Sei sulla pagina 1di 10

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 89-54

Consideration of Questions about Beam-Column Joints

by Stavroula Pantazopoulou and John Bonacci

The mechanics of beam-column joints in laterally loaded frame formance demand: should it be stated in terms of a
structures are investigated in this paper. The formulation presented ductility ratio? a drift ratio? or should joint design be
establishes compatibility of strains and stress equilibrium. Factors
made a slave to the forces associated with inexhaustible
known to affect the behavior of connections, including the effect of
lateral restraint (present in connections of indeterminate structures), beam hinging? There has been considerable discussion
as well as the reduction of concrete strength associated with diagonal about whether to design for horizontal shear, vertical
tensile strains, are considered. The role of stirrups and axial load on shear, or both. 9-u There are conflicting views about the
the behavior of the joint are illustrated clearly. In addition to provid- function of transverse reinforcement and whether or
ing an improved understanding of joint behavior, the derivation re-
not column axial force is beneficiaJ.9· 12- 15 There is little
flects the relationship between the design limit states and correspond-
ing deformations, giving an opportunity to link joint design to the known about how connections would fare in actual
overall lateral drift of the structure. (statically indeterminate) frames, for which it is proba-
ble that internal forces could be redistributed and hor-
Keywords: beams (supports); connections; cyclic loads; deformation; earth- izontal restraining forces would develop as a result of
quake-resistant structures; joints Gunctions); reinforced concrete; stirrups. inelastic member growth. Finally-but not to say that
it is the least troubling issue-there are several views
A measure of uncertainty exists about connections in about how bond softening influences connection be-
reinforced concrete frame structures designed to with- havior.
stand earthquake motions. There are several reasons This paper was inspired by the existence of this im-
why this has been such a stubborn problem. Rein- pressive list of uncertainties. It does not pretend to set-
forced concrete connection behavior involves interac- tle finally all or any of these issues, nor does it lead to
tion of several key phenomena (shear, bond, fatigue, infallible design recommendations. The paper begins by
confinement), which, even independently, are not well exploring the mechanics of connections for the sake of
understood. Further, the kinds of tests made to pro- improved understanding. It shows that a number of the
vide experimental evidence have imposed a specialized unresolved issues can be addressed satisfactorily by ap-
reference frame through which the problem is usually plying relatively familiar formal mechanical principles.
viewed. 1•8 Another issue is that investigation of this The mathematical constructs adopted to record me-
problem has been pursued on two different fronts. chanical arguments are then applied to investigate how
Some are concerned with describing mechanical truths. certain design variables influence the behavior of con-
Others want merely to develop dependable design rec- nections.
ommendations. It is likely that the level of uncertainty The discussion of mechanics is built around two im-
may have been amplified by debate that mixes the lan- portant assertions: 1) that many of the unsettled ques-
guage and objectives of these two separate pursuits. It tions result from selection of mechanically inconven-
is plausible that a structure can be designed adequately ient (though perhaps humanly convenient) coordinate
without full knowledge of every aspect of its inner systems for viewing the problem, and 2) that consider-
workings, just as it is plausible that the mechanics of a ation of joint deformation characteristics is essential if
system can be described in a way that would leave no performance requirements are to be related to struc-
impression on a designer. tural response or if the effects of indeterminacy are to
Uncertainty or differences of opinion still pervade
almost every aspect of connection behavior and de-
sign.9-u To start off with, reconnaissance studies have ACI Structural Journal, V. 89, No. I, January-February 1992.
Received Jan. 15 1991 and reviewed under Institute publication policies.
not always shown that joints make actual structures as Copyright © 1992, Ameri~an Concrete Institute. All rights reserved, includin~
vulnerable as laboratory studies might suggest. There is the making of copies unless permission is obtained from the copyright propn-
etors. Pertinent discussion will be published in the November-December 1992
also much debate over what constitutes reasonable per- ACI Structural Journal if received by July I, 1992. ·

ACI Structural Journal I January-February 1992 27


this time on the fronts of experimental research and
Stavroula Pantazopoulou is assistant professor of civil engineering at the Uni-
versity of Toronto. She holds a degree in civil engineering from the National design, many new unanswered questions have also
Technical University of Athens, Greece, and MS and PhD degrees from the emerged.
University of California at Berkeley. She is secretary of ACI Committee 341, In addition to p1aterial nonlinearity, understanding
Earthquake-Resistant Concrete Bridges, and is a member of ACI Committee
368, Earthquake Resisting Concrete Structural Elements and Systems. of joints is made difficult because of the fact that large
variations in both geometry and distribution of loads
occur over a relatively small volume. This feature com-
John Bonacci is assistant professor of civil engineering at the University of To-
ronto. He received his BS, MS, and PhD degrees in civil engineering at the plicates the task of idealization for the purpose of
University of Illinois-Urbana. He is a member of joint ACI-ASCE Committee analysis.
352, Joints and Connections in Monolithic Concrete Structures; and A CI
Much of the current knowledge about joint behavior
Committee 408, Bond and Development of Reinforcement.
is based on experiments which, for practical reasons,
are usually done on statically determinate assemblies
be accounted for. If the goal of this paper were to representing frame connections. 1-8 Ironically, though the
achieve pinpoint accuracy, the influence of bond slip form of these specimens has been greatly simplified,
would have to be considered in detail. But rigorous interpretation of test results is very demanding and is
treatment of bond was purposely omitted because it subject to considerable numerical scatter. This is espe-
adds a unique dimension that might obscure readily cially true of efforts to establish design limits for joint
understandable results. shear strength, for which a lower-bound empirical ap-
proach has been taken. In this section, these issues of
experimental idealization and scatter are reviewed,
RESEARCH SIGNIFICANCE along with several others that impede consistent inter-
Current design code requirements for beam-column pretation of experimental evidence. The arguments set
connections rely on empirical interpretation of special- forth here are meant to provide a list of specific fea-
ized experiments. In this paper, a description of the ba- tures to be incorporated into the interpretation of joint
sic mechanics of joint behavior is used to provide an mechanics that follows.
alternative approach for generalizing from available
experimental evidence. The derivation establishes equi-
librium of stress resultants, satisfies compatibility of Choosing a coordinate system
deformations within the joint, and incorporates avail- Experimental behavior of joints is commonly char-
able models of material nonlinear behavior. This com- acterized by an average shear stress (horizontal and/ or
prehensive interpretation of joint mechanics constitutes vertical) introduced to the joint by the adjacent beams
a significant step toward development of unified gen- and columns (Fig. 1). The same approach is adopted by
eral design standards for beam-column connections re- joint ACI-ASCE Committee 3529 and is characteristic
sisting earthquake forces. of the North American design practice.
However, the value of shear stress is associated with
PRESCRIPTIONS FOR DESCRIBING JOINT the specific coordinate system in which it is evaluated.
MECHANICS By appropriate choice of the coordinate system, shear
Understanding the behavior of reinforced concrete stress can be eliminated, and therefore it alone cannot
beam-column joints has become a topic of particular adequately describe the complete stress state of the ele-
challenge to researchers throughout the world for the ment under consideration. While it is admittedly most
last 30 years. While much progress has been made in practical to compute shear stresses along the main di-
rections of connection geometry from the forces acting

....
.<
_l\
- at the boundaries of the joint (Fig. 1), quantifying this
aspect of behavior without the associated angle from
the plane of principal stresses within the joint is incom-
column plete. This is undoubtedly a factor that contributes to
joint scatter observed in summaries of experimentally meas-
-~ ured joint shear strength.

~ ./ >
At any point in a structure, it is possible to deter-
mine the principal directions for which no shear is re-
quired to describe the state of stress and for which all
stresses are normal, in either tension or compression.
Because these stresses can be compared directly with
conventionally obtained material strengths, it would

longitudinal bearn
~ .II
seem that the corresponding coordinate system is the
least arbitrary of the possible choices.
Several attempts have been made to approximate the

- ...
Fig. 1 - Actions at the boundaries of the joint
..,. directions of principal stress in beam-column joints us-
ing "truss" models, wherein the reference frame is ide-
ally related to the path of loads within the structure. 12•15
28 ACI Structural Journal I January-February 1992
Typically, this path is established thro1,1gh engineering uniformly distributed longitudinal column reinforce-
judgment and experimental observation, leaving room ment (to form a closed cage), significantly enhanced the
for uncertainty about whether the selected reference shear resistance of joints. 1 Since then, much research
frame is associated with zero shear. has been con~ucted with an aim toward establishing re-
lationships between the degree of deterioration of shear
resistance under cyclic loads and the amount of lateral
Consideration of joint deformations
reinforcement provided in the joint. 2•6 Recent tests (as
The idea of evaluating the behavior of joints by ex-
part of a U.S.-Japan-New Zealand-China cooperative
amining the principal stresses is not novel. However, its
research effort) of connection specimens simulating
realization has been impeded by the fact that even
Japanese, American, and New Zealand practice (of
computation of average principal stresses requires
which the first and last pose the lowest and highest re-
knowledge of the pattern and magnitude of deforma-
quirements of joint reinforcement, respectively), clearly
tion, an aspect of the response usually omitted from
suggest that there is an upper limit in the amount of
design methods because, when accounted for, it gener-
joint reinforcement, beyond which the overall resis-
ally complicates the extent of computations required
tance of beam-column assemblies is not improved. 7
for a solution. Nevertheless, there is a slowly converg-
This manifests the obvious fact that, although large
ing consensus that the nature and size of joint defor-
amounts of joint reinforcement can increase the avail-
mations associated with a given level of forces must be
able shear resistance of the joint itself, this additional
better understood. Even if, by some means, the direc-
strength is not likely to be usable because the dema~d
tion and magnitude of principal stresses could be deter- on the joint is eventually limited by the flexural resis-
mined, it would be difficult to know if these stresses tance of adjacent members. All reported specimens
were too high or whether they are even likely to de-
(which were designed to fail in beam yielding and had
velop considering the response of the overall structure. favorable bond conditions) demonstrated similar over-
For example, the compressive resistance of concrete all resistance, with the only significant differences be-
along the principal diagonal depends on the amount of
ing apparent stiffness of the connections and the
tension in the orthogonal direction. 16 Such effects can amount of deformation. Increased amounts of lateral
only be addressed quantitatively if information about reinforcement delayed the initiation and propagation of
the kinematics of joint response is available. As the cracking in the joint, and reduced the amount of joint
forces at the joint boundary increase, several possible distortion that occurred under a given level of joint
milestones of response can develop, such as yielding of shear stress. These observations suggest that an impor-
transverse reinforcement, crushing the concrete along tant consequence of adding joint reinforcement is an
the diagonal, or yielding of the column reinforcement.
increase of joint stiffness.
If the deformations throughout the progress of re-
sponse could be "monitored," it would be possible to Based on this experimental evidence, it is possible to
establish the sequence in which these phenomena would idealize joints as two-dimensional (2-D) panels rein-
occur. Furthermore, if the deformations of the joint forced in two orthogonal directions, and acted upon by
could be linked to the overall structural response, then, in-plane shear and normal stresses. The total shear
in addition, the very likelihood of these occurrences stiffness K of the panel comprises concrete Kc and steel
could be established. K contributions according to: K = Kc + Ks. However,
With respect to the overall displacements of beam- c~ntrary to familiar 2-D panels, 16 the concrete stiffness
column connections, experiments show that joint de- contribution Kc is not independent of Ks. This is sup-
formation is a strong indicator of joint performance. ported by experimental evidence obtained from tests of
For example, it is generally understood that satisfac- beam-column joints in which the closed stirrups were
tory joint behavior is associated with minimal contri- replaced by longitudinal beam reinforcement that was
bution of joint distortion to the overall lateral drift of uniformly distributed along the height and anchored
the structure. 7 Ideally, the fraction of joint contribu- outside of the joint. 8 Although the joints in these spec-
tion would decrease with increasing amount of total imens were able to resist the shear demanded to de-
lateral drift as the adjacent beams develop plastic velop beam hinging, a rapid deterioration of joint shear
hinges and undertake the bulk of inelastic action. By resistance was observed with cycling. Evidently, the
contrast, for cases of inadequate joint design, the con- stiffening action of closed stirrups occurs not only in
tribution of joint distortion to the total drift has been the direction of the load (as is represented by the value
observed to increase with increasing magnitude of total of Ks) but in the perpendicular direction as well, mak-
displacement. In these cases, the joint is often respon- ing the confining role of stirrups a more transparent
sible for much of the inelasticity detected in the overall phenomenon. Therefore, the experimental evidence
response. suggests that stirrups contribute to the shear resistance
of joints directly (by resisting part of the joint shear)
and indirectly (by confining the concrete core, thus en-
Role of stirrups hancing its diagonal compressive strength). However,
Early tests conducted in the 1960s on isolated con- these two functions are not independent or mutually
nection specimens illustrated that joint reinforcement in exclusive of each other - a point that currently fuels
the form of horizontal stirrups, in combination with debate between differing design philosophies. 13 ' 14
ACI Structural Journal I January-February 1992 29
the confining effect of transverse beams on the joint is
likely to be significant in actual (indeterminate) struc-
tures, where each connection is restrained by the pres-
ence of adjacent fr~mes.
longitudinal beam

BASIC MECHANICS OF REINFORCED


CONCRETE JOINTS
Definition of the problem
The typical loading system considered in analysis and
design of interior beam-column joints is illustrated in
Fig. 1. In most contemporary design codes, the magni-
tudes of these forces are associated with a beam flex-
ural hinging mechanism, implying that beams and col-
transverse beam umns are dimensioned first. Because a large portion of
the forces loading the joint are introduced by bond
stresses that develop between concrete and reinforce-
Fig. 2 - Internal axial forces at the connection
ment, codes require that the magnitude of bond stresses
be regulated by controlling the size of longitudinal bar
(a)
diameter with respect to the available development
length (column and/or beam depth). In the following
t {b)
derivation, it will be assumed that the development

tJl~1I1
length requirements are satisfied a priori, and that bond
deterioration is not significant.
Furthermore, it is assumed that the joint is properly
detailed and that reinforcement is present in quantities
~ t sufficient to provide adequate crack control. Stresses
ldw(l + €~)~ and strains are averaged 19 over the dimensions of the
entire joint. The joint is idealized as a three-dimen-
Fig. 3 -Kinematics of joints: (a) joint geometry; (b) sional solid, with dimensions dw, b, h [Fig. 3(a)]. Lat-
definition of average strains eral loads are considered parallel to one principal di-
rection, identified by axis t in the figure.
Role of transverse beams and static redundancy
From tests of interior beam-column joints with Kinematics
transverse beams, it has been established that even After deformation, the overall geometry of the joint
joints without any stirrup reinforcement can perform is described by the average angle of shear distortion 'Y
satisfactorily within realistic levels of lateral displace- and by the average longitudinal and transverse strains,
ment. 7 This suggests that transverse beams at interior denoted by £:1 and £: 1, respectively [Fig. 3(b)]. The aver-
connections and closed-hoop reinforcement affect the age tensor in the (t,/,3) system is
behavior of joints in a similar manner by restraining
volumetric expansion, which eventually leads to deteri-
oration of joint shear resistance. It has been suggested
that this is the result of insufficient modeling of f: = (
f:,
0.5"(
0.5"(
f:1
0)
0 (1)
boundary conditions in the experiments, since the en- 0 0 f:J

hancement of joint performance was observed only


when transverse beams were free of load during the where £: 3 represents the hoop strain in the direction per-
testsY Indeed, experiments in which transverse beams pendicular to the loading plane. Some useful relation-
were loaded have been carried out and, in these cases, ships between the entries of the tensor expressed in var-
transverse beams had negligible confining contribu- ious coordinate systems are
tion. 3 Nevertheless, tests conducted on indeterminate
specimens have shown that the excessive deformation (2a)
that would occur within the beam plastic hinge regions
and the joint if the assembly were statically determi-
nate is partially restrained by the presence of adjacent 'V
,
-- 2(€1 - €,) -- 2("'1
" -
) tan a
E1 (2b)
tan a
members. This restraint has been measured experimen-
tally as internal axial forces that developed in beams
€1 -f:, €2 - f:,
experiencing inelastic deformation near the connec- t a2 n a = - - = - - (2c)
tion.17·18 The internal forces, N 3 and Nh (Fig. 2), repre- €1- f:l f:2- f:,

sent the reactions of adjacent columns to the lateral


displacement required to accommodate beam expan- where a is the direction of p~incipal strains in the joint
sion. Because of the presence of these internal actions, and £: 1 and £: 2 represent the values of principal strains.
30 ACI Structural Journal I January-February 1992
Eq. (1) describes the deformation component of the
joint kinematics. If the bond conditions are favorable,
these deformations constitute the major contribution of
the joint to the overall displacements of the structure.
If, on the other hand, bond deterioration in the joint is
significant, then, in addition to the deformation com-
ponents described by Eq. (1), a pure "rigid-body" ro-
tation term must be considered to establish the total
contribution of the joint to the overall structural dis-
J[J
___ 1----l.
l
1'
v

!t
a,

placement. I!Utttttttttflttlt
(a)
az !
---
(b)

Equilibrium requirements
Average stresses in the joint are depicted in Fig. 4. Fig. 4- Stress equilibrium: (a) equilibrium of vertical
Shear stresses are introduced by direct member action forces; (b) equilibrium of horizontal forces
and by bond that develops between the main reinforce-
ment and the joint core concrete. For simplicity, the
shear stress v is assumed to be uniformly distributed
over the boundaries of the joint [Fig. 4(a)].
Equilibrium of vertical forces [Fig. 4(a)] requires that idealized strains
the average longitudinal compressive stress in the con-
crete a1 satisfy the following relationship
actual strains

(3)

where }; is the average stress of the longitudinal rein-


forcement of the column at the midheight of the joint, t
p 1 is the available reinforcement ratio, dw is the depth of longitudinal beam
the column, and Nv is the compressive axial load of the reinforcement
column.
Similarly, using equilibrium of forces in the trans- Fig. 5- Actual and idealized strains
verse direction [Fig. 4(b)], the average transverse com-
pressive stress in the concrete (here horizontal) a, is of the joint may be as high as the value at the face of
evaluated the support. To reflect this difference in the formula-
tion and to limit the number of variables of the prob-
lem, Eq. (4) is simplified as follows
(4)

(5)
where Ps and Pb are the area ratios of horizontal stir-
rups in the joint and total longitudinal beam reinforce-
ment, respecitvely. Nh is the beam axial force, which where}; = J,; and p, = Ps + f3Pb· The factor (3 relates
could develop, for instance, from partial restraint to the magnitude of stresses in the longitudinal beam re-
beam expansion provided by adjacent columns in inde- inforcement to the average stirrup stresses at the col-
terminate frames. The terms j;s and j;b represent the av- umn centerline (Fig. 5). For perfect bond, (3 is zero. For
erage stress in the stirrups and the beam reinforcement, negligible bond resistance, (3 = 1. The actual condi-
respectively. If the bond conditions along the longitu- tions of a joint are represented by a value of (3 between
dinal beam reinforcement are favorable, then stresses these two limits. In the remainder of the study, (3 will
and strains in this reinforcement attenuate sharply from be considered an independent parameter, although it
the face of the support to the column centerline. has been assumed that bond conditions are generally
Therefore, although cracking may develop at the mid- favorable.
height of the joint as a result of shear, the concrete in An additional equation may be obtained by consid-
the vicinity of the longitudinal beam bars is practically ering equilibrium of stresses in the out-of-plane [or 3,
uncracked. Thus, the actual profile of lateral reinforce- as named in Eq. (6)] direction. If the joint reinforce-
ment strains or stresses over the height of the joint has ment is of the closed-hoop type, then the concrete of
the shape shown by the solid line in Fig. 5, although in the joint is subjected to passive confining stress a3 =
this study it is idealized by a single average value. On - psf3 • Here, f 3 represents the hoop stress in a direction
the other hand, if the bond resistance has deteriorated, normal to the line of action of the applied shear force.
then a significant amount of cracking is likely to occur Based on the preceding definitions of average joint
in the vicinity of the bars thro\lghout the depth of the stresses, the average stress tensor associated with the
column, and in such a case the bar stress at the center joint concrete is
ACI Structural Journal I January-February 1992 31
0 = a, it is possible to compute the angle of principal
stresses 0 from the entries of the two tensors just indi-
cated. For this purpose, the strains in Eq. (2c) are writ-
ten in terms of corresponding
., stresses as follows
confined concrete
tan2 0 = tan2 a = e2 - e1
E2 - e, (9)
= (~ - };) (~ - };)_,
Ec(e) E, Ec(e) E,

The stress components};, j,, and u2 in Eq. (9) are


written in terms of the average joint shear stress v us-
unconfined concrete ing Eq. (3), (5), and (8); vis then replaced by (- u,!tan
0) = [p,j, + (Nhlbh)ltan OJ.
This procedure leads to a quadratic polynomial of
tan 0
Fig. 6 - Transition of stress-strain curve of concrete

n~Jn:,+ r)] tan 0


after yielding of hoops
I + ;;;. -
(1 =(v
u, v
(11
0)
0 (6)
[
1 +-
np1
4

(10)
0 0 (13

+[ eJe, ] tan2 0- 1 =0
(1 + npl)(np1 + r)
The maximum principal stress associated with the stress
tensor given by Eq. (6) cannot exceed the tensile capac-
ity of concrete. If, for the sake of simplicity, it is as- where n = n(e) = E/E, the ratio of material moduli
sumed that this capacity is negligible, then, since nor- (a function of the state of strain under consideration).
mal stresses are either negative (compressive) or zero in The strain ratio r ( = ehle,) reflects the amount of lat-
any plane, it is evident that eral restraint to joint growth, which is likely to be sig-
nificant for indeterminate structures. It is evident from
u, = 0 and u2 = u1 + u, (first stress invariant) (7a) Eq. (10) that such restraint plays the same role algebra-
ically as horizontal joint reinforcement, which parallels
u1 - u, = v tan 0, and since u1 experimental observation of improved joint perform-
ance when transverse beams were present in test of in-
= 0, u, = -v tan 0 (7b)
terior connections. The quantities ev = NJ Ec(e)bdw and
eh = Nh!Ec(e)hb have units of strain, and represent the
and
deformations occurring in the joint under purely axial
v v forces.
u, - ul = --,
tan 0
or u1 =
tan 0
(7c)
Behavior before yielding of joint reinforcement
Reported experimental results indicate that joint per-
where u, and u2 are the principal stresses of concrete,
formance deteriorates significantly after the initiation
and 0 is the angle of inclination (from the t-axis) of the
of yielding in the joint hoops. For example, it is ex-
maximum principal stress u1• Upon substitution of Eq.
pected that the confinement available for the joint con-
(7b) and (7c) in Eq. (7a), the nonzero principal stress of
crete, which results from the restraint of volumetric ex-
concrete is expressed in terms of the average joint shear
pansion provided by the stirrups, will become ineffec-
stress v
tive when the stirrups yield. Fig. 6 illustrates the
transition that occurs from the stress-strain curve cor-
u2 = - v(tan 0 + -tan1-)
0
(8) responding to confined concrete action to that corre-
sponding to the state of confinement after yielding of
the hoops. Therefore, from a practical point of view, it
Eq. (1) through (8) describe fully the state of strain is important to establish the state of stress and strain in
and stress in the joint in an average sense. However, the the joint prior to hoop yielding.
directions of principal strain and stress (a and 0) that If the lateral reinforcement is within the elastic range,
enter the preceding equations are in general unknown. then the magnitude of joint shear stress is related to
Considering the behavior before yielding of hoops in hoop strain as follows
the joint (of primary interest from the design point of
view), it is reasonable to assume that, if the reinforce-
ment has not yielded, the direction pf principal strains e, = j, = - 1- (v tan 0 - Nh) (lla)
a is closely related to that of stresses 0. Assuming that E, p, E, bh

32 ACI Structural Journal I January-February 1992


and therefore

(11b)

In a similar manner, expressions for the remaining


elements of the strain tensor [Eq. (1)] may be obtained
From the equations, it is evident that after yielding of
hoops a dramatic increase occurs in the values of u1, u2 ,
(12a) and e1 for small increases in the value of joint shear,
since all quantities except v in the preceding expres-
e, - e, tan' 0 sions remain constant after hoop yield.
e, = I - tan' 0 Hoop yielding is likely to be succeeded by either a)
I [ p, - p, N,tan' N• 0] yielding of the longitudinal column reinforcement, or
=
E, (I - tan' 0)
vtan O---- + -
p1 p,
-
bhp, bd.p,
(12b) b) crushing of concrete in the direction of principal
compressive stress. Upper limits to the shear capacity
Et- e1 tan2 0 associated with these two mechanisms can be estab-
e2 = 1 - tan2 0 (12c)
lished. For the first case, the stress in the longitudinal
steel reaches the yield stress J;,. Thus
2(e 1 - e,)
'Y=--
, tan 0 N v'
a, = - p, /, - = -
___!_ therefore
p,j, + N.l bh
= E,(l -
2 [v
tan' 0) p, p, (p, tan
2
°- p,)
(N, N•) °]
+ bd.p, - bhp1 tan (12d)
bd.

(15)
Eq. (12b) indicates that the principal tensile strain in-
creases with increasing vertical axial load as well as For the second case, failure occurs when the principal
shear stress, while the effect of lateral restraint on the compressive stress u2 reaches the crushing strength of
joint is reverse. This effect is significant because the concrete, fr:=. This crushing strength, however, de-
magnitude of diagonal (principal) compression that can pends upon the amount of restraint to volumetric ex-
develop in the compressive strut of the concrete de- pansion, which here is represented by stress u3 • Fur-
creases with increasing magnitude of the tensile strain thermore, f~ also depends on the amount of tensile
in the normal direction. deformation in the perpendicular direction, 16 character-
Eq. (12d) provides a relationship between average ized by e1• It is assumed here that the relationship be-
joint shear stress and the amount of associated joint tween stress and strain along the principal compressive
distortion. It is evident from the precediQg that the col- direction can be described by
umn and beam axial loads have opposite effects on the
joint distortion at a given level of shear stress. a2 =/7"' [ 2E2 - -
Enua
E2 ) ' ] where
( -
fnurx
t/m"' --::, A/:]
EnutX XE
0
'

Behavior after yielding of joint reinforcement and A = _ __:K.:..___ (16)


0.8 - 0.34e 1/e0
As hoops yield, the pattern of deformations in the
joint is likely to change dramatically. In terms of
stresses, it is evident from Eq. (5) that where K = 1 + p,!J;,If:! (note that/: and e0 carry neg-
ative signs for calculations). Upon substitution of Eq.
(16) in Eq. (14b), the following alternative expression
for the limiting joint shear stress is established

which can be solved for the angle of principal stresses


tan 0: (tan 0 = [p1 .f;, + Nh/bh]lv).
This result can be used to obtain expressions for the NUMERICAL CASE STUDY OF JOINT BEHAVIOR
average longitudinal (vertical) stress [from Eq. (3)], the AFTER HOOP YIELD
average nonzero principal stress [from Eq. (8)], and the In the preceding sections, the mechanics of interior
amount of hoop strain, e1 [from Eq. (2c)] in terms of connections were studied in accordance with the pre-
the joint shear v scriptions set forth at the beginning of the paper. Phys-
ical suppositions have been recorded in the form of al-
gebraic expressions [Eq. (1) through (17), which con-
(14a) tain many of the design variables believed to influence
joint behavior. In this section, the equations for joint
shear strength after hoop yield [Eq. (15) and (17)] are
Nh v2 applied to two example problems (Table 1) to illustrate
u2 = -Pth - bh - Pth + Nh/bh (1 4b) their sensitivity to several design parameters.
ACI Structural Journal I January-February 1992 33
30.r---~--~----~---T----~--~ Table 1 - Summary of design parameters for
sample joints
N, N.
t: !,, J:bdw J:bh
"'
Case ,.p, P• p,

A 5 60 0.04 60 0.01 varies 0.1 0.05


10. Yield (Eqn. 15) B 7.5 60 0.025 60 0.01 varies 0.1 0.05
Crushing (Eqn. 17)
Units ksi ksi - ksi · - - - -
ACI (interior)
Note: 1 ksi = 6.9 MPa.
0.~--~--~~--~--~~--~--~
0.00 0.01 0.02 0.03
Ratio of hoops, Ps
(a) Discussion of numerical results
The expressions for limiting joint shear stress [Eq.
30.r---~--~----~--~--~~---, (15) and (17)] were applied to example joints A and B
with varied amounts of hoop reinforcement. The re-
sulting shear resistance normalized by .JJ[ for each
equation is shown in Fig. 7, along with the value of
20.JJ[ recommended by ACI-ASCE Committee 352 for
a similar connection type. 9 The equations suggest that
10.
the strength of Joint A [Fig. 7(a)] is limited by crushing
along the principal diagonal after hoops yield, while ..for
0.~--~~~~--~--~~--~--~
Joint B [Fig. 7(b)] crushing is somewhat less likely than
~oo ~m ~~ ~ro yielding of vertical reinforcement. This shifting of fail-
Ratio of hoops, Ps ure modes was achieved by increasing J: and decreasing
(b) p1 for Joint B. It can be observed that, as the likelihood
Fig. 7- Design shear strength: (a) Case A; (b) Case B of crushing is diminished, a larger quantity of hoops
can be made to yield. Thus, as more hoops are added
to Joint B, it can be made to carry a larger force than
Example connections was possible for Joint A.
Eq. (15) and (17) were applied to two sample interior Fig. 7 suggests that the supply of joint shear resis-
connections summarized in Table 1. For both exam- tance increases steadily with increasing amounts of
ples, {3 was taken as 0.5, which implies (but not by def- hoop reinforcement. This may be difficult to observe
inition) that stress in the beam reinforcement at the from a series of experiments on beam-column assem-
column center would be half as much as at the critical blies if the beams or columns yield, because the supply
section for bending when the joint reaches its limiting is hidden by the limited demand.
resistance. In keeping with the main disclaimer of this The preceding discussion is not intended to be a
paper, this would require unusually good bond condi- nomination of new design equations or an affirmation
tions. Absolute dimensions of the joint need not be of existing recommendations. Rather, the examples are
specified, as they serve only to normalize steel quanti- presented to show how the formulation developed in
ties (as in Pt> Pb• and p,), axial loads (as in Nvlbdw and this paper provides a device to investigate the sensitiv-
Nh/bh), and internal forces. . ity of joint behavior to various design parameters in a
Examples A and B differ only in the ratio of vertical way that violates neither physical intuition nor past ob-
reinforcement and compressive strength of concrete. servation.
For each example, the ratio of hoop reinforcement was
'varied from zero up to a value at which Eq. (15) and CONCLUSIONS AND LIMITATIONS
(17) no longer apply (crushing occurs before hoops Through persistent debate and careful scrutiny of ex-
yield). periments over the past several decades, recommenda-
tions for reliable design of beam-column joints have
Solution procedure evolved. This study was an effort to expand the under-
Unlike Eq. (15), which depends only on given con- standing of the mechanics of joint response with the
stants, Eq. (17) requires determination of the magni- objectives of improving confidence in design ap-
tude and direction of principal stresses and strain to proaches, expanding the ability to assess joint perform-
evaluate the term f~. Because the stress-strain rela- ance and demand, and providing a vehicle to sort out
tionship assumed for concrete [Eq. (16)] is nonlinear, the influence of various design parameters. The ap-
iteration was required to determine the appropriate proach taken was to compile a list of recognizable
value of Ec. Though the secant-modulus iteration shortfalls in existing interpretations of joint behavior
scheme used was quite simple, it is worthwhile to note and then to use this list as specifications for develop-
that very similar results can be obtained without itera- ment of an improved interpretation. The result is prob-
tion by choosing Ec to be the seca,:tt modulus at peak ably not, nor was it intended to be, an answer to all or
stress, which is insensitive to the factor A in Eq. (16). any of the nagging questions about beam-.column
34 ACI Structural Journal I January-February 1992
joints. The result is a collection of algebraic expres- 6. It is important to remember that the size and pat-
sions the record physical arguments. It is apparent that tern of forces that will act on a joint depend on the na-
these expressions manage to link recognized design var- ture of bond for beam and column longitudinal rein-
iables to familiar indexes of joint behavior and design forcement along the joint boundaries. While some at-
in a manner that is physically sensible. tempt was made " here to account for the effectiveness of
It is shown that the shear strength of a joint depends bond, further development is required to account for
on the usable compressive strength of concrete, which the total rotation attributable to connections, which is
decreases with increasing principal tensile strain e1• Eq. a sum of joint deformation and a rigid-body rotation
(12b) indicates that e1 increases (and so it follows that that becomes prevalent when bond is poor or nonexist-
joint shear strength would decrease) with increasing ent.
column axial force Nv. Axial compression in the beam 7. The ability to use the formulation presented here
(Nh), on the other hand, would reduce e1• Similar ob- to study response of joints under reversed cyclic load-
servations about joint shear distortion are apparent ing depends only on the development of a concrete
from the algebra of Eq. (12d). Nv and v tend to encour- stress-strain relationship that can account for strain
age shear distortion, while Nh tends to work against it. softening and confinement under load reversals.
Additional specific conclusions and limitations perti-
nent to the development and interpretation of the for-
mulation presented in this study are: ACKNOWLEDGMENTS
1. To describe joint forces in a concise manner, it is The work presented in this paper was carried out at the University
essential to consider both the kinematics and equilib- of Toronto, Canada. Financial support for the study was provided by
the Natural Sciences and Engineering Research Council of Canada
rium associated with joint response. (NSERC Grants No. OGP0042033 and OGP0042154.
2. It was not necessary to declare whether joint
hoops serve to confine or to carry shear to include them
in the formulation. When hoops are closed around the
NOTATION
joint core, their role is sufficiently described from con- N. = horizontal axial load acting in the main beam of the connec-
sideration of equilibrium and deformations. If j9int re- tion (positive for compression) ·
inforcement is not closed, compatibility with core con- N, = column axial load (positive for compression)
crete will depend on bond, which may explain why the N, = horizontal axial load acting in the transverse beam of the con-
nection (positive for compression)
benefit from intermediate beam bars diminishes quickly }; average stress in the longitudinal column reinforcement (ten-
under cyclic loading. sion positive)
3. By determining principal stresses and strain, the }; average stress in the transverse reinforcement through the joint
strength of a connection can be evaluated from the (tension positive)
compressive stress-strain behavior of concrete. It is v average joint shear stress
a = angle of the first principal strain direction from the longitudi-
possible to do this in a manner that accounts for the nal axis t
effects of softening from transverse tensile strain and /1 ratio of average beam reinforcement stress to average stirrup
for confinement. The brief case studies suggest that the stress at the column centerline
results of this formulation are not at variance with the 'Y average shear distortion of the joint panel
E, = average strain along the direction I of the joint panel (tension
empirical basis of current design recommendations.
positive)
4. By establishing a relationship between loads ap- E, average strain along the direction t of the joint panel (tension
plied to the joint and the resulting deformations, the positive)
formulation makes it possible to consider the mode of () angle of the first principal stress direction from the longitudi-
joint failure in addition to the associated strength. Be- nal t-axis
sides bond failure, joint capacity can be limited by A coefficient to account for the effects of softening and confine-
ment on concrete compressive strength
crushing along the principal diagonal (either before or p, column longitudinal reinforcement percentage
after hoop yield), or by the yield of vertical reinforce- p, percent of total horizontal reinforcement in the direction of
ment after hoop yield. The need to be able to discern shear action through the joint (im;luding beam reinforcement)
failure modes is underscored by the importance of hoop p, percent of stirrup reinforcement in the direction of shear
yield as a milestone in response. As illustrated by com- through the joint
p6 percent of main beam reinforcement through the joint
paring Eq. (llb) and (14c), the sensitivity of joint be- q, average normal concrete stress in the /-direction (tension posi-
havior to variations of the stress state changes mark- tive)
edly after hoops yield. q, average normal concrete stress in the t-direction (tension pos-
5. With the formulation presented, it is possible that itive)
joint shear deformations can be linked to the displace-
ments of a structure so that the demand level of joint
design in a particular system can be considered. Though REFERENCES
such an assessment of demand would be no more ac- 1. Hanson, N. W., and Conner, H. W., "Seismic Resistance of
curate than the estimate of displacement response, the Reinforced Concrete Beam-Column Joints," Journal of the Struc-
tural Division, ASCE, V. 93, ST5, Oct. 1967, pp. 533-560.
ability to consider it adds a layer of perspective that 2. Meinheit, D. F., and Jirsa, J. 0., "Shear Strength of R.C.
would be useful to developmeqt of joint design philos- Bea!ll-Column Connections," Journal of the Structural Division,
ophy. ASCE, V. 107, STll, Nov. 1981, pp. 2227-2244.

ACI Structural Journal I January-February 1992 35


3. Uzumeri, S. M., and Seckin, M., "Behavior of Reinforced 11. CEB-FIP Model Code for Seismic Design of Concrete Struc-
Concrete Beam-Column Joints Subjected to Slow Load Reversals," tures," Comite Euro-lnternational du Beton, Bulletin d'Information
Publication No. 74-05, Department of Civil Engineering, University No. 160, Paris, 1983, 117 pp.
of Toronto, Mar. 1984, 84 pp. 12. Paulay, T.; Park, R.; and Priestley, M. J. N., "Reinforced
4. Durrani, A. J ., and Wight, J. K., "Behavior of Interior Beam- Concrete Beam-Column ~oints Under Seismic Actions," ACI JouR-
to-Column Connections Under Earthquake Type Loading," ACI NAL, Proceedings V. 75, No. 11, Nov. 1978, pp. 585-593.
JOURNAL, Proceedings V. 82, No. 3, May-June 1985, pp. 343-350. 13. Paulay, T., "Critique of the Special Provisions for Seismic
5. Ehsani, M. R., and Wight, J. K., "Effect of Transverse Beams Design of the Building Code Requirements for Reinforced Concrete
and Slab on Reinforced Concrete Beam-to-Column Connections," (ACI 318-83)," ACI JOURNAL, Proceedings V. 83, No.2, Mar.-Apr.
ACI JouRNAL, Proceedings V. 82, No.2, Mar.-Apr. 1985, pp. 188- 1986, pp. 274-283.
195. 14. Paulay, T., "Equilibrium Criteria for Reinforced Concrete
6. Joglekar, M.; Murry, P.; Jirsa, J. 0.; and Klinger, R., "Full Beam-Column Joints," ACI Structural Journal, V. 86, No.6, Nov.-
Scale Tests of Beam-Column Joints," Earthquake Effects on Rein- Dec. 1989, pp. 635-643.
forced Concrete Structures-U.S.-Japan Research, SP 84, American
15. Schlaich, J.; Schafer, K.; and Jennewein, M., "Towards a
Concrete Institute, Detroit, 1984, pp. 271-304.
Consistent Design of Reinforced Concrete Structures," PC/ Journal,
7. Kurose, Y ., "Recent Studies on Reinforced Concrete Beam
V. 32, No.3, May-June 1987, pp. 74-150.
Column Joints in Japan," PMFSEL Report No. 87-8, Phil M. Fer-
guson Structural Engineering Laboratory, Department of Civil Engi- 16. Vecchio, F. J., and Collins, M. P., "Modified Compression-
neering, University of Texas at Austin, Dec 1987, and references Field Theory for Reinforced Concrete Elements Subjected to Shear,''
thereof. ACI JOURNAL, Proceedings V. 83, No.2, Mar.-Apr. 1986, pp. 219-
8. Wong, P. K. C.; Priestley, M. J. N.; and Park, R., "Seismic 231.
Resistance of Frames with Vertically Distributed Longitudinal Rein- 17. Pantazopoulou, S. J., and Qi, X., "Lateral Load Behavior of
forcement in Beams," ACI Structural Journal, V. 87, No. 4, July- an R.C. Frame with Floor Slabs," Proceedings, Fourth U.S. Na-
Aug. 1990, pp. 488-498. tional Conference on Earthquake Engineering, EERI, Palm Springs,
9. ACI-ASCE Committee 352, "Recommendations for Design of V. II, May 1990, pp. 137-146.
Beam Column Joints in Monolithic Reinforced Concrete Struc- 18. Zerbe, H. E., and Durrani, A. J., "Seismic Response of Con-
tures," ACI JOURNAL, Proceedings V. 82, No. 3, May-June 1985, pp. nections in Two-Bay R.C. Frame Subassemblies," Journal of Struc-
266-283. tural Engineering, ASCE, V. 115, No. 11 Nov. 1989, pp. 2829-2844.
10. "Code of Practice for the Design of Concrete Structures," 19. Collins, M.P., "Towards a Rational Theory for RC Members
(NZS 3101:1982), Standards Association of New Zealand, Welling- in Shear," Journal of the Structural Division. ASCE, V. 104, No.
ton, 1982, Part 1, 127 pp. and Part 2, 156 pp. ST4, Apr. 1978, pp. 649-666.

36 ACt Structural Journal I January-February 1992

Potrebbero piacerti anche