Beruflich Dokumente
Kultur Dokumente
net/publication/259221644
CITATIONS READS
10 310
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Pier Paolo Rossi on 14 April 2015.
P. P. Rossi1
2
A. Recupero
ABSTRACT3
Over the last few decades a significant effort has been made to achieve accurate evaluation of the
resistance of reinforced concrete elements subjected to pure shear or combined internal forces
including shear. As regards the latter, continuum models characterized by simplified stress fields
have recently been applied by one of the Authors for the evaluation of the ultimate capacity
This paper illustrates the natural progress of these studies and describes an analytical tool for the
calculation of the ultimate strength of reinforced concrete columns with circular cross-section. The
proposed method is based on the application of the static theorem of limit analysis and takes into
account both truss and arch effects. To ascertain the accuracy and reliability of the method, the
relations developed are applied with reference to a large number of tests reported in the literature
and a comparison between the theoretical and experimental results is drawn. Finally, the predictions
of the proposed method are compared with those of other simplified methods present in the
literature.
Keywords: Circular cross-section, shear force, axial force, bending moment, reinforced concrete
1
Associate Professor, Dept. of Civil and Environmental Engineering, Univ. of Catania, Catania 95125, Italy
(corresponding author). E-mail: prossi@dica.unict.it
2
Assistant Professor, Dept. of Civil Engineering, Univ. of Messina, Messina, S. Agata (Messina) 98166, Italy
E-mail: antonino.recupero@unime.it
INTRODUCTION
According to a traditional approach, the design shear strength VRd of beams and columns is often
as the sum of two contributions: the strength of the concrete shear-resisting mechanism VcRd and the
The first term, VcRd, takes into account the shear force transmitted across the compression zone of
the element, the dowel action of the longitudinal reinforcement and the vertical component of the
shear force transferred across the crack by aggregate interlock effect. As also reported by some
researchers (Turmo et al. 2009) , its evaluation is controversial and relies on empirical methods.
The second term, VsRd, represents the shear force corresponding to yielding of the transverse
reinforcement. In the case of rectangular, T and I beams with stirrups orthogonal to the longitudinal
Asw
VsRd z f ywd cot θ (2)
s
where z is the internal lever arm, i.e. the distance between the compressive and tensile resultant
forces, Asw is the total transverse reinforcement area per layer, s is the spacing of the transverse
reinforcement along the longitudinal member axis, fywd is the yield strength of the transverse
reinforcement and is the angle of inclination between the compressive stresses of the concrete in
A relation similar to Equation (2) has also been derived by Ang et al. (1989) for reinforced concrete
elements with circular cross-section. In particular, these researchers assumed that diagonal cracks
were characterised by an inclination of 45 degrees with respect to the longitudinal member axis and
that cracks were able to mobilize the transverse reinforcement along a length extending the full
width of the confined core of the concrete. The truss mechanism contribution was evaluated as
Asw π
VsRd D ' f ywd (3)
s 4
where D’ is the diameter of the circular hoops. This relation has subsequently been modified in
Priestley et al. (1994a; 1996) and the term VsRd calculated by the relation
Asw π
VsRd D ' f ywd cot θ (4)
s 4
where the angle is assumed equal to the maximum between 30 degrees and the corner-to-corner
Recently, Kowalsky and Priestley (2000) have modified Equation (4) so as to consider that cracks
are closed in the compression zone of the member and thus that shear cannot be transferred across
cracks by tension strain in the transverse reinforcement. The steel truss mechanism is evaluated by
D 2
2
D' 2
Asw
'
y2
VsRd f ywd cot θ dy (5)
xn D 2
s D' 2
where xn is the depth of the compression zone and D is the diameter of the cross-section. In the
same paper Kowalsky and Priestley also propose the following more simple relation where the
integration is removed and the effective transverse reinforcement area is approximated as Asw /4
VsRd D 2 D ' 2 xn
Asw
f ywd cot θ (6)
s 4
More recently, Turmo et al. (2009) have proposed a new expression for the calculation of the
Asw
VsRd χ z f ywd cot θ (7)
s
This expression is formally similar to that provided by Priestley et al. (1994a; 1996) but it is
modulated through the efficiency factor . The value of this parameter is variable with the extension
of the compression zone of the cross-section (i.e. depending on the position of the compressive and
where z0 is the distance from the centroid of the tensile forces to the geometric centre of the
cross-section. It is worth noting that Equation (7) is equal to Equation (4) if the height of the hoops
is considered to be fully effective. In fact, in this case the lever arm of the internal forces is equal to
the diameter of the hoops, the ratio z0/D’ is equal to 0.50 and thus the efficiency factor is equal to
/4.
Turmo et al. (2009) recognize that the use of Equation (8) is not simple in real situations because
the internal lever arm and the efficiency factor are not constant throughout the length of the member.
To overcome this problem, they propose a design oriented formula of the efficiency factor in which
z/D’ is assumed equal to 0.9 and z0/D’ is equal to 0.45. These simplifications lead to a value of the
efficiency factor equal to 0.85 and to the following expression of the term VsRd
Asw
VsRd 0.765D ' f ywd cot θ (9)
s
The same researchers also affirm that the design oriented formula leads to inaccurate results for
columns with low z/D’ ratios and high z0/D’ ratios and that, in these cases, the more complex
In the attempt to overcome some drawbacks of the traditional approach and achieve more accurate
estimates of the shear strength many researchers have also addressed the study of continuum
structural models. As attention is usually focused on the ultimate load capacity of structural
members, the approach based on continuum models often makes use of the fundamental theorems
of limit analysis and considers simplified continuous stress fields to simulate the response of steel
and concrete. Ever since the first studies on this topic (e.g. Nielsen et al. 1978; Nielsen 1984) the
use of continuum models has gradually been perfected and allowed the achievement of important
results with moderately onerous computational tools. In particular, continuum models have often
been used to compensate for the lack of valid code provisions for the assessment of the resistance of
concrete members subjected to the combined action of forces including shear. In this very context,
this approach has been adopted by one of the Authors to define the ultimate capacity interaction
diagrams of reinforced (Recupero et al. 2003) and prestressed concrete members (Recupero et al.
2005) characterised by rectangular, T or I shaped cross-section and subjected to axial force, shear
The present paper constitutes the natural progress of these studies and leads to the evaluation of the
ultimate strength of reinforced concrete circular columns subjected to axial force, bending moment
and shear force. Unlike similar methods proposed in the past (Recupero et al. 2003; 2005), the
method described in this paper considers both truss and arch mechanisms. This method is in step
with the guidelines reported in the Fib Structural Concrete Textbook (2009) which, in the case of
structural elements subjected to axial forces, proposes the superposition of statically allowable
partial models.
The non-linear problem is described from the mathematical point of view and applied to a large
number of columns tested in the past. The comparison between the numerical and laboratory results
is drawn to assess the accuracy and reliability of the approach. The resisting shear forces resulting
from the proposed method are also compared with those from other methods in order to highlight
the advantages for adopting the proposed method instead of existing and more simplified ones.
The proposed method is based on the application of the static theorem of limit analysis and
considers both truss and arch actions. The mathematical programming problem resulting from the
application of the static theorem is nonlinear and defined by means of equilibrium equations and
constraint conditions for geometric and mechanical parameters. All these relations are reported later
assuming that the longitudinal axis of the reinforced concrete member is parallel to the z-axis of a
reference system and that the external shear force is oriented along the positive y-axis.
To achieve accurate evaluations of the ultimate strength of the column, the generic cross-section of
this member is divided into three parts, named F1, F2 and F3 (Fig. 1). In each of these parts the
response of the basic materials (concrete and steel) is defined by means of simplified stress fields
which are described later for truss and arch mechanisms separately considered. The stress-strain
constitutive behaviour of both concrete and steel is considered to be rigid-perfectly plastic; however,
while steel bars are assumed to resist both compression and tension, concrete is assumed to resist
compression only. The geometry of the zones F1, F2 and F3 of the cross-section is completely
defined once the position of the separation lines of the central part F3 is known. The position of
these two lines is determined through the angles, 1 and 2, that the x-axis forms with the segments
The column is endowed with longitudinal and transverse reinforcements. In particular, the area Asl
of the longitudinal steel bars is assumed to be distributed over the circumference passing through
the centroids of the longitudinal steel bars. The longitudinal reinforcement per unit of length of
this circumference is
Asl A
sl (10)
2 R c 2Rl
where R is the radius of the column cross-section, Rl is the radius of the circle passing through the
centroid of the longitudinal steel bars and c is the mechanical cover of the longitudinal
reinforcement, i.e. the distance between the external surface of the cross-section and the centroid of
the longitudinal bars. The geometry of the distributed reinforcement included in the zones F1, F2
and F3 is defined through the angles and subtended at the geometric centre by the circle arc
The transverse reinforcement is constituted by circular hoops orthogonal to the longitudinal member
axis and placed at a distance cs from the external surface of the column cross-section. The spacing
of hoops and the mechanical properties of steel of hoops are assumed to be constant along the
length of the column. The geometry of the transverse reinforcement included in the zones F1, F2 and
F3 is defined through the angles and subtended at the geometric centre by the hoop segment
where
rl R ( R c) (15)
rs R ( R cs ) (16)
The Authors hypothesize that in the outermost parts of the cross-section (called F1 and F2)
longitudinal reinforcement and concrete are subjected to “stress fields” parallel to the longitudinal
member axis and that these stresses are constant within the single part of the cross-section. In the
central part (called F3) longitudinal and transverse reinforcements are assumed to experience
stresses which are constant and parallel to the axis of the steel bars. Still in the central zone,
with respect to the longitudinal member axis. Similar stress fields have already been suggested by
many other researchers (e.g. Nielsen et al. 1978; Hsu 1993) and are particularly justified in
To define the equilibrium conditions relative to the cross-section parts named F1 and F2, let us
consider the elements E1 and E2 (Fig. 2a) obtained by cutting the reinforced concrete member by
means of two couples of planes which are parallel and orthogonal to the longitudinal member axis.
In addition, let us consider that those planes which are parallel to each other are set apart at an
infinitesimal quantity. The elements E1 and E2 have, therefore, size equal to dy and dz along the y
and z-axes and depth equal to 2Rcosalong the x-axis. Further, they are subjected to the stresses l
The equilibrium equation of the element E1 along the z-axis may be written as
l1 1 R cos c1 R cos 0 (17)
cos
where 1 is the equivalent normal stress due to the response of steel and concrete. It is worth noting
that in all the equations in this paper the normal stress of concrete is assumed to be positive when
compressive and that the normal stress of longitudinal and transverse reinforcements is assumed to
1 l1 c1 (18)
R cos 1 rl 2 sin 2
As is evident, this stress is not constant within the part F1 because it depends on the angle .
Equation (17) simplifies further when referring to elements which are within the mechanical cover
Similar mathematical expressions may be obtained with reference to the element E2 in the part F2 of
the cross-section. Specifically, the equilibrium along the z-axis may be expressed by means of the
relation
l 2 2 R cos c 2 R cos 0 (20)
cos
When referring to elements within the mechanical cover of the longitudinal steel bars, Equation (20)
In the central part (F3) of the cross-section let us consider two elements (E3 and E4) which are
obtained by cutting the reinforced concrete member by means of three planes parallel to the x-axis
(see Fig. 2b). Like E1 and E2, the elements E3 and E4 have depth equal to 2Rcosalong the x-axis.
First, let us focus attention on the element E3 which is obtained by cutting the member with one
plane parallel to the compressive stress of the concrete and with other two planes parallel and
orthogonal to the longitudinal member axis. This element is subjected to the stress l3 of the
longitudinal reinforcement, to the stress s3 of the circular hoops, to the tangential stress and to the
equivalent normal stress 3 of concrete and steel bars (Fig. 4a). The stress of the longitudinal and
transverse reinforcements is assumed constant in this part of the cross-section. However, unlike
rectangular or I-shaped cross-sections, in circular cross-sections the normal stress of the hoop is
inclined to the y-axis at an angle which depends on the position of the hoop cross-section
considered.
The equilibrium equations along the y and z-axes (see Fig. 4a) are
Asw
(transl. along the y-axis) s3 cos cos 2 R cos sin 0 (23)
s
2
(transl. along the z-axis) l 3 sin 2 R cos cos 3 2 R cos sin 0 (24)
cos
If the angle is expressed as a function of the angle , Equation (23) gives the shear stress value
Hence, the stress 3 may be defined by substituting Equation (25) into Equation (24)
Asw 1 rs sin
2 2
3 l 3 s3 cot 2 (26)
R cos 1 rl sin
2 2 s 2 R cos
As is evident in Equations (25-26), the equivalent normal stress 3 and the shear stress are not
To evaluate the normal stress c3 of concrete, let us consider the element E4 which is obtained by
cutting the member with two planes parallel and orthogonal to the compressive stress of the
concrete and with a plane parallel to the longitudinal member axis (Fig. 2b). This element is
subjected to the compressive stress c3 of concrete and to the component s3 of the stress of the
The constraint condition y = 0 applied to the element under examination states that
Asw
s3 cos c 3 2 R cos sin 2 0 (27)
s
and thus
where 0c 3 is the stress of concrete for =0. As evident from this equation the stress c 3 is
non-uniform in F3. In particular, it is maximum on the y-axis and symmetric with respect to this
axis.
Internal forces
The axial force of the generic cross-section is obtained by the following relation
N 1dA 2 dA 3 dA (29)
A1 A2 A3
N1 N2 N3
where N1, N2 and N3 are the axial force contributions of the parts F1, F2 and F3 and A1, A2 and A3 are
The contribution N1 to the axial force may be calculated by means of the relation
2 llim 2
cos
N1 1 2 R cos d l1 2 R d c1 2 R 2 cos d
2 2 2
(30)
1 1* 1 rl sin
2 2
1
where
and llim is the angle corresponding to the separation line passing through the longitudinal steel
The consideration of angles llim and 1* is necessary because the contribution of the longitudinal
reinforcement is possible only for values of the angle in the range llim to llim .
*2 2
cos
N 2 l 2 2 R 1 rl sin
2 2
dc 2 2 R 2
2
cos 2 d (33)
llim
where
1*
cos A 1
llim 2
cos sin
M 1 l 1 2 R d c1 2 R cos sin d
2 3 2
(37)
1* 1 rl sin
2 2
1
*2 2
cos sin
M 2 l 2 2 R d c 2 2 R cos 2 sin d
2 3
(38)
llim 1 rl sin
2 2
2
1*
cos sin A 1
M 3 l 3 2 R 2
d s 3 sw R 2 cot 2 sin cos 1 rs2 sin 2 d (39)
*2 1 rl 2 sin 2 s 2
2
V 2R cos d
2 2
(40)
1
1 A
V s 3 sw Rs cot 1 sin 1 cos 1 2 sin 2 cos 2 (41)
2 s
where Rs is the radius of the circular hoops. In particular, it is worth noting that if 1=-2=π/2 the
As explained by some researchers (Priestley et al. 1994; Kowalsky and Priestley 2000), in the
presence of external axial load some lateral force can be transmitted directly to the base of the
member through inclined compressive stresses (this type of arch effect is later named as type 1).
The magnitude of the resisting shear force produced by the external axial load is equal to the
horizontal component of the diagonal compression strut that forms between the top and bottom of
the column (Fig. 5a). However, as remarked by Priestley et al. (1994a) the contribution of this effect
to the axial resistance cannot be greater than the external axial load.
An arch effect may also exist without any contribution being made by the external axial load (this
type of arch effect is later named as type 2) (e.g. see Watanabe and Ichinose 1991). In this case a
diagonal compressive strut can directly transfer the lateral load to the base if the member is
endowed with longitudinal reinforcement able to resist the vertical component of the diagonal
compressive force. The diagonal compressive strut can form whether the shear force is applied to
the lateral surface of the member (Fig. 5b) or to the geometric centre of the cross-section (Fig. 5c).
It is worth noting that the arch effect type 2 can develop both in the presence and in the absence of
axial load. Owing to this, in the subsequent analyses both types of arch actions are simultaneously
considered.
To include the arch effect type 1 in the proposed model, the Authors assume that in single bending
columns the resultant of the diagonal strut passes through the point of application of the external
axial load at the top of the column and through the centre Gc2 of the zone F2 at the base of the
column (Fig. 5a); even if not shown in the figure, in double bending columns the strut is assumed to
enter and leave the column through the centre of the zone F2 at the column top and bottom.
The internal forces (N’, M’, V’) produced by this type of arch action are:
2
2
M y dAc 2 R
' '
c2
'
c2
3
cos 2 sin d (43)
A2 2
2
where 'c 2 is the vertical stress produced in concrete at the toe of the column base and ' is the
angle of inclination of the strut with respect to the longitudinal member axis. The latter parameter is
where yGc 2 is the y-coordinate of the centroid of the area A2 and LV is the shear span of the
member. In regard to the possible values of the normal stress 'c 2 , it should be noted that the
contribution of this arch mechanism to the axial resistance cannot not be greater than the external
To describe the arch effect type 2, the Authors consider the case in which the shear force is applied
to the lateral surface of the member separately from the case in which the shear force is applied to
the cross-section. Referring to the former case (frequent in laboratory tests), the axis of the strut is
assumed to pass through the centroid of the zone named F2 and the intersection point between the
directions of the lateral force and the resultant of the tensile stresses in the longitudinal bars of the
zone named F1 (Fig. 5b). The geometric characteristics of this strut are completely defined through
the geometric properties of the concrete in the zone F2 and the longitudinal reinforcement in the zone
F1. The internal forces (N’’, M’’, V’’) produced by this type of arch action are
llim 2
cos sin
M y dAl y dAc 2 R d 2 R cos 2 sin d
'' '' '' '' 2 '' 3
l1 c2 l1 c2 (47)
A1 A2 1* 1 rl sin
2
2
2
where ''c 2 is the vertical compressive stress in the concrete of the zone F2, l''1 is the tensile stress
in the longitudinal bars of the zone F1 and '' is the angle of inclination of the strut with respect to
R
1 rl 2 sin 2
d
2 1*
'' arctan c 2 arctan 2
llim (49)
LV LV LV cos
cos 2 d d
1 rl sin
2 2
2 1*
where yGl1 is the y-coordinate of the centroid of the longitudinal reinforcement in the zone F1.
In the case the shear force is applied to the cross-section the strut is assumed to pass through the
centroid of the compression zone at the column base and the centroid of the member cross-section
** 2
cos sin
M y dAl y dAc 2 R d c'' 2 2 R3 cos 2 sin d
'' '' '' '' 2
l3 c2 l3 (51)
A3 A2 ** 1 rl sin
2
2
2
In Equation (51), ** is the angle which identifies the area of the longitudinal bars subjected to the
stress l''3 due to the arch effect type 2. The angle ** is given by the following relation supposing
that the longitudinal bars subjected to the stress l''3 are symmetric with respect to the longitudinal
2
2
cos sin d
yGc 2 R 2
'' arctan arctan 2 (54)
LV LV
cos 2 d
2
MECHANICAL AND GEOMETRIC CONSTRAINT CONDITIONS
The constraint conditions reported in this section refer to geometric and mechanical parameters. In
particular, in regard to the discretization of the generic cross-section the solution of the
programming problem must ensure that the zones F1 and F2 do not overlap. To this end, angles 1
1 2 (55)
As evident from Equation (28), diagonal stresses caused by truss action can develop in the concrete
of a layer orthogonal to the bending plane and belonging to the central part F3 if the circular hoop
intersects this layer. The maximum extension of the central zone is thus equal to the diameter of the
-lim
s 1 s
lim
(56)
-lim
s 2 s
lim
(57)
circular hoop with minimum y-coordinate and is calculated by means of the relation
lim
s arcsin 1 rs (58)
The constraint conditions involving mechanical parameters refer to the simplified stress fields
considered for concrete and steel bars. The relations reported for longitudinal and transverse
reinforcements are simple and rigorous because the stresses produced by truss and arch actions are
parallel to the steel bars. The maximum normal stresses must be lower than the yield strength of
f yl l 2 f yl (60)
f yl l 3 l''3 f yl (61)
f yw s 3 f yw (62)
fyl being the yield strength of the longitudinal reinforcement and fyw the yield strength of the circular
hoops.
Different and slightly approximate constraint conditions are considered for concrete in zones F1, F2
and F3. Specifically, in zones F1 and F2 the normal stress of concrete is assumed to be positive and
bounded above by the compressive strength of concrete fc1 , as shown in the following inequalities
0 c1 f c1 (63)
As evident in the last relation, to simplify the solution of the nonlinear programming problem, the
maximum normal stress considered in the zone F2 is assumed equal to the sum of the maximum
normal stresses caused by truss and arch effects, i.e. the three maximum normal stresses are added
In the central zone F3, instead, the normal stresses of concrete must be positive and bounded above
by the reduced compressive strength of concrete under biaxial state of stress fc2 . The normal
stresses relative to this zone of the cross-section must verify the relations
c 3,min 0 (65)
c 3,max ''c 2 cos 2 '' f c 2 (shear force applied to the lateral surface of the column) (66)
where c 3,min and c 3,max are the minimum and maximum diagonal stresses produced by truss
action in the central zone of the cross-section. The first of these relations is intended to ensure that
the minimum normal stress is positive. It is worth noting that in this relation only the contribution
due to truss action is considered, i.e. to be on the safe side, no compressive stress produced by arch
mechanism is added. As apparent from Equation (28), the stress c 3,min produced by truss action is
developed on either separation line of the central zone F3. The calculation of the maximum normal
stress considers, instead, the contributions of both truss and arch actions. In particular, the normal
stress c 3,max produced by truss action is obtained in the concrete layer orthogonal to the bending
plane and closer than others to the geometric centre of the cross-section. If the shear force is directly
applied to the lateral surface of the member the principal stress produced by the arch mechanism
type 2 is added to the truss contribution because the strut caused by this arch mechanism in squat
columns is often characterised by fairly high angles of inclination with respect to the longitudinal
member axis. Therefore, this strut entirely crosses the central part of the member close to the
column base, where the bending moment is practically equal to the maximum value. Conversely,
the normal stress produced by the other arch actions considered in this paper are not included in the
inequality because the strut produced by these arch actions is generally characterised by low angles
of inclination with respect to the longitudinal member and thus crosses the zone F3 in a part of the
column where the bending moment is low. Similarly to the inequality (64), the approximate
maximum value of the compressive stress in F3 is calculated ignoring the difference between the
Still with reference to the arch effects the solution of the mathematical problem must ensure that the
'c 2 0 (68)
''c 2 0 (69)
and that the contribution of the arch effect type 1 to the resisting axial force is not greater than the
external axial load Nexp (Priestley et al. 1994a; Kowalsky and Priestley 2000), i.e.
Further, to avoid that in the longitudinal reinforcement the normal stresses caused by truss and arch
effects may have opposite sign the following relations must be verified
l 3 l''3 0 (72)
Finally, to consider that the angle of inclination of failure cracks in the zone F3 does not coincide
with that of first cracks and depends on geometric and mechanical parameters, the cotangent of the
angle of inclination of the compressive stress c 3 is assumed to be variable between two minimum
To verify the reliability and accuracy of the proposed method, the maximum shear forces resisted
by 67 columns are compared with those derived from the application of the numerical method. The
columns under examination were tested in the past in either single or double bending under a
constant value of the axial force. The laboratory tests and the relative results are described in detail
in Ang et al. 1985, 1989; Priestley et al. 1994a, 1994b, 1994c; Kowalsky and Priestley 2000; Stone
and Cheok 1989; Wong et al. 1990, 1993; Chai et al. 1991; Ascheim and Moehle 1992; Priestley
and Benzoni 1996; Kunnath et al. 1997; Hose et al. 1997; Lehman and Moehle 1998; Calderone et
al. 2000; Berry et al. 2004. Some geometric and mechanical characteristics of the members are
reported in Table 1 to show that the selected set of laboratory tests represents a suitable basis for the
validation of the proposed method. In particular, the columns considered are characterized by very
different geometric and mechanical properties as well as by different shear span to diameter ratios;
also, some of the columns were tested in the absence of the axial force. It should be noted that the
maximum internal forces recorded during the tests have been slightly corrected in Berry et al. (2004)
to take account of P- effects and that, in this paper, the modified values are considered for
comparison.
The ultimate internal actions of these columns have been numerically predicted by means of the
method described in the previous sections. The variables of the mathematical problem
are c1 , l1 , c 2 , l 2 , 0c 3 , l 3 , s , , 1, 2, (regarding the truss action), 'c 2 (arch action type 1)
and either ''c 2 l''1 (in case the arch action type 2 is due to an external shear force applied to the
lateral surface of the member) or ''c 2 l''3 (in case the arch action type 2 is due to shear force applied
to the cross-section). The programming problem is constituted by one objective function and
equality and inequality conditions. The total shear force (sum of the shear forces due to truss and
arch actions) is considered as the objective function of the mathematical programming problem and
is maximized to obtain the maximum positive value of this internal force. Therefore, the function to
be maximized is
where the truss and arch contributions to the resisting shear force are obtained by Equations (40, 44,
48, 52). As is evident from these equations the objective function is non-linear in the variables of
the problem.
The constraint conditions of the problem are represented by the equilibrium equation (23); the
equilibrium equation (27) calculated for =0 so that the stress c 3 in this relation is equal to the
variable 0c 3 ; inequalities (55-57) relative to the geometry of the parts F1, F2 and F3; inequalities
(59-72) relative to the stresses produced by truss and arch actions; inequalities (73-74) relative to
the angle of inclination of the stress c 3 in the truss mechanism; equations (46, 50) and two
equality conditions regarding the total internal forces of the base cross-section. The first of these
latter equality conditions states that the total axial force is equal to the assigned axial force Nexp, i.e.
The second states that the bending moment and shear force resulting from the numerical approach
are in the same proportion as the maximum internal forces of the laboratory test (Mexp and Vexp), i.e.
force are obtained by Equations (29, 36, 40, 42-44, 47-48, 51-52).
The mathematical problem is non-linear and, therefore, its solution depends on the starting values of
all the variables of the problem. To find an accurate estimate of the shear strength, the mathematical
problem is solved more than once with reference to different starting values of these variables. The
highest value of the shear force is assumed as the ultimate shear capacity of the member. The Solver
tool of the program Excel has been used to solve the single programming problem.
The mechanical properties of concrete and steel are assumed equal to those reported in papers and
thus derived in the past from laboratory tests. In particular, the concrete compressive strengths
considered in the constraint conditions are derived from the experimental values through the
following relations
f c1 f c (78)
f
f c 2 0.6 1 c f c (79)
250
In accordance with other researchers (Walther and Miehlbradt 1990; SIA 1989), the minimum and
maximum values of the cotangent of the angle θ are calculated assuming that the failure crack angle
may be 20-25 degrees lower or higher than the angle of inclination I of the first crack. From this
where I is the angle of inclination of the first crack with respect to the longitudinal axis of the
member (Norme tecniche per le Costruzioni 2008) and is the maximum excursion allowed for
the angle . The angle I is obtained from the relation cot I I where and are respectively
the shear stress and the principal tensile stress which develop at first crack on the chord parallel to
the neutral axis and passing through the centroid of the uncracked, homogeneous cross-section.
where A is the cross-sectional area of concrete and f ct 0.7 0.3 f c21 3 . The maximum value of the
cotangent of the angle of inclination of the compressive stress is given by the relation
In this relation the cotangent of the angle has been limited below to the value L/y3, where y3 is the
dimension of the zone F3 parallel to the y-axis, to consider that the direction of the compressive
stresses of the zone F3 must intersect the zones F1 and F2 (see Fig. 5). In both Equations (80, 82) the
parameter i.e. the maximum excursion allowed for the angle is assumed equal to 23.2°. So, in
the absence of external axial loads, I =45° and thus the values obtained by Equations (80, 82) are
0.4 and 2.5, as sometimes considered in codes (e.g. see Eurocode 2, 1993).
Vexp
RV (83)
Vnum
where Vnum is the ultimate shear force predicted by means of the proposed method and Vexp is the
maximum shear force recorded during the laboratory test. The values of the parameter RV are first
calculated by ignoring the arch effects and reported in Figure 6 as a function of the ratio M/VD (the
As shown in the figure, in the case of high values of the ratio M/VD the numerical results are in
accord with those obtained from the laboratory tests. Conversely, the accuracy of the numerical
results is poor for members with low values of the ratio M/VD. This finding is not surprising
because the resisting mechanism of members with low aspect ratios is chiefly characterized by
direct travel of the force to the support, i.e. by arch action. To highlight the influence of the arch
action, the values of the parameter RV are then calculated considering both truss and arch
mechanisms in the numerical approach. The resisting mechanisms resulting from the arch actions
type 1 and 2 are considered together with that resulting from the truss action. The comparison
between Figures 6 and 7 shows that in the case of high values of the ratio M/VD the results obtained
by means of the evaluation of the sole truss action are virtually equal to those achieved by means of
the evaluation of both truss and arch mechanisms (see also Table 1). On the contrary, the results
corresponding to low aspect ratios change remarkably when the arch effects are considered. Further,
as is evident in Figure 7, the values RV corresponding to low values of the ratio M/VD are only a
little higher than those corresponding to high values of the ratio M/VD, i.e. the accuracy of the
results is virtually independent of the value of the aspect ratio. The minimum and maximum values
of the parameter RV are 0.99 and 1.38 and the standard deviation is 0.086. Even if not shown in
figure, there is no appreciable trend with increasing external axial force or longitudinal and
transverse reinforcements. In particular, the negligible correlation between the parameter Rv and the
external axial force demonstrates that the method may be successfully applied to columns and
beams as well.
It should be noted that the proposed formulation of the problem is particularly suitable for
estimating the shear strength of the abovementioned columns because the transverse reinforcement
ratio is constant along the length of these columns. In case the transverse reinforcement ratio varies
along the length of the member the shear resistance should be theoretically verified in all the
cross-sections where the transverse reinforcement ratio changes. The stress verification in more
than one cross-section would make, however, the mathematical problem very complicated. This
complication is not justified at all if the transverse reinforcement ratio is only slightly different
along the length of the member because in this case a reliable estimate of the shear strength can be
obtained assuming that the transverse reinforcement ratio of the column is equal to the value at the
column base. If some greater approximation is accepted with regard to the modelling of the column,
the abovementioned complication can be avoided even if the transverse reinforcement ratio changes
very sharply along the length of the member. In this case the proposed formulation can still provide
a noticeable result if the transverse reinforcement ratio of the whole column is defined as
representative of the different transverse reinforcement ratios of the member, e.g. equal to the
average value.
The accuracy and reliability of the proposed method are finally compared with those of other
existing and less complicated methods. Specifically, the results described in the previous section are
compared with those deriving from the methods proposed by Ang et al. (1978), Priestley et al.
(1994a) and Kowalsky and Priestley (2000). The relations proposed by Turmo et al. (2009) are not
considered for comparison because no particular relation is suggested for the concrete contribution
to the shear strength. In all the comparison methods the shear force corresponding to the column
Table 2 reports the results which are deemed to be significant for the comparison, i.e. the results
relative to columns for which the predicted shear strengths are not all equal or virtually equal to the
shear force corresponding to the flexural capacity of the column. The mean value, the standard
deviation and the coefficient of variation of RV are also reported in the table.
As is evident from the latter two parameters, the proposed method (with arch action) is more
reliable than the other simplified methods. In all truth, the mean values of RV show that the
proposed method is less accurate than the other methods. However, this result is neither surprising
nor alarming. In fact, the predictions of the comparison methods are sometimes unconservative and
thus the higher accuracy is obtained at the expense of the structural safety. Further, it should be
noted that all the simplified methods considered here were adjusted in the past on the basis of most
CONCLUSION
The paper proposes a mathematical model for the evaluation of the strength of columns with
circular cross-section subjected to combined axial force, shear force and bending moment. The
model considers both truss and arch effects and hypothesizes simplified stress fields to simulate the
response of steel and concrete. The shear strength is evaluated by means of limit analysis.
The study leads to the following conclusions:
- the proposed method highlights the mutual influence of the internal forces in circular
reinforced concrete members and, in particular, explains the level of importance of the shear
- the comparison between experimental and theoretical results highlights that the proposed
model provides good estimates of the experimental results for low and high aspect ratios. The
accuracy and reliability of the method has been proven on many columns with different
- the comparison with some simplified methods available in the literature demonstrates the
The Authors underline that the described method can be applied to columns with other cross-sections.
This extension is not difficult to accomplish and only needs that the relations considered in this paper
REFERENCES
Ang Beng Ghee, Priestley M.J.N., Paulay T. 1985. Seismic shear strength of circular bridge piers.
Zealand.
Ang B.G., Priestley M.J.N., Paulay T. 1989. Seismic shear strength of circular concrete columns.
Ascheim M.A., Moehle J.P. 1992. Shear Strength and Deformability of RC Bridge Columns
Subjected to Inelastic Cyclic Displacement. Report n° UCB/EERC 92/04, Earth. Eng. Research
Berry M., Parrish M., Eberhard M. 2004. PEER Structural Performance Database. Pacific Earth.
Having Aspect Ratios and Varying Lengths of Confinement. PEER Report 2000/08.
Chai Y.H., Priestley M.J.N., Seible F. 1991. Seismic Retrofit of Circular Bridge Columns for
Eurocode 2. Design of concrete structures – Part 1-1: general rules and rules for buildings. European
Fib 2009. Structural Concrete Textbook on behaviour, design and performance – DCC Siegmar Kästl
Hose Y.D., Seible F., Priestley M.J.N. 1997. Strategic relocation of plastic hinges in bridge columns,
Structural Systems Research Project, 97/05, University of California, San Diego, La Jolla.
Kowalsky M.J., Priestley M.J.N. 2000. Improved analytical model for shear strength of circular
reinforced concrete columns in seismic regions. ACI Structural Journal, 97 (3): pp. 388-396.
Kunnath S.K., El-Bahy A., Taylor A.W., Stone W.C. 1997. Cumulative Seismic Damage of
Reinforced Concrete Bridge Piers. NISTIR 6075, Building and Fire Research Laboratory,
Lehman D.E., Moehle J.P. 1998. Seismic Performance of Well-Confined Concrete Bridge Columns.
Pacific Earthquake Engineering Research Center, PEER Report 1998/01, pp. 205.
Nielsen M.P., Braestrup M.W., Bach F. 1978. Rational analysis of shear in reinforced concrete
Nielsen M.P. 1984. Limit analysis and concrete plasticity. Prentice-Hall Series in Civil Engineering,
Norme Tecniche per le Costruzioni. Decreto Ministeriale 14/01/2008. Gazzetta Ufficiale n. 29,
Priestley M.J.N., Seible F., Xiao Y., Verma R. 1994a. Steel jacket retrofitting of reinforced concrete
bridge columns for enhanced shear strength- Part 1: Theoretical consideration and test design.
ACI Structural Journal; 91: pp. 394 – 405.
Priestley M.J.N., Seible F., Xiao Y., Verma R. 1994b. Steel jacket retrofitting of reinforced concrete
bridge columns for enhanced shear strength- Part 2: Test results and comparison with theory.
Priestley M.J.N., Verma R., Xiao Y. 1994c. Seismic shear strength of reinforced concrete columns.
Priestley M.J.N., Benzoni G. 1996. Seismic Performance of Circular Columns with Low
Longitudinal Reinforcement Ratios. ACI Structural Journal; 93 (4): pp. 474 – 485.
Priestley M.J.N., Seible F., Calvi M. 1996. Seismic design and retrofit of bridges. John Wiley &
Recupero A., D’Aveni A., Ghersi A. 2003. N-M-V Interaction Domains for Box I-shaped Reinforced
Recupero A., D’Aveni, A., Ghersi, A. 2005. N-M-V Interaction Domains for Prestressed Concrete
Stone W.C., Cheok G.S. 1989. Inelastic behavior of full-scale bridge columns subjected to cycling
loading, NIST BSS 166, Building Science Series, Center for building Technology, National
Turmo J., Ramos G., Aparicio A.C. 2009. Shear truss analogy for concrete members of solid and
Walther R., Miehlbradt M. 1990. Dimensionnement des structures en béton: Bases et technologie.
Wong Y., Paulay T., Priestley M.J.N. 1990. Squat circular bridge piers under multi-directional
Wong Y., Paulay T., Priestley M.J.N. 1993. Response of Circular Reinforced Concrete Columns to
Multi-Directional Seismic Attack. ACI Structural Journal; 90 (2): pp. 180 – 191.
NOTATION
Greek letters
** = min 1 ; 2
llim angle corresponding to the separation line passing through the longitudinal steel
lim
s angle corresponding to the separation line passing through the point of the circular
, angles subtended by the hoop segment delimited by the x-axis and the separation
efficiency factor
c1 ,c2 ,c3 normal stress of concrete in the zones F1 , F2 and F3 (truss action)
'c 2 , ''c 2 vertical compressive stress in the zone F2 (arch actions type 1 and 2)
principal tensile stress developed at first crack on the chord parallel to the neutral
axis and passing through the centroid of the uncracked, homogeneous cross-section
l 1 , l 2 , l 3 normal stress of the longitudinal reinforcement in the zones F1, F2 and F3 (truss action)
l''1 tensile stress in the longitudinal bars of the zone F1 (arch effect type 2)
Roman letters
M1 , M2 , M3 contributions of zones F1, F2 and F3 to the bending moment by means of truss action
N1, N2, N3 contributions of zones F1, F2 and F3 to the axial force produced by truss action
N1'' , N 2'' contributions of zones F1 and F2 to the axial force produced by the arch action type 2
Nexp axial force applied to the column during the laboratory test
rl R ( R c)
rs R ( R cs )
Rl radius of the circle passing through the centroids of the longitudinal steel bars
RV = Vexp/ Vnum
s spacing of the transverse reinforcement
yGl 1 y-coordinate of the centroid of the area of the longitudinal reinforcement in the zone F1
z0 distance from the centroid of the tensile forces to the centre of the cross-section
Table 1. Laboratory tests
Reference Label Setup (1) D L/D fyl fyw fc1 l s M/VD Nexp Vexp (2)
Vnum (3)
Vnum
2 2 2
(cm) (kN/cm ) (kN/cm ) (kN/cm ) (%) (%) (kN) (kN) (kN) (kN)
Ang et al. (4) 6 SB 60 1.5 43.6 32.8 3.01 3.20 0.51 1.50 0 392 171 296
Priestley et al. (5) C7A DB 183 3.0 46.9 32.4 3.07 2.53 0.17 1.51 592 787 246 600
Priestley and Benzoni (6) 2 DB 183 3.0 46.2 36.1 3.00 1.04 0.17 1.52 503 579 260 422
Ang et al. (4) 19 SB 60 1.5 43.6 32.6 3.44 3.20 0.38 1.52 432 437 134 384
Priestley and Benzoni (6) 1 DB 183 3.0 46.2 36.1 3.00 0.52 0.28 1.52 503 393 226 285
Ang et al. (4) 12 SB 60 1.5 43.6 32.8 2.86 3.20 1.02 1.53 359 526 340 403
Ang et al. (4) 18 SB 60 1.5 43.6 32.6 3.50 3.20 0.51 1.53 440 505 179 402
Ang et al. (4) 20 SB 70 1.8 48.2 32.6 3.67 3.20 0.38 1.80 807 487 157 428
Ang et al. (4) 21 SB 80 2.0 43.6 32.6 3.32 3.20 0.38 1.99 0 271 156 247
Ang et al. (4) 1 SB 80 2.0 43.6 32.8 3.75 3.20 0.51 1.99 0 321 207 263
Ang et al. (4) 24 SB 80 2.0 43.6 31.0 3.31 3.20 0.77 1.99 0 341 239 266
Ang et al. (4) 23 SB 80 2.0 43.6 33.2 3.23 3.20 0.76 2.00 0 333 240 266
Ang et al. (4) 2 SB 80 2.0 29.6 32.8 3.72 3.20 0.51 2.00 0 219 170 201
Ang et al. (4) 4 SB 80 2.0 43.6 31.6 3.06 3.20 0.51 2.00 0 289 198 246
Ang et al. (4) 22 SB 80 2.0 43.6 31.0 3.09 3.20 0.39 2.00 0 285 148 237
Ang et al. (4) 15 SB 80 2.0 43.6 32.6 3.48 1.92 0.51 2.00 0 230 154 181
Ang et al. (4) 5 SB 80 2.0 43.6 32.8 3.11 3.20 0.76 2.00 0 331 243 267
Ang et al. (4) 14 SB 80 2.0 42.4 32.6 3.37 3.24 0.51 2.00 0 316 204 251
Ang et al. (4) 7 SB 80 2.0 44.8 37.2 3.01 3.20 0.38 2.00 0 281 228 257
Priestley et al. (5) C5A DB 244 4.0 46.9 32.4 3.59 2.53 0.17 2.01 592 610 238 524
Priestley et al. (5) C1A DB 244 4.0 32.4 35.9 3.10 2.53 0.17 2.02 592 567 265 454
Priestley et al. (5) C3A DB 244 4.0 32.4 32.4 3.45 2.53 0.17 2.04 1779 718 344 572
Ang et al. (4) 16 SB 80 2.0 43.6 32.6 3.34 3.20 0.51 2.04 420 352 232 325
Ang et al. (4) 11 SB 80 2.0 44.8 37.2 2.99 3.20 0.51 2.08 751 407 266 355
Ang et al. (4) 13 SB 80 2.0 43.6 32.6 3.62 3.20 1.02 2.09 455 436 294 344
Vu et al. (7) NH4 DB 183 4.0 46.8 43.4 3.50 5.21 2.70 2.10 850 905 638 673
Wong et al. (8) 2 SB 80 2.0 47.5 34.0 3.70 3.20 0.47 2.11 1813 489 229 442
Ang et al. (4) 8 SB 80 2.0 44.8 37.2 2.87 3.20 1.02 2.12 721 445 311 357
Ang et al. (4) 10 SB 80 2.0 44.8 33.2 3.12 3.20 1.02 2.12 784 437 299 361
Wong et al. (9) 1 SB 80 2.0 42.3 30.0 3.80 3.20 1.42 2.14 907 461 328 379
Vu et. al (7) NH3 DB 183 4.0 42.8 43.0 3.94 2.41 1.14 2.15 970 510 365 402
Wong et al. (8) 3 SB 80 2.0 47.5 30.0 3.70 3.20 1.42 2.15 1813 579 380 444
Vu et.al. (7) NH1 DB 183 4.0 42.8 43.0 3.83 2.41 1.14 2.17 1928 535 421 467
Vu et al. (7) NH6 DB 183 4.0 48.6 43.4 3.50 5.21 3.04 2.23 1914 957 674 710
Ang et al. (4) 3 SB 100 2.5 43.6 32.8 3.60 3.20 0.51 2.50 0 276 196 222
Ang et al. (4) 17 SB 100 2.5 43.6 32.6 3.43 3.20 0.51 2.56 431 312 219 273
Ang et al. (4) 9 SB 100 2.5 44.8 37.2 2.99 3.20 1.02 2.75 751 364 252 285
Sritharan et al. (7) IC3 SB 180 3.0 46.1 43.4 3.30 1.92 0.81 3.14 400 433 318 337
Sritharan et al. (7) IC2 SB 180 3.0 44.8 43.1 3.46 1.92 0.54 3.14 400 411 300 331
Sritharan et al. (7) IC1 SB 180 3.0 44.8 43.1 3.14 1.92 0.54 3.17 400 387 295 323
Stone and Cheok (7) 2 SB 457 3.0 47.5 43.5 3.43 1.99 1.49 3.21 4450 2968 2355 2454
Calderone et al. (10) 328 SB 183 3.0 44.1 60.7 3.45 2.73 0.89 3.22 454 525 425 442
Stone and Cheok (11) N4 SB 75 3.0 44.6 44.1 2.44 1.98 1.41 3.24 120 63 57 59
Stone and Cheok (11) N5 SB 75 3.0 44.6 44.1 2.43 1.98 1.41 3.32 239 77 62 64
Stone and Cheok (11) N1 SB 75 3.0 44.6 44.1 2.41 1.98 1.41 3.39 120 59 55 56
Stone and Cheok (11) N2 SB 75 3.0 44.6 44.1 2.31 1.98 1.41 3.45 239 73 60 62
Lehman and Moehle (7) 407 SB 244 4.0 46.2 60.7 3.10 0.75 0.70 4.23 654 172 151 154
Lehman and Moehle (7) 415 SB 244 4.0 46.2 60.7 3.10 1.49 0.70 4.32 654 269 226 231
Lehman and Moehle (12) 430 SB 244 4.0 46.2 60.7 3.10 2.98 0.70 4.32 654 448 364 380
Henry et al. (7) 415s SB 244 4.0 46.2 60.7 3.72 1.49 0.35 4.53 654 259 214 226
Henry et al. (7) 415p SB 244 4.0 46.2 60.7 3.72 1.49 0.70 4.92 1308 277 239 245
Kunnath et al. (13) A4 SB 137 4.5 44.8 43.4 3.55 2.04 0.94 5.05 222 72 62 63
Kunnath et al. (13) A2 SB 137 4.5 44.8 43.4 2.90 2.04 0.94 5.10 200 74 59 60
Kunnath et al. (13) A5 SB 137 4.5 44.8 43.4 3.55 2.04 0.94 5.24 222 77 59 61
Kunnath et al. (13) A3 SB 137 4.5 44.8 43.4 2.90 2.04 0.94 5.25 200 75 57 58
Stone and Cheok (7) 1 SB 914 6.0 47.5 49.3 3.58 1.99 0.63 6.79 4450 1289 1161 1182
Stone and Cheok (11) N6 SB 150 6.0 44.6 47.6 2.33 1.98 0.68 6.93 120 30 27 28
Chai et al. (14) 3 SB 366 6.0 31.5 35.2 3.26 2.54 0.17 7.05 1779 207 178 196
Stone and Cheok (11) N3 SB 150 6.0 44.6 47.6 2.54 1.98 0.68 7.13 120 32 27 27
Hose et al. (15) SRPH1 SB 366 6.0 45.5 41.4 4.11 2.66 0.89 7.48 1780 285 236 240
Calderone et al. (10) 828 SB 488 8.0 44.1 60.7 3.45 2.73 0.89 9.30 454 172 155 156
Lehman and Moehle (12) 815 SB 488 8.0 46.2 60.7 3.10 1.49 0.70 9.40 654 130 106 107
Calderone et al. (10) 1028 SB 610 10.0 44.1 60.7 3.45 2.73 0.89 12.12 912 157 128 129
Lehman and Moehle (12) 1015 SB 610 10.0 46.2 60.7 3.10 1.49 0.70 12.39 654 80 81 81
Notes: (1) SB=single bending; DB=double bending; (2) proposed method without arch action; (3) proposed method with arch action; (4) Ang et al.
1989; (5) Priestley et al. 1994b; (6) Priestley and Benzoni 1996; (7) Berry et al. 2004; (8) Wong et al. 1993; (9) Wong et al. 1990; (10) Calderone et al.
2000; (11) Stone and Cheok 1989; (12) Lehman and Moehle 1998; (13) Kunnath et al. 1997; (14) Chai et al. 1991; (15) Hose et al. 1997
Table 2. Comparison with simplified methods
Reference Label M/VD RV (1) RV (2) RV (3) RV (4)
Notes: (1) proposed method with arch action; (2) method by Ang et al.
1989; (3) method by Priestley et al. 1994a; (4) method by Kowalsky and
Priestley 2000
1 2
(a)
F1 F2 y
F3
direction of the
external force
1 2
(b)
y
direction of the
external force
1 2
(a) (b)
direction of the E2 direction of the E4
external force dy external force
A A B B
E1 dz E3
y y
y1 y3 y2 y1 y3 y2
2R cos
E1 y y
E3
axis of
dy the element E1 dy
x x
LV ' ''
''
y y y y y y y y y
D D D
3.0
2.0
1.0
0.0
0 3 6 9 12 15
M/VD
3.0
2.0
1.0
0.0
0 3 6 9 12 15
M/VD