Documente Academic
Documente Profesional
Documente Cultură
net/publication/267765818
CITATIONS READS
8 476
3 authors, including:
Some of the authors of this publication are also working on these related projects:
Dear Hasan, I'm working on nonlocal models and limit analysis View project
All content following this page was uploaded by Dario De Domenico on 05 November 2014.
An advanced version of a recently developed numerical limit analysis procedure for the
prediction of peak loads and failure modes of steel-reinforced concrete elements is pro-
posed. The modified procedure allows to take into account possible yielding of reinforce-
ment thus capturing the actual behavior at the collapse of both steel and concrete. This
implies a finite element (FE) modeling of the reinforced concrete (RC) elements in which
concrete is governed by a Menétrey–Willam-type yield criterion, with cap in compres-
sion, while steel bars are governed by a von Mises yield criterion. The peak load of a
wide range of RC structures whose behavior at ultimate state is dominated either by
the concrete crushing or by the steel bars yielding is then predicted with a very good
accuracy.
Keywords: Reinforced concrete; FE-based limit analysis; two yield criteria formulation.
1. Introduction
Reinforced concrete (RC) structures, because of their widespread use in civil engi-
neering applications, have received great attention from the scientific community
giving rise to several analytical and experimental research studies, dealing either
with the constitutive modeling of the constituent materials of such structures or
with their mechanical characterization up to failure. Although plain concrete is not
a ductile material, the presence of longitudinal web or stirrups reinforcement might
render the confined concrete and consequently the RC structural elements quite
ductile thus justifying the applicability, for their analysis, of approaches based on
plasticity theory [Chen, 1982; Bažant, 1986]. The so-called direct methods belong
1450058-1
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
to such approaches embracing limit analysis, see e.g. Spiliopoulos and Weichert
[2013], or shakedown analysis, also endowed with apposite constitutive assumptions
[Fuschi, 1999]. Several contributions concern direct methods based on finite ele-
ments method (FEM) in conjunction with optimization algorithms such as linear
[Sloan, 1988] and nonlinear programming [Le et al., 2010], [Garcea and Leonetti,
2011]. These approaches indeed do not allow the treatment of post elastic phenom-
ena that may arise in concrete structures, such as localization, fracture, damage,
creep, etc. and that can be faced by coupling plasticity with fracture or damage
mechanical theories within step-by-step analyses (see e.g., Lubliner et al. [1989],
Lee and Fenves [1998] and Zhang et al. [2010]); however, they can give information
on the behavior at limit (collapse) states of structures, which are very useful for
design purposes. In this context, with an approach not based on an optimization
algorithm but grounding on sequences of FE elastic analyses, it has to be inserted
to the present study.
The promoted approach belongs to a wider research program started by the
authors in the context of laminates of fibres reinforced polymers [Pisano and Fuschi,
2007; Pisano et al., 2012, 2013b; De Domenico et al., 2014] and then extended in
the context of RC structures with reference to a Menétrey–Willam (M–W)-type
yield criterion with cap in compression [Pisano et al., 2013a, 2013c; De Domenico
et al., 2014]. In both the latter quoted studies, two limit analysis methods, namely
the linear matching method (LMM), and the elastic compensation method (ECM),
have been applied under the simplified hypothesis where reinforcements behave as
indefinitely elastic. This hypothesis implies, as explicitly declared, the limitation of
applicability to those structures whose behavior, at incipient collapse, is dominated
by crushing of concrete, being the steel bars far from yielding which is the case of
the so-called over-reinforced structures. To overcome this limitation and to improve
the overall modeling of the RC structural elements at collapse, the present study
proposes an advanced version of the above numerical limit analysis procedure for
dealing with possible steel bar yielding. To this purpose, the paper proposes a two
yield criteria limit analysis formulation in which concrete is governed by a M–W-
type yield criterion with cap in compression, while steel bars are governed by a von
Mises yield criterion. This allows, actually, to predict also the peak loads of those
structures in which the steel reinforcements play an important role in determining
the post elastic behavior of the RC elements that is the case of the so-called under
reinforced elements. The effectiveness of the proposed refined approach, also with
respect to the previous one, is shown by comparing the obtained numerical results
with the ones experimentally detected on several RC beams and slabs.
The paper is organized as follows: after this introductory section, Sec. 2 recalls
few analytical expressions of the von Mises yield criterion and of the M–W-type yield
criterion enriched with a cap in compression to set the constitutive assumptions;
Sec. 3 gives the principles of the LMM and ECM employed in the limit analysis
procedure with reference to both steel and concrete; Sec. 4 gives the results obtained
1450058-2
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
by analyzing large-scale RC beams and slabs (14 in all) and compares these results
with the corresponding experimental tests; finally Sec. 5 draws the conclusions.
1450058-3
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
1450058-4
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
where ε̇cj = λ̇∂f /∂σj are the components of the outward normal to the yield surface
(with λ̇ > 0 a positive scalar multiplier); σjY are the stresses at yield associated
to given compatible strain rates ε̇cj ; u̇ci are the related displacement rates. p̄i are
the surface force components of the reference load vector p̄ acting on the external
portion ∂Vt of the body surface.
More precisely, the LMM is an iterative procedure involving one sequence of
linear analyses on the structure made, by hypothesis, of a linear viscous fictitious
material with spatially varying moduli and imposed initial stresses. At each iteration
an adjustment of the fictitious moduli is carried out so that the computed fictitious
stresses, say σj (where stands for linear viscous), are brought on the yield surface
at a fixed strain rate distribution, say ε̇j . Using a geometrical interpretation of the
procedure this simply states that at each point where the solution is computed (i.e.,
at each Gauss point (GP), within a FE modeling) the complementary dissipation
rate referred to the fictitious viscous material, say W (σj ) = const ., must be brought
to be tangential to (i.e., to match) the yield surface at a stress point σ Y whose
external normal is ε̇ . Indeed, for the formal analogy existing between the linear
viscous problem and the linear elastic problem, the strain rates ε̇ to be interpreted
as ε̇c , can be evaluated as linear elastic strain rates, viewing W (σj ) = const as
the related complementary energy equipotential surface. Operatively, the matching
procedure performed at generic GP starts with the search of a stress point on the
yield surface having an assigned strain rate ε̇ ≡ ε̇c as outward normal. Obviously,
this point is unique only if the yield surface is strictly convex, which is then an
essential requisite to apply the matching. All the above allows one to define a
collapse mechanism, (ε̇cj , u̇ci ), the related stresses at yield, σjY , and, consequently,
by Eq. (7), to evaluate an upper bound to the collapse load multiplier. In the
following, the LMM is applied simultaneously to steel bars and to concrete.
The fundamental concepts of the LMM were originally proposed by Ponter and
Carter [1997] specifically with reference to von Mises materials. In the principal
stress space the von Mises yield surface is represented by a cylinder, indefinite along
the hydrostatic axis, which does not satisfy the required condition of strict convexity.
However, the procedure can be carried on with referenceto the deviatoric plane
2
where the von Mises criterion is a circle of radius ρy = 3 σy while, considering
2
that ρ = 2J2 , Eq. (1) can be rewritten as:
3 2
f (ρ) = ρ − σy2 . (8)
2
Under the hypothesis of incompressible material, the complementary energy
potential function for a von Mises material in the deviatoric plane can be also
1450058-5
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
'
1
W , E (0 ) W (0)
(0)
(0)
V-M y L
surface M
W (1) , E (1) W (0) O
' '
3 2
Fig. 1. Geometrical sketch, in the deviatoric plane, of the matching procedure performed on von
Mises yield surface from iteration #0 to #1 at the generic GP. QL = stress point pertaining to
ε̇(0) ; QM = stress (matching) point at yield of
the fictitious effective linear strain rate solution ¯
given normal ¯ε̇(0) assumed as strain rate at collapse ¯ε̇c(0) .
written as:
3ρ2
W (ρ) = (9)
4E
which represents a circle concentric to the von Mises one, circumstance that signif-
icantly simplifies the matching procedures.
In fact, with reference to Fig. 1 and for a given initial value of the fictitious Young
modulus E, say E (0) , the linear solution at the generic GP can be represented by the
effective stress point QL ≡ σ̄ (0) , having outward normal ¯ε̇(0) on the complementary
3ρ2
energy equipotential function W (0) (ρ) = 4E (0) = const.
To find the corresponding matching point QM on the yield surface (having out-
ward normal ¯ε̇(0) ) it is sufficient to move along the radius QL -O that is to rescale
the complementary energy surface by modifying the Young elastic modulus E (0) in
such a way that W (0) coincides with the von Mises circle. To this purpose, if W (1)
denotes such rescaled complementary energy surface, i.e.,:
3ρ2
W (1) (ρ) =
= W̄ (0) = const., (10)
4E (1)
by comparing Eq. (8) with Eq. (10) it appears evident that the two surfaces can be
brought to be coincident if
σy2 σy2
E (1) = = . (11)
2W̄ (0) σ̄ (0) ¯ε̇(0)
Moreover, by interpreting the effective strain rate ¯ε̇(0) as effective strain rate at
collapse, i.e., on posing ¯ε̇(0) = ¯ε̇c = (0)
σy
, Eq. (11) becomes:
E
σy
E (1) = E (0) (12)
σ̄ (0)
which provides the updating matching formula for the elastic modulus. The elastic
modulus E (1) is then used to perform a new elastic analysis, as in fact the stresses
at yield have been determined geometrically and so they do not meet the global
1450058-6
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
equilibrium conditions with the load acting at the first analysis. The procedure to
determine the upper bound PUB has to be performed iteratively. At the generic
iteration, say k − 1, and for each GP in the mesh, Eq. (12) generalizes as:
σy
E (k) = E (k−1) (k−1)
(13)
σ̄
which, in the case of uniaxial stress condition, simplifies as:
σy
E (k) = E (k−1) (k−1) . (14)
σ1
The rationale of the LMM expounded with reference to von Mises materials can be
applied considering other materials obeying to more complicated yield conditions. In
practice, this extension is not straightforward and in the case of concrete addressed
here essentially requires the solution of an intricate geometrical problem. The key
idea is always that the kinematic solution referred to a fictitious material, is forced
to represent, at the current GP of the current element, a collapse mechanism while
the fictitious stresses are brought on the M–W-type yield surface so identifying with
the pertinent stresses at yield. The problem has been deeply discussed in a recent
paper by the authors Pisano et al. [2013a] and, hereafter, it is summarized with the
aid of its geometrical interpretation.
In the H–W coordinate, the complementary energy equipotential surface for
concrete depends on the bulk modulus K, on the shear elastic modulus G as well as
on initial stresses, ξ¯ and ρ̄. For an arbitrary choice of the initial elastic parameters
and of the initial stresses, the above complementary energy surface can be given
the shape:
(ξ − ξ¯(0) )2 (ρ − ρ̄ (0) )2
W (0) (ξ, ρ) = (0)
+ . (15)
6K 4G(0)
With reference to Fig. 2, Eq. (15) represents, in the principal stress space, a prolate
spheroid having semi axes proportional to the elastic parameters (G (0) , K (0) ) and
coordinate of the center depending on the initial stresses (ξ¯(0) , ρ̄ (0) ). Similarly to
the von Mises material, the elastic moduli and the initial stresses can be modified
in such a way that the fictitious linear solution can be interpreted as the solution at
yield. In Fig. 2, the initial fictitious linear solution computed at the generic GP can
be located at the stress point PL , having outward normal ε̇(0) on the complementary
energy equipotential surface.
The updating of the fictitious moduli to values K (1) , G(1) and of the initial
(1) (1)
stresses to values (ξ¯(1) , ρ̄x , ρ̄y ) is carried out in such a way that the spheroid
is modified in shape and position, that is the complementary energy surface of the
fictitious material is modified so that PL is brought onto the point PM of normal
ε̇(0) . Due to the strict convexity of the M–W surface, the point PM is uniquely
determined by the given normal ε̇(0) . Once again, such outward normal at PM can
then be viewed as an ε̇c(0) , i.e., as a strain rate at collapse, the related displacement
rates u̇c being those pertaining to the collapse mechanism while stress coordinates
1450058-7
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
(0) c (0)
1
(1)
2
(1)
M-W-type
yield surface O
4G (1)
6K (1) 3
Y (0) Y (0)
of PM (ξ Y (0) , ρx , ρy ) are the associated stresses at yield. Moreover, on taking
into account a sufficient condition for convergence of the iterative procedure [Ponter
et al., 2000], at matching W has to be tangential to the M–W surface in PM but
it has to be also completely outside the yield surface. These constraints bring to
a nonlinear system of five equations [see e.g., Pisano et al., 2013a] whose solution
provides the searched matching point.
When, at the generic iteration, (k − 1), and for each GP of the discrete model
of the whole structure (i.e., within the FEs describing concrete and steel bars), the
stresses at yield corresponding to the matching points QM and PM are known, it is
(k)
possible to compute an upper bound multiplier, say PUB , by applying Eq. (7) that
1450058-8
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
particularizes as follows:
Y (k−1) c(k−1) Y (k−1) c(k−1)
(σy ¯ε̇c(k−1) + ξ Y (k−1) ε̇v
c(k−1)
(k) V
+ ρx ε̇dx + ρy ε̇dy )dV
PUB = c(k−1)
.
∂Vt p̄i u̇i d(∂V )
(16)
The iterative procedure is carried on until the difference between two subsequent
PUB values is less than a fixed tolerance.
1450058-9
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
'
1
( k 1)
1
( k 1) ( k 1)
e R
y
( k 1)
von Mises 2
yield surface O
' '
3 ( k 1) 2
n
(a)
e( k 1) ( k 1)
e R
e( k 1)
1
Y ( k 1) 1
1
e( k 1)
e( k 1)
Y ( k 1) n
2 2 Y ( k 1) Y ( k 1)
e R
2
M-W-type Y ( k 1)
yield surface n
b O v
(b)
Fig. 3. Geometrical sketch of the ECM at iteration #(k − 1) of the current sequence and for the
generic element within the mesh. (a) Effective stresses points in the deviatoric plane within the steel
bar elements #1, #2, . . . , #e, . . . , #n, with σ̄R “maximum stress” among all the bar elements; (b)
stresses points in the principal stress space within the concrete elements #1, #2, . . . , #e, . . . , #n,
with PR “maximum stress” among all the concrete elements.
(k−1)
the convergence rate increases if the square of the updating ratio, σy /σ̄#e , is used
so that the discussed Young’s modulus updating formula becomes:
2
(k) (k−1) σy
E#e = E#e (k−1)
. (17)
σ̄# e
The same reasoning can be followed to apply the ECM to concrete FEs. With refer-
ence to Fig. 3(b), at the generic iteration (k−1), the (element) elastic stress solution,
1450058-10
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
after transformation in the H–W coordinates, locates in the principal stress space a
e(k−1) Y (k−1)
stress point, say P# e , while P# e denotes the corresponding stress point at
−→ e
−→ e
yield (i.e., on the M–W-type surface), measured on the direction OP# e |OP# e |. In
the same figure are reported also other stress points, representing the average stress
elastic solution within the elements #1, #2, . . . , #e, . . . , #n. If the elastic solution
−→ e (k−1) −→ Y (k−1)
at the #eth element is such that |OP# e| > |OP# e| then the element’s
Young’s modulus is reduced according to the formula:
−→ 2
Y (k−1)
(k) (k−1) |OP# e|
E#e = E#e −→ e (k−1) (18)
|OP# e|
which is in all similar to Eq. (17). Assuming that the concrete Poisson ratio v
remains constant, the updating of the Young’s modulus by Eq. (18) is equivalent
to modify the bulk modulus K#e and shear modulus G#e by the same reducing
factor.
Within the current sequence, after the above described modulus variation on
steel bars and concrete, the maximum stress values have to be detected in the whole
(k−1)
FE mesh. Precisely, in the bar elements the “maximum effective stress”, say σ̄R ,
(that is the stress point farthest away from the von Mises surface) is searched.
Likewise, among all the concrete elements, the “maximum stress” (that is the stress
(k−1)
point farthest away from the M–W-type yield surface) say PR , is also searched.
(k−1) −→ (k−1) −→Y (k−1)
If σ̄R is greater than σy or |OPR | is greater than |OPR | (see again
Fig. 3) a new FE analysis is performed within the current sequence keeping the
(k)
applied load fixed but with the updated E#e given by Eqs. (17) and (18). The
iterations are carried on, inside the given sequence, until all the stress points just
reach or are below their corresponding yield values, which means that an admissible
stress field has been built. Increased values of loads are then considered in the
(s)
subsequent sequences of analyses, each one with an increased value of PD , until
further load increase does not allow the stress points PR and σ̄R to be brought below
yield by the redistribution procedure. A PLB load multiplier can then be evaluated
(s)
at last admissible stress field attained for a maximum acting load PD p̄i , say at
s = S, and at last FE analysis, say at k = K, as the minimum between the two
values:
P (S) −→ (K) P (S)
PLB = min σy (K)
D
, |OPYR | D
. (19)
σ̄ −→ (K)
R |OPR |
A final general remark concerns the elastic analyses performed in both LMM
and ECM. They can be carried out by any commercial FE code. In the following
applications, the ADINA code has been used while a Fortran main program has
been created to drive the FE analyses within the sequences.
1450058-11
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
1450058-12
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
P Ptot
2
h
ux = 0
roller support
b /2 uz = 0
L0
L/2
rocker support Ptot = 100 kN
y at mid-span for all specimens
x
z
(a)
20 20 20
2T8 2T20 2T16
H1 S1 E1
while the fourth one, namely A3, is a part of the Vecchio and Shim [2004] test pro-
gram. In particular, beam A3 has been already studied by Pisano et al. [2013a] and
is here reconsidered to highlight the better performance of the proposed two yield
criteria approach compared to that presented in the above quoted paper. Figure 5
shows the mechanical scheme used for the simply supported beams; only half speci-
men is modeled due to symmetry with respect to the longitudinal direction. For all
the considered specimens, P denotes the load multiplier while P̄tot , assumed equal
to 100 kN, denotes the reference load. The beams are strengthened with different
arrangement of internal steel bars and stirrups, as reported in Table 1 and illus-
trated in Figs. 4(b) and 5(b) which also provide their geometrical details. Finally,
1450058-13
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
PPtot
b /2 uz = 0
L0
y L
Ptot = 100 kN
x for all specimens
z
(a)
20 20
2R6 2R6
MD1.3−A90 MD2.1−A90
380 380
R8@100 80
R8@100 80
4MD20 4MD25
240 20 240 20
280 280
20
3M10 50
2R6
A3
T0.2 −A135
D4@168
552
380
91.5
R8@100
2M25
4M30 64
2T12
64
240 20 178
280 305
(b)
Fig. 5. Mechanical model of the analyzed simply-supported beams MD1.3-A90, MD2.1-A90,
T0.2-A135, A3: (a) geometry, loading and boundary conditions; (b) cross section geometry (dimen-
sions in mm) with reinforcement arrangement (after Lau and Pam [2010] and Vecchio and Shim
[2004]).
1450058-14
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
the relevant material properties for reinforcement and concrete are given in Tables 2
and 3, respectively.
Among the seven slabs, once again prototypes with different geometrical config-
urations, reinforcement arrangements, loading and boundary conditions are inves-
tigated. The first three slabs, namely specimens SL15, SL30 and SL45, are taken
from the experimental work of Breveglieri et al. [2012]. These slabs are continuous-
supported and comprise two equal spans each of which are subjected to the same line
1450058-15
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
load P p̄/2 (with p̄ reference line load) at the two mid-spans. Using the same sym-
metry considerations made for continuous-supported beams, (with respect to the
longitudinal axis x and to the central support in the z direction), only a quarter of
each slab has been analyzed, as sketched in the mechanical scheme of Fig. 6(a), which
Pp
ux = 0 S1'
roller support
sagging region
S 2' S1
L1
z b
hogging region
x S2
L1 L/2
y rocker support
at slab center
(a)
z z
26 2 φ12 SL15 (S1 - S1' ) 26 5 φ12 SL15 (S2 - S'2 )
68 120 68 120
26 y 26 y
4 φ12 3φ8 53.8 26 2 φ12 1φ8 80.75 26
375 375
z z
26 2 φ12 SL30 (S1 - S1' ) 26 4 φ12 SL30 (S2 - S'2 )
68 120 68 120
26 y 26 y
4 φ10 3φ12 53.8 26 2 φ10 1φ12 107.7 26
375 375
z z
26 2 φ10 SL45 (S1 - S1' ) 26 3 φ10 2 φ8 SL45 (S2 - S'2 )
68 120 68 120
26 y 26 y
6 φ12 1φ8 53.8 26 2 φ12 1φ8 80.75 26
375 375
(b)
Fig. 6. Mechanical model of the analyzed continuous-supported slabs SL15, SL30 and SL45: (a)
geometry, loading and boundary conditions; (b) cross section geometry (dimensions in mm) with
reinforcement arrangement (after Breveglieri et al. [2012]).
1450058-16
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
reports also constraints and loading conditions. Three additional slabs, labeled as
S1, S2 and S3 refer to the report of Gilbert and Nejadi [2004], and one more slab,
indicated as prototype C-1, refers to the paper of Al-Rousan et al. [2012]. Slabs S1,
S2, S3 and C-1 are simply-supported and are subjected to two equal line loads P p̄/2
symmetrically placed about mid-span. In this case, only half of the slab has been
modeled exploiting the symmetry with respect to x as shown in Fig. 7(a) which
report geometrical data constrains and loading conditions. For all the examined
examples, the resultant of the reference line load is assumed equal to 100 kN.
Table 4 provides the geometrical details of the slabs together with the arrange-
ment of internal steel bars, (see also Figs. 6(b) and 7(b)). Table 5 gives the properties
Pp
2
Pp
2
z t
x S1'
y
S1 L1
b
L2
L1 L
(a)
400 400
z z
40 320 40 40 160 160 40
S1 S2
130 155 130 155
25 y 25 y
2N12 3N12
400 600
z z
40 106.6 106.6 106.6 40 50 125 125 125 125 50
S3 C -1
130 155 100 125
25 y 25 y
4N12 5 #4 #3@ 300 mm c/c
(b)
Fig. 7. Mechanical model of the analyzed simply-supported slabs S1, S2, S3 and C−1: (a) geom-
etry, loading and boundary conditions; (b) cross section geometry (dimensions in mm) with rein-
forcement arrangement (after Gilbert and Nejadi [2004], Al-Rousan et al. [2012]).
1450058-17
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
of the reinforcement, while Table 6 reports the concrete properties. It has to be noted
that in Table 3 as well as in Table 6 the uniaxial compressive strength, fc , when
not directly available, has been derived from the compressive strength on cubes fcu
as fc = 0.83fcu according to Eurocode 2; the uniaxial tensile strength value has
been computed as ft = 0.33 fc as suggested by Bresler and Scordelis [1963], and
0.3
the elastic concrete modulus has been assumed as Ec = 22(fc /10) according to
again Eurocode 2. Moreover, for what concerns the choices related to the adopted
concrete constitutive model, the value of the eccentricity parameter e of the M–W-
type yield surface has been evaluated by the expression e = [2 + ft /fc ]/[4 − ft /fc ],
the ft /fc being assumed as a measure of the material brittleness. Finally, the value
1450058-18
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
√
of ξv can be expressed as ξv = 3fc /m, while ξa and ξb have been set equal to
ξa = 0.7923 fc and ξb = 1.8964 fc as suggested by Li and Crouch [2010].
5. Numerical Results
The LMM and ECM have been applied to all the specimens described in Sec. 4. In
order to highlight the improvements achieved by using the present two yield criteria
approach, upper and lower bounds to the peak load multiplier that were obtained
assuming bars as indefinitely elastic are computed. The obtained results are sum-
marized in Tables 7 and 8 for beams and slabs, respectively. In such tables, for
each analyzed specimen, are given: the peak load multiplier experimentally detected
PEXP ; the predictions in terms of upper and lower bound values, PUB and PLB , as
given by the present approach; the predictions, shown at columns labeled PUB eb
and PLB eb , respectively, obtained under the hypothesis of indefinitely elastic behav-
ior of the bars. Finally, Table 9 reports the errors (in %) computed as the difference
between the numerically detected values and the experimental ones over the exper-
imental one. It is evident that, the upper bound values are expected to have a
positive relative error, which means that the numerically predicted upper bound is
greater than the experimental one, while the lower bound values are expected to
Table 7. Peak load multipliers for the analyzed RC beams: values experimentally
detected, PEXP ; values of upper and lower bounds, PUB and PLB , allowing bars
yielding; values PUB eb and PLB eb computed keeping bars elastic.
Table 8. Peak load multipliers for the analyzed RC slabs: values experimentally
detected, PEXP ; values of upper and lower bounds, PUB and PLB , allowing bars
yielding; values PUB eb and PLB eb computed keeping bars elastic.
1450058-19
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Table 9. Relative errors of the numerically predicted peak load multipliers for the
analyzed RC specimens: values of upper and lower bounds, PUB and PLB , allowing
bars yielding; values PUB eb and PLB eb computed keeping bars elastic.
Beam label
H1 1.60 −5.30 8.52 −4.93
S1 9.04 −5.92 18.09 −9.58
E1 1.22 −6.51 11.14 3.21a
MD1.3-A90 8.76 −6.59 21.73 11.58a
MD2.1-A190 6.73 −5.57 13.45 1.47a
T0.2-A135 7.71 −12.46 16.11 −7.22
A3 5.23 −3.62 24.92 7.10a
Slab label
SL15 9.40 −2.87 14.75 −13.24
SL30 8.30 −6.14 17.14 −11.87
SL45 8.22 −10.09 21.87 9.97a
S1 7.61 −9.95 26.76 −6.49
S2 4.72 −11.70 18.88 −15.31
S3 5.28 −9.72 7.47 −14.30
C−1 7.99 −6.93 36.27 14.48a
a Wrong prediction (PLB > PEXP ).
eb
have a negative relative error, which means that the numerically predicted lower
bound is less than the experimental one.
The inspection of the numerical findings highlights the actual improvement, for
all the examined RC beams and slabs, obtained with the proposed two yield criteria
limit analysis approach. In particular, the upper bound values given by the proposed
approach (PUB values) show a relative error much smaller than the one given by
the previous formulation (PUB eb values). This error never exceeds 10% for both
beams and slabs, while the same error with the previous formulation comes up to
20% for the beams and up to 30% for plates. The better performance of the present
approach is more evident with regard to the lower bound values except for beams
T0.2-A135 and H1 and slab S1. The relevant result is that all the computed PLB
values are below the experimental ones, as it has to be, while the procedure which
does not allow the yielding of the bars leads to incorrect results for 4 beams and 2
slabs. In the cited cases, the PUB eb value exceeds the value of PEXP .
Figures 8(a)–8(d) show, for sake of brevity only with reference to beams E1 and
A3 and to slabs SL15 and SL45, the plots of the upper and lower bounds to the
peak load multiplier versus iteration numbers. A rapid and monotonic convergence
of the procedure can be actually observed in all the run examples.
Furthermore, the procedure gives also some information on the state of speci-
mens at collapse giving some hints on the type of failure mechanisms. A possibility
to predict the failure mechanism of beams or slabs is in fact given by the possibility
to point out the plastic zone (collapse mechanism) at “last converged solution” of
the LMM. To this aim, the plots of the principal (compressive) strain rates ε̇c3 have
been considered on the beams loaded by PUB P̄tot and on the slabs loaded by PUB p̄,
1450058-20
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
2.5 10.0
Beam E1 Beam A3
2.0 8.0
load multiplier P
load multiplier P
1.5 6.0
1.0 4.0
PUB PUB_eb PUB PUB_eb
0.5 PLB PLB_eb 2.0 PLB PLB_eb
PEXP PEXP
0.0 0.0
1 2 4 6 8 10 12 14 1 2 4 6 8 10 12 14 16
iteration number iteration number
(a) (b)
load multiplier P
0.8
0.8
0.6
0.6
0.4
PUB PUB_eb 0.4 PUB PUB_eb
0.2 PLB PLB_eb 0.2 PLB PLB_eb
PEXP PEXP
0.0 0.0
1 2 4 6 8 10 12 14 16 1 2 4 6 8 10 12 14 16
iteration number iteration number
(c) (d)
Fig. 8. Values of the upper and lower bounds to the peak load multiplier versus iteration number
obtained by the proposed two-yield-criteria approach (solid lines with filled markers) and keeping
bars elastic (dashed lines with empty markers) against the collapse experimental values (dashed
lines).
at the “last” distribution of fictitious parameters and initial stresses. For sake of
brevity, only the band plots of ε̇c3 for beam E1 and slab SL15 are reported in Figs. 9
and 10. By inspection of Fig. 9 it is possible to observe two plasticized zones corre-
sponding to two plastic hinges, one over the rocker support, at beam center, and one
in the sagging region, at the mid-span. These zones appear reasonably narrow and
located where the damaged zones have been experimentally detected. Analogous
considerations can be made by observing Fig. 10 where again the detected plastic
hinges, a sort of cylindrical hinges spread over the slab width and thickness, meet
the experimental evidences. It is worth to mention that the band plots of Figs. 9
and 10 provide only qualitative information of the failure mechanisms, but these
plots can be anyway useful to localize critical zones or weaker members within larger
structural elements.
1450058-21
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
rocker support
at beam center
x continuous Beam E1
Fig. 9. Prediction of the failure mechanism for the continuous beam E1. Band plot of principal
compressive strain rates, in the deformed configuration, at the ultimate value of the acting load
obtained at last converged solution of the LMM.
z
x
Fig. 10. Prediction of the failure mechanism for the continuous slab SL15. Band plot of principal
compressive strain rates, in the deformed configuration, at the ultimate value of the acting load
obtained at last converged solution of the LMM.
6. Conclusions
A two yield criteria numerical limit analysis procedure for the prediction of peak
loads and failure modes of steel-RC elements has been proposed. The behavior at
ultimate state of both concrete and steel has been taken into account so that the
numerical procedure involves a von Mises yield criterion for bars and a M–W-type
yield criterion, with cap in compression, for concrete. The lack of associativity,
postulated for concrete, to deal with its dilatancy, has led to search for an upper
and a lower bound to the peak load multiplier of the whole RC structural element.
The former has been pursued by the LMM, the latter by the ECM. Both methods
have been reformulated to take into account possible yielding of the steel bars. The
reliability and effectiveness of the proposed numerical methodology has been proved
by analyzing large scale prototypes of steel-RC beams and slabs experimentally
tested up to failure and available in the literature. The obtained numerical results, in
terms of peak load multiplier and collapse mechanism of the analyzed specimens, are
1450058-22
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
Limit State Evaluation of Steel-RC Elements by von Mises and M–W-Type Yield Criteria
very satisfactory either with respect to the experimental findings or with respect to
the results obtained using a formulation utilized in a previous work where steel bars
were assumed indefinitely elastic. The improvements achieved are very encouraging
especially when analyzing RC beams and slabs in which pronounced yielding of the
steel bars occurs, a circumstance which is often exhibited by under-reinforced RC
structures employed in civil engineering applications. Finally, the LMM allows to
predict, even if qualitatively, the failure mechanism type of the analyzed structural
element so resulting in an useful tool to localize critical zones or weaker members
of real structures.
References
Al-Rousan, R., Issa, M. and Shabila, H. [2012] “Performance of reinforced concrete slabs
strengthened with different types and configurations of CFRP,” Composites: Part B
43, 510–521.
Ashour, A. F., El-Refaie, S. A. and Garrity, S. W. [2004] “Flexural strengthening of
RC continuous beams using CFRP laminates,” Cement and Concrete Composites 26,
765–775.
Bažant, Z. P. [1986] “Mechanics of distributed cracking,” Applied Mechania Reviews 39,
675–705.
Bresler, B. and Scordelis, A. C. [1963] “Shear strength of reinforced concrete beams,”
Journal of American Concrete Institute 60(1), 51–72.
Breveglieri, M., Barros, J. A. O., Dalfré, G. M. and Aprile, A. [2012] “A parametric study
on the effectiveness of the NSM technique for the flexural strengthening of continuous
RC slabs,” Composites: Part B 43, 1970–1987.
Chen, W. F. [1982] Plasticity in Reinforced Concrete (McGraw-Hill, USA).
De Domenico, D., Pisano, A. A. and Fuschi, P. [2014] “A FE-based limit analysis approach
for concrete elements reinforced with FRP bars,” Composite Structures 107, 594–603.
Fuschi, P. [1999] “Structural shakedown for elastic-plastic materials with hardening satu-
ration surface,” International Journal of Solids and Structures 36, 219–240.
Garcea, G. and Leonetti, L. [2011] “A unified mathematical programming formulation of
strain driven and interior point algorithms for shakedown and limit analysis,” Inter-
national Journal for Numerical Methods Engineering 88, 1085–1111.
Gilbert, R. I. and Nejadi, S. [2004] An Experimental Study of Flexural Cracking in Rein-
forced Concrete Members Under Short Term Loads (UNICIV REPORT No. R-434,
School of Civil and Environmental Engineering, The University of New South Wales,
Sydney, Australia).
Lau, D. and Pam, H. J. [2010] “Experimental study of hybrid FRP reinforced concrete
beams,” Engineering Structures 32, 3857–3865.
Le, C. V., Nguyen-Xuan, H. and Nguyen-Dang, H. [2010] “Upper and lower bound limit
analysis of plates using FEM and second-order cone programming,” Computers and
Structures 88, 65–73.
Lee, J. and Fenves, G. L. [1998] “Plastic-damage model for cyclic loading of concrete
structures,” Journal of Engineering Mechanics, ASCE 124, 892–900.
Li, T. and Crouch, R. [2010] “A C2 plasticity model for structural concrete,” Computers
and Structures 88, 1322–1332.
Lubliner, J., Oliver, J., Oller, S. and Oñate, E. [1989] “A plastic-damage model for con-
crete,” International Journal of Solids and Structures 25(3), 299–326.
1450058-23
2nd Reading
October 1, 2014 12:38 WSPC-255-IJAM S1758-8251 1450058
1450058-24