Sei sulla pagina 1di 10

ACI Structural Journal/November-December 2008 675

ACI Structural Journal, V. 105, No. 6, November-December 2008.


MS No. S-2006-156.R2 received December 19, 2007, and reviewed under Institute
publication policies. Copyright 2008, American Concrete Institute. All rights reserved,
including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including authors closure, if any, will be published in the September-
October 2009 ACI Structural Journal if the discussion is received by May 1, 2009.
ACI STRUCTURAL JOURNAL TECHNICAL PAPER
An enhanced fiber element is formulated to model the effects of
shear-flexure interaction in reinforced concrete (RC) elements
subjected to cyclic loading. In this stiffness-based element,
adopting the Timoshenko beam theory, shear and flexural behavior
are linked by means of kinematical assumptions. In bending,
different from standard fiber elements, the cross-sectional fibers
have the direction of the compressive principal stress and are not
aligned with the element longitudinal axis. This accounts for the
contributions to shear strength due to both the arch action and the
inclined thrust-line developing in squat elements. Additionally, in
shear, the so-called Mrschs truss is explicitly modeled considering
both the tension and compression concrete diagonal, allowing for
adopting only one truss for cyclic loading. The nonlinear behavior
of materials is described by means of appropriate constitutive
relations for which the critical implementation issues are high-
lighted. The proposed element has been validated by comparison
with selected experimental results.
Keywords: cyclic shear; fiber element; reinforced concrete; shear-flexure
interaction.
INTRODUCTION
Reinforced concrete (RC) elements can suffer from
different failures when subjected to strong cyclic transverse
loading, such as those induced by earthquakes. Tracing back
to the work of Leonhardt,
1
it appears that, for RC elements
having an intermediate ratio (2 to 6) between the unsupported
length and the cross section depth, the failure can be of
mixed type, that is, involving both flexural and shear
mechanisms; coupling of the two within the element end
zones is further enhanced by the arch effect.
The finite element (FE) method provides, in principle, the
best modeling approach in these cases, but suffers from two
main drawbacks when applied to large structures subjected
to strong seismic demands. The large number of elements
required to describe the geometry with sufficient precision,
along with the need for accurate, multi-axial, nonlinear,
cyclic constitutive relations, require a tremendous compu-
tational effort, further complicated by the problem of
analyzing and processing a large amount of output data.
Fiber models can overcome these problems by means of a
simplified modeling approach. Each structural member
(beam, column, and bridge pier) is discretized in fiber
elements and the stress-strain history for both steel and
concrete is evaluated throughout the analysis by means of
uniaxial constitutive laws at different positions within
selected cross sections. This approach allows for the reduction of
the number of degrees of freedom and retains the advantage of
an output more familiar to designers.
At present, only a limited number of fiber elements that
account for shear in different ways have been formulated,
both in the direct stiffness framework
2,3
and in a flexibility
context.
4-6
The first flexibility-based fiber models
5,6
were
mainly devoted to capturing the axial force-bending moment
interaction. More recently, a refined model was proposed by
Petrangeli et al.,
7
in which the appropriate kinematic
assumptions and constitutive laws enable for a description of
the shear effects. The flexibility-based approach adopted by
some of the quoted models poses the computational problem
of the element state determination
8
when inserted in the
familiar context of the direct stiffness method, though
correctly describing the shape of the bending moment
diagram along the element. In the works of Spacone et al.
5
and Petrangeli et al.,
7
this problem has been solved by
nesting a second iterative procedure, at the element level,
inside that required to enforce equilibrium at the end of each
load step at the structure level.
In this work, a different modeling approach is proposed
starting from the consideration that, even if shear effects
actually spread throughout the element, the shear-flexure
interaction is more pronounced in limited zones, for
example, the fixed-end region in a cantilever. Only regions
where the shear-flexure coupling takes place, both for
strength and stiffness, are modeled with the proposed ad hoc
fiber model (the reinforced concrete inelastic zone [RCIZ]
model). The limited length of the fiber element makes the
choice of the shape functions easier and allows for the adoption
of a stiffness-based approach. The stiffness-based approach,
in turn, eliminates the need for the iteration at the element
level in the state determination phase, and is therefore
advantageous from the computational point for the adoption
in seismic nonlinear analyses.
Two main features characterize the proposed fiber
model, both deemed to reduce the computational effort in
view of the application to seismic problems. First of all, a
uniaxial constitutive relation for concrete is assumed.
Different from the models proposed to date, the principal
direction of the compressive stress rotates during the analysis to
account for the arch action, and thus it may be non-normal
to the cross section. Second, the so-called Mrschs truss
9
shear-resisting mechanism is explicitly modeled on the
basis of a mechanical model.
In the following, the FE formulation, including the
element kinematics; the modeling of the shear resisting
mechanisms; and the constitutive relations for the material
models will be described. Finally, the element will be validated
by comparison with some experimental results.
Title no. 105-S61
Modeling Shear-Flexure Interaction in Reinforced Concrete
Elements Subjected to Cyclic Lateral Loading
by Luca Martinelli
ACI Structural Journal/November-December 2008 676
RESEARCH SIGNIFICANCE
This paper presents a fiber element capable to account for
the cyclic nonlinear shear behavior, and its coupling with
bending, by means of shear resisting mechanisms. The
proposed element is of simple implementation in existing
displacement-based FE codes, being formulated in the
stiffness-based framework for FE. Its efficiency and
computational economy allow for the coarse modeling of
entire structures, with significant shear effects in the
nonlinear response under cyclic or seismic forces, and for
the economical evaluation of the reanalysis cycles
required by performance-based design.
FIBER ELEMENT FORMULATION
The element follows the approach for stiffness-based fiber
elements representative of RC members. The geometry and
the position of the reinforcement are described by subdividing
selected cross sections in small portions (fibers) where the
cyclic nonlinear material behavior is followed, distinct for
concrete and reinforcement. The nodal restoring forces and
the stiffness matrix are computed by area integrals over these
cross sections and subsequent line integrals along the
element length. The proposed element is characterized by
the way the shear force is obtained, superimposing in the
cross section different shear resisting mechanisms derived
starting from the Timoshenko beam theory to describe shear.
A three-node element is adopted to obtain a linear variation
of the curvature (Fig. 1(a)). To avoid locking, which may
also arise in short nonlinear beam elements using this kinematic
model, the formulation adopts the shear constraints
method.
10,11
Consequently, in addition to axial displacements
u
an
= {u
13
}
T
the transverse displacements (w
1
, w
3
, v
1
, and
v
3
) of the end nodes and the nodal rotations
zn
= {
z,13
}
T
,

yn
= {
y,13
}
T
are assumed as independent parameters,
herein listed in vector u
n
= {u
an
T
,v
1
,v
2
,w
1
,w
2
,
zn
T
,
yn
T
}
T
.
Note that the element does not include torsional degrees of
freedom.
According to the adopted beam kinematics, the only
nonzero components of the engineering strain tensor, listed
in = {
x
,
xy
,
xz
}
T
, are related to the displacements and
rotation fields inside the element in the following way
(1a)
(1b)
(1c)
where
x
is the normal strain;
xy
and
xz
are the shear distortions;
u
a
(x) is the displacement along the x axis of the element of a
reference point in the cross section, usually the centroid; v(x)
and w(x) are the transverse displacements of the same point;

y
(x) and
z
(x) the rotations of the cross section, defined
according to Fig. 1(a); and the curvatures
z
and
y
have
been introduced. Herein the shear strains
xy
and
xz
are
assumed constant over the cross section, consistent with the
Timoshenko beam model. Denoting with
g
= {
a
(x),
xy
(x),

xz
(x),
z
(x),
y
(x)}
T
the vector of the generalized strain
components, which can be used equally well to describe the
strain fields inside the element, from Eq. (1) the following
matrix relation is derived
= b(y,z)
g
(x) = b(y,z)C
x
(u(x)) (2)
where C
x
is the linear differential (with respect to x) compatibility
operator that transforms the displacements and rotations vector
u(x) = {u
a
(x), v(x), w(x),
z
(x),
y
(x)}
T
into
g
.
Mapping the x axis domain, which spans the element, in
r [1 1] and using an isoparametric formulation, u(x(r))
can be computed from u
n
using appropriate shape functions
N
1
(r)N
6
(r)
(3)
as
u
a
(r) = n
1
u
an
(3a)

z
(r) = n
1

zn
(3b)

y
(r) = n
1

yn
(3c)
v(r) = (N
1
(r)v
1
+ N
3
(r)v
3
+ n
2

zn
(3d)
w(r) = (N
1
(r)w
1
+ N
3
(r)w
3
+ n
2

yn
(3e)
or, in matrix form u(r) = N(r)u
n
. In Eq. (3), n
1
= {N
1
(r),
N
2
(r), N
3
(r)} and n
2
= {N
4
(r), N
5
(r), N
6
(r)}.
From Eq. (3),
g
(x) = C
x
(u(x)) = C
r
(N(r))|J|u
n
, where |J| is
the Jacobian of the transformation x(r) and the variable x or

x
x y z , , ( )
u
a
x ( )
x
---------------- y

z
x ( )
x
---------------- z

y
x ( )
x
---------------- +
a
x ( ) y
z
x ( ) z
y
x ( ) + = =

xy
x ( )
z
x ( )
v x ( )
x
------------- + =

xz
x ( )
y
x ( )
w x ( )
x
--------------- + =
N
1
r ( ) 1/2 ( )r 1 r ( ) ; = N
2
r ( ) 1 r ( ) 1 r + ( ) =
N
3
r ( ) 1/2 ( )r 1 r + ( ); = N
4
r ( )
L
24
------


3 1 r
2
( ) 2r 1 r
2
( ) + [ ] =
N
5
r ( )
L
24
----- -


4r 1 r
2
( ) [ ]; = N
6
r ( )
L
24
------


3 1 r
2
( ) 2r 1 r
2
( ) + [ ] =
Luca Martinelli is an Assistant Professor in the Structural Engineering Department
of Politecnico di Milano, Milano, Italy, where he received his MS and PhD. His
research interests include the seismic analysis and design of concrete buildings and
the dynamics of flexible structures.
Fig. 1(a) Nodal displacements and control sections of
RCIZ element; and (b) shear-resisting contributions in RC
beams. (Note: F
sl
= force in flexural reinforcement; F
c
=
compressive force in concrete above neutral axis; T
ag
and
N
ag
= components of force due to aggregate interlock;
V
dw
= force due to dowel action; and F
s
= force in transverse
shear reinforcement.)
ACI Structural Journal/November-December 2008 677
r in the subscript denotes derivation with respect to the same
variable; the following definition is obtained
= b(y,z)C
r
(N(r))|J|u
n
= B
r
u
n
(4)
where B
r
is the strain-displacement matrix.
For stiffness-based FE, the element stiffness matrix K and
the nodal restoring forces vector r are computed as the
results of volume integrals; by use of Eq. (4), these can be
conveniently decomposed in area integrals over the element
cross section and subsequent line integral along the element
length, allowing K and r to be viewed as line integrals of
cross-sectional stiffness and internal forces, respectively
(5)
(6)
=
where D = Diag(E, G, G) is the diagonal matrix listing the
tangent values for the Youngs modulus E and shear modulus
G at the position of the material point inside the element; K
s
(r) =
b(y, z)
T
Db(y, z)dA is the cross section

stiffness matrix; =
{
x
,

xy
,
xz
}
T
is the vector listing the stress components;
r
s
(r) =
{N, V
y
, V
z
, M
z
, M
y
}
T
is the vector listing the sectional axial
force N, shear forces V
y
, V
z
, and bending moments M
y
, M
z
at r.
The following sections will illustrate the computation of the
area integrals appearing in Eq. (5) and (6), whereas the line
integrals are evaluated using a five-point Gauss-Lobatto
scheme; although less efficient than an equal number of Gauss
points, this allows the positioning of integration points (herein,
sectional quantities) at the ends of the integration interval where
a more significant nonlinear material behavior is expected.
STATE DETERMINATION OF CROSS SECTIONS
In the proposed element, shear and flexural behavior are
coupled, so that it is not possible to present them independently.
Ideally, by separating the structural element along a shear
crack in two parts, it is possible to consider the shear force
12
as the sum of two contributions, V
s
and V
c
, transmitted by the
Mrschs truss action and by other resisting actions,
respectively, as depicted in Fig. 1(b). Herein, F
sl
and F
c
are
the force in the flexural reinforcement and the compression
force in the concrete above the neutral axis, respectively; T
ag
and N
ag
are the components of the force due to aggregate
interlock; V
dw
is the force due to dowel action, whereas F
s
is
the force in the transverse shear reinforcement. The so-called
concrete contribution V
c
can be considered grouping several
mechanisms, as aggregate interlock (V
IN
), compression concrete
above the neutral axis V
cc
, arch action V
Fc
due to F
c
being
inclined of , and dowel action V
dw
. Of these, only the truss
action V
s
, the arch action V
Fc
, and V
cc
will be explicitly
modeled in the proposed formulation; the first is normally
the predominant term in elements with web reinforcement and
low axial force, the second becomes important in short elements or
when a compressive axial load is present, as for columns or walls.
K B
r
T
DB
r
J A d r d
A

1
1

C N r ( ) ( )
T
K
s
r ( )C N r ( ) ( ) J r d
1
1

= =
r B
r
T
A J d r d
A

1
1

=
C N r ( ) ( )
T
b y z , ( )
T
A J d r C N r ( ) ( )
T
r
s
J r d
A

1
1

d
A

1
1

For the computation of the shear force resultant V = V


s
+
V
Fc
+ V
cc
, a distinction is made for the contributions of the
concrete above and below the neutral axis because it is
assumed that different mechanisms are involved. Compression
concrete above the neutral axis participates to the transfer of
V
Fc
and V
cc
, whereas concrete below the neutral axis only
participates to the transfer of V
s
. It follows that there is not
an explicit constitutive model for shear behavior because
the final effect is obtained summing different partially
independent contributions.
Truss mechanism
According to the Mrschs truss analogy,
9
the contribution V
s
can be derived assuming that the concrete between two shear
cracks is a strut of an idealized truss that also comprises the
shear reinforcement acting as a tie. The angle formed by
the shear cracks with the element longitudinal axis defines
the inclination of the strut. Experimental results show that
the truss mechanism and the aggregate interlock effect are
coupled; this is considered in some of the more refined
models of the local behavior in the neighborhood of a
crack
13
and in models applied to a larger portion of the
structural element.
14
The quoted models, however, do not
describe the element behavior under cyclic loading. The
coupling influences the relative displacements at the crack
faces and the principal stresses direction in concrete between
cracks due to the presence of shear and normal stresses at the
crack face. The direction of the principal compression
deviates slightly from the crack direction, generally
reducing the angle with the element axis at increasing shear
distortion. In the model proposed, however, the relative
angle of the concrete struts in the truss mechanism remains
constant throughout the analysis. This choice is acceptable
because it introduces inaccuracies that decrease with the
increase of the shear distortions; this aspect will be further
discussed in a following section.
The shear force V
s
transferred by truss action is computed
assuming that constant shear stresses
xz,t
and
xy,t
act on the
area A
t
of concrete in tension in the cross section
(7)
A constant shear stress over A
t
is an acceptable approxi-
mation of the true shear stress distribution (for example,
refer to Fig. 3 in Reference 15) that is almost constant below
the neutral axis and has lower values above the neutral axis.
For the sake of brevity, only the computation of V
y,s
will
be presented. The portion of the structure close to the cross
section upon which V
s
is computed is depicted in Fig. 2(a).
It is assumed that inside the element, a modified idealized
multiple Mrsch truss exists but, different from the traditional
truss model
15
by Mrsch, this is composed not only of the
concrete compression strut and the transverse reinforcement
(nonlinear springs K
d1
and K
s
, respectively, in Fig. 2(a)) but
also of a second concrete diagonal (nonlinear spring K
d2
)
accounting for tensile stresses in the concrete. In the model
in Fig. 2(a), K
d1
, K
d2
, and K
s
represent the concrete diagonals
and the transverse reinforcement over the distance s between
two cracks, respectively; the angle of the diagonals, a
parameter of the model, equals the crack inclination and a
rigid link connects the concrete diagonals with K
s
.
V
z s ,

xz t ,
A V
y s ,
; d
At


xy t ,
A d
At

= =
678 ACI Structural Journal/November-December 2008
The presence of a second diagonal (which is in tension, as
illustrated in Fig. 2(b) for a negative shear distortion) has two
positive effects. First of all, the tension strength of the
concrete can represent the effect of the aggregate interlock
on the rotation of principal compression stresses. In fact, this
effect shows degradation with the increase of the shear
distortion and with the accumulation of strain in the transverse
steel, as it happens for the aggregate interlock. Second, it
avoids the ambiguities present in models
2
in which only the
active compressed diagonal is present, related to switching
from one compressed diagonal to the other whenever the
shear distortion changes in sign.
The stress
xy,t
is derived considering the effects of two
different diagonal stress fields (one for each concrete diagonal).
The values of the stresses in these diagonal fields are derived
from the analysis of a single module of the ideal truss
(Fig. 2(a)), assuming this is in the x-y plane for the
computation of V
y,s
. The following procedure is followed:
1) Compatibility of the shear strain in the plane of the
truss with the shear strain
xy
(r) from the Timoshenko beam
model in the cross section is imposed assuming (Fig. 2(b))
=
xy
(r).
2) Strains
d1
and
d2
in the concrete diagonals are computed
from the shear strain in the truss plane and the strain
s
of the
transverse reinforcement by means of Mohrs circle under the
assumption that, for a truss in the x-y plane,
y
=
s
(8)
where
i,j
denotes the cosine of the angle between directions
i and j. A similar transformation is used to compute the strain

in the direction normal to each diagonal.


3) The stress
s
of the transverse reinforcement and
d1
and
d2
in the concrete diagonals are computed using
uniaxial constitutive laws for steel and concrete, according
to the current value of
s
,
d1
, and
d2
coming from Eq. (8)
and those

normal to the diagonals; the latter are required


by the model for the strain softening in compression of the
concrete constitutive law.

d1

y

y d1 ,

y d1 ,

xy

x d1 ,

y d1 ,
+ =

d2

y

y d2 ,

y d2 ,

xy

x d2 ,

y d2 ,
+ =
4) The equilibrium condition in the y-direction at the
reinforcement-diagonals node (Fig. 2(c)) provides
F
s
+ F
d1

s,d1
+ F
d2

s,d2
= 0 (9)
where
s,d
denotes the cosine of the angle between the diagonal
direction and the shear reinforcement.
The forces F
d1
, F
d2
, and F
s
can be expressed as functions of
the normal stresses in K
s
, K
d1
, and K
d2
exploiting the geometry
of Fig. 2(c). The equilibrium Eq. (9) can then be recast more
conveniently in terms of stresses in the following form

s
+
d1

y,d1

y,d1
+
d2

y,d2

y,d2
= 0 (10)
where
s
denote the reinforcement ratio of the shear reinforcement.
5) If the unbalanced force in Eq. (9) is greater than a fixed
tolerance, the strain
s
is modified and Steps 2 to 5 are repeated.
Finally, the shear stress
xy,t
is computed superimposing
the effects of the uniaxial stress fields in the concrete diagonals

xy,t
=
d1

x,d1

y,d1
+
d2

x,d2

y,d2
= 0 (11)
Due to the kinematical model adopted for the truss, the
state of the longitudinal reinforcement has no direct effect on
the shear stress
xy,t
and the coupling with flexure is
produced both by the kinematic field of the element and by
the assumption that the shear stress
xy,t
acts on the area A
t
of
concrete in tension in the cross section, which descends from
the flexural behavior of the element.
The truss analogy adopted is simple in use when cyclic
deformations are concerned, but it is restricted to the case of
shear acting in a fixed plane. This may not be a limit for
some cross section shapes if it is possible to identify a few
planes in which trusses may form, for example, shear walls
or box-hollow cross sections.
Also, the contribution of the truss to the shear stiffness is
derived from the model of Fig. 3, once the tangent stiffness
E
s
t
, E
d1
t
, E
d2
t
of the transverse reinforcement and the
concrete diagonals, respectively, are known. The following
linear incremental relations hold
(12)
Substituting Eq. (12) in Eq. (11); using Eq. (8), (10), and
(11) and solving for
xy
yields the value of the tangent shear
modulus equivalent to the truss action
= (13)
where = .
Arch action and compression concrete
contribution
In RC beams, shear can be sustained by inclined internal
compression force in the concrete above the neutral axis
(Fig. 1(b)); this mechanism is termed arch action and
requires a reaction component at the beam ends, which is

s
E
s
t

y

d1
; E
d1
t

d1

d2
; E
d2
t

d2
= = =
G
t

xy t ,
/
xy
=
E
d2
t
E
s
t

x d2 ,
2

y d2 ,
2
E
d1
t

y d1 ,
2
( ) +
E
s
t

s
E
d1
t

y d1 ,
4
E
d2
t

y d2 ,
4
+ +
---------------------------------------------------------------------------------
E
s
t

x d1 ,
2
E
d2
t

y d2 ,
2

x d2 ,

y d1 ,

x d1 ,

y d2 ,
( )
2
+
Fig. 2Truss mechanism: (a) module of truss used in
computing shear force; (b) deformed truss configuration;
and (c) concrete portion, equivalent to hinge of roller, used
in imposing equilibrium in y direction.
ACI Structural Journal/November-December 2008 679
provided by the tie action of the longitudinal reinforcing
bars. A similar mechanism was pointed out by Priestley et al.
16
for squat bridge piers, and is already well known for squat
shear walls,
17
where part of the shear forces is transferred
directly from one end of the structural element to the other
through a compression field. This resembles the effect of an
inclined compression strut at an angle with the element
longitudinal axis (Fig. 3(a)). In the model by Priestley et al.,
16
the tie action is the axial force N and the portion V
Fc
of shear
force transferred is limited to V
Fc
= Ntan(). Experimental
results
18
point out that the importance of this resisting mechanism
increases as the angle increases but, at the same time,
the available ductility decreases more than proportionally. For
large values of , the collapse is due to crushing of the
compression concrete.
In fiber elements adopting uniaxial constitutive laws, the
direction of the concrete fibers is generally normal to the
cross section. In the proposed fiber model, however, which
adopts a uniaxial stress-strain relation for the concrete, the
effect of the inclined internal compression force on the shear
strength is reproduced by a suitable rotation of the response
direction of the fibers, that may then become non-normal to
the cross section during the loading history. The fibers are
assumed aligned with the direction of the compression strut
of Fig. 3(a) to reproduce the arch effect. From the mechanical
point of view, this corresponds to two common assumptions: the
directions of the principal stresses and strains are coincident and
the strut direction determines that of the principal compressive
stress
2
in the concrete.
The shear force corresponding to this mechanism is
derived in the cross section from the strain
x
at each fiber
initially with reference to the case of the strut in the x-y
plane. From a Mohrs circle for strains, under the assumption
that
y
= 0, it is possible to compute the principal strain
2
for
the generic fiber

2
=
x
/2 +
x
/2cos(2) (14)
Figure 3(b) and (c) illustrate the application of Eq. (14) for
increasing values of the angle . For a fixed value of
x
, the
compressive strain
2
increases with ; concrete damage
also increases, depending on plastic part of
2
.
The stress
2
, corresponding to the strain
2
, is computed from
a uniaxial material model (to be described in the following). It
must be recognized that
2
is consistent (refer to Fig. 3(b) and
(c)) with a shear strain that is different from the one given by
the beam element kinematics; this last, however, is only an
average approximation of the behavior of an RC element.
Given
2
, the corresponding normal stress
x
and tangential
stress
xy,s
are obtained from a Mohrs circle for stresses
under the assumption of null principal tension stress,
1
= 0.
The general three-dimensional case is treated in a similar way,
being now the inclined strut not necessarily in the x-y plane.
Indicating with
i
the director cosine of the strut with
element axis i = x, y, z, the following stresses are computed
in the cross section

x
=
2

x
(15a)

xy,s
=
2

y
(15b)

xz,s
=
2

z
(15c)
The integration of
x
over the concrete area A
c
and of
xy,s
and
xz,s
over the compression concrete area A
cc
yields the
concrete contribution to the generalized internal forces at the
cross section
(16a)
(16b)
(16c)
(16d)
(16e)
From Eq. (15) and (16), it appears that the fiber direction
directly couples shear with bending; however, the process
requires an a priori knowledge of the angle .
The axial force value is not usually high in squat elements:
therefore, it is assumed that the inclined strut connects the
center of compression of the end section having the highest
bending moment with the centroid of the cross section
having zero bending moment. The value of for each
N
x
A d
Ac

=
M
y

x
z A d
Ac

=
M
z

x
y A d
Ac

=
V
z Fc ,

xz s ,
A d
Acc

=
V
y Fc ,

xy s ,
A d
Acc

=
Fig. 3(a) Arch-action-resisting mechanism in squat RC
elements; (b) and (c) Mohrs circles for strains at increasing
values of angle ; and (d) model used in computing strut
angle .
680 ACI Structural Journal/November-December 2008
element is hence computed according to the following steps:
1) the nodal reaction forces are computed, and bending
moments (M
zi
, M
yi
, M
zf
, M
yf
) are derived at the end nodes;
2) the position vector of the centroid of the compression
stresses is computed in the end cross sections; and 3) for each
component (y or z) of the position vector the problem is
considered planar: (a) the cross section with zero bending
moment is found by linear extrapolation (or interpolation)
from the values of the moment at the end nodes; and (b) a
new value of the component is computed, at the end with the
smaller absolute value of the bending moment, from a
triangle similitude. The process is illustrated in Fig. 3(d) in
the x-y plane assuming M
zi
is the maximum moment along
the element; the triangle similitude yields for the y component
y
cpi
:y
cpf
= L
i
:L
f
, where y
cpi
and y
cpf
are the y coordinates of
the compression stresses centroid in the initial and final
section of the element. The angle represents the inclination
of the line connecting the compression centers at the ends of
the element. This approach can also be used when the
member is in double bending.
The computations show that, in a step-by-step dynamic
analysis, is not sensitive to the stress level attained in the
cross section fibers, especially when the cross section is
heavily loaded in flexure with a small portion reacting in
compression. Accordingly, can be taken, in a simplified
way, as the value at the end of the previous load step.
The contributions V
z,cc
and V
y,cc
due to the shear distortion
of the compression concrete above the neutral axis are
explicitly modeled also. In a simple way, they are computed
as the elastic shear force induced by the shear distortion,
having denoted with G the shear modulus of the concrete
V
z,cc
= A
cc
G
xz
; V
y,cc
= A
cc
G
xy
(17)
Finally, to obtain the resultant shear forces V
z
and V
y
in the
cross section, the contributions V
z,s
and V
y,s
due to the truss
action and those V
z,Fc
and V
y,Fc
due to the arch action, Eq. (7)
and (16), respectively, are added to those given by Eq. (17).
Cross section stiffness matrix
The stiffness matrix of the cross section, K
s
, which is
required for the volume integrals in Eq. (5), relates the vector
r
s
of the increments of the internal forces to the vector
g
of the increments of the generalized strain components. For
the vectors r
s
and
g
, as defined, the elements of matrix K
s
are listed in the Appendix. As a due remark, terms k
24
, k
25
and k
34
, k
35
arise because, according to the proposed formu-
lation for the arch action, curvatures are directly coupled
with shear forces even though shear distortions
xy
or
xz
are
not directly coupled with bending moments.
Materials models
The derivation of the proposed fiber model allows for the
adoption of uniaxial models for both steel and concrete, thus
enhancing the computational efficiency.
To have a simple but capable model from the beginning
and to simplify further developments, the model by Monti
and Nuti
19
is adopted for the steel reinforcement because it
is explicit and because it appears capable to account for the
post-elastic buckling behavior of the reinforcing bar with
little additional computational cost. Perfect bond is assumed
between steel bars and concrete. The same constitutive relations
are followed by concrete in the cross section fibers and in the
truss mechanism. For concrete, several uniaxial constitutive
models are available in the literature. Most of them,
however, lack the capability of reproducing the complex
phenomena of crack faces misalignment (crack bridging) and
the compressive strain softening due to coexisting principal
tensile strains; these are key factors in successfully simulating
cyclic concrete behavior, especially in the truss mechanism.
For these reasons, the model proposed by Stevens et al.
20
to
represent the average behavior of cracked RC has been
adopted. The model accounts for crack bridging through the
early reloading in compression following a large tensile
strain and considers both the effects of confinement and the
influence of the strain state in the direction normal to the
material response. This last feature has been applied only to
the concrete of the diagonals inside the truss mechanism.
The model, originally formulated
20
as an incremental
uniaxial relation in the direction of the principal strain
increments, is herein applied in a finite form in the direction
of each concrete fiber (or truss diagonal) assuming they
define the direction of the principal strain increment.
NUMERICAL TESTS
The proposed fiber column model has been implemented
inside an established research computer code for the analysis
of nonlinear structures.
21
The comparison of numerical and
experimental results first of all aims to highlight the accuracy of
the truss mechanism model and the consequences of the
choice of the material models. Subsequently, under
monotonic loading, the element ability to account for the
brittle concrete collapse induced by the arch action in squat
members will be investigated while, for cyclic loading, the
capability of reproducing the pinching and stiffness degradation
in more slender elements, principally governed by the truss
mechanism response, will be studied. Finally, the model is
used to simulate a full-scale element, a short bridge pier,
subject to axial force and imposed cyclic displacements. The
shortcomings of the element, descending from its base
hypotheses, are discussed as well.
Cyclic testing of truss model
An example of the predictions that can be obtained using
the proposed truss model is given by Fig. 4(a), which
compares the response in terms of shear stress versus shear
distortion for Specimen SE8, studied by Stevens et al.,
20
with that obtained from a single module of the proposed truss
mechanism. The specimen was a 1524 x 1524 x 285 mm (60 x
60 x 11.2 in.) RC panel tested in pure shear; two layers of
reinforcing bars were present on each side of the specimen,
each one formed by bars at a 90-degree angle. The reinforcement
ratio was 1% and 3% in the two directions, respectively.
Specimen SE8 has been chosen for its different reinforcement
ratios, as is the case in beam or column structural elements.
The transverse reinforcement ratio was sufficient to induce a
large inelastic strain in the concrete of the diagonals of the truss
model, thus helping to show the limits of the model
assumptions. For the truss model, the reported
20
material
properties was adopted and the transverse reinforcement,
represented by K
s
, was assumed in the direction having a 1%
reinforcement ratio. Two different values (45 and 30 degrees)
of the angle of the diagonals were considered, derived from
those measured
20
for the principal compression stresses
showing that the angle was larger for lower strain distortion.
The computed results highlight the dependence of the
response on this parameter.
ACI Structural Journal/November-December 2008 681
The numerical results of Fig. 4(a) are in good agreement
with the experiment, showing the satisfactory performance
of the proposed truss model. The initial response is better
captured by the model with the diagonals at 45 degrees, up
to a shear distortion of 0.0023. In this range the shallower
( = 30 degrees) truss overshoots both the strength and the
stiffness due to both the enhanced contribution of the
transverse reinforcement to the strength and the dependence
on the concrete material model, particularly the tension
strength of the concrete. Both these aspects descend from the
lower value of . For larger values of the shear distortion,
the shallower truss gives better results; this is consistent with
the evolution of the angle of the principal stresses measured
during the physical experiment and suggests the way for the
choice of when large values of strains are expected. The
strain range in which the concrete model has to be effective
in a truss model, like the one herein adopted, can be appreciated
in Fig. 4(b), which depicts the nondimensional stress-strain
curve followed during the test of Fig. 4(a) by the concrete of
one of the truss diagonals. In this figure, the normalization
values are the concrete strength f
c
and the corresponding
strain
c0
. In Fig. 4(b), it can be noted that a large strain
softening takes place in the last cycles, due to the tensile
strain in the direction normal to the concrete diagonal
induced by the accumulation of plastic strain for the transverse
steel (that is, in K
s
). The paramount importance of the
concrete material model, particularly in the transition from
tensile to compressive stresses, is also highlighted in Fig. 4(c),
which compares the effects of the concrete model of Stevens et
al.
20
(solid line) with that of Mander et al.
22
(dashed line) on
the response of a truss module, though for a different
imposed strain history from that of Fig. 4(a). The latter
material model lacks, in the original formulation, a proper
description of the tension-compression transition, leading to
phenomenologically unacceptable results. This confirms the
importance of the concrete model for the successful simulation
of the shear behavior of RC elements whenever truss models
are adopted, as already noted by Garstka
2
with reference to
a different truss model.
Monotonic loading
To investigate the interaction effects between flexure and
shear in short elements, the monotonic test series SBV by
Garstka
18
on short columns under predominating shear
loading were reproduced. The flexural reinforcement had
f
y
= 566 MPa (82.09 ksi), whereas the stirrups f
y
= 514 MPa
(74.54 ksi). All the specimens had the same square cross
section and shear reinforcement, differing only in the shear
span-depth ratio (a/h) equal to 1.9, 2.2, and 2.5, respectively. In
the numerical simulations, the specimen was reproduced with
two RCIZ elements and the cross section was satisfactorily
modeled with nine fibers. In selecting the value of , it must
be considered that the elements model a diffusion zone
within the structural member, where the shear cracks are
more inclined. The truss angle was assumed at 60 degrees
with the element longitudinal axis, following the suggestions
given by Crist
23
for the inclination of the crack as a function
of a/h, and a very low value G = 1 GPa (145 ksi) for the shear
modulus was adopted.
Figure 5 illustrates the experimental and numerical
monotonic force-deflection curves at the column top. During
the experimental tests, failure took place due to crushing of
the compression concrete after the base cross section had
attained its ultimate moment. This implies that a lower value
of the horizontal force corresponds to a higher a/h as a
consequence of the greater lever arm with respect to the base
of the specimen. According to the standard method of
Eurocode 2,
24
the shear reinforcement and the other shear
resisting mechanisms descending from beam action do not
provide sufficient shear strength to attain the full bending
strength of the cross section, and arch action is required to
reach the ultimate moment. As the influence of flexure (that
is, a/h) on the failure mode increased, the importance of the
arch action decreased and, consequently, the behavior
became more ductile. In the numerical simulations, the
higher stiffness at the beginning of the curve was mainly due
to two aspects of the model for the arch action. First, the
model for the concrete fibers was applied disregarding the
strain softening due to coexisting transversal strain; second,
all the fibers had the same inclination, with an excessive
Fig. 4(a) Experimental (gray) and numerical (black) shear stress-strain relationship for
RC panel tested in Reference 20; (b) nondimensional normal stress-strain curve followed
by one of the truss concrete diagonals; and (c) shear stress predicted by truss model with
concrete model of Stevens et al.
20
(solid line) and of Mander et al.
21
(broken line).
682 ACI Structural Journal/November-December 2008
contribution of the external compression fibers. Nevertheless,
the experimental trend was reproduced, although the strain
softening was much less pronounced. When the strut
mechanism was explicitly inhibited, so that the shear
strength was provided by the truss mechanism only, an
almost perfect elastic-plastic behavior was obtained due to
yielding of the shear reinforcement in the truss model.
Cyclic loading
The cyclic behavior in more slender structural elements
was reproduced for the test of Element R5 in the work by Ma
et al.
25
The specimen was a cantilever of span length a =
38.5 in. (977.9 mm) with rectangular cross section of effective
depth h = 14 in. (355.6 mm), with an aspect ratio a/h = 2.75.
The flexural reinforcement was made of deformed No. 6 bars
(d = 0.75 in. [19 mm]), f
y
= 65.6 ksi (452.3 MPa), and the
web reinforcement of deformed No. 2 bars (d = 0.25 in.
[6.3 mm]), f
y
= 60 ksi (413.7 MPa), at 3.5 in. (89 mm)
spacing, for a shear reinforcement ratio of 0.311%. The
corresponding Youngs moduli were 29,110 ksi (200.7 GPa)
and 28,170 ksi (194.2 GPa), respectively, whereas the
concrete strength was 4.58 ksi (31.5 MPa). The experimental
test involved a few initial cycles under force control until the
tip deflection value
y
, corresponding to yielding of the flexural
reinforcement, was reached and subsequent displacement
controlled cycles, at increasing values of the ductility factor
/
y
up to 4. At the end of the experiment, the tip deflection
due to shear was
25
approximately 36% of the total deflection
and more than 50% of that due to bending.
In the numerical model, the specimen was described with
only one element; the angle of the diagonals was assumed
as 53 degrees following the value given by Crist
23
for a/h =
2.75. The overall performance of the fiber element can be
judged from Fig. 6, which depicts the experimental and
numerical curves for the tip deflection versus the applied
tip force. The main discrepancies were an excessive
strength and stiffness for the first two cycles, an excessive
degradation of the strength for the following cycles at a
constant ductility factor, and some imperfections in the
simulation of the crack closing phase, which took place
when the tip deflection changed in sign. The model reproduced
the main phenomenological aspects of the experimental
response. The initial overestimation of the experimental
results in terms of strength and stiffness was partially due to
the tensile strength of the concrete in the truss diagonals, as
shown by the fit obtained when this strength was neglected;
a higher initial value of the truss angle, similarly to Fig. 4(a),
and the perfect bond assumption between concrete and
reinforcement also give a contribution. To highlight the
capability of the model, the response computed for the last
load cycle assuming an unrealistic angle ( = 30 degrees) for
the concrete diagonals is also shown in Fig. 6 with a dashed
curve; the enhancement of the shear strength induces an
eminently flexural response.
Finally, the model was adopted to simulate the response of
a short bridge pier, with an a/h of 1.75, subjected to imposed
displacements at the top and to an axial force of 1700 kN
(382 kip), corresponding to a normalized axial load =
0.1.
26
The specimen (Fig. 7) had a hollow-box cross section,
termed Type 3, with a main reinforcement ratio of 0.919%.
The span length was a = 2.8 m (110.23 in.), whereas the
overall depth was h = 1.6 m (62.99 in.). The imposed
displacement history was based on cycles of increasing
amplitude up to yielding of the bending reinforcement at
y
,
followed by three cycles at a displacement ductility factor
/
y
= 1.5, 3, and 6. The test ended at a displacement ductility
equal to 6 after rupture of some flexural bars that had
previously buckled. This specimen was selected because
more than 30% of the tip displacement was due to shear
26
and the axial load contributed to the strut mechanism. This fact
allowed the validity of the model to be checked in reproducing
the interaction between bending, axial force, and shear.
The specimen was modeled, with equally satisfactory
results, using one, two, or three of the proposed elements. In
Fig. 7, the numerical and experimental results are compared
for the two-elements case. To set a comparison with the perfor-
mance of fiber elements not including shear deformability, a
numerical model where the truss mechanisms were inhibited
and an unrealistically high shear stiffness was adopted was
analyzed. Its response is only partially reported in Fig. 7, for
the sake of clarity. This purely flexural model had a behavior
not too dissimilar from the experiment, but showed a higher
initial stiffness; a less pronounced pinching; and, character-
istically, an unloading stiffness much higher from that
depicted in Fig. 7, that was well simulated by the model with
shear deformability. These aspects were consistent with testing
modalities in displacement control. For large displacements,
the flexural behavior controlled the forces because the longi-
tudinal steel had yielded, whereas the effects of the shear
Fig. 5Force-deflection curve for SBV test series by Garstka et
al.
18
; experimental (gray) and numerical (black) model.
Fig. 6Force-deflection curve for Element R5 tested by Ma
et al.
25
; experimental (gray) and numerical (black) model.
ACI Structural Journal/November-December 2008 683
response were more evident when the flexural reinforcement
was elastic: the response of the model differed at the beginning
of the test and when the specimen was unloading. The test
confirmed that the subdivision of the shear force between the
different mechanisms modeled gave reasonable results,
enhancing those that could be computed by a flexural
element. Up to a displacement ductility factor equal to three, the
shear distortion predicted by the lower element in the model
with shear deformability showed a relative error of 30%
when compared with the experimental value measured on the
corresponding length of the specimen. The relative error
increased up to 100% when the ductility was 6. When using
the model with three elements, these relative errors became 34
and 60%, respectively. In judging these values, the relative
simplicity of the element, the paucity of the models (few
elements), and the intended purpose of the RCIZ element
have to be taken into consideration. Apparently, the model
did not capture the collapse because the failure mode for the
bars (low-cycle fatigue) is not directly reproduced by the
Monti and Nuti model and infinite ductility was adopted for
the reinforcement.
Shortcomings
The proposed element has, naturally, some shortcomings.
The kinematic model of the Timoshenko beam is an
economic though effective way of considering the shear
distortion, but it is based on the assumption that plane
sections remain plane, which limits the element capability
when the cross section warps (for example, T sections when
the flange is in tension). The perfect bond assumption for the
reinforcement implies that anchorage failures are not
directly captured. They normally take place outside the
structural element, however, and an assessment can be
indirectly carried out from the predicted strain and stress
levels. The mechanical models for the shear behavior have
some limits as well; the truss model does not consider the
interaction between the mechanisms in different planes and,
whereas it is based only on the knowledge of the mechanical
properties of the materials and the geometry of the element,
it also requires the knowledge of the inclination angle of the
concrete diagonals, that has to be estimated. For the arch
mechanism, collapse due to compression crushing is not
exactly reproduced, even though hints on the collapse are
given and the precollapse response is reproduced. An
improvement for this point can be exploiting for the fiber of
the cross section the dependence of the material model for
concrete on the transversal strain.
SUMMARY AND CONCLUSIONS
This paper presents a beam-column fiber element able to
describe the interaction between the bending moment and
the axial and shear forces. In RC elements, shear forces are
due to many complex interacting mechanisms, involving a
significant part of the volume of the element; in this work,
however, these are considered in an independent way and are
modeled mainly at a cross section level. This strategy aims
to reduce the computational effort in view of the application
to seismic problems, recalling that the purpose of the element is
rather to capture the behavior of a relevant portion of the
structural element than to model its complex local mechanics.
The shear force is computed in the cross section by super-
position of several contributions. The most important are due
to the truss and the arch mechanisms and are reproduced
with mechanical models coupled with the behavior in
bending of the element and depending only on the mechanical
properties of the materials and geometric parameters. The
truss is oriented as in the classical truss analogy, and
comprises the transverse reinforcement and two concrete
diagonals, reproducing the tensile strength of the concrete.
This avoids the need to identify which diagonal is in
compression and allows for the presence of only one truss for
cyclic loading also. The model for the arch effect, coupling
the axial force with bending, can be, in principle, also
adopted for an element kinematics different from the one
chosen herein. Another distinctive feature of this element is
that uniaxial material models are needed both for the flexural
and the shear behavior, reducing the computational costs.
Despite the limitations, the overall performance shows that
the proposed element is able to reasonably represent the
experimental response in selected test cases, strongly influ-
enced by shear. A limited number of elements is required and
the computational efficiency allows for the study of the
dynamic behavior of complete three-dimensional structures
with very short computer times. This aspect is of interest in
view of the diffusion that nonlinear analysis has gained in
seismic design regulations.
Fig. 7(a) Experimental (gray) and numerical (black) force-deflection relationships at
column top for Type 3 pier in Reference 26; and (b) cross section reinforcement.
ACI Structural Journal/November-December 2008 684
NOTATION
A
t
= area of concrete that is in tension in cross section
a/h = ratio of cantilever length and cross section depth
d = diameter of reinforcement bar
E
t
= tangential Youngs modulus
f
y
= reinforcement yielding strength
K
d1
, K
d2
= stiffness of nonlinear springs representing concrete diagonals
in truss
K
s
= stiffness of nonlinear spring representing transverse
reinforcement present over length s of element
L
i
, L
f
= distance of starting and ending cross section, respectively,
form ideal zero moment cross section
M
ji
, M
jf
= bending moment about axis j at element starting and ending
cross section, respectively
s = distance between two consecutive cracks
V
t
= shear force transmitted due to truss mechanism
V
i,t
= shear force transferred in direction i by truss mechanism
= strain increment
= stress increment

s
= average strain of transverse reinforcement
= angle formed by shear cracks with element longitudinal axis

xi,t
= shear stress in direction i representing in cross section the
effect of truss mechanisms
REFERENCES
1. Leonhardt, F., Reducing Shear Reinforcement in Reinforced
Concrete Beam and Slabs, Magazine of Concrete Research, V. 17, No. 53,
1965, pp. 187-198.
2. Garstka, B., Investigation on Resistance and Damage Behaviour of
Reinforced Concrete Linear Elements Considering Shear Effects under
Cyclic Nonlinear Loading, Technical Report TWM 93-2, Institut fr
Konstruktiven Ingenierbau, Ruhr University, Bochum, 1993, 136 pp.
3. Guedes, J.; Pegon, P.; and Pinto, A. V., A Fibre/Timoshenko
Beam Element in CASTEM 2000, Technical Report, Special Publication
No. I.94.31, EC, JRC, Ispra, 1994, 55 pp.
4. Ranzo, G., and Petrangeli, M., A Fibre Finite Beam Element with
Section Shear Modelling for Seismic Analysis of RC Structures, Journal
of Earthquake Engineering, V. 3, No. 2, 1998, pp. 443-473.
5. Spacone, E.; Filippou, F. C.; and Taucer, F. F., Fibre Beam-Column
Model for Non-Linear Analysis of R/C Frames: Part IFormulation,
Earthquake Engineering and Structural Dynamics, V. 25, No. 7, 1996,
pp. 711-725.
6. Kaba, S. A., and Mahin, S. A., Refined Modelling of Reinforced
Concrete Columns for Seismic Analysis, Technical Report UCB/EERC-
84/03, University of California-Berkeley, Berkeley, CA, 1984, 104 pp.
7. Petrangeli, M.; Pinto, P. E.; and Ciampi, V., Fiber Element for Cyclic
Bending and Shear of RC Structures, I: Theory, Journal of Engineering
Mechanics, V. 125, No. 9, 1999, pp. 994-1001.
8. Spacone, E.; Ciampi, V.; and Filippou, F. C., Mixed Formulation of
Nonlinear Beam Finite Element, Computers and Structures, V. 58, No. I,
1996, pp. 71-83.
9. Mrsch, E., Der Eisenbetonbau, seine Theorie und Anwendung,
Verlag Konrad Wittwer, Stuttgart, Germany, 1908, 376 pp.
10. Tessler, A., and Dong, S. B., On a Hierarchy of Conforming
Timoshenko Beam Elements, Computers and Structures, V. 14, No. 3-4,
1981, pp. 335-344.
11. Crisfield, M. A., Finite Elements and Solution Procedures for
Structural Analysis, Linear Analysis, V. 1, Pineridge Press, Swansea, 1986,
272 pp.
12. Park, R., and Paulay, T., Reinforced Concrete Structures, John Wiley
& Sons, New York, 1975, 769 pp.
13. Kupfer, H., and Bulicek, H., A Consistent Model for the Design of
Shear Reinforcement in Slender Beams with I or Box Shaped Cross
Section, Proceedings Concrete Shear in Earthquake, Elsevier Science
Publisher, Barking, 1992, pp. 256-265.
14. Di Prisco, M., and Gambarova, P. G., Comprehensive Model for
Study of Shear in Thin-Webbed RC and PC Beams, Journal of Structural
Engineering, ASCE, V. 121, No. 12, 1995, pp. 1822-1831.
15. Joint ASCE-ACI Committee 445, Recent Approaches to Shear
Design of Structural Concrete, Journal of Structural Engineering, ASCE,
V. 124, No. 12, Dec. 1998, pp. 1375-1417.
16. Priestley, M. J. N.; Verma, R.; and Xiao, Y., Seismic Shear Strength
of Reinforced Concrete Columns, Journal of Structural Engineering,
ASCE, V. 120, No. 8, 1994, pp. 2310-2329.
17. Inoue, N., and Suzuki, N., Microscopic and Macroscopic Analyses of
Reinforced Concrete Framed Shear Walls, Concrete Shear in Earthquake, T. Hsu
and S. Mau, eds., Elsevier Applied Science, 1992, pp. 333-342.
18. Garstka, B.; Krtzig, W. B.; and Stangenberg, F., Damage Assessment
in Cyclically Loaded Reinforced Concrete Members, Proceedings of the
Third European Conference on Structural Dynamics: EURODYN93,
Balkema, Rotterdam, 1993, pp. 121-128.
19. Monti, G., and Nuti, C., Nonlinear Cyclic Behavior of Reinforcing
Bars Including Buckling, Journal of Structural Engineering, ASCE, V. 118,
No. 12, 1992, pp. 3268-3284.
20. Stevens, N. J.; Uzumeri, S. M.; and Collins, M. P., Analytical
Modeling of Reinforced Concrete Subjected to Monotonic and Reversed
Loadings, Technical Report 87-1, Department of Civil Engineering,
University of Toronto, Toronto, ON, Canada, 1987, 201 pp.
21. Martinelli, L.; Mulas, M. G.; and Perotti, F., The Seismic Response
of Concentrically Braced Moment-Resisting Steel Frames, Earthquake
Engineering and Structural Dynamics, V. 25, No. 11, 1996, pp. 1275-1299.
22 Mander, J. B.; Priestley, M. J. N.; and Park, R., Theoretical Stress-
Strain Model for Confined Concrete, Journal of Structural Engineering,
V. 114, No. 8, Aug. 1988, pp. 1804-1826.
23. Crist, R. A., Static and Dynamic Shear Behavior of Uniformly
Reinforced Concrete Deep Beams, Technical Report AFWL-TR-71-74,
University of New Mexico (CERF) Kirtland Air Force Base, NM, Nov. 1971,
204 pp.
24. CEN, Eurocode 2: Design of Concrete StructuresPart 1-1:
General Rules and Rules for Buildings, ENV 1992-1-1:1991, European
Committee for Standardization, Brussels, 1991, 253 pp.
25. Ma, S. M.; Bertero, V. V.; and Popov, E. P., Experimental and
Analytical Studies on Hysteretic Behaviour or Reinforced Concrete in
Rectangular and T-Beams, Technical Report 02/76, Earthquake Engineering
Research Center, University of California-Berkeley, Berkeley, CA, 1976,
254 pp.
26. Pinto, A. V.; Verzeletti, G.; Negro, P.; and Guedes, J., Cycling
Testing of a Squat Bridge Pier, Report EUR 16247 EN, Ispra, 1995, 53 pp.
APPENDIX: ELEMENTS OF CROSS SECTION
STIFFNESS MATRIX
k
11
= k
s,i
+ k
c, j
+ k
cc,k

x
4
; k
12
= k
cc,k

x
3

y
; k
13
= k
cc,k

x
3

z
;
k
14
= (k
s,i
y
i
+ k
c,j
y
j
+ k
cc,k
y
k

x
4
) ; k
15
= k
s,i
z
i
+ k
c,j
z
j
+ k
cc,k
z
k

x
4
k
22
= GA
cc
+ G
t
A
ct
+ k
cc,k

x
2

y
2
; k
23
= k
cc,k

x
2

z
;
k
24
= (k
cc,k
y
k

x
3

y
) ; k
25
= k
cc,k
z
k

x
3

y
k
33
= GA
cc
+ G
t
A
ct
+ k
cc,k

x
2

z
2
; k
34
= (k
cc,k
y
k

x
3

z
); k
35
= k
cc,k
z
k

x
3

z
k
44
= k
s,i
y
i
2
+ k
c,j
y
j
2
+ k
cc,k
y
k
2

x
4
; k
45
= k
s,i
y
i
z
i
+ k
c,j
y
j
z
j
+ k
cc,k
y
k
z
k

x
4
;
k
55
= k
s,i
z
i
2
+ k
c,j
z
j
2
+ k
cc,k
z
k
2

x
4
where
i
is the director cosine of the strut with element axis
i = x,y,z; GA
cc
is the shear stiffness due to the compressed
concrete, G
t
A
ct
= G
t
(A
c
A
cc
) is that due to truss action, k
s,i
=
E
s,i
A
s,i
is the stiffness of the i-th steel bar, k
c,j
= E
c,j
A
c,j
that
of the j-th tensioned concrete fiber, k
cc,k
= E
cc,k
A
cc,k
that of
the k-th compressed concrete fiber and summation is implied
over the subscripts i, j, and k.

Potrebbero piacerti anche