Академический Документы
Профессиональный Документы
Культура Документы
Structures
journal homepage: www.elsevier.com/locate/structures
A R T I C LE I N FO A B S T R A C T
Keywords: The purpose of this article to design a low-weight cantilever reinforced concrete retaining wall with shear key by
Metaheuristic algorithms using an optimization algorithm, which is programmed in MATLAB. Grey Wolf Optimization (GWO) is an op-
Optimization timization technique which is developed by mimicking hierarchy and hunting methods of grey wolves. The
Retaining wall design Retaining Wall was formulated as an optimization problem based on ACI 318–05 code and Rankine's theory for
Grey Wolf optimization
lateral earth pressure. The constrains based on safety factors and design factors were used to reach the best
design. The geotechnical constraints were determined as the factor of safety against overturning, bending,
bearing capacity failure and the structural constraints were determined as the moment and shear capacities of for
elements of the wall. Two numerical examples for optimal design of retaining wall with/without shear key were
considered to evaluate the efficiency of GWO algorithm compared to similar studies.
1. Introduction structural restrictions. Once the dimensions are decided, designers must
check sufficiency of the wall for resistance to sliding and tipping,
Earth mass may lose its stability due to natural or artificial effects. bearing capacity of foundation, strength against bending and shear
One of the earth retaining structures is the retaining wall which sup- moment. These designing and analyze processes repeat with iterations
ports material behind it and resists to the pressure. In this study, the until the designer reach the final solution. Although this kind of rig-
cantilever earth retaining wall was chosen. orous approach, there is no guarantee that the last design is the best
Cantilever walls of reinforced concrete are still fairly common in solution. The infeasibility of this method is an unignorable fact. At this
urban areas because they are less susceptible to vandalism and often do point, optimization techniques may be helpful to prevent misusing
not require select backfill. Typically, they compete well in costs where material and time.
the wall is short (20 to 50 m in length) and not very high. They are also Optimization refers to the process of finding the best possible so-
widely used for basement walls and the like in buildings [1]. lution(s) for a particular problem [2]. One can achieve the optimum
A cantilever retaining wall is a common application for highway, design by formulating the process as a maximization or minimization
bridge and railway constructions and many other civil engineering problem. There are many different problems and algorithms for opti-
activities. These types of constructions entail cut and fill processes mization and thus this subject has been studied in various fields at
which can form soil slopes with the angle of inclination. A vertical stem several times. Optimization techniques for cantilever retaining walls
and a base slab are basically the elements which comprise a cantilever were developed in many years [3–7].
retaining wall. The Stem provides lateral support to the fill. In this case Metaheuristic algorithms, which they are highly popular recently,
there is a shear key on the bottom of the base slab. Soil may be retained can be convenient for this kind of complex structural optimization
at the two sides of the cantilever retaining wall; therefore, soil exert problems. In this method, sort of natural phenomenon is utilized to find
pressure from both sides, passive and active earth pressures. The shear the best candidates around all the other solutions. Kaveh and Soleimani
key increases stability of the wall with passive earth pressure arising [8] used Democratic Particle Swarm Optimization (DPSO) and Col-
from the soil in front of it. liding Bodies Optimization (CBO) algorithms, then compared to results
The designing process of cantilever wall is mostly based on people. with Improved Harmony Search (HIS) and Particle Swarm Optimization
Designers scientific intuition, vision, and experience are essential for (PSO) algorithms. Mohammed and Hemin [9] implemented to the
assuming the dimensions which they meet the geotechnical and problem the evolutionary method which is a combination of genetic
⁎
Corresponding author.
E-mail address: banuh@ktu.edu.tr (S.B. İkizler).
https://doi.org/10.1016/j.istruc.2019.09.013
Received 5 September 2019; Received in revised form 26 September 2019; Accepted 30 September 2019
2352-0124/ © 2019 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
E.N. Kalemci, et al. Structures 23 (2020) 245–253
246
E.N. Kalemci, et al. Structures 23 (2020) 245–253
For sliding analyze, the forces acting on the base slab are con- Mn is nominal flexural strength. Nominal strength is the strength
sidered. which is calculated by using required assumptions and equations of a
The ratio of resisting and driving forces is the factor of safety: cross section. A safer design strength can be reached by multiplying
nominal strength by a reduction factor (ɸ ). Md is the design moment at
∑ FR critical section which is calculated by using applied factored loads and
FSS =
∑ FD (5) forces. The design strength must be equal or greater than required
strength.
The driving force is the sum of the horizontal components of active
Flexural strength reduction factor is 0.9 according to the ACI
force:
318–05 [27] for tension-controlled sections. Nominal flexural strength
∑ FD = PAcosβ (6) can be described as:
a
Resisting forces: Mn = T ⎛d − ⎞
⎝ 2⎠ (17)
2 2
∑ FR = (∑ N ) ∗ tan ⎛ ∗ φbase ⎞ + ∗ B ∗ cbase + ∑ PP (7)
T is the tension force acting on a nodal zone:
⎝3 ⎠ 3
T = As f y (18)
∑ N: Sum of the weight of the retaining wall, surcharge load and
247
E.N. Kalemci, et al. Structures 23 (2020) 245–253
wherec is distance from extreme compression fiber to neutral axis. 1.7q + 1.4γc X 5 + 1.4γfill H 1.4Wbs q + 2qmin ⎤ 2
Mdheel = ⎡
⎢⎜
⎛ ⎞
⎟ + ⎛ ⎞−⎛ 1 ⎞ LH
β1, which is a factor based on fc , is 0,85 for fc ≤ 28MPa. β1 shall be
⎣⎝ 2 ⎠ ⎝ 3 ⎠ ⎝ 6 ⎠⎥
⎦
reduced linearly at a rate of 0.05 for each 7 MPa of strength in excess of
(33)
28 MPa, but β1 shall not be taken less than 0.65 according to the ACI
318–05 code [27] Chapter 10. W + Wbsdh
With C = T Vdheel = ⎡ (1.7q + 1.4γc X 5 + 1.4γfill H ) + 1.4 ⎛ bs ⎞
⎢
⎣ ⎝ 2 ⎠
As f y = 0, 85. fc ba (21) qdh + qmin
− 0.9 ⎛ ⎞ ⎤ (Lh − dh)
⎝ 2 ⎠⎦⎥ (34)
As f y
a= Similarly, the q1 calculated by using qmax and qmin is pressure on the
0.85. fc b (22)
intersection of stem with the heel slab. Wbs is the weight of the soil at
For shear capacity design: the triangle area above stem. So, qdh is the soil pressure at the critical
section of heel slab, which is at the distance dh (=X5-cc) away from
ɸVn ≥ Vd (23) front face of the wall. Wbsdh is the weight of the soil at the triangle area
ɸ is 0,75 for shear and torsion according to the code. The nominal above stem at the critical section.
shear strength is: The shear key acts during the sliding, it is not considered as a safety
factor under other conditions.
Vn = ɸ0.17 fc bd (24) Moment and shear strength calculations for shear key;
Required shear and moment strengths of the structure (Md , Vd ) are ⎡ K p γbase (X 8) ⎞ ⎤
3
calculated for stem, toe, heel and the shear key separately base on the Mdkey = ⎢ ⎜⎛ 2 ⎟
⎥
3 (35)
code. ⎣⎝ ⎠⎦
Required shear and moment strengths of the stem, a vertical beam,
K p γbase dsh + K p2 γbase X 8 ⎞
are combination of horizontal earth pressure and surcharge load. Vdkey = ⎡ ⎜⎛ 2 ⎟ max ([(X 8 − dsh ) 0]) ⎤
⎢ 2 ⎥
According to ACI 318–05 [27] the required strength can be calculated ⎣⎝ ⎠ ⎦ (36)
with the load factors which are shown in the equation below. Critical section of shear key is selected at the dsh = (X 7 − cc ) dis-
U = 1.4D + 1.7L + 1.7H (25) tance away from junction of base slab and shear key for shear strength.
As , minimum area of flexural reinforcement, shall not be less than
Where D or L reduce the effect of H the equation become:
that given by
U = 0.9D + 1.7H (26)
fc
D is dead load, L is live load and the H is the load caused by earth A smin = 0.25 bd
fy (37)
pressure, groundwater pressure.
Stem; and not less than (1.4/ f y bd ) according to ACI 318–05 [27] Chapter
10.
(Hs + H )2 (Hs + H )3
Mdstem = 1.7 ⎡qK a cos β + K a γfill cos β ⎤ The minimum steel reinforcement ratio (ρmin ) can be calculated with
⎢
⎣ 2 6 ⎥
⎦ (27)
given formula:
(Hs + H − ds )2 ⎤ As
Vdstem = 1.7 ⎡qK a cos β (Hs + H − ds ) + K a γfill cos β ρ=
⎢
⎣ 2 ⎥
⎦ bd (38)
(28) The reinforcement ratio, ρb, which produces balanced strain condi-
Hs = (Lh)tanβ (29) tions under flexure, is the limit for maximum steel reinforcement ratio
ρmax.
ds = X 5 − cc (30)
⎛ 0.85β1 fc ⎞ ⎛ 600 ⎞
Lh is the heel part of the base slab and cc is the depth of concrete ρb = ⎜
f y ⎟ ⎜ 600 + f y ⎟ (39)
cover. ⎝ ⎠⎝ ⎠
The critical sections of the structure must be checked. For stem, due Development length, ld , must be adequate to make the desired bond
to the maximum moment, the base of the stem is a critical section for strength between concrete and steel. For the no.6 and smaller bars
moment and shear. (db < 19mm), ld is described by ACI 318–05 [27] as:
Moment and shear force at the base slab are resulted from the
weight of retained soil and concrete, and the surcharge load above the ⎛ 12f y ψt ψe λ ⎞
ld = ⎜ db
slab and the earth pressure below the slab. For the shear strength, the 25 fc ⎟ (40)
⎝ ⎠
critical section of the toe is at distance dt (X 5 − cc ) away from rear face
of the wall. where db is the diameter of the bar. For the no.7 and larger bars
Moment and shear strength calculations for toe; (db > 19mm):
248
E.N. Kalemci, et al. Structures 23 (2020) 245–253
⎛ 12f y ψt ψe λ ⎞ Table 2
ld = ⎜ db Input Parameters for Example 1.
20 fc . ⎟ (41)
⎝ ⎠
Input parameters Symbol Value Unit
In adding to this ld shall not be less than 300 mm. In this case ψt
Stem height H 3 m
(modification factor for casting location), ψe (modification factor for
Reinforcing steel yield strength fy 400 MPa
development length), λ (factor base on reinforcement coating) are
Concrete compressive strength fc 21 MPa
coefficients which are taken as 1.0.
Concrete cover cc 7 cm
The development length of a standard hook, ldh , defined by code as : Shrinkage and temperature reinforcement percentage ρst 0,002 –
Surcharge load Q 20 kPa
⎛ 0.24f y ⎞ Backfill slope Β 10 °
ldh = ⎜ db
fc ⎟ (42)
Internal friction angle of base soil φbase 0 °
⎝ ⎠ Internal friction angle of retained soil φ 36 °
but ldh shall not be less than the smaller 8db and 150 mm. Unit weight of retained soil γfill 17,5 kN/m3
Unit weight of base soil γbase 18,5 kN/m3
Unit weight of steel Gs 78,5 kN/m3
3. Constrains
Unit weight of concrete γc 23,5 kN/m3
Depth of soil in front of wall D 0,5 m
The possible candidates for solution of the optimization problem Cost of steel CS 0,4 $/kg
must be searched in a confined space. Constrains, which constitute the Cost of concrete CC 40 $/kg
search spaces, were determined for structural and geotechnical design. Factor of safety for overturning stability FSO
design
1,5 –
Factor of safety for sliding FS S 1,5 –
design
3.1. Geotechnical constrains Factor of safety for bearing capacity FS B 1,5 –
design
Cohesion of the base soil cbase 125 kPa
Factor of safety based on stability against slippage, overturning and
bearing capacity failure must be adequate compared to design factor of
safety. Table 3
FSO design The boundaries of design variables for Example 1 (Camp and Akin, [20]).
g (1) = −1≤0
FSO (43) Design variable Unit Lower bound Upper bound
Table 4
Low-weight design variables for Example 1.
X1 X2 X3 X4 X5 R1 R2 R3
GWO (This study) 1.80 0.67 0.21 0.20 0.28 82 (28 * 10 mm) 15 (3 * 18 mm) 15 (3 * 18 mm)
SGA 1.71 0.65 0.20 0.20 0.27 77 (27 * 10 mm) 23 (12 * 10 mm) 17 (10 * 10 mm)
BSA 1.71 0.64 0.20 0.20 0.27 77 (27 * 10 mm) 14 (9 * 10 mm) 14 (9 * 10 mm)
BB-BC 1.74 0.65 0.20 0.20 0.27 77 (27 * 10 mm) 17 (10 * 10 mm) 17 (10 * 10 mm)
ISA 1.84 0.75 0.29 0.20 0.27 34 (6 * 16 mm) 15 (3 * 18 mm) 15 (3 * 18 mm)
DE 1.87 0.62 0.29 0.20 0.27 34 (6 * 16 mm) 19 (4 * 16 mm) 16 (5 * 14 mm)
GA 1.91 0.58 0.27 0.20 0.28 40 (17 * 10 mm) 21 (8 * 12 mm) 15 (3 * 18 mm)
BBO 1.84 0.74 0.27 0.20 0.27 37 (16 * 10 mm) 14 (9 * 10 mm) 14 (9 * 10 mm)
ES 1.84 0.69 0.32 0.22 0.28 26 (9 * 12 mm) 22 (6 * 14 mm) 29 (7 * 14 mm)
APSO 1.84 0.57 0.27 0.20 0.27 40 (17 * 10 mm) 28 (13 * 10 mm) 17 (10 * 10 mm)
PSO 1.84 0.74 0.29 0.20 0.27 33 (15 * 10 mm) 14 (9 * 10 mm) 14 (9 * 10 mm)
Table 5 Table 6
Low-weight design final results for Example 1. Input Parameters for Example 2.
Steel Concrete Design Objective (kg/m ) Input Parameters Symbol Value Unit
kg/m m3/m Best Mean
Stem height H 4,5 m
GWO (This study) 94.3013 1.1181 2721.7915 2748.7809 Reinforcing steel yield strength fy 400 MPa
SGA 83.1023 1.0644 2584.46 2589.00 Concrete compressive strength fc 21 MPa
BSA 80.6929 1.0646 2582.2 2582.4 Concrete cover cc 7 cm
BB-BC 82.1383 1.0750 2608.38 – Shrinkage and temperature reinforcement percentage ρst 0,002 –
ISA – – 2665.8027 2677.5681 Surcharge load q 30 kPa
GA – – 2744.8 2850.9 Backfill slope β 0 °
DE – – 2726.5 2851.0 Internal friction angle of retained soil φ 28 °
ES – – 2762.4 2845.0 Unit weight of retained soil γfill 18,5 kN/m3
BBO – – 2665.8 2677.7
Internal friction angle of base soil φbase 34 °
PSO – – 2665.8 2668.3
APSO – – 2668.0 2687.6 Unit weight of concrete γc 23,5 kN/m3
FA – – 2666.5 2673.4 Unit weight of base soil γbase 17 kN/m3
CS – – 2665.8 2665.8 Unit weight of steel Gs 78,5 kN/m3
Depth of soil in front of wall D 0,3 m
Cost of steel CS 0,4 $/kg
Cost of concrete CC 40 $/kg
10000
Factor of safety for overturning stability FSO 1,5 –
design
Factor of safety for sliding FS S 1,5 –
8000 design
–
Weight (kg/m)
4000
Table 7
The boundaries of design variables for Example 2 (Camp and Akin, [20]).
2000
Design variable Unit Lower bound Upper bound
0 X1 m 1,96 5,5
0 400 800 1200 1600 2000 m
X2 0,65 1,16
Iteration X3 m 0,25 0,5
X4 m 0,25 0,5
X5 m 0,4 0,5
Fig. 3. Convergence history for low-weight design for Example 1.
X6 m 1,96 5,5
X7 m 0,20 0,5
X8 m 0,20 0,5
R1 – 1 264
2500 R2 – 1 264
R3 – 1 264
–
Weight (kg/m)
2000 R4 1 264
1500
The mathematical model which represents the encircling process is:
1000 → → → →→ →
X (t + 1) = Xp (t ) − A |C X p (t ) − X (t )| (57)
500 →
where t is the current iteration, x (t + 1) is the next location of the
→ →
wolf, x p (t ) is the location of the prey. A denotes a coefficient factor:
0
0 5 10 15 20 25 30 35 →
A = 2→
a r1 − →
a (58)
Run no
→
where a is a vector whose value decreases linearly from 2 to 0 over
Fig. 4. The low-weight design for each run for Example 1. the course of iterations.
→ →
X (t ) indicates the current location of the wolf. C is another
250
E.N. Kalemci, et al. Structures 23 (2020) 245–253
Table 8
Low-weight design variables for Example 2.
X1 X2 X3 X4 X5 X6 X7 X8 R1 R2 R3 R4
GWO (This study) 3,21 0,66 0,39 0,25 0,47 2,49 0,20 0,49 140 (19 * 16 mm) 51 (20 * 10 mm) 90 (21 * 12 mm) 5 (4 * 12 mm)
SGA 3,45 0,65 0,41 0,25 0,42 2,23 0,20 0,50 125 (22 * 14 mm) 51(20 * 10 mm) 102(24 * 12 mm) - (27 * 10 mm)
BSA 3,46 0,65 0,41 0,25 0,42 2,72 0,20 0,50 - (30 * 12 mm) 37(16 * 10 mm) 107(25 * 12 mm) 7 (6 * 10 mm)
BB-BC 3.76 0.68 0.41 0.25 0.40 3.22 0.20 0.49 124 (22 * 14 mm) 45 (18 * 10 mm) 116 (20 * 14 mm) 7 (6 * 10 mm)
Table 9 → → → →→ →
X1 = Xα − A1 |C1 X α − X | (60)
Low-weight design final results for Example 2.
→ → → →→ →
Steel Concrete Design objective (kg/m ) X2 = Xβ − A2 |C2 X β − X | (61)
kg/m m3/m Best Mean
→ → → →→ →
X3 = Xδ − A3 |C3 X δ − X | (62)
GWO (This study) 267.1167 3.0840 7514.6204 7606.1421
BB-BC 270.9573 3.087 7525.407 – → → →
SGA 263,1200 3,0552 7442.92 7483.89 → X + X2 + X3
X (t + 1) = 1
BSA 246,0707 3,0561 7428.5 7484.3 3 (63)
The grey wolves do not attack until the prey stop. This deceleration
→ →
10000
of the prey modelled by decreasing the value of a i.e. value of A .
→
|A | < 1 situation forces the wolves to attack towards the prey.
→
8000 |A | > 1 situation forces the wolves to find a better prey.
→
Weight (kg/m)
end
7650 Where dim, ub, lb and PopSize are the number design variables, upper
bound and lover bound for design variables and the population size,
7600 respectively. rand is a random number ranging from 0 to 1.
251
E.N. Kalemci, et al. Structures 23 (2020) 245–253
reinforcement (R1-R3), which are discrete variables, as indicated in example can be obtained in the iteration number 1000. To show the
Table 3. The upper and lower boundaries of design variables were convergence of the objective function the developed program was
shown in the Table 3. continued until the maximum iteration number 2000.
Objective function of the design is weight of the structure. Previous Fig. 6 shows the low weight design for each run.
studies considered for comparison based on same boundaries and de-
sign parameters. The best results obtained in this study using GWO and 6. Conclusions
the results obtained by Lopez [12] according to Search Group Algo-
rithm and Backtracking Search Algorithm (SGA), Camp and Akin [20] The RC cantilever retaining wall with the shear key was designed
according to Big Bang–Big Crunch Algorithm (BB-BC), Gandomi et al., and analyzed by using the opensource software of GWO algorithm
[11] according to Genetic Algorithm (GA), Differential Evolution (DE), MATLAB code. One of the metaheuristic algorithms, Grey Wolf
Evolutionary Strategy (ES), Biogeography Based Optimization Algo- Optimization algorithm proposed by Mirjalili et al. [26]. Because of the
rithm (BBO), Gandomi et al., [13] according to Particle Swarm Opti- simplicity of the method, mathematical formulations of the design were
mization (PSO), Accelerated Particle Swarm Optimization (APSO), coded easily. Low-weight design of the wall was aimed in this study.
Firefly Algorithm (FA), Cuckoo Search Algorithm (CS), Gandomi et al., Two numerical examples were studied to evaluate the performance of
[10] according to Interior Search Algorithm (ISA) were summarized in GWO algorithm. Each problem was executed 30 times. The results were
Table 4. examined and compared with the similar works. The results were
As listed in Table 4, similar results were gotten from stated algo- consistent with the results of the other similar studies. It shows that the
rithms as design variables. Table 5 shows the best and mean objective GWO algorithm is a competitive optimization method for a RC canti-
function value for each algorithm. Best objective function, lever retaining wall with the shear key design, obviously.
2721.7915 kg/m, obtained with proposed method is approximately
lighter 0.84% than the design presented by Gandomi et al., [11] ac- Declaration of Competing Interest
cording to GA, 1.5% lighter than ES, 0.17% lighter than DE. In com-
parison with similarity between results of GWO and the subjected op- The authors declare that they have no known competing financial
timization algorithms, the difference between them looks insignificant. interests or personal relationships that could have appeared to influ-
Still, it may be mentioned as the design of proposed method approxi- ence the work reported in this paper.
mately 1.6% heavier than the design of subjected algorithms. Dissim-
ilarity might result from many factors. For instance, the formulations References
were not stated clearly in the subjected studies. There may be some
inconsistencies, which effect the results directly, in the formulation [1] J.E. Bowles, Foundation Analysis and Design Ed. Mc Graw Hill, USA, 2002.
process. Secondly, in this study each example was run 30 times, while [2] Mirjalili S. Moth-flame optimization algorithm: a novel nature-inspired heuristic
paradigm. Knowl. Based Syst. 2015;89:228–49.
the others were run 100 times. So, it is possible that proposed method [3] Saribas A, Erbatur F. Optimization and sensitivity of retaining structures. J.
did not reach its best solution among the candidate solutions. Elapsed Geotech. Eng. 1996;122(8):649–56.
time for one run of this example is 269.988677s. [4] Singla S, Gupta S. Optimization of reinforced concrete retaining walls of varying
heights using relieving platforms. Int. J. Eng. Res. Technol. 2015;4(06):1071–7.
In Fig. 3, convergence history for low-weight design is shown. As [5] U.A. Mahmood, B. Alam, Design optimization of cantilever retaining wall using
seen from this figure, after initial solution, optimal results for this ex- direct optimal design simulation, Int. J. Adv. Struct. Geotech. Eng. 4 (4) (2015)
ample can be obtained in the iteration number 1000. To show the ISSN: 2319-5347.
[6] K.S. Sable, A. Patil Archana, Optimization of retaining wall by using optimtool in
convergence of the objective function the developed program was MATLAB, Int. J. Eng. Res. Technol. (IJERT) 1 (6) (2012) ISSN: 2278-0181.
continued until the maximum iteration number 2000. [7] M.A. Bhatti, Retaining Wall Design Optimization with MS Excel Solver 2006, 1–15.
Fig. 4 shows the low weight design for each run. 10.1061/40878(202)34.
[8] Kaveh A, Soleimani N. CBO and DPSO for optimum design of reinforced concrete
cantilever retaining walls. Asian J. Civ. Eng. 2015;16(6):751–74.
5.2. Example 2 [9] Mohammad FA, Ahmed HG. optimum design of reinforced concrete cantilever re-
taining walls according Eurocode 2. Athens J. Technol. Eng. 2018;5(3):277–96.
In this case, the RC retaining wall with base shear key was studied [10] Gandomi AH, Kashani AR, Zeighami F. Retaining wall optimization using interior
search algorithm with different bound constraint handling. Int. J. Numer. Anal.
for the same objective function. Cohesionless soil conceived for this Meth. Geomech. 2017;41:1304–31.
example to design a better structure for long-term. The design para- [11] Gandomi AH, Kashani AR, Roke DA, Mousavi M. Optimization of retaining wall
meters, dimensions of the wall, indicated in the Table 6. design using evolutionary algorithms. Struct. Multidisc. Optim. 2017;55(3):809.
[12] R. Holdorf Lopez, Optimum project of cantilever retaining wall using search group
The upper and lower boundaries were shown in the Table 7. Four algorithm and backtracking search algorithm, Rev. Int. métodos numér. cálc. diseño
new variables were arisen with the shear key: 3 geometric variables for ing (2017).
the geometrical dimensions of the wall (X1-X8) and one reinforcement [13] Gandomi AH, Kashani AR, Roke DA, Mousavi M. Optimization of retaining wall
design using recent swarm intelligence techniques. Eng. Struct. 2015;103:72–84.
of the shear key (R1-R4). [14] Aydogdu İ. Cost optimization of reinforced concrete cantilever retaining walls
Previous studies considered for comparison based on same bound- under seismic loading using a biogeography-based optimization algorithm with
aries and design parameters. The best results of proposed method Levy flights. Eng. Optim. 2017;49(3):381–400.
[15] Kaveh A, Jafarpour Laiden D. Optimal design of reinforced concrete cantilever re-
(GWO) and the results obtained by Lopez [12] according to Search
taining walls using CBO ECBO and VPS algorithms. Asian J. Civil Eng.
Group Algorithm and Backtracking Search Algorithm (SGA), Camp and 2017;18(4):657–71.
Akin [20] according to, Big Bang–Big Crunch Algorithm (BB-BC) were [16] Tonne VR, Mohite PM. Optimization and improvement in stability of counterfront
retaining wall with relief shelf. Int. J. Res. Eng. Technol. 2015;4(4):447–51.
indicated in Table 8.
[17] Molina-Moreno F, García-Segura T, Martí JV, Yepes V. Optimization of buttressed
Table 9 shows that similar results were gotten from stated algo- earth-retaining walls using hybrid harmony search algorithms. Eng. Struct.
rithms as design variables. The best and mean objective function value 2017;134:205–16.
for each algorithm were indicated in Table 8. Best objective function, [18] Khajehzadeh M, Taha MR, El-Shafie A, Eslami M. Modified particle swarm opti-
mization for optimum design of spread footing and retaining wall. J. Zhejiang
7514.6204 kg/m, obtained with proposed method is lighter 0.14% than University-Sci. A 2011;12(6):415–27.
the design presented by Camp and Akin [20] according to BB-BC and [19] Temur R, Bekdaş G. Teaching learning-based optimization for design of cantilever
approximately heavier %0.66 the design of subjected algorithms. Dis- retaining walls. Struct. Eng. Mech. 2016;57(4):763–83.
[20] Camp CV, Akin A. Design of retaining walls using big bang-big crunch optimization.
similarity might result from the factors which is mentioned before. The J. Struct. Eng. 2012;138(3):438–48.
elapsed time required for one run of this example is 372.074457s. [21] Sheikholeslami R, Gholipour Khalili B, Zahrai SM. Optimum cost design of re-
In Fig. 5, convergence history for low-weight design is shown. As inforced concrete retaining walls using hybrid firefly algorithm. IACSIT Int. J. Eng.
Technol. 2014;6(6):465–70.
seen from this figure, after initial solution, optimal results for this
252
E.N. Kalemci, et al. Structures 23 (2020) 245–253
[22] Sheikholeslami R, Khalili BG, Sadollah A, Kim JH. Optimization of reinforced [25] Kumar VN, Suribabu CR. Optimal design of cantilever retaining wall using differ-
concrete retaining walls via hybrid firefly algorithm with upper bound strategy. ential evolution algorithm. Int. J. Optim. Civil Eng. 2017;7(3):433–49.
KSCE J. Civ. Eng. 2016;20:2428. [26] Mirjalili S, Mirjalili SM, Lewis A. Grey Wolf optimizer. Adv. Eng. Softw.
[23] M. Ghazavi, B.S. Bonab, Optimization of reinforced concrete retaining walls using 2014;69:46–61.
ant colony method, in N. Vogt, B. Schuppener, D. Straub, G. Bräu (Hg.), [27] American Concrete Institute. Building Code Requirements for Structural Concrete
Geotechnical Safety and Risk. ISGSR 2011, Bundesanstalt für Wasserbau, Karlsruhe, and Commentary (ACI 318-05), Detroit, 2005.
2011, pp. 297–305. [28] Muro C, Escobedo R, Spector L, Coppinger RP. Wolf-pack (Canis lupus) hunting
[24] Jasim NA, Al-Yaqoobi AM. Optimum design of tied back retaining wall. Open J. Civ. strategies emerge from simple rules in computational simulations. Behav. Process.
Eng. 2016;6(2):139–55. 2011;88(3):192–7.
253