7

Vjh = Mcol /hb (8)

where Mcol is the column moment at the cap-beam soffit,


and hb is the cap beam depth. It has been suggested7 that
Fig. 14—Test unit hysteresis curve considering all cycles of
joint diagonal cracking is more correctly related to joint
loading strength [kips] (1 kip = 4.45 kN).
nominal principal tension stress pt , rather than joint shear
stress Vjh. Thus, when the value of pt calculated from
Cracking in the joint, and between the joint region and
deck slab continued to extend as the test unit was cycled to ( fv + fh ) fv – fh ⎞ 2
- – ⎛ -------------
2
μΔ = 2, 3, and 4.5, but other cracks in the column and cap p t = -------------------
⎝ 2 ⎠
+ v jh (9)
2
beam remained stable. Lateral load decreased with
increasing displacement, and it was evident that most of the exceeds the tensile strength of the concrete, joint cracking
lateral displacement was provided by joint shear deforma- is expected. Based on test observations, it has been
tion. By μΔ = three cracks as wide as 0.4 in. (10.2 mm) were suggested7 that joint cracking can be expected when
noted in the joint region. The final cycle at μΔ = three under ⏐pt⏐>3.5 f c′ psi (0.29 f c′ MPa), and that failure of
increased axial load resulted in increased lateral resistance. joints without special joint reinforcement must be
Figure 13(b) shows cracking at this stage. Final cycles of expected if ⏐pt⏐>5.0 f c′ psi (0.42 f c′ MPa).
testing were carried out at μΔ = 4.5 and 6.0 corresponding to In Eq. (9) fv and fh are direct stresses in the joint, in vertical
a final drift angle of nearly 6 percent. Strength by this stage and horizontal directions and
had degraded to about 30 percent of the maximum, which
was developed at μΔ = 1.5. Large pieces of joint cover V jh
concrete spalled off, and sliding of the column longitudinal v jh = ---------
- (10)
reinforcement anchored within the joint was observed. bb D
Figure 13(c) shows the condition after completion of testing,
with the column deformed to maximum drift. Despite the is the nominal shear stress, related to the cap beam width
damage to the joint, the test unit was able to support an axial bb and column diameter D. For cap beams without
load of 1.5PD without any signs of incipient punching failure. prestressing, the nominal horizontal direct stress fh may
Lateral force-displacement response—Lateral force be taken as zero, and fv taken as the average vertical stress
displacement response of SM3 is shown in Fig. 14, and formed by distributing the column axial load over an area
compared with predicted flexural response based on the of cap beam found by a 45 deg spread from the column
assumption of elastic response of the cap beam, with a stiff- dimensions up to cap beam mid-depth.7
ness = 0.4Igross, an effective column length, including strain Based on the above approximations, first joint cracking (at
penetration, of 118.1 in. (3000 mm) for elastic response, and H = 80 kips [356 kN]) corresponds to a principal tension
a plastic hinge length of Lp = 20.2 in. (513 mm) in accor- stress of only 2.0 f c′ psi (0.17 f c′ MPa), though a full
dance with Eq. (6). Also shown is the lateral force level pattern of diagonal joint cracking did not develop until
corresponding to first yield of the column longitudinal rein- H=120 kips (534 kN) corresponding to ⏐pt⏐ = 3.2 f c′ psi
forcement. Predicted flexural response was based on (0.27 f c′ MPa). The maximum lateral load sustained by
measured material properties, and the stress-strain curve for SM3, at H-163 kips (725 kN) corresponded to ⏐pt⏐ =
confined concrete of Mander, Priestley and Park.10 4.5 f c′ psi (0.375 f c′ MPa). It is thus seen that both
Despite the poor anchorage of the column longitudinal cracking and failure occurred somewhat below the predicted
reinforcement, the column was able to achieve the predicted levels, though the discrepancy is not large, particularly for
“first yield” load. However, cycling to increased displace- the failure condition. It is felt that the short anchorage of the
ment levels of μΔ = 1 and 1.5 resulted in very little increase column bars created more critical conditions, and higher
in strength. At larger displacements strength decreased, as localized stress levels than anticipated by the nominal stress
noted earlier, but the degradation during successive cycles to model of reference.7 For example, if the development length,
each given level of response was rather gradual. At μΔ = 3, rather than the total beam depth hb is used in Eq. (8) to calcu-
corresponding to a drift angle of about 2.9 percent, a late the nominal joint shear force, the shear stress is
maximum horizontal force close to the first yield load could increased by 33 percent, resulting in principal tension
still be resisted in both directions of response, despite the stresses which exceed the expected limit states of 3.5 f c′
extensive damage to the joint. and 5 f c′ psi.
At initial force levels less than 80 kips (356 kN), the test It is also of interest to investigate the comparative stability
unit stiffness agreed well with predicted stiffness. However, of response between μΔ = 1 and μΔ = 2, where the unit was
as soon as joint cracking developed, the test unit stiffness capable of sustaining 10 cycles of loading without significant

522 ACI Structural Journal/September-October 1997


Fig. 15—Joint force transfer for SM3: (a) resisting mechanism;
and (b) critical tension forces at nominal.

strength degradation, despite a fully developed pattern of


crossed diagonal cracks in the joint, as is apparent in Fig. 13(a).
This can be explained by a model developed for design of new
column cap-beam joint regions,7 and outlined in Fig. 15(a).
In the absence of internal joint reinforcement to transfer
forces from column to beam, a strut and tie mechanism is Fig. 16—Column longitudinal bar strains; bar #8 small
developed. The column tension stress resultant from rein- magnitude displacement cycles.
forcement on the tensile side of the centroidal axis is concep-
tually separated into two equal components T1 and T2. T1,
furthest from the centroidal axis, is anchored by struts Cl and mechanisms, which would normally imply zero stress in
C3. Cl points back through the joint towards the compression the stirrups.
zone of column and cap beam, and anchors T2 and any Finally, it will be noted that the mechanism of Fig. 15(a)
tension force on the compressive side of the centroidal axis implies that clamping of the column longitudinal reinforcement
by a change in inclination of the strut Cl (to C2, in Fig. 15[a]). by struts Cl and C3 occurs close to the end of the bars, implying
C3 is directed outward from the joint, and is stabilized near very high bond stresses, and short effective development
the cap-beam soffit by tension forces T3 developed in cap- lengths. This was also confirmed by strain measurements on
beam shear reinforcement external to the joint and additional column longitudinal bars within the joint region. An example is
tension force T4 in the cap-beam soffit reinforcement. For shown in Fig. 16 where strain profiles along a typical column
design purposes, it is assumed that T3 is centered a distance bar are shown for stages of response up to μΔ = 1. At μΔ = 1,
about hb/4 from the column face, leading to T4 = 0.25T1 and yield strength was developed less than 10 in. (5.9db) from
T3 = 0.5T1. If T3 is centered further from the column face, the free end of the bar, implying an average bond stress of uu
more cap-beam shear reinforcement can contribute to T3, = 26 f c′ psi (2.17 f c′ MPa). This agrees well with
but at the expense of an increase in T4. values of 30 f c′ psi (2.5 f c′ MPa) reported by Paulay
Figure 15(b) shows the tension force levels that correspond and Priestley14 for fully confined conditions.
to the above assumption, where all column reinforcement on Degradation of the joint is thus attributed to instability of
the tensile side of the column axis is at yield, but without the mechanism of Fig. 15 following yield of the T3 and T4
strain-hardening, based on measured yield strength. This ties, leading to flexibility in the joint region and crushing of
approximately corresponds to nominal flexural strength. the diagonal struts inside the joint as a consequence of high
Also shown, in parenthesis, are the capacities for T3 and T4. transverse strains.
It will be seen that T3 has only 74 percent of required
capacity, but T4 has some excess capacity. However, if the CONCLUSIONS
position of T3 is moved further from the column face to Full-scale tests of lap-splice details of columns of the
utilize more stirrup reinforcement, T4 rapidly becomes over- Santa Monica Viaduct established that flexural strength
loaded. It should also be noted that some of the 139 kip (619 would be achieved at the critical section at the base of the
kN) capacity of T4 is already used up in resisting the cap- lap-splice, despite the short lap lengths and the fact that all
beam positive moment. It thus appears that the mechanism reinforcement was lapped at the same location. In the case of
should be capable of resisting a moment close to, but less the #11 (D35.8) lap-splice test, competent response was
than, the nominal column capacity, which agrees well with achieved until the unit was cycled at a displacement ductility
observed behavior. of μΔ = 3. Unit SM2, reinforced with #18 (DS7.2) bars, also
Confirmation of the behavior suggested in Fig. 15 is also performed well, achieving flexural strength at the base of the
available from strain readings in the soffit longitudinal rein- lap-splice, but degrading at μΔ = 1.5. On the basis of the test
forcement (T4) and cap-beam stirrups (T3). At maximum results and the theoretical approach for lap-splice strength
recorded response, the calculated tension force in T4 for presented in the paper, a decision was made not to retrofit the
flexural resistance was only 19 kips (85 kN), or 14 percent of lap-splices of columns reinforced with #11 bars, which
capacity. Despite this, strains recorded on the soffit rein- constituted the majority of the columns.
forcement exceeded yield.11 Similarly, stirrups were Testing of a 3/4-scale column/cap-beam test unit estab-
observed to yield for a distance of 13 in. (330 mm) from the lished that a joint shear failure could be expected as a result
column face despite shear stress levels that were less than 50 of inadequate column-bar development length and a lack of
percent of nominal strength of concrete shear-resisting special joint reinforcement. However, degradation of

ACI Structural Journal/September-October 1997 523


strength was comparatively gradual, with the column “first- and F. Seible, eds., University of California, San Diego, Structural Systems
yield” moment capacity still attainable at a drift angle of Research Project, Report No. SSRP-91/03, July 1991, pp. 333-367.
6. Priestley, M. J. N.; Seible, F.; and Fyfe, E., “Column Seismic Retrofit
close to 3 percent. The initial strength and gradual degradation
Using Fibreglass/Epoxy Jackets,” Proceedings of the ACMBS-1 Conference,
were attributed to resistance to joint failure provided by a Quebec, Canada, Oct. 7-9, 1992, pp. 287-297. Reprinted, Proceedings of
mechanism involving cap-beam soffit reinforcement and the Third NSF Workshop on Bridge Engineering Research in Progress, La
stirrup reinforcement external to the joint. Jolla, CA, Nov. 16-17, 1992, pp. 247-251.
7. Priestley, M. J. N.; Seible, F.; and Calvi, M., Seismic Design and
ACKNOWLEDGMENTS Retrofit of Bridges, John Wiley and Sons, 1996, 686 pp.
The contribution of Ms. Rebecca Wong to testing units SM1 and SM2 is 8. Priestley, M. J. N.; Seible, F.; Chai, Y. H.; and Y. L. Wong, “Santa
gratefully acknowledged. The research described above was carried out Monica Viaduct Retrofit—Full-Scale Test on Column Lap Splice with #11
under Caltrans Contract No. S9T488, with contract monitor Mr. Ali Ashna- [35] Reinforcement,” University of California, San Diego, Structural
ashari. His comments and support are also gratefully acknowledged. Views Systems Research Project, Report No. SSRP-92/08, Sept. 1992, 38 pp.
expressed in this paper are those of the authors alone, and do not necessarily 9. Priestley, M. J. N.; Seible, F.; MacRae, G. A.; Chai, Y. H.; and Wong,
reflect the official views or policies of the State of California. R., “Santa Monica Viaduct Retrofit Full-Scale Test on Column Lap Splices
with #18 [57 mm] Reinforcement,” University of California, San Diego,
REFERENCES Charles Lee Powell Structural Systems Laboratory, Test Report No. TR-93/
1. Zelinski, R., “California Highway Bridge Retrofit Strategy and 02, Jan. 1993, 50 pp.
Details,” Proceedings of the 2nd Workshop on Bridge Engineering 10. Mander, J. B.; Priestley, M. J. N.; and Seible, F., “Theoretical Stress-
Research in Progress, National Science Foundation and Civil Engineering Strain Model for Confined Concrete,” Journal of the Structural Division,
Department, University of Nevada, Reno, Oct. 1990. ASCE, V. 114, No. 8, Aug. 1988, pp. 1804-1826.
2. Priestley, M. J. N.; Seible, F.; and Uang, C.-M., “The Northridge 11. MacRae, G. A.; Priestley, M. J. N.; and Seible, F., “Santa Monica
Earthquake of January 17, 1994—Damage Analysis of Selected Freeway Viaduct Retrofit—Large-Scale Column/Cap Beam Joint Transverse Test—
Bridges,” University of California, San Diego, Structural Systems Research Preliminary Report,” University of California, San Diego, Charles Lee
Project, Report No. SSRP-94/06, Feb. 1994, 260 pp. Powell Structural Systems Laboratory, Technical Report No. TR- 94/02,
3. Priestley, M. J. N.; Ely, A.; Scott, R. F.; and Seible, F., “Seismic Safety Aug. 1994, 117 pp.
Review of Santa Monica Viaduct Retrofit,” Report by the Seismic Safety 12. Bridge Design Specifications, California Department of Transporta-
Review Panel to the Caltrans Office of Structures Design, California tion, June 1990.
Department of Transportation, Division of Structures, Sacramento, CA, 13. ACI Committee 318, “Building Code Requirements for Reinforced
Jan. 1992, 410 pp. Concrete (ACI 318-89) and Commentary (318R-89),” American Concrete
4. “Loma Prieta Earthquake Reconnaissance Report,” Supplement to V. Institute, Detroit, 1989.
6, Earthquake Spectra, May 1990, pp. 151-187. 14. Paulay, T., and Priestley, M. J. N., eds., Seismic Design of Reinforced
5. Priestley, M. J. N., “Seismic Assessment of Existing Concrete Concrete and Masonry Buildings, John Wiley and Sons, Inc., New York,
Bridges,” in Seismic Assessment and Retrofit of Bridges, M. J. N. Priestley 1992, 744 pp.

524 ACI Structural Journal/September-October 1997

You might also like