Sunteți pe pagina 1din 12

Journal of Advanced Concrete Technology Vol. 4, No.

2, 325-336, June 2006 / Copyright 2006 Japan Concrete Institute 325


Scientific paper
Computer-Aided Analysis of Reinforced Concrete Using a Refined
Nonlinear Strut and Tie Model Approach
Hamed M. Salem
1
and Koichi Maekawa
2

Received 10 February 2006, accepted 27 April 2006
Abstract
This paper presents a computer program for implementing a refined nonlinear strut and tie model approach for the prac-
tical design and analysis of disturbed regions in structural concrete. Nonlinear techniques in the selection, analysis and
verification processes of a strut and tie model are incorporated in this program to eliminate the limitations of the con-
ventional strut and tie model relating to the behavior and strength prediction of reinforced concrete. For the verification
of the proposed model, the model results are compared to the experimental results of one-quarter-scale simply sup-
ported bottom-loaded deep beams. Analytical results showed a lower bound solution that agreed well with the experi-
mental results. It was concluded that the nonlinear strut and tie model allows more economical design than the conven-
tional strut and tie model. It was also concluded that for higher strength concrete, the strength of struts and nodal zones
given by the ACI-318 02 code is unconservative and needs refinement to account for the brittleness of high-strength
concrete.

1. Introduction
1.1 Strut and tie models (STM)
Strut and tie models are discrete representations of ac-
tual stress fields in reinforced concrete structures. They
represent the load-carrying mechanisms of structural
members by approximating the flow of internal forces
by means of struts representing the flow of compressive
stresses and ties representing the flow of tensile stresses.
These models are mainly applied to the zones where the
beam theory does not apply, such as geometrical discon-
tinuities, loading points, deep beams and corbels.
Strut and tie models were first proposed by Ritter
(1899) as a simple truss model to visualize the internal
forces in cracked beams. Such models were the basis for
Ritter (1899) and Morsch (1902) for the design of con-
crete beams. Afterwards, strut and tie models were re-
fined by Kupfer (1964) and Leonhardt (1965). Marti
(1985) created the scientific basis for a rational applica-
tion in tracing the theory back to the theory of plasticity.
Collins and Mitchell (1986) further considered the de-
formation of the truss models and derived a rational
method for shear and torsion. Further work by Schlaich
et al. (1987) extended the beam truss models with the
use of uniformly inclined diagonals, thereby enabling
application to all parts of the structure in the form of
generalized strut and tie systems.


1.2 Nonlinear strut and tie model (NLSTM)
Although the conventional strut and tie model approach
has proved to be useful in the analysis and design of
disturbed regions of reinforced concrete, it has limita-
tions not only in predicting strength and nonlinear be-
havior, but also in the precise design of reinforced con-
crete that experiences nonlinear behavior. The nonlinear
strut and tie model approach originally proposed by
Young (2000) presents an effective solution for predict-
ing the essential aspects of behavior and strength in re-
inforced concrete and provides accurate detailing in
complex design situations. In Youngs model, the cen-
terlines of struts and ties are determined based on
nonlinear finite element analysis of unreinforced con-
crete. The strut and tie model is then analyzed linearly
to get the force and hence the area of each member. This
process is carried out in an iterative procedure until the
converged areas for the members are reached. The strut
and tie model is then analyzed nonlinearly and the
amount of reinforcement is computed after checking the
nodal stresses.

1.3 Why nonlinear strut and tie model?
In nonlinear analysis of statically indeterminate STM,
the absorbed energy in potential plastic hinges in rein-
forcement allows internal redistribution of stresses and
hence enables the utilization of higher load carrying
capacities. This should translate into more economical
designs. Figure 1 shows an example of analysis of a
statically indeterminate STM. In this example, the au-
thors applied the analysis twice, once linearly and once
nonlinearly. From the results, an increase of 8% in ca-
pacity can be noticed when adopting nonlinear analysis.
This means that an 8% saving in the amount of rein-
forcement can be achieved by using the NLSTM.
In nonlinear analysis of STM, it is also possible to put

1
Associate professor, Structural Engineering Dept.,
Cairo University, Giza, Egypt.
E-mail: hhadhoud@steelnetwork.com
2
Professor, Civil Engineering Dept., The University of
Tokyo, Tokyo, Japan.
326 H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006

concrete ties and test their ability to contribute to the
capacity of the structure. If the concrete ties are located
within a highly stressed part of the structure, it is evi-
dent that they will be cracked and will have a minor
contribution to the structure strength through their post
cracking strength. However, if the concrete ties are lo-
cated within a moderately stressed part of the structure,
they may not crack and hence contribute to the structure
strength.

1.4 Why nonlinear FEM analysis?
For deciding the layout of the STM, the NLSTM uses
nonlinear FEM. This is in fact essential because using
nonlinear FEM allows the redistribution of internal
forces due to material nonlinearity (concrete cracking,
concrete softening and steel yielding). The redistribution
of internal stresses allows the reorientation of principal
stresses, leading to a higher angle of diagonal struts in-
clination (Song et al., 1998). Figure 2 shows a sample
of analysis of a bottom loaded deep beam. In this exam-
ple, the authors solved the beam twice, once with linear
FEM and once with nonlinear FEM. It is clear that the
principal compressive stress trajectories have a larger
inclination in the nonlinear case and hence the diago-
nal strut could be assumed at a higher angle of inclina-
tion. This would lead to a STM with a higher capacity
and hence more economical design.

2. Proposed refined NLSTM
The nonlinear strut and tie model approach has proved
to be a useful and consistent method for the analysis and
design of reinforced and prestressed concrete members
Ac = 5000 mm
2
, As = 100 mm
2
Es = 200000 MPa, fy = 483 MPa, fu = 570 MPa
fc = 20 MPa, ft = 2 MPa
Ac = 5000 mm
2
, As = 100 mm
2
Es = 200000 MPa, fy = 483 MPa, fu = 570 MPa
fc = 20 MPa, ft = 2 MPa
Linear Nonlinear

0
20
40
60
80
100
120
140
160
180
200
0 10 20 30 40
Deflection (mm)
L
o
a
d

(
k
N
)
Nonlinear
Linear

Fig.1 Effect of introducing nonlinearity into the strut and
tie model.
Linear FEM
Mesh discretization Principal compressive
stress trajectories
Linear FEM
Mesh discretization Principal compressive
stress trajectories
Nonlinear FEM
Mesh discretization Principal compressive
stress trajectories
Nonlinear FEM
Mesh discretization Principal compressive
stress trajectories
Fig.2 Principal compressive stress trajectories from lin-
ear and nonlinear finite element analysis.
H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006 327

including disturbed regions. However, the authors be-
lieve that some drawbacks of the original NLSTM re-
main, as follows:
a) The preliminary determination of centerlines of struts
and ties is based on nonlinear FEM analysis of unre-
inforced concrete (Young 2000). This would not be
appropriate in many applications because of the in-
stability after cracking of concrete or the resulting in-
accuracy if the concrete is assumed not to crack. The
authors believe that steel reinforcement should be in-
troduced for such analysis to allow for the redistribu-
tion of stresses after concrete cracking. However, this
gives rise to another problem, namely the fact that
the amount of steel reinforcement is in fact an output
of the design and not an input. Nevertheless, a rea-
sonable pre-assumption of the amount of reinforce-
ment would be helpful and during the design proce-
dure it should be checked whether this assumption is
acceptable or not.
b) The reinforcement amount is calculated after only
one trial of the nonlinear analysis of the STM. At a
matter of fact, the nonlinear analysis would exhibit
some internal stress redistribution in statically inde-
terminate STM because of the energy absorption es-
pecially in the yielded steel ties. Therefore, the au-
thors propose to perform trial calculations in the
nonlinear stage.
c) The constitutive models for steel and concrete used
by Young (2000) are somewhat simplified models.
The authors use more enhanced and comprehensive
experimentally derived constitutive models.
(Maekawa and Okamura, 1983, and Okamura and
Maekawa, 1991)
The flow chart of the proposed refined NLSTM is
shown in Fig. 3. First, a nonlinear FEM analysis, using
the WCOMD code (Okamura and Maekawa, 1991) is
carried out for the structure, and the principal compres-
sive stress trajectories are obtained. From the principal
stress trajectories, the centerlines of the struts and ties
can be developed. In straight- forward situations, an
experienced engineer is generally capable of developing
strut-tie models based on engineering common sense
and knowledge of the behavior of structural concrete.
However, in more complex cases, this practical knowl-
edge is often not enough to develop adequate strut-tie
models. In such cases, the principal compressive stress
trajectories are used to select the orientation of the
model struts. The members areas and the initial stiff-
ness of the members are then assumed and elastic analy-
sis of the STM is carried out using the code NLSTM
developed by the authors. This code can actually incor-
porate both material and geometrical nonlinearities.
From this elastic analysis, the members forces are cal-
culated and hence the members areas can be deter-
mined based on the allowable stresses for concrete and
steel members. If the STM is statically indeterminate,
the analysis has to be carried out again with the new
members area as a second iteration. The obtained
members areas in the second iteration are compared to
those obtained from the first iteration. If they are not
Nonlinear FEM for reinforced concrete
Principal stress trajectories
Centerlines of struts and ties
Assume member Areas A=Ao
and Youngs Modulus E=Eo
Elastic analysis of STM
Calculate member forces
Calculate member areas (A)
Yes
Members areas converge ?
No
Statically indeterminate truss?
No
Yes
A
Yes
Add concrete ties and steel struts (if necessary)
Geometric compatibilities of members satisfied ?
(boundaries, bearing plates, etc)
No
Adjust members widths in accordance with geometrical
compatibilities. If necessary, adjust members centerlines
and go to step (A) again
Nonlinear analysis of STM
Calculate reinforcement areas
No
Yes
End
Check strut stresses and nodal stresses
Reinforcement areas converge ?
Statically indeterminate truss?
No
Statically indeterminate truss?
No
Failure in a strut or a node
No
Yes
Increase member widths in the critical
zones if possible
(if necessary, adjust members centerlines)
Perform nonlinear analysis till failure
Yes

Fig.3 Flow chart of the refined nonlinear strut and tie
model approach.
328 H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006

close, a third iteration has to be carried out using the
areas obtained from the second iteration, and so on until
an area convergence is achieved. The geometric com-
patibilities of members are then checked. If any member
lies outside the structure boundaries or if the members
connected to the bearing plates constitute a nodal zone
wider than the bearing plates, the members widths
should be adjusted in accordance with the geometrical
compatibilities (if necessary, the member centerlines are
adjusted and the elastic analysis is then performed
again). The next step is the nonlinear analysis of the
STM. Before this step, it is possible to add concrete ties
or steel struts. The analysis is carried out and the stress
levels in struts and nodal zones are checked. If the stress
levels reach the failure criteria, the possibility of in-
creasing the member widths in the critical zones is
tested. If it is possible, the nonlinear analysis is per-
formed again. If it is not possible, the areas of rein-
forcement are calculated. Whether the calculated areas
of reinforcement are close to those used in the nonlinear
analysis is checked. If they are different, another itera-
tion incorporating nonlinear analysis is performed and
so on until a converged solution is obtained. Finally, the
overall performance of the designed structure is checked
using nonlinear analysis until the failure of struts and/or
nodal zones.

3. Implementation of the refined NLSTM
3.1 Nonlinear FEM analysis
The nonlinear FEM analysis is carried out using the
nonlinear finite element code WCOMD (Okamura and
Maekawa, 1991) developed in the concrete laboratory of
the University of Tokyo. The code accommodates path-
dependent constitutive models for in-plane reinforced
concrete with multi-directional fixed smeared crack
idealization. In that scheme, a multi-directional cracking
system with up to four independent orientations is de-
veloped.

3.2 NLSTM analysis
A computer program was developed by the authors for
the analysis of the nonlinear strut and tie model. The
code is based on a stiffness method incorporating both
material and geometrical nonlinearities.

3.2.1 Formulation of the stiffness matrix
The global stiffness matrix of each truss member can be
formulated directly by assuming unit displacements in
the global directions as shown in Fig. 4. The elements
of the stiffness matrix are represented as functions of the
angle of inclination ( ) with the horizontal as follows,

| |
2 2
2 2
g
2 2
2 2
. .
. .
. .
. .
c c s c c s
c s s c s s EA
S
L
c c s c c s
c s s c s s
(

(
(
=
(

(
(
(

(1)
where c = cos (), s = sin (), E = tangent stiffness of
the stress-strain curve of the constituent material, A =
cross sectional area of the member and L = length of the
member. It should be mentioned that the tangent stiff-
ness has to be limited so that it is not zero or negative in
order to avoid divergence during the analysis. In fact, a
limiting minimum stiffness of 0.05 times the initial
stiffness is used here. Once the individual stiffness ma-
trices [S]
g
for each member are determined, the overall
stiffness matrix [K] is assembled.

3.2.2 Material nonlinearity
The constitutive models used in the analysis are shown
in Fig. 5 for concrete and steel for both tension and
compression.

(1) Concrete
Concrete in compression follows the elasto-plastic and
fracture model (Maekawa and Okamura, 1983) as fol-
lows,
( )
( )
1.25
p
0.73 1
c
peak
0.35
peak
peak
( )
2
20
1
7
x
x e
x
p
Ko Eo
Ko e
f
Eo
x e
x

=
=

=
| |
=
|
\ .
=
(2)
where,
Ko = Fracture parameter representing the damage of
concrete.
Eo = Initial Stiffness of concrete.

p
= Plastic strain corresponding to the total strain

peak
= Peak strain for concrete under compression.
c
f = Specified concrete compressive strength

The conventional STM usually do not take into ac-
count the tensile contribution of concrete. In NLSTM,
the concrete tensile contribution could be taken into
consideration either in plain concrete or reinforced con-
crete ties. Reinforced concrete ties are concrete ties that
contain reinforcing bars inside, while plain concrete ties
have no reinforcing bars at all. In plain concrete ties,
after the concrete cracks, the tie strength drops very
1 =1
2 =1
3 =1 4 =1
L


Fig.4 Global stiffness matrix of truss member.

H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006 329

rapidly and becomes dependant mainly on the bridging
tensile stress transferred across the crack surface, while
in reinforced concrete ties the concrete between cracks
still has the ability to contribute in resisting tie deforma-
tions through the tension stiffening effect. The post-
cracking tensile strength is simulated here with the ten-
sion softening/stiffening model of Okamura and
Maekawa (1991), where power coefficient C equals 0.4
for reinforced concrete ties and is set approximately to
2.0 for plain concrete ties. The model yields,
cr
t
C
f

| |
=
|
\ .
(3)
where f
t
= concrete tensile strength of concrete,
cr
=
cracking strain and = strain.

(2) Strut stress limits according to ACI 318-02
(2002)
The stress limits in the struts have been adopted from
the ACI 318-02 (2002). The allowable stress limit for
the struts is given by
all s c
0.85 f f = (4)
where:
s
1.00 = for prismatic struts in uncracked compression
zones
s
0.75 = for bottle-shaped struts where crack control
reinforcement is included
s
0.60 = for bottle-shaped struts where crack control
reinforcement is not included
s
0.40 = for struts in tension members
s
0.60 = for all other cases
The crack control reinforcement requirement is
( /( )) 0.003 sin
si i i
A bS

, where
si
A = steel area of
the i
th
layer of reinforcement crossing the strut at an
angle
vi
between the axis of the strut and the bars, S
i

= spacing and b = beam thickness.

(3) Reinforcing bars
Reinforcing bars are simulated by Okamuras model for
bare bars (1991). The stress is linear elastic up to the
yielding point and after a certain yielding plateau strain
hardening starts in an exponential form, as shown in Fig.
5.
s y
y y sh
u y sh
( ) /
sh
, 0
,
{1 } (1.01 ) ,
y
k
E
f
f e f f



= < <
= <
= + >
(5)
y
2/ 3
0.047(4000/ ) k f =
3.2.3 Geometrical Nonlinearity
Geometrical nonlinearity is simply introduced by updat-
ing the displacements at every iteration and by comput-
ing the strains based on the most updated displacements.
As illustrated in Fig. 6, the new position of the member
is used to compute the elongation, hence the current
strain. An example is shown in Fig. 6 for a horizontal
member experiencing only relative vertical displace-
ment at its ends. If geometrical nonlinearity is not taken
into consideration, the strain will be zero. However, if
the geometrical nonlinearity is considered, the relative
vertical displacement will cause some elongation in the
member direction, which means that normal strains exist.
Concrete
Elements
(struts or ties)
Steel
Elements
(struts or ties)
Struts
Elasto-plastic &
Fracture Model
(Maekawa)
Ties
Tension-
Stiffening/Softening
Model (Okamura)

Tension-Stiffening
Reinforced concrete ties
Tension-Softening
Plain concrete ties

Bare Bar Model


(Okamura)
f
y

y
f
u

sh

sh
f
y
f
u
+
-
+
-

Bare Bar Model


(Okamura)
f
y

y
f
u

sh

sh
f
y
f
u
+
-
+
-
I
n
i
t
i
a
l

p
o
s
i
t
i
o
n
N
e
w

p
o
s
i
t
i
o
n
lo
lo
l
lo

l
Horizontal member General inclined
member
No Geometrical
Nonlinearity
= 0
Geometrical
Nonlinearity
= 0
I
n
i
t
i
a
l

p
o
s
i
t
i
o
n
N
e
w

p
o
s
i
t
i
o
n
lo
lo
l
lo

l
Horizontal member General inclined
member
No Geometrical
Nonlinearity
= 0
Geometrical
Nonlinearity
= 0
Fig.5 Nonlinear constitutive models of constituent mate-
rials (Maekawa and Okamura, 1983), (Okamura and
Maekawa, 1991).
Fig.6 Geometrical nonlinearity for STM members.
f
u
f
u
f
y

f
y

sh

y

sh



330 H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006

3.2.4 Algorithm of nonlinear analysis
A nonlinear algorithm is adopted in which a step-
iterative procedure is followed until an acceptable con-
vergence is obtained. The convergence criterion here is
to minimize the sum of the squares of the residual forces
(unbalanced forces) at the joints.
The essential steps in the analysis scheme are;
1) At the beginning of the analysis, the geometry, the
boundary conditions and the incremental load vec-
tor {f} are input to the solver.
2) The overall stiffness matrix [K] is calculated as a
function of the initial stiffness of the constituent
materials, i.e. stiffness at strain equals zero, as
mentioned above.
3) The system of equilibrium equations [K]{d}={f} is
solved and the nodal displacement vector {d} is
obtained.
4) Once the nodal displacements are determined, the
strain in each member is calculated from its end
displacements (the component of the end dis-
placements in the direction of the member), and
consequently both the stress and the tangent stiff-
ness can be calculated from the nonlinear constitu-
tive laws of the constituent materials.
5) The force in each member is calculated by multi-
plying the stress times the cross-sectional area.
6) The nodal force vector {f1} that corresponds to the
current displacement field {d} is then calculated.
At each node, the nodal forces are computed as the
algebraic sum of the horizontal and vertical com-
ponents of the forces in the truss members meeting
at the node of concern.
7) The unbalanced nodal force vector {f} is deter-
mined by subtracting the compatibility nodal force
vector {f1} from the load vector {f}.
8) The convergence is checked for the unbalanced
nodal force vector {f} and if the convergence cri-
terion is not met, another iteration is carried out.
9) In the new iteration, the tangent stiffness matrix
[K] is calculated based on the current displacement
field of the nodes {d}, and the equilibrium equa-
tions are again solved to get the displacement field
{d} that corresponds to the unbalanced nodal
force vector {f} as [K]{d}={f}.
10) The deflection is then updated by summing up the
displacement vectors {d} and {d}.
11) The unbalanced nodal force vector {f} is deter-
mined again based on the most updated displace-
ment vector {d} as explained above in steps 4, 5, 6
and 7.
12) The convergence is checked again for the unbal-
anced nodal force vector {f} and if the conver-
gence criterion is not met, another iteration is car-
ried out, and so on until the proposed accuracy for
the solution is obtained. When the convergence is
fulfilled, the next load increment is analyzed and so
on.
3.2.5 Dummy members
For the truss analysis, it is sometimes necessary to add
dummy members that are not actually contributing to
the structural strength but whose existence ensures the
stability of the analytical computations. For example,
the STM shown in Fig. 7 is stable under the shown ver-
tical loading, but when solving this problem with any
numerical model, instability in analysis will occur. To
avoid instability, the dashed diagonal member is added.
This member will be a zero member due to symmetry.
In some situations, the dummy members might not be
zero members. In this case, the area of these members
should be set to a relatively small area, especially in the
nonlinear stage, in order to make them too flexible to
contribute to the load carrying system, or in other words,
to force them to become zero members.

3.3 Check of nodal stresses
As explained by Marti (1985), the nodal zones are in a
biaxially stressed state defined by the nodal Mohrs cir-
cle shown in Fig. 8. The nodal Mohrs circle is defined
as the circle passing through the three points represent-
ing the state of stress of three faces of the nodal zone
(face 1, face 2 and face 3). The center of this circle must
lie on the -axis. To get the point representing the stress
state on a certain face of the nodal zone, the Mohr's cir-
cle for the strut facing it is drawn and from the pole of
that circle a line parallel to the face is drawn and its in-
tersection with the circle is defined. The minimum prin-
cipal stress obtained from the nodal Mohrs circle is
checked to see whether it exceeds the allowable com-
pressive stress. The ACI 318-02 (2002) code assumes
that the safety of nodal zones is achieved if the com-
pressive stresses acting on the nodal faces are less than
the limits given by Eq. (6) in the next section (4.3.1). In
this study, however, the principal stresses inside the
nodal zones are already calculated and compared, for
simplicity, to the limit given by the ACI code.

D
u
m
m
y

m
e
m
b
e
r
D
u
m
m
y

m
e
m
b
e
r


Fig.7 Dummy members used for analytical stability.
H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006 331

3.3.1 Nodal zone stress limits according to ACI
318-02 (2002)
The stress limits in the nodal zones have been adopted
from the ACI 318-02. The allowable stress limit for the
struts is given by
all n c
0.85 f f = (6)
where:
n
1.00 = when nodes are bounded by struts and/or bear-
ing areas
n
0.80 = when nodes anchor only one tie
n
0.60 = when nodes anchor more than one tie

4. Experimental verifications
For the verification of the proposed model, the analyti-
cal results were compared to the experimental results of
one-quarter-scale simply supported bottom-loaded deep
beams shown in Fig. 9 (Elattar et al., 2002). Four
specimens were used in the analysis, namely B3, B4, B5
and B11. The specimens had identical reinforcement.
Each of them had a length/depth ratio of 1.25 except
B11, which had a length/depth ratio of 1.75. The con-
crete characteristic strengths, f
cu
, of B3, B4, B5 and B11
were 29, 40.8, 21.3 and 27.8 MPa, respectively. The
specimens thickness was 90 mm for B3, B4, and B11
while it was 120 mm for B5. The shear-span/depth ratio
was 0.42 for all specimens. The yield stresses for the
reinforcing bars were 383, 395, 279 and 346 MPa for
Y12, Y10, R8 and R6, respectively.
Figure 10(a) shows the principal compressive stress
trajectories resulting from both linear and nonlinear
FEM analyses of specimen B11. The inclination of the
diagonal strut is determined based on the average incli-
nation of the stress trajectories as shown in Fig. 10(a). It
is obvious that a steeper diagonal strut (a strut with
higher inclination) is derived from the nonlinear FEM.
Cracking of the concrete especially at the bottom of the
specimen causes redistribution of internal forces leading
to steeper diagonal struts. From the stress trajectories
derived from the nonlinear FEM, it is clear that the di-
agonal struts may be modeled as a bottle-shaped strut
with the web reinforcement as the connecting ties of the
bottle. The allowable stresses for the struts here may be
assumed with
s
0.75 = since the web reinforcement con-
trols the cracking according to the ACI provisions. It is
worthy to mention that the 0.85- coefficient in Eq. (4)
need not be used in the analysis of the specimens be-
cause the loading in the laboratory is not a sustained
loading.
Figure 10(b) shows both the detailed LSTM and
NLSTM of B11. The failure of the beam was caused by
the failure of the diagonal struts shown in the figure,
which coincides with the experimental observations
(Elattar et al., 2002). Figures 10(c) and 10(d) show the
analytical forces in the members and the deformed
shapes of specimen B11 according to both the LSTM
and NLSTM.
min
max
+ -
Normal
Stress
Shear
Stress
Nodal Mohrs Circle
Face (3)
F
a
c
e

(
1
)
F
a
c
e
(
2
)
Face (1)
Face (2)
Face (3)
Pole (3)
Pole (1)
Pole (2)
min
max
+ -
Normal
Stress
Shear
Stress
Nodal Mohrs Circle
Face (3)
F
a
c
e

(
1
)
F
a
c
e
(
2
)
Face (1)
Face (2)
Face (3)
Pole (3)
Pole (1)
Pole (2)

Fig.8 Nodal zone Mohrs circle and principal stresses
inside the nodal zone.
2R8
2R8
2R8
B3, B4, B5
B11
655
150
490
2Y10
2R8
2R8
2Y10
2R6
2R6
90 for B11& B3 & B4
120 for B5
75 75 1365
150
655
805
R8@218
Y12
2Y10
2Y10
2R8
2R8
R6@163
75
150
805
655
75 975
2Y10
2R8
2R8
Y12 R6@163
2Y10
R8@218
Dimensions in mm
327 mm
100 mm
20 mm
cover
End
anchorage
Y
1
0
Y
1
0
2R8
2R8
2R8
B3, B4, B5
B11
655
150
655
150
490
2Y10
2R8
2R8
2Y10
2R6
2R6
90 for B11& B3 & B4
120 for B5
75 75 1365
150
655
805
R8@218
Y12
2Y10
2Y10
2R8
2R8
R6@163
75 75
150
805
655
75 975
2Y10
2R8
2R8
Y12 R6@163
2Y10
R8@218
Dimensions in mm
327 mm
100 mm
20 mm
cover
End
anchorage
Y
1
0
Y
1
0

Fig.9 Test specimens (Elattar et al., 2002) used in verifi-
cation of the analysis.
332 H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006


C L

Bottle-shaped strut
C L

Linear FEM Nonlinear FEM
(a) Principal compressive stress trajectories

8
8
.
8
4
5
0

m
m
7
9
.
0
6
78.91
6
6
.
6

m
m
6
6
.
5
8
66.09
Failed
member
Failed
member

LSTM NLSTM
(b) Strut and tie model

100 kN
+68 kN
-68 kN -
7
2

k
N
-
7
2 -
9
8

-
9
8

+
5
4 +146
+146
-
7
2

k
N
-
9
8

+
5
4
-
7
2
-
9
8

-70 kN
-
1
3
6

k
N
+70 kN
+114 kN
-
1
3
6

k
N
+114 kN
146 kN 146 kN 114 kN 114 kN
100 kN
+68 kN
-68 kN -
7
2

k
N
-
7
2 -
9
8

-
9
8

+
5
4 +146
+146
-
7
2

k
N
-
9
8

+
5
4
-
7
2
-
9
8

-70 kN
-
1
3
6

k
N
+70 kN
+114 kN
-
1
3
6

k
N
+114 kN
146 kN 146 kN 114 kN 114 kN

LSTM NLSTM
(c) Member forces

1
0

m
m


LSTM NLSTM
(d) Deformed shape

Fig.10 Analytical results for specimen B11.

H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006 333

The nodal stresses are checked in the critical nodal
zones. Although the bearing plate width was relatively
small, no failure occurred in the experiments in the
nodal zones over the bearing plates. This was due to the
use of a bottom- flanged beam for bottom loading pur-
poses. This helped increase the bearing area and consid-
erably reduce the bearing stresses. This fact was taken
into consideration in the analytical model. Since the
members used in the analysis are prismatic, the reduc-
tion in bearing stresses over the bearing plate was taken
into account by increasing the allowable nodal stresses
in these nodal zones in the ratio of the flange width to
the web width. The critical nodal zones are shown in
Fig. 11(a). The nodal zones are four-face nodes. To ana-
lyze these nodal zones, they is subdivided into two
three-face sub-nodal zones as shown in the figure and
checking of nodal stresses is performed for each sub-
node as shown in Fig. 11(b) for the upper sub-node. It is
also worthy to mention that the 0.85 coefficient in Eq.
(6) need not to be used in the analysis of the nodal zones
because the loading in the laboratory is not a sustained
loading.
Figure 12 shows the total applied load-midspan de-
flection relation for both the LSTM and NLSTM of B11
in comparison with the experimental results. The ex-
perimental failure load was 326 kN, while the LSTM
and NLSTM failure loads were 228 kN and 292.9 kN,
respectively. This comparison shows that either the
LSTM or NLSTM is a lower bound model. The LSTM
is 42% conservative while the NLSTM is only 11%
conservative. Design using either the LSTM or NLSTM
will then be acceptable, however, the NLSTM will al-
low more economical design.
Figure 13 shows the effect of taking the tension stiff-
ening of the cracked concrete into consideration in
analysis. It can be seen that considering the tension
stiffening reduces the deformability of the structure in
the post-cracking stage. This is explained by the contri-
bution of uncracked concrete between adjacent cracks in
a tension member in increasing its stiffness.
Figures 14, 15 and 16 show the analytical and ex-
perimental results for specimens B3, B5 and B4, respec-
tively. As can be seen, the LSTM is 30% and 45% con-
servative, while the NLSTM is 8% and 10% conserva-
tive for B3 and B5, respectively. This agrees with the
result of specimen B11. However, for specimen B4,
with relatively higher concrete strength, it was found
that the calculated beam capacity using the NLSTM was
19% higher than the experimental results. This is ex-
plained by the brittle behavior of the high-strength con-
crete especially when failure is governed by shear, due
to its relatively smooth crack surface and its inability to
have good shear transfer after cracking. In fact, the al-
lowable stresses given by ACI 318-02 (2002) code do
not take into account the brittleness of high-strength
concrete. The strength used by ACI is linearly propor-
tional to the concrete strength, which may overestimate
the strength in higher strength concrete. This fact was
recorded earlier by Ramirez and Breen (1983), who
suggested that the strut strength is proportional to the
square root of f
c
. When the strut strength was specified
according to Ramirez and Breen (20% less than the ACI
strength), the results of the NLSTM became very close
to the experimental results, as shown in Fig. 17. The
authors recommend reducing the efficiency factor in
the higher strength concrete. In other words, it seems
7
9
.
0
6
78.91
6
6
.
6

m
m
6
6
.
5
8
66.09
Critical Nodal Zone
1
8
.5
M
P
a
15.3
12.1
1
1
.
3
2
11.9
15.3
12.1
1
1
.
3
2
11.9
A
B
A
B
1
8
.5
M
P
a
15.3
12.1
1
1
.
3
2
11.9
15.3
12.1
1
1
.
3
2
11.9
A
B
A
B


(a) Critical nodal zones


Fig.11 Check of stresses in critical nodal zone in specimen B11.
(b) Upper part of the critical nodal zone (Part A)
334 H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006


0
100
200
300
400
500
0 5 10 15 20
Deflection (mm)
L
o
a
d

(
k
N
)
NLSTM
LSTM
Experiment
Yielding of
main RFT
Strain hardening
of main RFT

Fig.12 Total applied load-midspan deflection relationship
for B11.

0
100
200
300
400
500
0 5 10 15 20
Deflection (mm)
L
o
a
d

(
k
N
)
NLSTM (with tension stiffening)
NLSTM
Experiment
Fig.13 Effect of tension stiffening on analytical results for
B11.

0
100
200
300
400
500
0 5 10 15
Deflection (mm)
L
o
a
d

(
k
N
)
NLSTM
LSTM
Experiment
(a) Total applied load-midspan deflection relationship
79.93
44.4
Failed
member
LSTM

89.87 mm
67.64
77.79
55.64
55.6
Failed
member
NLSTM
(b) Linear and nonlinear STM
Fig.14 Analytical and experimental results of specimen
B3.
0
100
200
300
400
500
0 5 10 15
Deflection (mm)
L
o
a
d

(
k
N
)
NLSTM
LSTM
Experiment

(a) Total applied load-midspan deflection relationship

79.93
44.4
Failed
member
LSTM

89.87 mm
67.64
77.79
55.64
55.6
Failed
member
NLSTM
(b) Linear and Nonlinear STM

Fig.15 Analytical and experimental results of specimen.


0
100
200
300
400
500
0 5 10 15 20
Deflection (mm)
L
o
a
d

(
k
N
)
NLSTM
LSTM
Experiment
Yielding of
main RFT
Strain hardening
of main RFT

(a) Total applied load-midspan deflection relationship

79.92
44.4
Failed
member
LSTM

89.87 mm
67.64
77.79
55.66
55.6
Failed
member
NLSTM
(b) Linear and Nonlinear STM

Fig.16 Analytical and experimental results of specimen
B4.
H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006 335

essential to introduce a versatile efficiency factor for
the strength of struts and nodal zones in the STM.
Table 1 shows the summary of the analytical and ex-
perimental ultimate loads of the four specimens studied
here. Generally speaking, both the LSTM and NLSTM
are lower bound models. On average, the LSTM is 37%
conservative in predicting the ultimate load while the
NLSTM is only 8% conservative. Designs using either
the LSTM or NLSTM will then be acceptable, however,
the NLSTM allows more economical design.

5. Conclusions
This paper presents a computer program for implement-
ing a refined nonlinear strut and tie model approach for
the practical analysis and design of disturbed regions in
structural concrete. Nonlinear techniques in the selec-
tion (nonlinear FEM), analysis (nonlinear STM with
both material and geometrical nonlinearities) and veri-
fication processes of nonlinear strut and tie model are
incorporated in this program to eliminate the limitations
of the conventional strut and tie model relating to the
behavior predictions of reinforced concrete. Implement-
ing the nonlinear strut and tie model approach has been
shown to be efficient in the development, analysis, and
detailing of strut and tie models for the analysis and
design of reinforced concrete in disturbed regions.
For the studied cases of bottom-loaded deep beams,
the LSTM was 37% conservative in predicting the ulti-
mate load while the NLSTM was only 8% conservative.
In other words, design using either the LSTM or
NLSTM will then be acceptable, however, the NLSTM
allows more economical design. The NLSTM was
proved to allow economical design for the following
reasons:
1. The proper selection of the layout of the STM
through the use of nonlinear FEM analysis for de-
tecting the principal compressive stress trajectories.
The nonlinear FEM allows the redistribution of in-
ternal forces due to material nonlinearity (concrete
cracking, concrete softening and steel yielding).
The redistribution of internal stresses allows the
reorientation of principal stress trajectories, leading
to higher inclination angles of diagonal struts and
consequently higher load carrying capacity for the
NLSTM.
2. The NLSTM allows the redistribution of internal
forces in statically indeterminate STM particularly
due to the creation of plastic hinges in steel ties.
This leads to higher load carrying capacity.
The efficiency factor given by ACI-318 02 (2002)
to determine the strut and nodal zone strength is uncon-
servative in the case of higher strength concrete. Use of
the model suggested by Ramirez and Breen (1983),
which takes into consideration the brittle behavior of
high-strength concrete, is recommended instead.

References
ACI Committee 318-02 (2002). Building Code
Requirements for Reinforced Concrete (ACI 318-
02). American Concrete Institute, Detroit.
Collins, M., P. and Mitchell, D. (1986). A rational
approach for shear designThe 1984 Canadian code
provisions. ACI Journal, 83(6), 925-933.
El-Attar, A., Ghoneim, M., El-Ibiari, S. and Abdel
Rahman, A. (2002). Behavior of bottom loaded R.C.
deep beams: an experimental investigation. Journal
of Engineering and Applied Science, Faculty of
Engineering, Cairo University, 49(2), 337-356.
Kupfer, H. (1964). Expansion of Morschs truss
analogy by application of the principle of minimum
strain energy. CEB bulletin No. 40.
Leonhardt, F. (1965). Reducing the shear
reinforcement in reinforced concrete beams and
slabs. Magazine of Concrete Research, 17(53), 187-
198.
Maekawa, K. and Okamura, H. (1983). The
deformational behavior and constitutive equation of
Table 1 Ultimate load prediction.
Specimen Test LSTM NLSTM Test
LSTM
Test
NLSTM
B3 280.0 kN 215.0 kN 259.2 kN 1.30 1.08
B4 310.0 kN 235.4 kN 302.8 kN 1.31 1.03
B5 301.0 kN 207.4 kN 273.6 kN 1.45 1.10
B11 326.0 kN 228.0 kN 292.9 kN 1.42 1.11

0
100
200
300
400
500
0 5 10 15 20
Deflection (mm)
L
o
a
d

(
k
N
)
Strut strength acc. to ACI-02
Strut strength acc. to Ramirez
Experiment
Fig.17 Effect of failure criteria of high-strength struts on
the analytical results.
336 H. M. Salem and K. Maekawa / Journal of Advanced Concrete Technology Vol. 4, No. 2, 325-336, 2006

concrete using the elasto-plastic and fracture model.
Journal of the Faculty of Engineering, The University
of Tokyo (B), 37(2), 253-328.
Marti, P. (1985). Basic tools of reinforced concrete
beam design. ACI Journal, 83(1), 36-42.
Morsch, E. (1902). Der Eisenbetonbau, seine
Anwendung und Theorie. 1st ed., Wayss and Freytag,
A. G., Im Selbstverlag der Firma, Neustadt a. d.
Haardt, 118 pp. (adopted from Schlaich et al., 1987).
Okamura, H. and Maekawa, K. (1991). Nonlinear
analysis and constitutive models of reinforced
concrete. Gihodo, Tokyo.
Ramirez, J. and Breen, J. (1983). Proposed design
procedures for shear and torsion in reinforced and
prestressed concrete. research report 248-4F, Center
for Transportation Research, The University of Texas
at Austin, 254 pp.
Ritter, W. (1899). Die Bauweise Hennebique.
Schweizerische Bauzeitung, Zurich, (adopted from
Schlaich et al., 1987).
Schlaich, J., Shafar, K. and Jennewin, K. (1987).
Toward a consistent design of structural concrete.
PCI Journal, 32(3), 74-150.
Song, H., Bae, H. and Byun, K. (1998). Design and
analysis of anchorage zone in prestressed concrete
girder using nonlinear strut and tie model. The Sixth
East Asia-Pacific Conference on Structural
Engineering & Construction, Taipei, Taiwan, 871-876.
Young, M. (2000). Nonlinear strut-tie model approach
for structural concrete. ACI Structural Journal, 97(4),
581-590.

S-ar putea să vă placă și