J. Cent. South Univ. Technol.

(2011) 18: 1685−1692

DOI: 10.1007/s11771−011−0889−x

Comparison on construction of strut-and-tie models for

reinforced concrete deep beams

QIU Yi-ke(仇一颗), LIU Xia(刘霞)

School of Civil Engineering, Hunan University, Changsha 410082, China
© Central South University Press and Springer-Verlag Berlin Heidelberg 2011

Abstract: With consideration of the differences between concrete and steel, three solutions using genetic evolutionary structural
optimization algorithm were presented to automatically develop optimal strut-and-tie model for deep beams. In the finite element
analysis of the first method, the concrete and steel rebar are modeled by a plane element and a bar element, respectively. In the
second method, the concrete and steel are assigned to two different plane elements, whereas in the third method only one kind of
plane element is used with no consideration of the differences of the two materials. A simply supported beam under two point loads
was presented as an example to verify the validity of the three proposed methods. The results indicate that all the three methods can
generate optimal strut-and-tie models and the third algorithm has powerful capability in searching more optimal results with less
computational effort. The effectiveness of the proposed algorithm III has also been demonstrated by other two examples.

Key words: reinforced concrete deep beam; topology optimization; strut-and-tie model; genetic evolutionary structural optimization

maximize the plastic capacity of the proposed strut-and-

1 Introduction tie model. ALSHEGEIR [5] developed a software
utilizing finite-element analysis in which the user should
Reinforced concrete deep beams, whose span-depth align the strut-and-tie model with the directions of the
ratio is less than or equals 5, are frequently used in principal stress. However, ALSHEGEIR did not justify
basement facades, frame-supported shear walls and the use of the elastic analysis. In fact, many designs
technical equipment beam-type transition floors. deviate from the elastic strain distribution and yet they
Experimental and theoretical studies have revealed very are deemed to be good designs [6]. ALI and WHITE [6]
different mechanical behaviors of deep beams in presented a consistent and rational approach to generate
comparison with shallow beams with large span-depth optimal strut-and-tie models for disturbed D-regions in
ratio. Shear failure model of reinforced concrete deep concrete structures, including criteria for practical
beam is dominated by baroclinic model [1]. Experiments reinforcement layout. This approach also moved the
[2−3] in 1980s showed that strut-and-tie models can be D-region design process much closer to the philosophy
introduced for deep beam to simulate its stress used in the existing flexural design methods. BRUGGI
mechanism. For instance, the stress mechanism of simply [7] proposed a simple implementation for minimum
supported deep beams with no hole is usually simulated compliance optimization relying on the finite element
using the tie-arch model. In practice, holes with different library. GUEST and MOEN [8] employed topology
sizes are often opened in deep beams in order to meet the optimization techniques to automate the design of
needs posed by pedestrian passage or installation of reinforced concrete members. KWAK and NOH [9]
equipment channels, which makes the stress distribution introduced the basic idea of the ESO method to
of deep beams with holes more complex. Currently, a determine more rational strut-and-tie models. In their
combination of experiment and finite element method is works, in order to prevent the structural instability that
widely used to analyze deep beams with holes. However, may occur during the evolutionary optimization process,
no effective algorithm is yet available for building a brick element composed of six truss elements was
strut-and-tie models of deep beams. designed as a basic element unit. Several RC structures
Available tools for generating strut-and-tie models have been used as examples to demonstrate the capability
include the program of ANDERHEGGEW and of the proposed method and to verify the efficiency in
SCHLAICH [4] where a linear programming is used to application to real design problems. PERERA and VIQUE

Foundation item: Project(50908082) supported by the National Natural Science Foundation of China; Project(2009ZK3111) supported by the Science and
Technology Department of Hunan Province, China
Received date: 2011−06−13; Accepted date: 2011−08−31
Corresponding author: LIU Xia, Associate Professor, PhD; Tel: +86−13017387719; E-mail: [email protected]
1686 J. Cent. South Univ. Technol. (2011) 18: 1685−1692
[10] presented genetic algorithms to automatically element analysis constitute the whole population, i.e.
produce optimal strut-and-tie models for the design of each element is an individual of the population. As the
reinforced concrete beams. In the optimal configuration evolution progresses, the fittest elements with higher
of their methods, compressive struts are not enforced to sensitivity values will be conserved while the elements
be parallel, which allows for representing the physical with lower sensitivity values will be deleted. GESO is
reality of the flow of forces more consistently. heuristic, rather than adding and/or removing elements
Furthermore, they claimed that the method was more deterministically (according to optimal criteria) in ESO
simple and easier than the methods based on the concepts and BESO. An element that satisfies the removing
of evolutionary structural optimization. YUN and KIM criterion will be deleted or be reserved by chance
[11] presented a grid strut-and-tie model approach which according to probabilistic mechanism in evolutionary
allowed for a consistent and effective design of structural generations. The basic procedure of GESO can be
concrete, and employed a single type of grid strut-tie summarized as follows:
model where various load combinations can be 1) Build up an original finite element model and
considered. The approach performed an automatic determine boundary and load conditions.
selection of the optimal strut-and-tie model by evaluating 2) Impose an n-bits length chromosome whose
the capacities of struts and ties using a simple values of genes are all ‘1’ to each element in the case of
optimization algorithm. Eight reinforced concrete deep the present FEA mesh (n is selected arbitrarily).
beams were tested to failure and the shear wall with two 3) Solve structural static equation.
openings were used to verify the validity and 4) Calculate sensitivity number αi for each element.
effectiveness of the presented approach. 5) Rank individuals according to the sensitivity
Evolutionary structural method (ESO) [12] is based number αi;
on the simple idea by slowly removing the inefficient 6) Operate mutation over the last m individuals
material from a ground structure, and then the residual according to ranking, where mutation is slightly different
part evolves towards an optimum. Currently, with the aid from that of GA and only changes a nonzero gene to
of powerful finite element software, evolutionary zero.
structural method can be conveniently applied to various 7) Operate crossover in population.
structural topological optimization problems, and can 8) Reduce the depths of the elements whose genes
provide satisfactory optimal solution. The research on have only one ‘0’ to interim depth if the problem is
ESO is quite extensive and covers problems with stress, difficult for GESO or skip this step.
stiffness/displacement, frequency and buckling load 9) Remove elements whose values of genes are all
‘0’.
constraints [13−16]. The ESO method is considered to be
10) Repeat steps 3)−9) until an optimum is reached,
practical and effective optimal method to solve different
or one of the constraints reaches its limit.
problems. LIANG et al [17−19] applied this method to
construct strut-and-tie models for reinforced concrete
3 Three algorithms
structures and pre-stressed concrete structures. To get
stronger ability to find the optimal solution, LIU et al
3.1 Algorithm I
[20] improved the classic ESO method and proposed
3.1.1 Analysis model
genetic ESO algorithm (GESO). This work focuses on
In the finite element model of reinforcement and
the application of GESO algorithm on building concrete, non-coordinated plane elements with four
strut-and-tie models of deep reinforced concrete beams. nodes are used, and initially the member is full of
In particular, based on the difference in the elements reinforcement elements and concrete elements, as shown
used in finite element analysis and the treatment in in Fig.1.
constitutive relation of concrete and cracking 3.1.2 Concrete cracking and treatment of cracking
characteristics, three different optimization algorithms When the tensile stress of the concrete under
were proposed to build strut-and-tie models of deep loading exceeds the ultimate tolerable stress, the concrete
beams. Besides, the performances of the three algorithms cracks. Cracks reduce the rigidness of the concrete,
were compared to select the algorithm which is most because the tensile stress can no longer be carried along
suitable for the problem. the surface of cracks, although it can still be carried in
the direction parallel to the surface of cracks. In other
2 Genetic evolutionary structural optimiza- words, the material gets orthotropic. The modulus of
tion (GESO) elasticity becomes zero in the direction perpendicular to
the surface of crack, and remains the same in the
GESO combines the genetic algorithm with ESO, in direction parallel to the surface of crack. Therefore, the
which all elements of a structure generated by finite elasticity matrix Dc of concrete after cracking is [14]
J. Cent. South Univ. Technol. (2011) 18: 1685−1692 1687

Fig.1 Initial finite element model of reinforced concrete member in Algorithm I

E 0 0  failure and cracking failure. After cracks appear in

Dc   0 0 0  (1) concrete elements, the strength of the material is not
 0 0  G  completely lost, and this situation can be dealt with the
method presented in Section 3.1.2. However, with
where βG (0≤β≤1) is the shearing modulus along the deepening of the evolution process and reduction of
surface of cracks, reflecting the interlocking of reinforcement elements, cracks might appear in new
aggregates along the surface, β is dependent on the direction. This indicates that the concrete will be divided
roughness of the surface and the width of crack, which by two groups of cracks which are along different
decreases gradually with the increase of crack width, and directions, thus the strength of the material is completely
equals zero when the crack width is large enough and lost and the elastic matrix of the concrete becomes zero.
there is no interlocking among aggregates. When the principal compressive strain of concrete
Equation (1) holds in the local coordinate system element is beyond the ultimate tolerable strain, the
with x′ axis parallel to the surface of crack (as shown in concrete is crushed and cannot bear stress any more,
Fig.2). In the global coordinate system, the elasticity which implies that the elastic matrix becomes zero. The
matrix of concrete is concrete member is considered to reach the state of
Dc=T Dc T T (2) failure when a certain number of concrete elements are
crushed.
where T is the coordinate transformation matrix, 3.1.4 Stress and yielding of reinforcement
c 2 Considering that plane unit is used for
s2  2sc 
  reinforcement, Mises stress s of reinforcement is
T  s 2 c 2
2 sc  (3)
calculated according to Mises yielding conditions in
 sc  sc c 2  s 2 
  plastic pattern. Ideal elastic-plastic model is used for the
constitutive relation of reinforcement:
where c=cosθ; s=sinθ.
3.1.3 Failure and treatment after failure of concrete 1
s  ( x   y ) 2  6 xy
2
(4)
There are two failure modes for concrete: crushing 2
The element does not yield and its modulus of
elasticity remains unchanged when s<fy, while the
element yields with the modulus of elasticity equal to
zero when s≥fy.
3.1.5 Optimization model
The aim of building strut-and-tie models of deep
reinforced concrete beams is to find the main force
transmission path. In reinforced concrete member, the
tensile strength of the concrete varies largely. Therefore,
in the evolution process, the concrete elements with
tensile stress beyond tolerable the strength will
automatically be out of work, and the remaining
elements under compression resemble the compressive
bars in truss model. Besides, the compression strength of
the reinforcement almost remains unchanged in the
Fig.2 Stress states of concrete before and after cracking: evolution process, and will not be out of work due to the
(a) Before cracking; (b) After cracking similar compressive and tensile strength. Therefore, it is
1688 J. Cent. South Univ. Technol. (2011) 18: 1685−1692
optimal to set reinforcement with high working
efficiency as tensile bars in truss model. Algorithm I
starts with aligning reinforcement elements all over the
member, then eliminates the reinforcement elements with
low working efficiency and retains more effective
elements through calculation, achieving the total
elements of reinforcements at last.
The whole process can be formulated as solving an
optimization problem:
n
Fig.3 Simple supported beam (Unit: mm)
min f  W  Wi xi
i 1 is taken to be Es=210 GPa; the Poisson ratio is taken to
subject to  g1  u  u  0 * be ν=0.15; the width of the beam is b=250 mm; the
 thickness of the reinforcement unit is t=25 mm; εcu=
 g 2   c   cu  0
 (5) 0.003 3；εtu=0.000 15. In the finite element model, the
g3   s  f y  0 square length of 100 mm is used to divide the units. The
 x  {0, 1}
 i stopping criterion is taken to be the exceed of deflection
where Wi is the weight of the reinforcement element i; xi at the mid-span over the tolerable deflection u*=l/1 500=
is the indicator of the state of the reinforcement element, 12 mm.
and taken to be 1 if reinforcement element exists, and Figure 4 shows the strut-and-tie model obtained by
algorithm I, where the compressive bars in truss model
taken to be 0 otherwise; n is the total number of
represent concrete elements and the tensile bars represent
reinforcement elements in the initial model; u is the
reinforcement elements. Figure 4 also shows that the
displacement of the constrained point in the structure; u*
force transmission system consists of concrete
is the maximum tolerable displacement of the
compressive arc, longitudinal tensile reinforcement bars
constrained point in the structure, and can be determined
at the bottom and the inclined reinforcement bars at the
according to specific codes ； εc and εcu are the
bottom.
compressive strain and the tolerable compressive strain
of concrete, respectively; σs and fy are the tensile stress 3.2 Algorithm II
and tensile strength of the reinforcement, respectively. 3.2.1 Analysis model
3.1.6 Example of simply supported deep flexural In the reinforced concrete member, different
member dividing elements are used for reinforcement and
The example involves a simply supported beam concrete, i.e. non-coordinated plane unit with four nodes
subjected to two concentrated loads, as shown in Fig.3. used for concrete and bar unit used for reinforcement.
The two concentrated loads are P1=P2=1 200 kN; the Initially, the reinforcement units are aligned along the
modulus of elasticity of the concrete is taken to be Ec= four sides and the diagonal of the concrete unit, as shown
28.6 GPa; the modulus of elasticity of the reinforcement in Fig.5.

Fig.4 Strut-and-tie model obtained by Algorithm I: (a) Result obtained by GESO: (b) Truss model

Fig.5 Initial finite element model used in algorithm II

J. Cent. South Univ. Technol. (2011) 18: 1685−1692 1689
The treatment for cracking and failure of concrete is and be out of bearing loading. Thus, the loading can be
the same as that presented in Section 2. The evolution considered to pass through the uncracked tensile
process stops if any yielding is found in the reinforcement bars and compressive concrete bars.
reinforcement. Therefore, in the application of GESO algorithm to build
3.2.2 Optimization model strut-and-tie model, nonlinear property of the concrete
The aim of topology optimization of reinforcement can be neglected and the whole concrete and
is to find the optimal reinforcement bars layout satisfying reinforcement is considered as linearly elastic in finite
the strength and deformation requirements. To achieve element analysis. Moreover, performance index is
this, GESO algorithm starts with aligning reinforcement introduced to guide the progress of GESO algorithm. For
units all over the member, then eliminates the a displacement optimization problem, the performance
reinforcement units with low working efficiency and index can be calculated by
retains more effective units through calculation,
W0s u 0W0
achieving the total of reinforcements at last. IP  s
 (7)
The whole process can be formulated as solving an Wi uiWi
optimization problem: where Ip is the performance index; W0 and Wi are the
n
weights of the original design and the i-th step design,
min f  W  Wi xi
i 1 respectively; u0 and ui are the original and the i-th step
displacement of controlled point, respectively.
subject to  g1  u  u *  0
 The topology with maximal value of performance
 g 2   c   cu  0 index provides the optimal strut-and-tie model. This
 (6)
g3   s  f y  0 means that in algorithm III, it is unnecessary to
 x  {0, 1} distinguish concrete units with reinforcement units. In
 i
algorithm III, the dimension of the cross section of the
3.2.3 Example of simply supported deep flexural concrete and the quantity of the reinforcement bars can
member be designed according to the stress in the compressive
The same model involving simply supported deep and tensile bars in the obtained strut-and-tie model.
beam in Section 3.1.6 is used here, with the same 3.3.2 Optimization process
parameters for the model and the optimization algorithm. The process of algorithm III can be summarized as
Also, the same stopping criterion is used for the follows:
optimization algorithm. Figure 6 shows the layout of the 1) Considering all the boundary conditions and
reinforcement bars obtained by algorithm II, with the loading conditions, use plane units to establish the initial
ratio of volume of 5.2%. finite element model of the structure.
2) Use finite element analysis to evaluate the
sensitivity of each unit.
3) Delete the units with low sensitivity after
mutation and crossover operator.
4) Compute the structural performance parameter.
5) If the performance parameter is not less than 1,
repeat steps 2)−4) until the performance parameter is less
than 1. Then, topology with maximal performance
parameter value is selected.
6) Build the strut-and-tie model based on the
optimal topology, and compute the internal forces in the
bars.
7) Based on the calculated internal forces,
determine the type of material, the quantity of
Fig.6 Results obtained by algorithm II: (a) Calculation result; reinforcement bars and the dimension of the cross
(b) Distribution of steel bars section of concrete, which provides the optimal design.
3.3.3 Example of simply supported deep beam
3.3 Algorithm III The same model involving simply supported deep
3.3.1 Analysis model beam in Section 3.1.6 is used here, with the same
According to GB 50010—2002 [21], the limit state geometric parameters. The results obtained by algorithm
design method is adopted. When structural member III are shown in Fig.7(a), and the corresponding strut-
reaches its limit state, the concrete in tension will crack and-tie model is shown in Fig.7(b). The bending moments
1690 J. Cent. South Univ. Technol. (2011) 18: 1685−1692
0.029 m3 for the reinforcement used in the member.
Compared with the total volume of 0.055 m3 for the
reinforcement designed according to GB 50010—2002
[21], it can be seen that nearly half of the reinforcement
can be reduced using algorithm III.

3.4 Comparison on three algorithms

In this work, three algorithms using GESO are
proposed to find the optimal reinforcement alignment,
and their performances are illustrated by examples
involving a simply supported deep beam. Table 1 shows
the comparison results of the three algorithms.
On the whole, algorithm III is the best one among
the three algorithms because of its simplicity,
practicability and best fitness to GESO algorithm. In
algorithm I, although reinforcement is expected to be in
tension, and concrete is expected to be in compression,
GESO algorithm cannot automatically distinguish
Fig.7 Strut-and-tie model obtained by algorithm III: (a) Result
between units in tension and those in compression. This
obtained by GESO; (b) Strut-and-tie model (Unit: mm)
means that there will be large quantity of reinforcement
in compression if additional condition is not added. On
at the joints are calculated to be very small, implying that
the contrary, the result obtained by GESO will get much
the joints in the strut-and-tie model can be assumed to be
worse if additional condition is added to discard the
rigid, which is consistent with the real situation. The reinforcement element in compression. The topology
model can be treated as compressive and tensile bar changes continuously in evolution process, and the
model. Also the internal forces in the bars are calculated element, which is in compression in one evolution step,
and shown in Fig.7(b). It is noted that the model in may be in tension in another evolution step, thus it is not
Fig.7(b) is not a truss model, because the joints involved reasonable to discard the element just because it is once
are not hinge joints. in compression during the evolution process.
Based on the obtained strut-and-tie model and In algorithm II, bar and plane are respectively used
relevant concrete design codes, the dimension of the for reinforcement and concrete, and the influence of
cross section of the member is designed to be 210 mm × cracks is also considered, which makes the modeling in
2 400 mm, HRB400 class of reinforcement bar (fy= this algorithm more consistent with the real simulation,
360 N/mm2) and C30 concrete (fc=14.3 N/mm2, ft= compared with algorithms I and III. However, the result
1.43 N/mm2) are used for the member. The cross obtained by algorithm II is the worst among the three
sectional area of the reinforcement at the mid-span of the algorithms, because the result is greatly affected by the
member is designed to be 3 333 mm2, and the area values meshing and the initial cross-sectional area of the unit. In
of the three inclined reinforcement bars are 2 564, 894, the initial finite element model adopted by GESO, the
and 1 719 mm2, respectively, leading to the volume of meshing and the direction of the reinforcement unit are

Table 1 Comparison of three algorithms

Assessment of calculation process Result evaluation

Algorithm Considering Specification of

Element Stopping Strut-and-tie
nonlinearity of thickness of Complexity Practicality
type criterion model
material unit
Reinforcement: plane;
I Yes By experience Deflection High Simple Medium
Concrete: plane
Reinforcement: bar;
II Yes By experience Deflection High Complex Poor
Concrete: plane
Performance
parameter with
III All use plane No Arbitrary Low Simple High
respect to
strain energy
J. Cent. South Univ. Technol. (2011) 18: 1685−1692 1691
both specific, and the cross sectional area of the bar experience, which violates the intension of optimization.
(determined from experience) remains the same during According to GB 50010—2002 [21], the quantity of the
the evolution process. As a result, less reinforcement reinforcement is determined by the ultimate limit state of
elements are aligned at the places with higher the structure, thus it is acceptable not to consider the
reinforcement demand, more reinforcement elements are stiffness of the reinforcement and the nonlinearity of
aligned at the places with lower reinforcement demand, concrete when establishing strut-and-tie model. With this
and the evolution process stops earlier than expected. insight, in algorithm III, the performance index is used to
Although GESO algorithm can adjust the cross sectional control the evolution process, leading to the simplicity in
area of the bar in evolution process, that is, adjust the calculation process while ensuring the correct alignment
quantity of used reinforcement, the resulting effect is still of compressive and tensile bars. In conclusion, algorithm
not satisfactory. The reason is that the topology in the III is the best one (among the three algorithms) for
process is not the final optimal topology, and the change building the strut-and-tie model of reinforced concrete
of cross sectional area in evolution process may change member.
the force transmission path of the member, leading to the
derivation of the search path from the optimal solution. 4 Two examples
In algorithm II, the optimization in topology and cross
sectional area is expected to be operated simultaneously, Two examples of the algorithm III are presented in
and the result is not up to satisfaction. this section. The first one is a simply supported beam in
Furthermore, in both algorithm I and algorithm II, it which the ratio of span to height is 2 (see Fig.8(a)). The
cannot be guaranteed that all reinforcements are in GESO result is shown in Fig.8(b) and Fig.8(c) is the
tension in the final topology. Besides, in these two strut-and-tie model which consists of semi-circle
algorithms, the deflection is used as the stopping compression and radial tensile bars. The second one is
criterion, and it is required to specify the thickness of the two-pin supported beam in which the ratio of span to
unit or the cross sectional area of the unit from height is 2/3 (see Fig.9(a)), the GESO result is shown in

Fig.8 Simply supported beam—Example 1: (a) Ground field; (b) GESO result; (c) Strut-and-tie model

Fig.9 Simply supported beam—Example 2: (a) Ground field; (b) GESO result; (c) Strut-and-tie model
1692 J. Cent. South Univ. Technol. (2011) 18: 1685−1692
Fig.9(b) and Fig.9(c) is its strut-and-tie model which optimal design of reinforced concrete structures [J]. ACI Structural
Journal, 2001, 98(4): 431−442.
consists of three compressive bars. The difference of the
[7] BRUGGI M. Generating strut-and-tie patterns for reinforced concrete
strut-and-tie models between these two deep beams is structures using topology optimization [J]. Computers and Structures
stemmed from the relative location between the loadings 2009, 87(23/24): 1483−1495.
and the bearings. According to the two strut-and-tie [8] GUEST J K, MOEN C D. Reinforced concrete design with topology
model, reinforcements location can be determined. For optimization [C]// Proceedings of the 19th Analysis & Computation
Specialty Conference. Reston: American Society of Civil Engineers,
the beam in Fig.8, the reinforcements should be placed
2010: 445−454.
along the radius of the semi-circle; while for the beam in [9] KWAK H G, NOH S H. Determination of strut-and-tie models using
Fig.9, there is no need for any reinforcement to support evolutionary structural optimization [J]. Engineering Structures, 2006,
the loading acting on the top. Then, it can be concluded 28(10): 1440−1449.
[10] PERERA R, VIQUE J. Strut-and-tie modeling of reinforced concrete
that it is convenient to acquire practical reinforcement
beams using genetic algorithms optimization [J]. Construction and
layout for designers with the aid of algorithm III. Building Materials, 2009, 23(8): 2914−2925.
[11] YUN Y M, KIM B H. Two-dimensional grid strut-tie model approach
5 Conclusions for structural concrete [J]. Journal of Structural Engineering, 2008,
134(7): 1199−1214.
[12] XIE Y M, STEVEN G P. A simple evolutionary procedure for
1) Algorithm III has powerful capacity in searching
structural optimization [J]. Computers & Structures, 1993, 49(5):
more optimal results with less computational effort. 885−896.
2) The effectiveness of the proposed algorithm III [13] XIE Y M, STEVEN G P. A simple approach to structural frequency
has also been demonstrated from the strut-and-tie models optimization [J]. Computers & Structures, 1994, 53(6): 1487−1491.
which can be easily used to determine the places of [14] XIE Y M, STEVEN G P. Evolutionary structural optimization for
dynamic problems [J]. Computers & Structures, 1996, 58(6):
reinforcements of deep beams.
1067−1073.
3) Algorithms I and II also have the ability to [15] CHU D N. Evolutionary structural optimization method for systems
acquire the optimal reinforcement layout. But, the two with stiffness and displacement constraints [D]. Melbourne, Australia:
algorithms still need improvement to fit the practical Victoria University of Technology, 1997: 25−35.
[16] MANICKARAJAH D, XIE Y M, STEVEN G P. An evolutionary
purposes.
method for optimization of plate bucking resistance [J]. Finite
Elements in Analysis and Design, 1998, 29(3/4): 205−230.
References [17] LIANG Q Q, XIE Y M, STEVEN G P. Generating optimal
strut-and-tie models in pre-stressed concrete beams by
[1] DING Da-jun. Deep beam industrial building [J]. Construction performance-based optimization [J]. ACI Structural Journal, 2001,
Machines of Science, 1995, 25(3): 41−46. 98(2): 226−232.
[2] ROGOWSKY D, MACGREGOR J. The design of reinforced [18] LIANG Q Q, XIE Y M, STEVEN G P. Topology optimization of
concrete deep beams [J]. Concrete International, 1986, 8(8): 49−58. strut-and-tie models in reinforced concrete structures using an
[3] COLLINS M, ITCHELL D. A rational approach for shear design— evolutionary procedure [J]. ACI Structural Journal, 2000, 97(2):
The 1984 canadian code provisions [J]. ACI Structural Journal, 1986, 322−332.
83(6): 925−933. [19] LIANG Q Q. Performance-based optimization of structures [M].
[4] ANDERHEGGEN E, SCHLAICH M. Computer-aided design of London and New York: Spon Press, 2005: 134−254.
reinforced concrete structures using the truss model approach [C]// [20] LIU Xia, YI Wei-jian, SHEN Pu-sheng. Genetic evolutionary
Computer Aided Analysis and Design of Concrete Structures. structural optimization [J]. Journal of Constructional Steel Research,
Swansea UK: Pineridge Press, 1990: 539−550. 2008, 64(3): 305−311.
[5] ALSHEGEIR A. Analysis and design of disturbed regions with [21] GB 50010—2002. Ministry of Housing and Urban-Rural Development
strut-tie models [D]. West Lafayette: Purdue University, 1992: 11− of China. Code for design of concrete structure [S]. (in Chinese)
19. (Edited by HE Yun-bin)
[6] ALI A A, WHITE R N. Automatic generation of truss model for