Sie sind auf Seite 1von 14

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/288316253

Shear and flexural behavior of lightweight concrete bridge columns in seismic


regions

Article  in  Aci Structural Journal · January 1999

CITATIONS READS

54 719

3 authors, including:

Mervyn Kowalsky
North Carolina State University
130 PUBLICATIONS   3,347 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Seismic Load Path Effects in Reinforced Concrete Bridge Columns and Wall Piers View project

Reinforced Concrete Filled Steel Tubes in Soil View project

All content following this page was uploaded by Mervyn Kowalsky on 22 March 2016.

The user has requested enhancement of the downloaded file.


ACI STRUCTURAL JOURNAL TECHNICAL PAPER
Title no. 96-S16

Shear and Flexural Behavior of Lightweight Concrete


Bridge Columns in Seismic Regions
by Mervyn J. Kowalsky, M. J. Nigel Priestley, and Frieder Seible

This paper describes research related to the use of lightweight has been questionable, such as in the Olive View Hospital
concrete for bridge column construction in seismic regions. Of Complex2 during the 1971 San Fernando Earthquake, where
interest is the behavior of columns dominated by shear as well as several lightweight concrete columns failed, apparently as a re-
those dominated by flexure. A series of large-scale experimental sult of the combination of poor detailing and the potentially brit-
tests was performed to assess the response of reinforced light- tle failure of lightweight concrete.
weight concrete columns in the inelastic range. Results indicate As part of the capacity design approach,3 bridge columns are
that a reduction in shear and flexural strength is appropriate for designed for ductile response in earthquakes.4 Therefore, in or-
lightweight concrete, while displacement capacity and energy dis- der to use lightweight concrete for members expected to form
sipation are not affected in a significant manner. flexural hinges, the inelastic response must be assessed. Specif-
ically, we seek to determine the strength of the lightweight con-
Keywords: bridges; ductility; lightweight concrete; seismic design; shear. crete shear resisting mechanism across a wide range of ductility
such that the structure can be protected against shear failure. It
INTRODUCTION is also important to quantify flexural response such that the de-
The use of lightweight concrete in structural design dates pendability of lightweight concrete flexural hinges can be as-
back to almost two millennia. Portions of the Pantheon were sessed. Specifically, we are interested in flexural strength,
constructed from lightweight concrete where the aggregate con- stiffness, displacement capacity, and energy dissipation ability.
sisted of pumice.1 With the advent of artificially made light- To meet these research objectives, a series of large scale col-
weight aggregate early in the twentieth century, it became
umn tests was undertaken as part of a lightweight concrete re-
possible to obtain lightweight aggregate concrete with compres- search program. In this paper, the results from the first two
sive strengths similar to normal weight concrete. This type of phases of testing are discussed where the shear and flexural be-
lightweight concrete has been used recently for the design of
havior are of interest.5,6 In all, five columns were constructed and
building structures as well as for bridge deck pavement and in a
tested during the first two phases. The third phase covers shake
more limited role for entire bridge superstructures.
table testing, which will be reported separately.
The main advantage that lightweight concrete accords the en-
gineer is a reduction in weight. In bridge design, this means re-
duced gravity load and seismic inertial mass, resulting in RESEARCH REVIEW
reduced member sizes and foundation forces. Generally, only Although there has been extensive work done on material as-
superstructures are constructed of lightweight concrete with the pects of lightweight concrete, there has been comparatively lit-
substructure in normal weight concrete. tle relevant research regarding the inelastic behavior of the
material under seismic conditions.
It is felt that the advantages that lightweight concrete provide
the engineer can be further enhanced by applying the use of the Research done in Italy by Monti and Nuti7 included cyclic
material to the design of bridge substructure elements such as testing of 14 beam column subassemblies. The authors conclud-
cap-beams, joints, and columns. While the mass of the super- ed that the performance of the lightweight specimens was com-
structure is generally much more significant than the column parable to the equivalent normal weight specimens in terms of
masses, there are cases such as deep narrow valley crossings or flexural strength and displacement capacity.
river crossings with navigation channels where the pier height In 1980 Bertero et al. published a paper8 describing experi-
results in the column mass becoming significant when com- mental results of two lightweight concrete beam-column
pared to the superstructure mass. Also, in the case where span sub-assemblages. Results from the first test unit, which was
lengths are comparatively small, the column mass can become tested monotonically, compared very favorably with an earlier
significant. Lightweight concrete can also be used in retrofit ap- test unit constructed from normal weight concrete as displace-
plications where a concrete column jacket is desired due to ar- ment capacity, strength, and energy absorption were essentially
chitectural reasons over other methods such as steel or identical. Results from the cyclic test were considerably less fa-
composite jackets. In that situation, a conventional weight con- vorable. From that test, a comparison with normal weight con-
crete jacket might result in foundation forces that would require crete indicated lower strength and energy absorption.
expensive foundation retrofit whereas a lightweight concrete Displacement capacity was essentially the same. Bertero et al.
jacket may not require a footing retrofit.

ACI Structural Journal, V. 96, No. 1, January-February 1999.


RESEARCH OBJECTIVE Received September 12, 1997, and reviewed under Institute publication policies.
The objective of the research described in this paper is to Copyright  1999, American Concrete Institute. All rights reserved, including the
making of copies unless permission is obtained from the copyright proprietors. Perti-
quantify the seismic performance of lightweight concrete bridg- nent discussion will be published in the November-December 1999 ACI Structural
es. Past seismic performance of lightweight concrete buildings Journal if received by July 1, 1999.

136 ACI StructuralJournal/January-February1999


Mervyn J. Kowalsky is an assistant professor in the department of civil engineering
at North Carolina State University, Raleigh, NC. He is a member of ACI Committee
213, Lightweight Aggregate and Concrete. His research interests include earthquake
engineering and the seismic design and analysis of concrete structures.

M. J. Nigel Priestly is a professor of structural engineering at the University of Cali-


fornia, San Diego, Calif.

Frieder Seible is a professor of structural engineering at the University of California,


San Diego.

attributed this difference to bond deterioration in earlier cycles


as a result of the weaker aggregate.
The objective of the experimental work performed by Shah et
al.9 was to characterize the effect of confinement on lightweight
concrete. Their testing consisted of both lightweight and normal
weight specimens such that a direct comparison could be made.
The test units were conventional test cylinders tested under con- Fig. 1—Strength of concrete shear resisting mechanism for nor-
mal weight concrete.12
centric axial loading. Their results indicated that confinement
has little effect on lightweight concrete compressive strength.
However, the confinement still provided enhanced strain capac- many code approaches. The concrete mechanism strength is
ity similar to that exhibited by normal weight concrete. thus expressed as Eq. (2)

LIGHTWEIGHT CONCRETE MATERIAL ASPECTS V c = 0.8A g αβ K f c′ (MPa) (2a)


The criteria for mix proportioning selection was that the con-
crete should be readily available, economical, and not require
any reactive admixtures. The mix proportioning used included V c = 0.8A α β ( 12K ) f c′ (psi) (2b)
g
only lightweight coarse aggregates (Hydrolite) and had a target
density of 1760 kg/m3 and target strength of 35 MPa at 28 days. In Eq. (2), the variable α is given by Eq. (3) and accounts for
The target slump was 76 mm. Complete mix proportioning the observed effect of column aspect ratio on the strength of the
specifications can be found in Reference 5. concrete shear resisting mechanism. The variable β considers
the effect of longitudinal steel ratio on the strength of the con-
SHEAR DOMINATED COLUMN TESTS crete shear resisting mechanism and is given by Eq. (4). The
Objective variable K is given in Fig. 1 and is meant to characterize the re-
The brittle nature of shear failure is classified as a mode of de- duction in strength with increasing ductility due to the increase
formation that is to be avoided in bridge design. In addition, the in crack width and corresponding decrease in aggregate inter-
use of lightweight concrete in members experiencing high levels lock. Lastly, Ag is the column gross area, and f ′ c is the concrete
of inelastic deformation in a large earthquake could compound compressive strength in MPa units
the problem due to the potentially brittle nature of the concrete it-
self. The objective of the research on shear strength of light-
1 ≤ α =  3 – -------- ≤ 1 . 5
M (3)
weight concrete columns is to quantify any reduction in shear  V D
strength required when compared to normal weight concrete.

Normal weight concrete model β = ( 0.5 + 20ρ l ) ≤ 1 (4)


The basis for comparison was the predictive reinforced con-
crete shear model proposed by Priestley et al. 10 This model, Recent research has also resulted in a modification to the steel
which shows good agreement with experimental results for nor- truss mechanism.5 The revised relation, which is shown as Eq.
mal weight concrete, defines the shear strength of a concrete (5), considers the effect of the concrete compression zone on
member as the sum of three components in accordance with Eq. mobilization of the transverse steel. The interested reader is re-
(1) ferred to Ref. 5 for details regarding development of the model.
In Eq. (5), A sp is the spiral area, f y is the spiral yield stress, D′ is
V n = Vc + Vp + Vs (1) the confined core diameter, c is the section neutral axis depth, s
is the spiral spacing, and θ is the assumed crack angle measured
from the column vertical axis (usually assumed to be 30 deg for
In Eq. (1), V c represents the concrete shear resisting mecha-
assessment of existing structures).
nism, Vs represents the truss mechanism depending on trans-
verse reinforcement, and Vp represents the strength provided by
π A sp D ′ – c
the axial compression in the member. V s =  -----------f y -------------- cot ( θ ) (5)
Recent research11 has suggested that the concrete mechanism  2 s 
strength, as well as being dependent on the ductility of the sec-
tion, should reflect the influence of longitudinal reinforcement The last component comprising shear strength of a reinforced
ratio ρ l , and member effective aspect ratio M/VD, which were concrete member in the UCSD model is that due to the horizon-
not included in the original model, but that are recognized in tal component of the diagonal axial load strut which forms be-

ACI StructuralJournal/January-February1999 137


Table 1—Material properties for all tests
Unit γ, kg/m3 f ′′c , MPa f ′′r , MPa fyt , MPa fu l , MPa fyh, MPa fuh , MPa
SL1 1835 47.6 2.90 444 662 — 476
SL2 1835 50.6 2.83 444 662 — 674
FL1 2035 36.2 — 477 777 448 538
FL2 2035 40.0 3.09 477 777 437 514
FL3 2484 38.6 2.94 477 777 448 538

tween the top and bottom of the column. Eq. (6) represents the
axial load component of shear strength for a column in double
bending. In Eq. (6), P is the column axial load, D is column di-
ameter, and c is the neutral axis depth. For new design, each of
Eq. (2) and (6) are multiplied by 0.85, and a crack angle of θ =
35 deg is assumed in Eq. (5) to provide an effective lower bound
to experimental results.4
(a)
P( D – 2 ( c ⁄ 2 ) )
V p = ------------------------------------- (6)
L

Test unit details


To investigate the entire envelope of shear strength, two cir-
cular column test units were constructed. The first specimen,
SL1, was designed with transverse reinforcement such that a
brittle shear failure at low ductility would be likely. The goal of
this specimen was to investigate the nonductile strength of the
concrete shear resisting mechanism, as well as providing a
means for a direct comparison with a column, NC5, constructed
from normal weight concrete and tested earlier. 13 The second
test unit, SL2, was designed for a ductile shear failure at high
ductility to investigate the rate of degradation of the concrete
shear resisting mechanism with increasing ductility.
In both cases, the test units were designed by selecting a rein-
forcement configuration whereby the theoretical flexural re- (b)
sponse intersects the shear capacity envelope at a specified
ductility level. Both test units were 2.44 m tall and 610 mm in di- Fig. 2—Theoretical flexural response and shear capacity enve-
ameter. Longitudinal reinforcement for both test units consisted lope for unit: (a) SL1; and (b) SL2.
of continuous 26 no. 6 (19.1 mm) Grade 60 (f y = 414 MPa) rein-
forcing bars (ρ l = 2.53 percent). The transverse reinforcement strain penetration into the footing and loading stub which is given
was varied in each of the test units. Unit SL1, which was designed by Eq. (8)4 where Lclear is the clear column height, fy is the longi-
for a brittle shear failure contained no. 2 (6.35 mm) Grade 40 (f y tudinal bar yield stress, and dbl is the longitudinal bar diameter.
= 275 MPa) smooth hoops at 127 mm spacing (ρ s = 0.17 percent), The shear deformation at first yield, ∆′st, is then calculated using
while unit SL2 contained no. 3 (9.5 mm) Grade 60 deformed spi- the procedure described by Priestley et al.16 which is given by Eq.
ral at a pitch of 76 mm ( ρs = 0.68 percent). Concrete and steel ma- (9). In Eq. (9), the variable εt represents the spiral strain at first
terial properties can be found in Table 1. yield of the column longitudinal reinforcement, that can be ob-
Fig. 2 illustrates the theoretical flexural response and shear ca- tained by solving for the spiral stress in Eq. (5) where V s is ob-
pacity envelopes for units SL1 and SL2. The theoretical flexural tained by subtracting the concrete and axial load components of
response shown was obtained from a section analysis 14 using the shear strength from the column first yield force. The total dis-
Mander model for confined concrete15 modified for lightweight placement at first yield is then given by Eq. (10). For levels be-
concrete.5 The modification consisted of a reduced elastic mod- yond first yield, the plastic curvature is assumed to be constant
ulus of 18,200 MPa for unit SL1 and 18,800 MPa for unit SL2 across a plastic hinge length given by the greater of Eq. (11).4 In
(≈ 35,000 MPa for normal weight concrete of similar strength) Eq. (11), the variable L is the column height to the point of inflec-
and a strain at maximum concrete compressive stress of 0.004. tion, which for a column in double bending is given by 50 percent
In each case the modifications were obtained from cylinder tests of the clear column length, Lclear. The plastic curvature φp is cal-
on the concrete placed in units SL1 and SL2. culated with Eq. (12) where φ is the column curvature, while the
The calculation of the predicted response is performed by con- plastic displacement ∆ p is obtained by Eq. (13). The total dis-
verting the moment-curvature response into a force-displacement placement for a given curvature level is then obtained by Eq. (14),
response. The displacement of the column at first yield of the lon- where F is the lateral force at the curvature level φ , and Fy′ is the
gitudinal reinforcement is obtained by application of the mo- column first yield force. In addition to an envelope prediction, a
ment-area method whereby the first yield curvature, φ′y, is used to bilinear prediction is easily obtained as well. The first point is es-
obtain the first yield flexural displacement, ∆′f . For a column in tablished by extrapolating the column first yield displacement to
double bending, this is obtained as in Eq. (7). In Eq. (7), Leff rep- the ideal column lateral force [Eq. (15)] which is defined as the
resents the effective column length, including a component for force at five times the first yield curvature. The maximum dis-

138 ACIStructuralJournal/January-February 1999


placement capacity is obtained by establishing the maximum
concrete compression strain, εcu , which is given by Eq. (16).12
In Eq. (16), ρ s represents the transverse volumetric steel ratio, fyh
the transverse spiral yield stress, ε su the maximum steel strain
(usually 0.12) and f c′ c the confined concrete compressive
strength determined from the Mander et al.15 energy balance ap-
proach. Then, the column maximum curvature φm is obtained
with Eq. (17) where c is the neutral axis depth at maximum re-
sponse. Displacement capacity is then assessed by applying Eq.
(12) through (14).
Column displacement ductility, which will be referred to
throughout this paper, is defined by the column displacement
divided by the displacement given by Eq. (15). However, in
terms of experimental response, the displacement at first yield
is obtained directly from test results

φ′y Leff
2 Fig. 3—Shear test setup.
∆ ′f = ---------------- (7)
6 is reduced by 15 percent as suggested by the ACI code17 to ac-
count for the use of lightweight concrete. The steel truss mech-
Leff = L clear + 0.044f y d b l (MPa) (8a) anism strength was taken to be that without the neutral axis
modification since at the time of the design, the revised equation
[Eq. (5)] had yet to be developed. As seen from Fig. 2, unit SL1
L eff = L clear + 0.30f y d bl (ksi) (8b) is designed for a shear failure near a displacement ductility of
1.0, and unit SL2 was expected to fail in shear at a ductility of
approximately 5.0.
2 ( V c + V p ) L clear
∆′s = --------------------------------------- + εt Lc l e a r (9) The specimens were tested in double bending along a north
( 0.4E c) ( 0.8A g ) (push) south (pull) axis in an existing shear test setup 13 as shown
in Fig. 3. Axial load was applied through a pair of hollow core
∆′ y = ∆′ f + ∆′ s (10) jacks that were attached beneath the strong floor to metal rods
that extend to a cross beam at the column top. The axial load lev-
el was 592 kN. External column instrumentation consisted of lin-
L p = 0.08L + 0.022f y d b l (11a)
ear displacement transducers placed on the north and south faces
to measure column curvature, and on the east and west faces to
(MPa)
L p = 0.30f y d bl measure shear deformation. Internal instrumentation consisted
of strain gages on the east and west faces of every hoop in unit
(MPa) SL1, and every spiral on SL2. Additionally, strain gages were
Lp = 0.08L + 0.15f y d bl (11b)
placed on the north and south faces of SL2 at the top and bottom
of the column to measure confinement induced hoop strains. The
(ksi) interested reader is referred to Reference 5 for a detailed discus-
L p = 0.30 f y d b l
sion of the instrumentation setup.
(ksi)
φ p = φ – φ′y (12)
Test observations—SL1
The overall behavior of unit SL1 was dominated by extensive
∆ p = φ p L p L clear (13) shear cracking, as expected. The specimen failed to reach its pre-
dicted flexural strength due to the onset of a shear failure at duc-
tility 1.0 (column displacement of 22 mm). The displacement at
F ductility 1.0 was obtained by applying Eq. (15) where ∆′y is the
∆ = ∆′y ----- + ∆p (14)
Fy experimental displacement corresponding to the applied lateral
force, F y′ . The test unit achieved one cycle at ductility 1.0 after
F which significant strength degradation occurred. Fig. 4 illustrates
∆y = ∆′y -----i (15) a close-up of cleanly fractured aggregate and an overall view of
Fy
the test unit after removal of instrumentation.

1.4ρ s f yh ε s u Test observations—SL2


εc u = 0.004 + ---------------------------- (16)
f ′c c Unit SL2, which contained significantly more transverse steel
than unit SL1, performed well. Stable response was achieved
until failure, which consisted of a fracture of the transverse re-
εc u
φ m = ------- (17) inforcement at ductility 6.0. SL2 nearly reached its predicted
c lateral flexural strength, and failed at the expected
displacementductilitylevel.
Shear capacity envelopes in Fig. 2 are shown for normal In the first push cycle to ductility 6.0, a large shear crack
weight concrete, as well as one where the concrete component formed at the top of the column, and extended down 600 mm at

ACI StructuralJournal/January-February1999 139


(a) (b)

Fig. 4—(a) Close-up of cleanly fractured aggregate; and (b) SL1 after testing.

(a) (b)

Fig. 5—(a) West face of SL2 at ductility 6; and (b) shear failure of SL2.

140 ACIStructuralJournal/January-February 1999


an angle of approximately 30 deg [Fig. 5(a), west side]. Fig. flexural and shear deformation components [Fig. 7(f)] from
5(b) provides a clear view of the crack that caused fracture of measured potentiometer readings illustrates that flexural defor-
the spirals on the west face, allowing an accurate measurement mation dominated to a displacement ductility of 1.0 at which
of the failure angle. The critical angle had an inclination of point the shear deformation became very significant.
about 30 deg. The force-displacement response of SL2 is shown in Fig.
8(a), along with the theoretical flexural response. The column
Test results behaved dependably through ductility 4.0, exhibiting excellent
In terms of the actual column response, SL1 first exhibited energy absorption and strength gain. Degradation due to cycling
severe strength degradation on the return cycle from ductility was no more than normally expected of a well-confined flexural
1.0, indicative of a shear failure [Fig. 6(a)]. SL1 failed to reach response. The failure at ductility 6.0 was marked by some insta-
its predicted ideal flexural strength of 775 kN in either direc- bility in the response and loss of capacity. The failure mode was
tion of loading. The hysteresis loops exhibited severe pinch- very brittle, typical of shear failure, and gave little warning. The
ing, and a significantly reduced stiffness after ductility 1.0, theoretical flexural response matched the actual response very
which is as expected. A significant departure from the predict- well up to the predicted ultimate ductility of µ∆ = 4. Overall, the
ed flexural response occurred beyond 400 kN. The discrepan- column obtained a dependable drift of 3.3 percent and failed at
cy was due to the underestimation of shear deformation at a drift of 5 percent. A summary of experimental and analytical
higher levels of response. An identical column was construct- parameters is shown in Table 2. The variable F u represents the
ed from normal weight concrete (Unit NC5 13) and its response maximum force assuming flexural response, while ∆ u repre-
is shown in Fig. 6(b),13 which indicates similar response to sents the maximum flexural displacement based on the value for
SL1. the maximum concrete compression strain from Mander’s ener-
A summary of experimental results for Unit SL1 are shown gy balance approach.15 The variable ∆y represents the column
in Fig. 7. Transverse steel strain profiles at column response displacement corresponding to the lateral force at first yield of
levels of 85 percent F y , Fy , and µ ∆ =1 are shown for the east the longitudinal reinforcement. The analytical first yield dis-
face in each direction of loading as well as the west face in the placement included the reduced modulus for lightweight con-
pull direction. Fy represents the theoretical column first yield crete as well as a component for shear deformation in
force. The profiles indicate concentration of shear strains at ap- accordance with Eq. (9). The
proximately 450 mm (0.75 D) from the column end regions, procedureforthecalculationwaspreviouslydiscussed.
which is consistent with previous experimental results.4 The Transverse steel strain profiles are shown in Fig. 8(b) for unit
strain-displacement history of the gage located at 450 mm from SL2 for response levels from ductility 1.0 to 6.0. Note the drastic
the base is shown in Fig. 7(d). From this figure, it is clear that increase in strain at ductility 6.0, indicating the onset of a shear
the steel truss mechanism did not contribute resistance until a failure. The curvature profiles for SL2 shown in Fig. 8(c) indicate
column displacement of 8 mm (approximately 52 percent of significantly different behavior from SL1, as the inelastic defor-
experimental column first yield). At that time, rapid straining mation is concentrated in the end regions. A breakdown of shear
occurred in the transverse steel to a level near the yield strain. and flexural deformation is shown in Fig. 8(d). Although flexural
Extensive demand on the hoop involving hoop yielding oc- deformation always constituted the majority of the deformation,
curred at 13 mm of column displacement, which corresponded the shear deformation throughout the loading history was
to 85 percent of the experimental column first approximately33percentofthetotaldeformation.
yielddisplacement.
The curvature distribution of Unit SL1 is shown in Fig. 7(e) Discussion of concrete shear resisting mechanism
for each direction of loading. From this figure, it is noted that The strength of the concrete shear resisting mechanism con-
yielding occurred near the base region with an essentially lin- sists of two distinct regions: 1) pre-diagonal cracking, and 2)
ear distribution throughout the column height. Calculation of post-diagonal cracking. In considering the response of a column

(a) (b)
Fig. 6—(a) SL1 hysteretic response (1 kN = 225 lb; 1 mm = 0.039 in.); and (b) NC5 hysteretic response.12

ACI StructuralJournal/January-February1999 141


(c)
(a) (b)

(d) (f)

(e)

Fig. 7—SL1 experimental results summary: (a) shear induced strains east face, push; (b) shear induced strains east face, pull; (c) shear
induced strains west face, pull; (d) strain history, hoop gage at 450 mm from base; (e) curvature profile; and (f) deformation components.
(1 mm = 0.039 in.)
under an increasing cyclic load, there is a certain stage when a As the response is increased further, there comes a point
column exhibits diagonal shear cracks. This occurs when the where the strength of the concrete shear resisting mechanism is
principal tension stress in the concrete near the centroidal axis maximized. This is commonly referred to as the maximum non-
exceeds the tension strength of the concrete. The shear force ductile strength of the concrete shear resisting mechanism in the
that coincides with this level of response V cd is known as the model.10
shear force to initiate diagonal cracking. For shear forces less Further increases in seismic attack result in larger ductility
than V cd, the concrete shear resisting mechanism consists of levels. As the ductility level increases, the diagonal cracks be-
the diagonal tension strength of the concrete. Since no diago- come larger and the strength of the concrete shear resisting
nal cracks form below Vcd , the steel truss component is zero. mechanism degrades as aggregate interlock breaks down. The
The axial load component, however, does exist and contrib- maximum strength of the concrete shear resisting mechanism at
utes to the overall shear strength. Diagonal cracking is charac- high ductility is known as the ductile strength of the concrete
terized by the activation of transverse reinforcement as shown shear resisting mechanism. In this paper, the nonductile and
ductile strengths of the lightweight concrete shear resisting
in Fig. 7(d) for Unit SL1.
mechanism will be discussed.
Once diagonal cracking is initiated in a column, the concrete
shear resisting mechanism changes completely. After crack-
Nonductile strength of lightweight concrete shear
ing, the tension strength of the concrete clearly has little influ-
resisting mechanism
ence on the strength of the shear resisting mechanism. Instead, As a means of quantifying the reduction in the strength of the
the strength of the concrete shear resisting mechanism is pro- nonductile concrete shear resisting mechanism, unit SL1 was
vided mainly by aggregate interlock due to shear friction compared with an equivalent normal weight concrete test unit
across the crack. Also, after diagonal cracking occurs, the tested earlier (NC5).13 The configuration of SL1 was identical to
transverse reinforcement becomes active, and the steel truss that of NC5, including only small differences in steel properties.
mechanism contributes to the overall shear capacity of the sec- However, there was a significant difference in concrete compres-
tion. sive strength (47.6 MPa for SL1 and 35.9 MPa for NC5).

142 ACIStructuralJournal/January-February 1999


(a)

(c)

(b) (d)

Fig. 8—SL2 experimental results summary: (a) force-displacement hysteretic response; (b) shear induced strain profile; (c) curva-
ture profiles; (d) deformation components. (1 kN = 225 lb; 1 mm = 0.039 in.)

Table 2—Expected and recorded parameters Ductile strength of lightweight concrete shear
Fu ∆u ∆y
resisting mechanism
analytic, F u exp., analytic, ∆u exp., analytic, ∆y exp., To compare the ductile strength of the lightweight concrete
Unit kN kN mm mm mm mm shear resisting mechanism with that of normal weight concrete, a
SL1 775 606 33.5 22 13.1 15.3 column tested by Priestley et al., referred to as N1,18 was ana-
SL2 869 847 69.5 117 12.6 13.9 lyzed and compared to SL2. The analysis was performed by sub-
tracting the truss mechanism component [Eq. (4)] and the axial
Test results from Unit NC5 indicate that the maximum lateral load component from the total lateral force at discrete levels of
force before shear failure was 614 kN, which compares with 606 response. Details regarding the approach can be found in Ref. 5.
The results are shown in Fig. 9, where at high levels of ductility
kN for Unit SL1. Of the total shear force, 409 kN was resisted by
the strength of the lightweight concrete shear resisting mecha-
the strength of the concrete shear resisting mechanism in Unit
nism for SL2 is as much as 30 percent lower than for N1.
NC5. Assuming that the strength of the concrete shear resisting
mechanism is proportional to the square root of the compressive
Recommendations for shear design
strength, as is generally suggested, this results in a normalized
Based on the results of the experimental testing, it is suggest-
lateral strength of 676 kN for NC5. Therefore, considering the
ed that the strength of the concrete shear resisting mechanism in
difference in concrete strength between NC5 and SL1, we can in- the model be reduced by 15 percent for the nonductile strength
fer that a reduction in the nonductile strength of the concrete and by 30 percent for the ductile strength. The suggested model
shear resisting mechanism of 15 percent is appropriate for light- as a function of displacement ductility is shown in Fig. 10. In
weight concrete, which is consistent with the ACI recommenda- each case, K is the factor which accounts for ductility level in
tions for lightweight aggregate concrete. the strength of the concrete shear resisting mechanism, and is

ACI StructuralJournal/January-February1999 143


Fig. 9—Comparison of concrete shear strength. 5

Fig. 11—Flexural test setup.

Table 3(a)—Summary of predicted strength and


displacement for flexurally dominated columns
φ′y , ∆′ y,, M′ y,, Mu , φu, ∆u,
Unit 1/m mm kNm kNm 1/m mm
FL1 0.01159 57 511 675 0.09780 203
FL2 0.01137 56 522 641 0.06947 155
FL3 0.01146 56 519 683 0.09572 199

Table 3(b)—Summary of experimental strength and


displacement for flexurally dominated columns
∆′ y,, Mu , ∆u ,
Unit mm kNm mm
FL1 68.3 589 332
Fig. 10—Lightweight and normal weight concrete models. FL2 58.4 641 210
FL3 54.5 673 340
employed directly in Eq. (2). The values of α and β remain the
same as in Eq. (3) and (4). The steel truss mechanism, Eq. (5), Test unit details
and axial load component, Eq. (6), remain the same as presented All three test specimens were 3.66 m tall and 457 mm in diam-
earlier in this paper. eter, resulting in an aspect ratio of 8. Longitudinal reinforcement
in all cases consisted of 30 no. 5 (15.9 mm) Grade 60 (414 MPa)
FLEXURAL DOMINATED COLUMN TESTS continuous bars (ρ l = 3.62 percent). The only difference between
Objective FL1 and FL2 was the amount of confinement reinforcement. FL1
Inelastic flexural deformation with low levels of strength deg- contained no. 3 (9.5 mm) Grade 60 deformed spiral at 76 mm
radation is the desired mode of energy dissipation for reinforced
spacing ( ρs = 0.87 percent), while FL2 contained no. 2 (6.4 mm)
concrete columns in seismic regions. The basis of capacity
Grade 60 deformed spiral at 51 mm spacing ( ρs = 0.60 percent).
design3 is that flexural hinges must be detailed to ensure ade-
Unit FL3 is as Unit FL1 except it was constructed from normal
quate and dependable strength and deformation capacity, while
weight concrete. Axial load was applied through the same setup
all other modes of deformation are protected against.
as described for the shear dominated columns. The axial load lev-
To use lightweight concrete in bridge members designed for in-
el for the three flexural tests was 1780 kN. The test setup for these
elastic flexural deformation, it is necessary to quantify the behav-
test units is shown in Fig. 11.
ior of the material at the survival limit state in terms of strength,
stiffness, displacement capacity, and energy absorption charac- Through the use of Chai’s modification to Mander’s energy
teristics. Additionally, the service limit state is of interest in terms balance approach [Eq. (16)],12 transverse reinforcement was
of evaluating the concrete strain that initiates crushing. sized and an approximate displacement capacity obtained. Flex-
To accomplish the objectives stated above, three slender col- ural strength was calculated through the use of moment-curva-
umns were designed and tested in single bending. Two of the ture analysis14 where stress-strain models for confined and
columns, FL1 and FL2, were constructed from lightweight con- unconfined concrete were considered.15 Table 3 represents a
crete, and FL3 was a duplicate of FL1 constructed in normal summary of the predicted and experimental quantities for units
weight concrete for comparison purposes. Results are present- FL1, FL2, and FL3 using day-of-test material properties. Con-
ed in terms of test observations as well as an examination of the crete and steel material properties are listed in Table 1.
overall hysteretic response. Comparisons between FL1 and Instrumentation for these test units consisted of external dis-
FL3 illustrate any differences between the concrete types. placement devices for measuring column curvature and top dis-

144 ACIStructuralJournal/January-February 1999


(a) (b) (c)
Fig. 12—(a) FL1 at ductility 4; (b) FL2 at ductility 3; and (c) FL3 at ductility 5.

placement, as well as internal strain gage devices. Four spirals cycling to 332 mm in FL1. Fig. 12 represents photos of each of
in the hinge region of all three test units were instrumented with the test units at maximum capacity.
eight gages each in an effort to obtain the strain distribution
along the spiral. Test results
The overall force displacement hysteretic response for the
Test observations—FL1 three test units are shown in Fig. 13. Note that the response of
The first test conducted was that of Unit FL1. Unit FL1 start- Units FL1 and FL3 indicate very similar behavior between
ed to show signs of crushing at first yield of the longitudinal re- lightweight and normal weight concrete. In each case, the re-
inforcement, which coincided with an extreme fiber sponse is stable and dependable. Also shown in Fig. 13 is a plot
compression strain of 0.0025. This is less than would be expect- of the envelope response for Units FL1 and FL3. From this fig-
ed for normal weight concrete, where 0.005 is generally used as ure, it is clear that FL1 exhibited a slight reduction in strength
a lower bound. 4 However, the column exhibited dependable be- as well as initial stiffness when compared with FL3.
havior with little strength degradation after three cycles at each The strain profiles along the spiral located at 150 mm from
ductility level. Failure in FL1 was a result of buckling and sub- the base for FL1 (lightweight) and FL3 (normal weight) are also
sequent rupture of five longitudinal reinforcement bars at duc- shown in Fig. 13. A comparison of these plots indicates that a
tility 4.0. FL1 exceeded its theoretical maximum displacement higher demand on the transverse reinforcement is apparent for
Unit FL1 than for FL3 which may indicate earlier exhaustion of
capacity of 203 mm by a factor of 1.64. The maximum drift
achieved was 9.1 percent. the transverse steel strain capacity, and hence, slightly reduced
displacement capacity for lightweight concrete. The implication
is that a lightweight concrete column will dilate more than a nor-
Test observations—FL2 mal weight concrete column, resulting in larger lateral strains
Unit FL2, which contained a lower level of transverse rein- for the same level of response.
forcement, exhibited concrete crushing at ductility 1.0, which
Column curvature profiles for the three test units are shown
coincided with a concrete strain of 0.003. Just as in Unit FL1, in Fig. 14. Note that the inelastic curvature was concentrated at
FL2 exhibited dependable behavior with little strength degrada- a level of 300 to 400 mm from the column base, indicating the
tion upon successive cycling. The mode of failure for FL2 was significant confinement effect provided by the footing.
both buckling and subsequent rupture of longitudinal reinforce-
ment as well as fracture of spiral reinforcement.
Elastic stiffness
To assess the influence of lightweight concrete on member
Test observations—FL3 stiffness, the experimental column stiffnesses at first yield were
Unit FL3, which was identical to FL1 except for the concrete calculated and compared to the analytical column stiffnesses as-
type, exhibited very dependable behavior. Splitting cracks suming normal weight concrete behavior. The results are shown
which occurred in FL1 were much less pronounced in FL3. in Table 4(a). The experimental stiffnesses were obtained by di-
Spalling of the concrete in FL3 occurred at ductility 1.5, which viding the average of the first yield forces in each direction of
is much later than in Unit FL1 and coincided with an extreme loading by the average of the first yield flexural displacement in
fiber compression strain of 0.0036. The failure mode of FL3 each direction of loading, as shown in Eq. (18). The analytical
was very similar to FL1. Longitudinal bars buckled at a dis- stiffnesses were obtained by dividing the first yield force by the
placement of 340 mm in FL3, whereas buckling occurred after first yield analytical flexural displacement as in Eq. (19).

ACI StructuralJournal/January-February1999 145


(a) (b) (c)

(d) (e) (f)


Fig. 13—Flexural test experimental results summary: (a) FL1 force-displacement hysteresis; (b) FL2 force-displacement hysteresis; (c)
FL3 force-displacement hysteresis; (d) FL1 and FL2 response envelopes; (e) FL1 spiral profile at 150 mm from base; and (f) FL3 spiral
profle at 150 mm from base. (1 kN = 225 lb; 1 mm = 0.039 in.)

(a) (b) (c)

Fig. 14—Column curvature profiles for flexurally dominated columns: (a) FL1; (b) FL2; and (c) FL3. (1 mm = 0.039 in.)

F ypos + F y neg FL3, the recorded and analytical stiffnesses are essentially the
k exp = ----------------------------------- (18)
∆′ypos + ∆′ yneg same, as expected.
For Units SL1 and SL2, the elastic modulus was measured us-
ing concrete test cylinders. 5 The results indicated an elastic
Fy
k a n a = ---------- (19) modulus of E c = 18,811 MPa for SL2. Using the suggested ACI
∆′y f
equation,17 which is shown in Eq. (20), results in an elastic
modulus of 23,887 MPa for SL2. The ACI equation was then re-
From Table 4(a), it is noted that the experimental stiffnesses
duced based on the cylinder data to that given by Eq. (21) and
for lightweight concrete columns SL1, SL2, FL1, FL2 are 61
the analysis performed again with the results shown in Table
percent to 92 percent of that for the analytical stiffness assuming
normal weight behavior (elastic modulus and strain at maxi- 4(b). The average ratio of experimental to analytical stiffness in-
mum compressive stress reduced as previously discussed) with creased to 91 percent indicating a better estimate of the light-
an average value of 80 percent. Note that for normal weight Unit weight concrete elastic flexural stiffness.

146 ACIStructuralJournal/January-February 1999


Table 4(a)—Elastic stiffness at first yield of test
columns—normal weight analytical model
∆′ y EXP ∆′y ANA Fy
flexural, Fy EXP, k exp, flexural, ANA, k ana ,
Test unit mm kN kN/m mm kN kN/m k exp/k ana
FL1 68.3 141 2064 57 139.7 2450 0.84
FL2 58.4 137 2346 56 142.7 2548 0.92
FL3: 54.5 138 2532 56 141.9 2534 1.00
NW
SL1 12.87 533.5 41,453 7.8 528 67,692 0.61
SL2 9.67 533.5 55,171 7.8 530 67,949 0.81

Table 4(b)—Elastic stiffness at first yield of test


columns—lightweight analytical model
∆′ y EXP ∆′ y ANA
flexural, F y EXP, k exp, flexural, Fy ANA, k ana ,
Test unit mm kN kN/m mm kN kN/m k exp/k ana
FL1 68.3 141 2064 58.7 130.6 2226 0.93
Fig. 15—Predicted force-displacement response for FL1 with and
FL2 58.4 137 2346 60 136.2 2266 1.04 without compression strength enhancement from comfinement. (1
FL3: kN = 225 lb; 1 mm = 0.039 in.)
N/A N/A N/A N/A N/A N/A N/A
NW
SL1 12.87 533.5 41,453 8.6 504 58,559 0.71
the theoretical maximum force using the revised concrete model
SL2 9.67 533.5 55,171 8.6 504 58,559 0.94 results in a value of 161 kN, which equals the experimentally re-
corded maximum average for the two directions of loading. For
1.5 3 Unit FL3, the theoretical maximum strength using the conven-
Ec = 0.0427 γ f ′c (MPa, kg/m ) (1MPa = 0.145 ksi) (20) tional normal weight concrete model was 187 kN, while the ex-
perimentally recorded maximum was 184 kN (average of two
1.5 3 directions of loading).
Ec = 0.0336 γ f ′c (MPa, kg/m ) (1MPa = 0.145 ksi) (21)

Displacement capacity
Based on the results, there are two possible approaches: 1) The Displacement capacity of a flexural member is governed by
elastic cracked section stiffness for lightweight concrete can be one of four criteria: 1) Buckling of the longitudinal reinforce-
reduced by 20 percent from that for normal weight concrete, or ment between layers of transverse reinforcement; 2) Buckling
2) The elastic concrete modulus can be calculated with Eq. (21) of longitudinal reinforcement over a series of transverse rein-
with reasonable accuracy. forcing bars; 3) Confinement failure due to crushing of the con-
crete/rupture of transverse steel; or 4) Rupture of longitudinal
Flexural strength steel. The majority of circular bridge columns generally fail by
In terms of a direct comparison in ultimate flexural strength, the second mechanism, thus never testing the true strain capac-
test results indicate that Unit FL1 achieved a flexural strength ity of the confined concrete.
approximately 12 percent less than Unit FL3. Since the strength For each of the test units discussed here, the failure mode con-
at maximum response is partly a function of the strength of the sisted of buckling of the longitudinal reinforcement. Although the
confined concrete, it is possible that the lower strength is a result ductility capacity of FL1 was lower than FL3, the displacement ca-
of a reduced effectiveness of confinement in increasing light- pacity was essentially the same. The difference in ductility capac-
weight concrete compression strength. Similar observations ity was due to the increased flexibility of the lightweight concrete
have been made by Shah et al. 9 test unit (FL1). It is also noted that the transverse reinforcement
The tests by Shah et al. concluded that the compressive strains were higher at a given inelastic displacement level for light-
strength enhancement for a heavily confined concrete cylinder weight concrete; however, the difference is minimal. It is also not-
(ρ s = 1.86 percent) constructed from normal weight concrete ed that the similar displacement capacity achieved by the two
was 44 percent greater than the unconfined compressive comparative test units agrees with the results from test performed
strength, while an identical lightweight cylinder achieved a con- by Shah et al.9 It is suggested that displacement capacity be based
fined compressive strength only 7 percent larger than the uncon- on the maximum concrete compression strain given by Eq. (16).12
fined strength. The confined concrete strength should be that used for normal
To determine if this can account for the difference in flexural weight concrete since experimental evidence does not indicate an
strength implied by Fig. 13(d), a section analysis 14 using the increase in ultimate concrete strain capacity which would be ob-
Mander energy balance approach15 was employed with the mod- tained if the confined concrete strength were set equal to the un-
ification that the compressive strength of confined concrete was confined concrete strength. It is noted that such a result raises
set equal to the unconfined strength. This effectively neglected questions regarding the energy balance approach, as a higher strain
the 7 percent enhancement suggested by Shah et al. The result of capacity would be expected in order to compensate for the reduced
the analysis in terms of the force-deformation response is shown confined concrete compressive strength.
in Fig. 15, along with the theoretical flexural response assuming
normal weight concrete behavior. Also shown is the envelope of Energy dissipation
response for the average of the two directions of loading for The area under the hysteretic loops is a measure of energy ab-
lightweight Unit FL1. In terms of expected maximum strengths, sorption, which is a good indicator of hysteretic damping. Re-

ACIStructuralJournal/January-February 1999 147


ACKNOWLEDGMENTS
The research described in this paper was funded by the California De-
partment of Transportation (Caltrans) under contract 59V-375. Their sup-
port is gratefully acknowledged. The assistance of Jose Fulginiti of Pacific
Custom Materials in aggregate selection and supply is also gratefully ac-
knowledged. We would also like to thank the entire staff of the Charles Lee
Powell Structures Laboratories for their assistance in the experimental stud-
ies. The conclusions presented are those of the authors alone and should not
be construed to imply endorsement by Caltrans.

REFERENCES
1. Harries, K. A., “Quake Resistance in Ancient Rome,” Concrete Inter-
national, V. 19, No. 1,. Jan. 1997, pp. 55-62.
2. Jennings, P. C., Engineering Features of the San Fernando Earth-
quake, California Institute of Technology, Pasadena, Calif., June 1971.
3. Paulay, T., and Priestley, M. J. N., Seismic Design of Reinforced Con-
crete and Masonry Buildings, Wiley-Interscience, New York, 1992.
Fig. 16—Comparision of hysteretic energy absorbtion for FL1 4. Priestley, M. J. N.; Seible, F.; and Calvi, G. M., Seismic Design and
and FL3. (1 mm = 0.039 in.) Retrofit of Bridge Structures, John Wiley and Sons, New York, 1995.
5. Kowalsky M. J.; Priestley, M. J. N.; and Seible, F., “Shear Behavior of
sults of such analysis are shown in damping versus ductility Lightweight Concrete Columns under Seismic Conditions,” Structural Sys-
curves for the test columns as shown in Fig. 16. From this fig- tems Research Project SSRP--95/10, Department of Applied Mechanics and
ure, it is noted that no appreciable difference exists between the Engineering Sciences, University of California, San Diego, Calif., July 1995.
concrete types. 6. Kowalsky, M. J.; Priestley, M. J. N.; and Seible, F., “Flexural Behavior
of Lightweight Concrete Columns Under Seismic Conditions,” Structural
Systems Research Project SSRP—96/08, Department of Applied Mechan-
CONCLUSIONS ics and Engineering Sciences, University of California, San Diego, Calif.,
Based on the results presented in this paper, it is concluded 1996.
that lightweight concrete can be employed for bridge structure 7. Monti, G., and Nuti, C., Cyclic Tests on Normal and Lightweight Con-
elements expected to undergo large inelastic deformation dur- crete Interior Beam-Column Subassemblages Designed to EC8 Code,
ing extreme earthquakes. The differences between normal Dipartimento di Ingegneria Strutturale e Geotecnica, Universita La Sapi-
enza, Roma, Italia, May 1993.
weight and lightweight concrete were presented in a qualitative
8. Bertero, V. V.; Popov E. P.; and Forzani, B., “Seismic Behavior of
manner through test observations, as well as in a quantitative as-
Lightweight Concrete Beam-Column Subassemblages,” ACI Structural
sessment of key design parameters, such as the strength of the Journal, V. 77, No. 6, Nov.-Dec. 1991.
concrete shear resisting mechanism, flexural stiffness, concrete 9. Shah, S. P.; Naaman, A. E.; and Moreno, J., “Effect of Confinement
crushing strain, flexural strength, displacement capacity, and on the Ductility of Lightweight Concrete,” The International Journal of
energy dissipation. Cement Composites and Lightweight Concrete, V. 5, No. 1, Feb. 1983.
It was concluded that the existing model for shear capacity 10. Priestley, M. J. N.; Verma, R.; and Xiao, Y., “Seismic Shear Strength
could be applied with the modification that the concrete compo- of Reinforced Concrete Columns,” Journal of the Structural Division,
ASCE, V. 120, No. 8, Aug. 1994.
nent be reduced by 15 percent in the nonductile region and 30
11. Kowalsky, M. J.; McDaniel, C. C.; Benzoni, G.; and Priestley, M. J. N.,
percent in the ductile region. In each case, the results agree rea-
“An Improved Analytical Model for Shear Strength of Circular RC Columns
sonably well with those suggested by the ACI code.17 It was in Seismic Regions,” ACI Spring Convention, Seattle, Wash., Apr. 5-10, 1997.
also noted that elastic modulus for lightweight concrete suggest- 12. Chai, Y. H.; Priestley, M. J. N.; and Seible, F., “Steel Jacketing of
ed by the ACI code17 provides a reasonable estimate although a Circular Reinforced Concrete Bridge Columns for Enhanced Flexural Per-
larger reduction may be appropriate. formance,” Structural Systems Research Project SSRP—91/06, University
Test results indicate that flexural strength be calculated with- of California, San Diego, Calif., Oct. 1991.
out any strength enhancement to the compressive strength from 13. Verma, R.; Priestley, M. J. N.; and Seible, F., “Assessment of Seismic
Response and Steel Jacket Retrofit of Squat Circular Reinforced Concrete
confinement reinforcement. This results in a reduction in ulti-
Bridge Columns,” Structural Systems Research Project SSRP—93/05
mate moment capacity that varies depending on the confine- Department of Applied Mechanics and Engineering Sciences, University of
ment level. However, flexural strength at yielding of the California, San Diego, Calif., June 1993.
reinforcement is not affected. 14. King, D. J.; Priestley, M. J. N.; and Park R., “Computer Programs for
The crushing strain for lightweight concrete was established Concrete Column Design,” Research Report 86/12, Department of Civil
as being as low as 0.0025, which is 50 percent of that usually as- Engineering, University of Canterbury, New Zealand, May 1986.
sumed for normal weight concrete and 70 percent of that mea- 15. Mander, J. B.; Priestley, M. J. N.; and Park R., “Theoretical
Stress-Strain Model for Confined Concrete,” Journal of the Structural
sured for the comparative normal weight test column. This will Division , ASCE , V. 114, No. 8, Aug. 1988.
clearly result in the onset of the service limit state at a much
16. Priestley, M. J. N.; Ranzo, G.; Benzoni, G.; and Kowalsky, M. J.,
lower level when lightweight concrete is employed. “Elastic Stiffness of Reinforced Concrete Bridge Columns,” Proceedings,
Test results also indicated that there was no appreciable dif- Fourth Caltrans Seismic Research Workshop, July 9-11, 1996, Sacramento,
ference in displacement capacity for lightweight concrete, par- Calif.
ticularly in the case where the maximum displacement is 17. “Building Code and Commentary (ACI 318-89R-92),” American
controlled by longitudinal bar buckling. No appreciable differ- Concrete Institute, Farmington Hills, Mich., 1992.
18. Priestley, M. J. N.; Seible, F.; Benzoni, G., “Seismic Performance of
ence was noted for energy absorption as well, which is as ex-
Circular Columns with Low Longitudinal Steel Ratios,” Structural Systems
pected since hysteretic damping is related more to steel Research Project SSRP—94/08, Department of Applied Mechanics and
reinforcing properties rather than concrete properties at high Engineering Sciences, University of California, San Diego, Calif. June
levels of inelastic response. 1994.

148 ACIStructuralJournal/January-February 1999


View publication stats

Das könnte Ihnen auch gefallen