Decision Analytics Journal 3 (2022) 100067

Contents lists available at ScienceDirect

Decision Analytics Journal

journal homepage: www.elsevier.com/locate/dajour

Optimal design of low- and high-rise building structures by Tribe-Harmony

Search algorithm
Mahdi Azizi a ,∗, Siamak Talatahari b , Mahla Basiri a , Milad Baghalzadeh Shishehgarkhaneh c
a Department of Civil Engineering, University of Tabriz, Tabriz, Iran
b
Data Science Institute, Faculty of Engineering & Information Technology, University of Technology Sydney, NSW 2007, Australia
c
Department of Civil Engineering, Islamic Azad University of Tabriz, Tabriz, Iran

ARTICLE INFO ABSTRACT

Keywords: In this paper, optimum design of building structures is conducted by metaheuristic algorithms due to the
Optimum design shortcomings of the conventional design methods in providing economical designs. The Harmony Search (HS)
Building structure is utilized as the main algorithm which was developed based on the musical process of searching for the
Tribe-Harmony Search algorithm
optimal condition of harmony to produce an appropriate search approach for design optimization purposes.
Metaheuristic
Besides, the Tribe-Harmony Search (Tribe-HS) algorithm is also proposed for the first time in this paper to
Optimization
improve the performance of the HS algorithm which divides the HS’s searching phase into three distinct phases
called ‘‘tribes’’, lead the primary algorithm to prioritize global search in the early iterations while resolving
local search in the later iterations. Three building structures with 135, 3860, and 8272 structural members
are used as design examples to demonstrate the suggested method’s capacity to solve challenging optimization
problems. The recommended method’s overall performance is compared to that of the conventional Harmony
Search algorithm and ten alternative metaheuristic algorithms using a total of 30 independent runs in each
instance for statistical reasons. The findings demonstrated that the suggested method outperformed the other
metaheuristics for the study design instances.

1. Introduction several decades. The Genetic Algorithm (GA) [1], Differential Evolu-
tion (DE) [2], Ant Colony Optimization (ACO) [3], Particle Swarm
Structural optimization is the process of finding the best configura- Optimization (PSO) [4], Charged System Search (CSS) [5–7], Material
tions of elements for a structural system with consideration of design Generation Algorithm (MGA) [8,9], and Chaos Game Optimization
constraints and a fully developed objective function. In most cases, the (CGO) [10,11] are some of these methods. Additionally, several of the
total construction cost of the structure is considered objective functions aforementioned metaheuristic algorithms have been used in various en-
in which the topology, size and shape of the structural systems have gineering problems, yet none of them has ever proven to be extremely
the main role in this purpose. Design constraints are the other aspect the optimal method. Regardless matter how strong the algorithms are,
of the structural optimization process which demonstrate the structural several improvements may be made to basic algorithms to provide
behavior, including the deformation, force, fatigue, and damping of more accurate results with less computing time. These advancements
structural members. Structural optimization considers these objective are intended to enhance current algorithms or hybridize two or three
functions and design constraints to provide a better configuration of of them to achieve reasonable outcomes for objective functions.
elements for a structural system. In other words, structural optimization Many of the significant optimization algorithm improvements in-
is an intelligently developed design concept in which the optimal clude: the enhanced PSO introduced by Wang, et al. [12], improved
configuration of the structural components is considered by means of ACO proposed by Kaveh and Talatahari [13], upgraded Whale
a fully-established optimization algorithm. Optimization Algorithm proposed by Azizi, et al. [14], hybrid GA-
The process of finding an optimal configuration of elements for a Imperialistic Competitive Algorithm developed by Fasahat and Pay-
structural system requires an optimization algorithm that should be vandy [15] and hybrid Ant Lion Optimizer-Jaya approach presented
capable of providing better results than the traditionally developed by Azizi, et al. [16]. Meanwhile, some other developed approaches are
design approaches. In this regard, the metaheuristic-based optimiza- mentioned in Refs. [10,17–24].
tion approaches could be considered as optimization algorithms that While designing diverse engineering structures, one of the most
have been utilized for optimization purposes in different fields for difficult challenges for structural engineers is optimizing the structure’s

∗ Corresponding author.
E-mail address: [email protected] (M. Azizi).

https://doi.org/10.1016/j.dajour.2022.100067
Received 1 April 2022; Received in revised form 2 May 2022; Accepted 6 May 2022
Available online 13 May 2022
2772-6622/© 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

weight, shape, cost of construction, topology, and manufacturing time design sections may be chosen to determine the search space and give
while considering numerous constraints. These features are frequently an appropriate structural design configuration. The Harmony Search
incorporated in an optimization problem that considers the best design (HS) algorithm is chosen as the primary optimization algorithm sug-
sections of structural elements for minimizing the structure’s weight by gested by Geem, et al. [40] and is based on the musical practice of
considering inequality and equality constraints. Farshchin, et al. [25] attempting to achieve perfect harmony. Zhang and Geem [41] reviewed
discussed the optimum design of multiple steel frame structures using and focused on the historical development of Harmony Search (HS)
School Based Optimization (SBO) algorithm. Khodadadi, et al. [26] pro- algorithm structure instead of applications; they elucidated adaption
posed the multi-objective version of a recently proposed metaheuristic of original operators of the basic harmony search, parameter adaption,
algorithm called Crystal Structure Algorithm (CryStAl) for engineering hybrid methods, handling multi-objective optimization problems and
constraint handling. There has been a growing interest in enhancing the
optimization problems. Kaveh and BolandGerami [27] proposed an up-
overall efficiency of this algorithm as a result of its many applications
graded version of Colliding Body Optimization (CBO) algorithm called
in various optimization domains [42–48]. Numerous improved, modi-
‘‘Cascade Enhanced CBO’’ for large-scale steel space frames’ optimiza-
fied, or hybridized variants of the HS algorithm have been suggested
tion. Talatahari, et al. [28] combined eagle strategy with DE algorithm
and used for engineering design optimization as search techniques.
for design optimization of different frame structures with steel sections.
Keshtegar, et al. [49] proposed a modified version of harmony search
Maheri, et al. [29] developed an improved version of the Honey Bee (HS) algorithm for improvisation procedure. Ouyang, et al. [50] sug-
Mating Optimization (HBMO) algorithm for optimum element determi- gested an improved version of HS algorithm for problems related to
nation of side-sway structural frames with steel design sections. Kaveh engineering design considering the general iteration models. Moon,
and Ghazaan [30] discussed size optimization of skeletal steel struc- et al. [51] proposed a novel approach to estimate the vanishing point
tures while an improved Whale Optimization Algorithm (WOA) has using a harmony search (HS) algorithm; they claimed that HS stably
been proposed for this purpose. Kazemzadeh Azad, et al. [31] utilized estimates vanishing points with respect to statistics when compared
the Big Bang–Big Crunch (BB–BC) as the main optimization algorithm with RANSAC. Yi, et al. [52] discussed the engineering design of
for optimum design of different frame structures, while the Upper different optimization problems by using parallel chaotic local search
Bound Strategy (UBS) is implemented for enhancing the computational improved HS algorithm. Keshtegar, et al. [53] used a dynamic harmony
complexity of the main algorithm. Kazemzadeh Azad, et al. [32] dis- search (DHS) algorithm for accurate calibration of strength and strain
cussed the optimum design of steel frame structures by combining the enhancement ratios of FRP-confined concrete. Hasanipanah, et al. [54]
BB–BC algorithm with UBS to reduce the number of structural analyses proposed an ANN-adaptive dynamical harmony search algorithm for
needed as much as possible throughout the optimization procedure. accurate prediction of blast-induced flyrock. Yi, et al. [55] developed
Hasançebi [33] utilized Evolutionary Strategy (ES) for economic design an improved HS algorithm considering a multi-level screening strat-
optimization of multiple frame structures. Tort, et al. [34] discussed egy for design optimization of engineering problems. Sheikholeslami,
the optimal design of towers in real-world engineering for lattice et al. [56] discussed the optimum design of water distribution systems
transmission by utilization of Simulated Annealing (SA). Furthermore, utilizing a hybrid optimization method developed based on cuckoo
search (CS) and HS algorithms. Keshtegar, et al. [57] proposed a bi-
Kundu and Garg [35] introduced an efficient hybrid approach for
loop optimization framework of stiffened panels is proposed to search
solving several types of engineering design and numerical optimization
the global optimum, including an adaptive response surface method
problems called enhanced teaching–learning Harris hawks optimization
(ARSM) loop and a Gaussian global-best harmony search (GGHS) loop.
(ITLHHO), which uses improved teaching–learning-based optimization.
Ouyang, et al. [58] proposed a hybrid metaheuristic approach by utiliz-
Kaveh and Vazirinia [36] introduced an improved sine cosine algorithm ing Teaching–Learning Based Optimization (TLBO) and HS algorithms
(USCA), which utilizes a harmony search-based operator to increase for optimum design of difficult problems in engineering. Keshtegar and
exploration while also dealing with changeable constraints, and saves Etedali [59] proposed based on the dynamical parameters that are ad-
the best answers in an archive. Brajević and Tuba [37] developed justed using the previous results of the harmony memory with a simple
an upgraded version of artificial bee colony (UABC) approach for formulation. Gholizadeh and Barzegar [60] developed a sequential HS
constrained optimization problems that improves the fine-tuning prop- algorithm for shape optimization of different structures by considering
erties of the modification rate parameter and uses the ABC algorithm’s frequency constraints. Jaberipour and Khorram [61] discussed mixed–
modified scout bee phase. To increase firefly algorithm’s efficiency discrete problems in the engineering optimization field by utilizing
in handling constrained engineering optimization problems, Brajević an enhanced HS algorithm. This study proposes the Tribe-Harmony
and Ignjatović [38] suggested an updated firefly algorithm (UFA). The Search (Tribe-HS) algorithm, in which the primary notion of improve-
suggested methodology employs a set of feasibility-based criteria to ment is derived from the ‘‘Tribe-CSS’’ method proposed by Talatahari
guide the search to the most feasible section of the search space, and Azizi [21]. These phases considered tribes lead the algorithm to
as well as an enhanced boundary constraint scheme and an equality concentrate on global searching in the early iterations while local
constraint approach. Pathak and Srivastava [39] proposed a new bat searching is handled in the later iterations in the Tribe-HS technique.
algorithm that includes a cuckoo search and Sugeno inertia weight These adjustments improve the exploration and exploitation rates of
(UBCSIW). The bat algorithm, which can exploit optimum solutions the standard algorithm. To assess the suggested method’s ability to deal
in search space, is merged with cuckoo search, which can explore the with complex optimization problems, three different building structures
best solution globally utilizing Levy flight in the search space, in the with 3, 20, and 60 stories with 135, 3860, and 8272 structural members
are deemed as design examples. The W-shaped design sections for struc-
proposed UBCSIW algorithm.
tural components in these structures are utilized to analyze the design
The critical contribution of most of the research studies reviewed
requirements, and the AISC-LRFD [62] code for steel structure design
is developing a preferable design method for optimum frame struc-
is applied. The suggested method’s overall performance is compared to
ture element configuration. Due to the shortcomings of conventional
that of the conventional HS algorithm and several metaheuristics, with
approaches with computational complexity difficulties, the importance 30 independent runs performed in each example for statistical reasons.
of offering a thoroughly defined optimization procedure is growing. The remainder of the paper is divided into the following sections.
Determining the appropriate search space is one of the most promi- In Section 2, the optimum design of steel frames, including objective
nent phases of designing an optimum design strategy for structural function and design constraints are presented. Section 3 describes the
optimization. From a structural standpoint, using trial and error to utilized optimization algorithm in detail. In Sections 4 and 5, design
create building structures using available wide-flange sections (W- examples, including 3, 20, and 60-story steel structures, alongside alter-
shaped sections) does not fulfill the affordable aspects of engineering native metaheuristic algorithms are illustrated, respectively. Numerical
projects. As a result, this research focuses on the optimum design of results have been reported in Section 6. Finally in Section 7, the core
real-size steel building structures, where more optimum and practical findings of this study are presented as concluding remarks.

2
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

2. Optimum design of steel frames

2.1. Objective function

There is an assumption in the optimum design of steel structural

frames that 𝑁𝑚 structural members are classified to 𝑁𝑑 design groups.
To reduce the total weight of the structure, the sequence numbers in
steel design sections given to 𝑁𝑑 member groups are calculated using
a vector of integer values. The integer vector and total weight of the
analyzed structure are summarized below.
[ ]
Find 𝐼 𝑇 = 𝐼1 , 𝐼2 , … , 𝐼𝑁𝑑 (1)

𝑁𝑑 𝑁𝑡
∑ ∑
To minimize 𝑊 = 𝜌𝑖 .𝐴𝑖 𝐿𝑗 (2)
𝑖=1 𝑗=1

where, 𝜌𝑖 and 𝐴𝑖 are the steel design section’s unit weight and length
established for member group i, respectively; 𝐿𝑗 shows the length of the
jth member associated with the ith group, and 𝑁𝑡 indicates the overall
number of all structural members in group 𝑖.

2.2. Design constraints

The AISC-LRFD [29] code for steel structure design specifies two
primary design criteria: strength and serviceability. When attempting
to minimize the structures’ weight, for the design sections’ strength
criteria, the following constraints must be met:
[ ] ( )
𝑖 𝑃𝑢𝐽 8 𝑀𝑢𝑥𝐽 𝑀𝑢𝑦𝐽
𝐶𝐼𝐸𝐿 = + +
𝜑𝑃𝑛 𝐼𝐸𝐿 9 𝜑𝑏 𝑀𝑛𝑥 𝜑𝑏 𝑀𝑛𝑦 𝐼𝐸𝐿
[ ]
𝑃𝑢𝐽 Fig. 1. Flowchart of the HS algorithm [63].
− 1 ≤ 0 𝑓 𝑜𝑟 ≥ 0.2 (3)
𝜑𝑃𝑛 𝐼𝐸𝐿
[ ] ( )
𝑖 𝑃𝑢𝐽 𝑀𝑢𝑥𝐽 𝑀𝑢𝑦𝐽
𝐶𝐼𝐸𝐿 = + +
2𝜑𝑃𝑛 𝐼𝐸𝐿 𝜑𝑏 𝑀𝑛𝑥 𝜑𝑏 𝑀𝑛𝑦 𝐼𝐸𝐿 performance process, a musician naturally conducts a proper searching
[ ]
𝑃𝑢𝐽 process to discover a better state of harmony with multiple tries. In
− 1 ≤ 0 𝑓 𝑜𝑟 < 0.2 (4) jazz improvisation, the player tends to achieve a musically pleasant
𝜑𝑃𝑛 𝐼𝐸𝐿
𝑣
( ) ( ) harmony as a perfect state by considering the aesthetic aspects. This
𝐶𝐼𝐸𝐿 = 𝑉𝑢𝐽 𝐼𝐸𝐿 + 𝜑𝑣 𝑉𝑛 𝐼𝐸𝐿 ≤ 0 (5) procedure is analogous to the optimization process, in which the
where, 𝐼𝐸𝐿 shows the element number as 𝐼𝐸𝐿 = 1, 2, … , 𝑁𝐸𝐿 and optimization algorithm strives to attain the global solution as a perfect
𝑁𝐸𝐿 is the total number of elements; 𝐽 indicates the number of state by taking the objective function evaluation into account. Each
load combination as 𝐽 = 1, 2, … , 𝑁 and 𝑁 is the overall number musical instrument’s pitch controls the aesthetic aspect of musical
of whole design load combinations; 𝑃𝑢𝐽 indicates the compressive or performance as the values of decision variables control the objective
tensile (axial) strength that is needed for the 𝐽 th design load; 𝑀𝑢𝑥𝐽 and function evaluation. The HS algorithm’s mathematical formulation is
𝑀𝑢𝑦𝐽 are the total flexural strengths needed for bending of structural constructed in five major phases, each of which is discussed in detail.
The initialization procedure is carried out in the first phase, in
elements concerning 𝑥 and 𝑦, under the 𝐽 th design load combination,
which the initial values for the harmony vectors (𝑋𝑖 ) consisting of
respectively; the 𝑥 and 𝑦 subscripts are used as related symbols for
different decision variables (𝑋𝑖 = {𝑥1 , 𝑥2 , … , 𝑥𝑖 , … , 𝑥𝑛 }) and their
strong and weak axes bending, respectively. 𝑃𝑛 , 𝑀𝑛𝑥 and 𝑀𝑛𝑦 are the
related objective function amounts (𝐹𝑖 ) are determined. The decision
nominal compressive or tensile (axial) and flexural (for bending of
variables demonstrate different musicians, and the objective function
structural elements about 𝑥 and 𝑦 axes) strengths of the 𝐼𝐸𝐿th member
evaluations demonstrate the harmony which these musicians achieve.
under consideration. 𝜑 clarifies the axial strength resistance factor
In this step, the crucial parameters of the HS algorithm such as the
formulated about the gross section yielding which for compression and
Harmony Memory Size (HMS), Pitch Adjusting Rate (PAR), Harmony
tension are 0.85 and 0.9, respectively. 𝜑𝑏 shows the flexural resistance Memory Considering Rate (HMCR), and the termination criteria, which
factor (0.9). The shear strength needed under the Jth design load is deemed as the maximum number of iterations (MaxIter) are deter-
combination is denoted by 𝑉𝑢𝐽 , and 𝑉𝑛 elucidates the nominal shear mined. The PAR and HMCR parameters are utilized to improve each
strength of the 𝐼𝐸𝐿th deemed elements and 𝜑𝑣 equals 0.9. solution vector’s quality in the optimization process.
The initial Harmony Memory (HM) is determined in the second
3. Utilized optimization algorithms phase, including the solution vectors generated randomly with the
harmony memory (HMS) size, classified regarding their objective func-
This part discusses the metaheuristic optimization algorithms that tion’s values. The mathematical presentation of the HM is as
were used, including the conventional Harmony Search method and its follows:
improved variant, named ‘‘Tribe-HS’’.
⎡ 𝑥1 ⎤
⎢ ⎥
3.1. Harmony search algorithm ⎢ 𝑥2 ⎥
𝐻𝑀 = ⎢ ⎥ (6)
⎢ ⋮ ⎥
The fundamental concept behind the development of a novel opti- ⎢ ⎥
mization method called ‘‘Harmony Search’’ is that in a musical ⎢ 𝐻𝑀𝑆 ⎥
⎣𝑥 ⎦

3
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 2. Schematic demonstration of the search procedure for the Tribe-HS in the first phase.

In step three, a new harmony vector (𝑋𝑖′ = {𝑥′1 , 𝑥′2 , … , 𝑥′𝑖 , … , 𝑥′𝑛 }) is 3.2. Tribe-harmony search algorithm
improvised from the harmony memory or initial harmony vectors based
on the pitch adjustment, memory consideration, and randomization Premature convergence is a possibility for a large number of opti-
process. The decision variables can be determined by choosing any mization algorithms. An intriguing endeavor has been made to
values from the HM in Eq. (6) or choosing from the initial harmony strengthen the general capacity of metaheuristic approaches and algo-
vectors. In this regard, a random number distributed uniformly in the rithms by providing appropriate solutions to the algorithms’ inadequa-
range of (0, 1) is produced to decide between two choices. If the cies in the past few decades. In this respect, this study proposed the
produced random number is higher than the previously determined notion of Tribe-HS to improve the HS algorithm’s potential of solving
HMCR, the novel harmony vector is selected from the HM, while challenging optimization problems. This idea relies on the fact that
for the random numbers lower than the HMCR, the novel vector is by separating the search space into many separate groupings known
determined to form the initial harmony vectors (𝑋𝑖 ). These aspects are as ‘‘Tribes’’, the searching process is carried out in an old-fashioned
mathematically represented as follows: manner in which the tribes may offer a civilized way of life without
{ ′
𝑥𝑖 𝜖 𝐻𝑀 𝑤𝑖𝑡ℎ 𝑡ℎ𝑒 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 (𝐻𝑀𝐶𝑅) connecting in the early ages. Nonetheless, These tribes try to exchange
𝑥′𝑖 → (7) information and, in later years, even unite for a better way of life. The
𝑥′𝑖 𝜖 𝑋𝑖 𝑤𝑖𝑡ℎ 𝑡ℎ𝑒 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 (1 − 𝐻𝑀𝐶𝑅) search area in the HS algorithm is separated into many tribes (search
Pitch adjustment is used to mathematically model the mutation phase spaces) based on the presented notion, with each tribe’s searching
of the procedure for the values obtained from the HM by creating procedure completed in a unique way that increases the standard
another random value spread equally within the range of (0, 1). If the algorithm’s performance.
created random number is more than the previously determined PAR, To mathematically represent the above notion, a maximum number
the novel harmony vector selected from the HM will choose a neigh- of tribes (𝑁𝑡 ) should be established, which will be used to divide
boring value with the PAR probability; however, no pitch adjustment the solution vectors in the search space into these tribes. Each of the
is made if the generated random number is less than the PAR. These aforementioned tribes has a random number of solution vectors (𝑁𝑠 ),
considerations are mathematically expressed as follows: and the searching process is carried out in these tribes in a particular
{ ′ manner to converge on a correct solution effectively. The algorithm’s
𝑥𝑖 + (𝑏𝑤 × 𝑢) 𝑤𝑖𝑡ℎ 𝑡ℎ𝑒 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 (𝑃 𝐴𝑅)
′ primary search phase is split into three distinct stages: the isolated, the
𝑥𝑖 → (8)
𝑥′𝑖 𝑤𝑖𝑡ℎ 𝑡ℎ𝑒 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 (1 − 𝑃 𝐴𝑅) communing, and the unified phase. The new formulation includes the
where 𝑏𝑤 shows an arbitrary distance bandwidth and 𝑢 is a uniformly stopping requirements, which divides the maximum number of function
distributed random number in the range of (−1, 1). evaluations or iterations into three parts.
In step four, the HM is upgraded, and if the newly created harmony The isolated phase of the algorithm is the initial phase in which
vector outperforms the worst harmony in the HM concerning the value the solution vectors in the search space and inside the HM are not
of the objective function, the novel harmony is replaced by the worst, allowed to exchange information or personal experiences with other
and the HM is sorted using the objective function values. The third tribes. This procedure is repeated until the predetermined stopping
and fourth phases are repeated in the fifth step until the termination conditions, which are separated into the three stages indicated above,
requirements are met. The flowchart of the suggested HS algorithm is are met. The second step, called the communing phase, allows tribes
demonstrated in Fig. 1. to utilize the solution vectors in each other’s HMs and update their

4
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 3. Schematic demonstration for the Tribe-HS in the second phase in the search procedure.

Fig. 4. Schematic demonstration of the search procedure for the Tribe-HS in the third Phase.

most recently discovered information. As the third phase, the unified 4. Design examples
phase brings together all of the solution vectors from separate tribes,
and this phase continues until the stated termination requirement is This part contains detailed information about the 3 real-size steel
fulfilled. The schematic representation of the Tribe-HS algorithm in its building structures that are used to assess the Tribe-HS’s capability
three distinct stages is shown in Figs. 2–4, while the pseudo-code for to evaluate the structural elements’ optimal design sections. Different
this algorithm is shown in Fig. 5. plans select these structures and in diverse heights to find out the

5
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 5. Pseudocode of the Tribe-HS optimization algorithm.

usefulness of the improved optimization algorithm in dealing with To assist in the design process, the deemed building structures
different sorts of building structures. The material characteristics used are exposed to ten different load combinations, as listed in Table 1.
in these structures are stainless steel with an elasticity modulus (E) of On typical floor beams, the acting dead and live loads are 14 and
200 GPa, yield stress (Fy) of 248.2 MPa, and steel unit weight (q) of 10 kN/m, respectively, while the dead and live loads on roof beams
7.85 ton/m3 . are 12 and 7 kN/m, respectively. The seismic and wind loads on the

6
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 6. The schematic representation and plan views of the 3-story steel structure.

Table 1 1064 columns, 1836 beams, and 960 bracing parts, with the columns,
Load combinations for steel structural design.
beams, and braces all having standard W-shaped design sections. This
No. Combination
structure’s lateral resistance is provided by cross-bracing systems in
1 1.4 D
the X and Y axes, in addition to moment-resisting connections. Fig. 9
2 1.2 D + 1.6 L
3 1.2 D + 1.0 (E𝑥 /W𝑥 ) + 0.5 L depicts the schematic and plan views of this structure.
4 1.2 D + 1.0 (E𝑒𝑥 /W𝑒𝑥 ) + 0.5 L Based on their fabrication requirements, all 3860 structural mem-
5 1.2 D + 1.0 (E𝑦 /W𝑦 ) + 0.5 L
bers of the 20-story steel structure are divided into 73 member groups.
6 1.2 D + 1.0 (E𝑒𝑦 /W𝑒𝑦 ) + 0.5 L
7 0.9 D + 1.0 (E𝑥 /W𝑥 ) Member grouping is evaluated at both the plan and elevation levels,
8 0.9 D + 1.0 (E𝑒𝑥 /W𝑒𝑥 ) with structural members at the elevation level grouped every two
9 0.9 D + 1.0 (E𝑦 /W𝑦 ) stories. Additionally, the columns are classified into five distinct groups
10 0.9 D + 1.0 (E𝑒𝑦 /W𝑒𝑦 )
at the plan level, as illustrated in Fig. 10. Two groups are examined for
D: Dead Load, L: Live Load, E: Earthquake Load, W: Wind Load.
beams: inner and outer beams, while one group is selected for each of
𝑥 and 𝑦: Loading directions without eccentricity.
𝑒𝑥 and 𝑒𝑦: Loading directions with eccentricity. the structure’s neighboring two stories. As a result, 43 column design
groups, 20 beam design groups, and ten bracing design groups are
examined concerning the structure’s plan and elevation levels.
structural systems under consideration are assessed in accordance with
ASCE [64], establishing the minimum design loads for buildings and
4.3. Example 3: 60-story, 8272-member steel structure
other structures.

4.1. Example 1: 3-story, 135-member steel structure The third design example is a 60-story steel structure with a struc-
tural tube system comprised of 8272 structural components. 3960
A three-story steel structure with 135 structural members is the first
columns, 3960 beams, and 352 bracing components are evaluated
design example. The structural members of this structure are composed
in this design example, in which the design sections for the beams,
of 45 column elements, 66 beam elements, and 24 brace elements
columns, and braces are considered as standard W-shaped sections.
deemed as standard W-shaped sections. The moment resisting connec-
tions alongside inverted V-bracings are utilized as a lateral resisting The mega-bracing systems with the X and Y directions, as well as the
system of the structure. The schematic and plan views of this structure moment-resisting connections, are used to prepare the lateral stability
are shown in Fig. 6, while the elevation views are shown in Fig. 7. of this structure for the first 24 stories, while the lateral resisting system
All 135 structural components of the three-story steel structure are is used as a standard bracing system alongside the moment-resisting
classified into ten member groups based on their practical fabrication connections for the 25th to 60th stories. Fig. 11 shows a schematic
requirements. Member grouping is evaluated at both the plan and representation of this structure.
elevation levels, whereas structural members at the elevation level are Regarding the practical fabrication requirements, all 8272 members
grouped within each story. Each story’s beams and braces are regarded
of the studied 60-story steel structure are divided into 103 member
to be part of a single beam and bracing group, whereas the columns
groups. The member grouping process is performed at both the ele-
are divided into four distinct groups at the plan level, as seen in Fig. 8.
vation and plan levels, with the structural member grouping process
4.2. Example 2: 20-story, 3860-member steel structure occurring every six stories at the elevation level. Two column groups,
the corner, and side column groups, are considered for tubes A to D in
The second design problem is the structural design of a 20-story plan levels (Fig. 12), while each tube has one beam group. Every six
steel structure with 3860 structural components. This design sample has stories, the bracing components are considered a separate group.

7
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 7. The elevation views of the 3-story steel structure in X and Y directions.

Fig. 8. Column grouping of the 3-story steel structure.

Fig. 9. The schematic and plan views of the 20-story steel structure with 3860 members.

8
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 10. Column grouping of the 20-story steel structure with 3860 members.

Table 2 Table 2 (continued).

Internal parameters for the alternative metaheuristic algorithms. Metaheuristic Parameter Description Value
Metaheuristic Parameter Description Value
𝐻𝑀𝑆 Harmony memory size 50
𝑝𝑐 Crossover percentage 0.8 𝑁𝑛𝑒𝑤 Number of new harmonies 20
𝑝𝑚 Mutation percentage 0.3 𝐻𝑀𝐶𝑅 Harmony memory consideration rate 0.9
GA
𝜇 Mutation rate 0.02 Tribe-HS 𝑃 𝐴𝑅 Pitch adjustment rate 0.1
𝛽 Roulette wheel selection pressure 1 𝐹𝑊 Fret width (bandwidth) ±0.02
𝐹 𝑊𝑑𝑎𝑚𝑝 Fret width damp ratio 0.995
𝑁𝑠 Sample size 50
𝑁𝑇 Number of considered tribes 10
ACO 𝑞 Intensification factor 0.5
𝜁 Deviation-distance ratio 1
𝑤 Inertia weight 1
𝑤𝑑 Inertia weight damping ratio 0.99
PSO 5. Alternative metaheuristics
𝑐1 Personal learning coefficient 2
𝑐2 Global learning coefficient 2
𝑁𝑒𝑚𝑝 Number of empires/imperialists 10 This paper utilizes 10 other metaheuristic algorithms as alternative
𝛼 Selection pressure 1 approaches for comparative purposes. The GA, PSO, ACO, ICA, and
𝛽 Assimilation coefficient 1.5 CSS are selected as classical methods which have been utilized in most
ICA
𝑝𝑟 Revolution probability 0.05
of the previous research, while the Butterfly Optimization Algorithm
𝜇 Revolution rate 0.1
𝜁 Colonies mean cost coefficient 0.2 (BOA) [65], Harris Hawks Optimization (HHO) [66], Multi-Verse Op-
timizer (MVO) [67], Galactic Swarm Optimization (GSO) [68], and
𝑝 Probability switch 0.8
BOA 𝑝𝑒 Power exponent 0.1 Krill Herd Algorithm (KHA) [69] are selected as some of the recently
𝑠𝑚 Sensory modality 0.01 developed novel metaheuristic algorithms. Some of the approaches
𝑉𝑓 Foraging speed 0.02 are classified as parameter-less optimization algorithms, meaning they
KH 𝐷𝑚𝑎𝑥 Maximum diffusion speed 0.005 do not have any internal parameters in their general formulation. At
𝑁𝑚𝑎𝑥 Maximum induced speed 0.01 the same time, for some of them, some internal parameters need to
𝑎 Radius of charged sphere 0.1 be determined in the optimization process. In Table 2, a parameter
𝐻𝑀𝐶𝑅 Harmony memory consideration rate 0.85
summary is provided for these alternative approaches alongside the HS
𝑃 𝐴𝑅 Pitch adjustment rate 0.15
CSS 𝑘𝑡 Attract-repel coefficient 0.9
and the proposed Tribe-HS algorithms, while for all of them, the initial
𝑁𝑐𝑚 Charged memory size 12 population size is utilized as 50.
𝑘𝑎 Acceleration coefficient 0.5
𝑘𝑣 Velocity coefficient 0.5
6. Numerical results
𝐻𝑀𝑆 Harmony memory size 50
𝑁𝑛𝑒𝑤 Number of new harmonies 20
𝐻𝑀𝐶𝑅 Harmony memory consideration rate 0.9 The numerical outcomes of the weight optimization procedure for
HS
𝑃 𝐴𝑅 Pitch adjustment rate 0.1 the 3-, 20- and 60-story steel structures are reported in this section.
𝐹𝑊 Fret width (bandwidth) ±0.02 For each of the HS, Tribe-HS, and considered alternative methods, a
𝐹 𝑊𝑑𝑎𝑚𝑝 Fret width damp ratio 0.995
total of 30 independent runs were undertaken. Figs. 13 to 15 show
the convergence history for the best results of these approaches for the
chosen 3-, 20-, and 60-story steel structures, respectively. It is worth
noting that the Tribe-HS can get better outcomes than the standard HS
with the minimum number of required structural analyses.

9
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 11. Schematic view of the 60-story steel structure with 8272 members.

Table 3 presents the optimal design sections for the HS, Tribe-HS, The overall weight of the 20-story steel structure computed using the
and chosen alternative approaches when considering the three-story HS standard method is 3236.38 tons, but the Tribe-HS algorithm calcu-
steel structure. It should be mentioned that for this reason, the best lates this value as 2809.63 tons, which is less than the HS determined
results from 30 independent runs in each metaheuristic method were value. It could be observed that the overall weight of the structure
reported. acquired using Tribe-HS is less than the weight obtained using HS,
The optimal design elements for the 20-story steel structures derived demonstrating the suggested Tribe-HS method’s capabilities for this
using the Tribe-HS and the conventional HS are provided in Table 4. purpose.

10
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 12. Column grouping in plan levels of the 60-story steel structure with 8272 members.

Fig. 13. Convergence history for the best results of different metaheuristics for 3-story structure.

Given that Kazemzadeh Azad, et al. [32] evaluated this design The optimal design sections for the 60-story steel structure derived
example using a variety of metaheuristic approaches, Table 5 compares using the Tribe-HS, and the conventional HS are provided in Table 6.
the HS and the proposed Tribe-HS with various approaches. It should The overall weight of the 60-story steel structure is determined using
be highlighted that the Tribe-HS approach has the potential to provide the HS standard method to be 6958.17 tons, whereas the Tribe-HS
superior outcomes than the other alternatives. algorithm calculates it to be 6766.89 tons, which is less than the

11
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 14. Convergence history for the HS and Tribe-HS of the 20-story structure.

Fig. 15. Convergence history for the HS and Tribe-HS of the 60-story structure.

HS computed value. It could be mentioned that the overall weight The stress ratio of the structural elements for the 3, 20-and 60-
of the structures acquired using Tribe-HS is less than that achieved story design examples are depicted in Figs. 16 and 18 respectively
using HS, demonstrating the suggested Tribe-HS method’s capabilities. for the standard HS and the proposed Tribe-HS algorithms. The stress
A comparative analysis is not appropriate since this instance is being ratios of structural components in the Tribe-HS optimized structural
described for the first time in this work. systems are greater, particularly near the allowable value, demonstrat-

12
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 16. Stress ratio of the structural elements for the 3-story design example.

Fig. 17. The stress ratio of the structural elements for the 20-story design example.

Fig. 18. The stress ratio of the structural elements for the 60-story design example.

13
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Table 3
Optimum design sections of different metaheuristics for the 3-story steel structure.
Groups GA PSO ACO ICA MVO GSO BOA KHA HHO HS Tribe-HS
CG1 W8 × 28 W18 × 60 W14 × 38 W16 × 36 W24 × 55 W24 × 68 W12 × 35 W10 × 33 W18 × 55 W14 × 43 W16 × 40
CG2 W30 × 90 W16 × 67 W12 × 45 W14 × 48 W18 × 60 W21 × 62 W16 × 57 W21 × 62 W18 × 55 W21 × 68 W18 × 60
CG3 W24 × 76 W27 × 84 W24 × 68 W18 × 50 W27 × 94 W27 × 84 W40 × 149 W21 × 73 W18 × 55 W24 × 68 W30 × 90
CG4 W33 × 130 W30 × 90 W27 × 84 W30 × 99 W24 × 94 W24 × 84 W24 × 68 W24 × 76 W30 × 90 W21 × 73 W21 × 62
B1 W21 × 44 W18 × 40 W21 × 68 W21 × 57 W18 × 35 W21 × 44 W16 × 31 W18 × 50 W21 × 44 W18 × 40 W21 × 44
B2 W8 × 18 W18 × 35 W18 × 40 W16 × 40 W18 × 40 W16 × 26 W24 × 55 W18 × 46 W16 × 36 W21 × 44 W18 × 40
B3 W21 × 50 W14 × 30 W10 × 22 W12 × 26 W16 × 26 W16 × 26 W8 × 21 W14 × 22 W16 × 26 W14 × 22 W10 × 22
BR1 W12 × 26 W6 × 25 W8 × 28 W8 × 24 W8 × 28 W8 × 24 W6 × 25 W8 × 24 W12 × 30 W12 × 30 W8 × 28
BR2 W5 × 16 W10 × 39 W6 × 20 W8 × 18 W6 × 20 W6 × 15 W6 × 15 W5 × 16 W6 × 15 W8 × 21 W8 × 21
BR3 W5 × 19 W8 × 18 W10 × 30 W8 × 18 W6 × 15 W5 × 19 W6 × 15 W4 × 13 W5 × 16 W10 × 19 W6 × 15
Weight (ton) 43.2073 41.1092 40.2583 38.3812 39.0215 38.2119 38.3569 38.1377 37.6269 38.1889 36.9721
Maximum drift ratio 1 0.9978 0.9599 1 1 0.9923 0.9915 1 1 0.9858 0.9965
Maximum stress ratio 1 0.9812 0.9505 0.9766 0.9683 0.9327 0.9971 1 0.9638 0.8613 0.9380

CG: Column Groups

B: Beam Group.
BR: Bracing Group.

Table 4
Optimum design sections for the 20-story steel structure with 3860 members.
Stories Groups HS sections Tribe-HS sections Stories Groups HS sections Tribe-HS sections
CG1 W12 × 50 W30 × 99 CG1 W24 × 68 W33 × 291
CG2 W33 × 201 W14 × 132 CG2 W21 × 132 W27 × 94
CG3 W21 × 182 W12 × 230 CG3 W18 × 311 W30 × 124
CG4 W14 × 283 W36 × 650 CG4 W27 × 194 W36 × 300
1–2 11–12
CG5 W24 × 229 W36 × 160 CG5 W27 × 235 W18 × 76
IB W14 × 82 W10 × 100 IB W21 × 83 W24 × 84
OB W14 × 74 W14 × 61 OB W8 × 67 W12 × 72
BR W30 × 173 W18 × 119 BR W14 × 26 W14 × 109
CG1 W40 × 249 W24 × 207 CG1 W44 × 262 W10 × 77
CG2 W33 × 169 W14 × 109 CG2 W21 × 132 W33 × 130
CG3 W36 × 160 W27 × 235 CG3 W12 × 87 W18 × 143
CG4 W14 × 730 W44 × 230 CG4 W27 × 161 W18 × 106
3–4 13–14
CG5 W33 × 130 W36 × 182 CG5 W18 × 76 W14 × 68
IB W33 × 152 W12 × 50 IB W8 × 58 W12 × 53
OB W24 × 103 W14 × 48 OB W24 × 207 W36 × 135
BR W18 × 86 W12 × 65 BR W8 × 48 W14 × 53
CG1 W14 × 176 W12 × 72 CG1 W27 × 161 W44 × 335
CG2 W30 × 132 W21 × 122 CG2 W18 × 258 W36 × 260
CG3 W12 × 210 W40 × 264 CG3 W36 × 182 W40 × 174
CG4 W14 × 193 W40 × 264 CG4 W44 × 262 W27 × 84
5–6 15–16
CG5 W14 × 233 W18 × 143 CG5 W18 × 97 W30 × 90
IB W10 × 68 W27 × 84 IB W21 × 122 W14 × 68
OB W18 × 55 W21 × 73 OB W21 × 147 W24 × 103
BR W36 × 182 W8 × 40 BR W8 × 40 W10 × 45
CG1 W14 × 68 W40 × 183 CG1 W24 × 62 W30 × 477
CG2 W12 × 106 W12 × 106 CG2 W27 × 258 W12 × 79
CG3 W30 × 173 W21 × 166 CG3 W40 × 321 W16 × 67
CG4 W33 × 152 W36 × 393 CG4 W24 × 94 W40 × 174
7–8 17–18
CG5 W14 × 132 W30 × 108 CG5 W18 × 175 W10 × 88
IB W18 × 76 W12 × 58 IB W21 × 93 W14 × 99
OB W14 × 68 W21 × 166 OB W12 × 87 W30 × 116
BR W24 × 103 W24 × 94 BR W8 × 40 W14 × 68
CG1 W18 × 60 W16 × 77 CG1 W18 × 158 W24 × 250
CG2 W24 × 104 W27 × 94 CG2 W27 × 161 W10 × 112
CG3 W33 × 221 W30 × 292 CG3 W16 × 50 W18 × 86
CG4 W21 × 182 W40 × 174 CG4 W40 × 174 W8 × 67
9–10 19–20
CG5 W21 × 101 W40 × 211 CG5 W12 × 152 W24 × 104
IB W27 × 84 W10 × 77 IB W16 × 67 W18 × 55
OB W14 × 109 W27 × 102 OB W16 × 57 W12 × 190
BR W27 × 94 W10 × 49 BR W10 × 49 W10 × 17
Total weight (ton) 3236.38 2809.63
Maximum drift 0.8819 0.9091

CG𝟏−𝟓 : Column Groups 1 to 5 (Fig. 7).

IB: Inner Beam Group.
OB: Outer Beam Group.
BR: Bracing Group.

14
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 19. Drift ratio of the structural elements for the 3-story design example.

structural systems, drift ratios of structural elements are greater, par-

Table 5 ticularly near the allowable value, demonstrating that the Tribe-HS-
Comparative results for the 20-story steel structure with 3860 members. provided optimal design sections have the smallest feasible design
Stories UBS [25] HS Tribe-HS cross-sections in terms of an affordable design approach.
Total weight (ton) 4117.43 3236.38 2809.63

UBS: Upper Bound Strategy. 7. Conclusion

This study proposes an improved metaheuristic method named

ing that the Tribe-HS offered optimal design sections have the lowest ‘‘Tribe-Harmony Search’’ for optimal steel structure design. This algo-
feasible design cross-sections in terms of an affordable design approach rithm is a modified variant of the regular Harmony Search algorithm.
(see Fig. 17). The Harmony Search algorithm is one the wellknown metaheuristic
The drift ratios of structural elements for the 3, 20, and 60 story algorithms utilizing the musical process of looking for the optimal
design examples are shown in Figs. 19 to 21 for the standard HS and state of harmony to produce an appropriate searching strategy. Due to
suggested Tribe-HS algorithms, respectively. For Tribe-HS-optimized the algorithm’s many uses in a variety of optimization domains, there

15
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Table 6
Optimum design sections for the 60-story steel structure with 8272 members.
Stories Groups HS sections Tribe-HS sections Stories Groups HS sections Tribe-HS sections
CC-A W27 × 94 W24 × 279 CC-A W36 × 280 W30 × 235
SC-A W21 × 111 W21 × 57 SC-A W14 × 257 W40 × 199
CC-B W27 × 84 W21 × 93 CC-B W27 × 84 W16 × 67
SC-B W24 × 55 W21 × 62 SC-B W16 × 50 W21 × 50
CC-C W21 × 147 W14 × 550 CC-C W14 × 145 W36 × 393
SC-C W40 × 277 W24 × 492 SC-C W36 × 328 W36 × 280
1–6 CC-D W36 × 194 W30 × 235 7–12 CC-D W18 × 311 W14 × 211
SC-D W40 × 249 W36 × 245 SC-D W33 × 354 W14 × 283
BM-A W30 × 148 W18 × 192 BM-A W24 × 492 W30 × 148
BM-B W40 × 199 W18 × 258 BM-B W33 × 141 W10 × 100
BM-C W27 × 129 W24 × 408 BM-C W18 × 55 W16 × 45
BM-D W40 × 211 W27 × 114 BM-D W33 × 130 W12 × 96
BR-D W18 × 65 W24 × 68 BR-D W24 × 103 W24 × 94
CC-A W21 × 132 W24 × 131 CC-A W24 × 94 W21 × 132
SC-A W24 × 103 W36 × 135 SC-A W40 × 199 W24 × 84
CC-B W14 × 159 W27 × 258 CC-B W21 × 101 W30 × 124
SC-B W33 × 318 W40 × 277 SC-B W27 × 178 W12 × 279
CC-C W24 × 162 W24 × 176 CC-C W30 × 261 W18 × 97
SC-C W36 × 260 W27 × 217 SC-C W36 × 245 W40 × 297
13–18 CC-D W36 × 280 W12 × 79 19–24 CC-D W30 × 99 W27 × 129
SC-D W30 × 326 W18 × 211 SC-D W40 × 174 W21 × 147
BM-A W14 × 43 W40 × 167 BM-A W12 × 152 W21 × 101
BM-B W36 × 280 W14 × 193 BM-B W14 × 132 W40 × 199
BM-C W36 × 170 W21 × 122 BM-C W16 × 77 W18 × 158
BM-D W21 × 44 W30 × 90 BM-D W18 × 35 W27 × 84
BR-D W27 × 102 W27 × 94 BR-D W24 × 55 W27 × 146
CC-A W40 × 167 W33 × 130 CC-A W36 × 182 W27 × 258
SC-A W12 × 120 W24 × 117 SC-A W14 × 132 W44 × 230
CC-B W27 × 235 W21 × 147 CC-B W14 × 132 W12 × 336
SC-B W30 × 124 W18 × 86 SC-B W24 × 104 W30 × 124
CC-C W40 × 277 W30 × 292 CC-C W40 × 174 W30 × 211
25–30 31–36
SC-C W33 × 118 W21 × 68 SC-C W14 × 120 W40 × 211
BM-A W24 × 146 W12 × 136 BM-A W21 × 166 W36 × 359
BM-B W40 × 235 W14 × 74 BM-B W14 × 90 W12 × 30
BM-C W21 × 68 W40 × 167 BM-C W24 × 76 W33 × 263
BR-C W36 × 182 W27 × 161 BR-C W40 × 264 W24 × 103
CC-A W30 × 132 W33 × 169 CC-A W36 × 245 W14 × 159
SC-A W10 × 112 W40 × 199 SC-A W36 × 160 W21 × 166
CC-B W40 × 431 W30 × 108 CC-B W44 × 290 W14 × 159
SC-B W14 × 398 W40 × 215 SC-B W14 × 370 W12 × 210
CC-C W12 × 65 W14 × 398 BM-A W24 × 207 W27 × 146
37–42 43–48
SC-C W27 × 178 W30 × 99 BM-B W14 × 53 W40 × 183
BM-A W27 × 194 W40 × 174 BR-B W21 × 182 W40 × 149
BM-B W12 × 96 W18 × 50 –
BM-C W8 × 24 W16 × 57 –
BR-C W33 × 318 W14 × 61 –
CC-A W14 × 550 W18 × 55 CC-A W8 × 40 W8 × 31
SC-A W18 × 50 W36 × 182 SC-A W18 × 40 W12 × 252
CC-B W36 × 150 W18 × 40 CC-B W40 × 149 W27 × 161
49–54 SC-B W24 × 76 W24 × 117 55–60 SC-B W12 × 35 W18 × 119
BM-A W18 × 97 W36 × 170 BM-A W16 × 67 W12 × 50
BM-B W16 × 67 W12 × 152 BM-B W21 × 68 W12 × 96
BR-B W36 × 170 W8 × 35 BR-B W33 × 221 W30 × 173
Total Weight (ton) 6958.17 6766.89
Maximum Drift 0.9985 0.9722

CC-A, CC-B, CC-C, CC-D: Corner Column Groups for Tubes A to D (Fig. 9).
SC-A, SC-B, SC-C, SC-D: Side Column Groups for Tubes A to D (Fig. 9).
BM-A, BM-B, BM-C, BM-D: Beam Member Groups for Tubes A to D.
BR-B, BR-C, BR-D: Bracing Member Groups for Tubes B to D.

has been an increased interest in improving the algorithm’s overall of the suggested methodology in dealing with complex optimization
performance. The traditional algorithm’s searching phase is broken problems. The suggested method’s overall performance is compared
into three distinct phases in the Tribe-HS. These stages, called tribes, to that of the standard Harmony Search algorithm and many meta-
lead the algorithm to prioritize global search in the early iterations heuristics. The acquired findings demonstrated that the recommended
and local search in the later iterations. These adjustments improve the technique is capable of producing superior outcomes than the other
conventional algorithm’s exploration and exploitation rates. 3 different metaheuristics for the investigated design examples. The total weight of
building structures with 3, 20, and 60 stories with 135, 3860, and 8272 the 20-story steel structure is obtained as 3236.38 tons using HS and
structural members are deemed as design examples to assess the ability 2809.63 tons using Tribe-HS, while the reduction rate is about 13%.

16
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 20. The drift ratio of the structural elements for the 20-story design example.

The overall weight of the 60-story steel structure is 6958.17 tons when entity in the biomedical field which may be perceived to have poten-
using HS and 6766.89 tons when employing Tribe-HS, with a 3 percent tial conflict of interest with this work. For full disclosure statements
decrease rate. The stress and drift ratios of the structural elements refer to https://doi.org/10.1016/j.dajour.2022.100067. This research
are higher in the Tribe-HS optimized structural systems, particularly is supported by a research grant from the University of Tabriz (Number:
near the allowable value, demonstrating that the Tribe-HS provided
1615)
optimum design sections have the lowest possible design cross-sections
in terms of an economical design process.

Declaration of competing interest Acknowledgments

One or more of the authors of this paper have disclosed potential or

pertinent conflicts of interest, which may include receipt of payment, This research is supported by a research grant from the University
either direct or indirect, institutional support, or association with an of Tabriz, Iran (Number: 1615).

17
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

Fig. 21. The drift ratio of the structural elements for the 60-story design example.

References [7] S. Talatahari, M. Azizi, M. Toloo, Fuzzy adaptive charged system search for
global optimization, Appl. Soft Comput. 109 (2021) 107518.
[1] J.H. Holland, Genetic algorithms and adaptation, in: O.G. Selfridge, E.L. Rissland, [8] S. Talatahari, M. Azizi, A.H. Gandomi, Material generation algorithm: a novel
M.A. Arbib (Eds.), Adaptive Control of Ill-Defined Systems, Springer US, Boston, metaheuristic algorithm for optimization of engineering problems, Processes 9
MA, 1984, pp. 317–333. (5) (2021) 859.
[2] R. Storn, K. Price, Differential evolution–a simple and efficient heuristic for [9] M. Azizi, M.B. Shishehgarkhaneh, M. Basiri, Optimum design of truss structures
global optimization over continuous spaces, J. Global Optim. 11 (4) (1997) by material generation algorithm with discrete variables, Decis. Anal. J. (2022)
341–359. 100043, http://dx.doi.org/10.1016/j.dajour.2022.100043.
[3] M. Dorigo, V. Maniezzo, A. Colorni, Ant system: optimization by a colony of [10] S. Talatahari, M. Azizi, Chaos game optimization: a novel metaheuristic algo-
cooperating agents, IEEE Trans. Syst. Man Cybern. B 26 (1) (1996) 29–41, rithm, Artif. Intell. Rev. 54 (2) (2021) 917–1004, http://dx.doi.org/10.1007/
http://dx.doi.org/10.1109/3477.484436. s10462-020-09867-w.
[4] R. Eberhart, J. Kennedy, A new optimizer using particle swarm theory, in: [11] M. Azizi, U. Aickelin, H.A. Khorshidi, M.B. Shishehgarkhaneh, Shape and size op-
Proceedings of the Sixth International Symposium on Micro Machine and Human timization of truss structures by chaos game optimization considering frequency
Science, in: MHS’95, 1995, pp. 39–43. constraints, J. Adv. Res. (2022) http://dx.doi.org/10.1016/j.jare.2022.01.002.
[5] A. Kaveh, S. Talatahari, A novel heuristic optimization method: charged system [12] G.-G. Wang, A.H. Gandomi, X.-S. Yang, A.H. Alavi, A novel improved accelerated
search, Acta Mech. 213 (3) (2010) 267–289, http://dx.doi.org/10.1007/s00707- particle swarm optimization algorithm for global numerical optimization, Eng.
009-0270-4. Comput. (2014).
[6] S. Talatahari, M. Azizi, M. Toloo, M. Baghalzadeh Shishehgarkhaneh, Opti- [13] A. Kaveh, S. Talatahari, An improved ant colony optimization for constrained
mization of large-scale frame structures using fuzzy adaptive quantum inspired engineering design problems, Eng. Comput. (2010).
charged system search, Int. J. Steel Struct. (2022) http://dx.doi.org/10.1007/ [14] M. Azizi, R.G. Ejlali, S.A. Mousavi Ghasemi, S. Talatahari, Upgraded whale
s13296-022-00598-y. optimization algorithm for fuzzy logic based vibration control of nonlinear steel

18
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

structure, Eng. Struct. 192 (2019) 53–70, http://dx.doi.org/10.1016/j.engstruct. [40] Z.W. Geem, J.H. Kim, G.V. Loganathan, A new heuristic optimization algorithm:
2019.05.007. Harmony search, Simulation 76 (2) (2001) 60–68, http://dx.doi.org/10.1177/
[15] F. Fasahat, P. Payvandy, A novel hybrid genetic and imperialist competitive 003754970107600201.
algorithm for structure extraction of woven fabric images, J. Textile Inst. 108 [41] T. Zhang, Z.W. Geem, Review of harmony search with respect to algorithm
(6) (2017) 893–905. structure, Swarm Evol. Comput. 48 (2019) 31–43, http://dx.doi.org/10.1016/
[16] M. Azizi, A. Mousavi, R. Ejlali, S. Talatahari, Optimum design of fuzzy controller j.swevo.2019.03.012.
using hybrid ant lion optimizer and Jaya algorithm, Artif. Intell. Rev. 53 (2020) [42] H. Rezaie, M.H. Kazemi-Rahbar, B. Vahidi, H. Rastegar, Solution of combined
1–32, http://dx.doi.org/10.1007/s10462-019-09713-8. economic and emission dispatch problem using a novel chaotic improved
[17] M. Azizi, Atomic orbital search: A novel metaheuristic algorithm, Appl. Math. harmony search algorithm, J. Comput. Des. Eng. 6 (3) (2019) 447–467.
Model. 93 (2021) 657–683, http://dx.doi.org/10.1016/j.apm.2020.12.021. [43] G. Li, X. Li, L. Gao, B. Zeng, Tasks assigning and sequencing of multiple AGVs
[18] S. Talatahari, M. Azizi, Optimization of constrained mathematical and engineer- based on an improved harmony search algorithm, J. Ambient Intell. Humaniz.
ing design problems using chaos game optimization, Comput. Ind. Eng. 145 Comput. 10 (11) (2019) 4533–4546.
(2020) 106560, http://dx.doi.org/10.1016/j.cie.2020.106560. [44] P. Monica, M. Kowsalya, Improved harmony search based optimization of droop
[19] S. Talatahari, M. Azizi, Optimal design of real-size building structures using control parameters for load sharing in DC microgrids, in: TENCON 2019-2019
quantum-behaved developed swarm optimizer, Struct. Des. Tall Special Build. IEEE Region 10 Conference (TENCON), IEEE, 2019, pp. 2199–2203.
29 (11) (2020) e1747. [45] B.F. Zohra, A.K. Lahouari, R. Mostefa, An improved harmony search algorithm
[20] S. Talatahari, M. Azizi, Optimum design of building structures using tribe-interior for solved the combined heat and power economic dispatch, Int. J. Electr. Eng.
search algorithm, Structures 28 (2020) 1616–1633, http://dx.doi.org/10.1016/j. Inform. 11 (2) (2019) 440–450.
istruc.2020.09.075. [46] W. Wu, H. Ouyang, A.W. Mohamed, C. Zhang, S. Li, Enhanced harmony search
[21] S. Talatahari, M. Azizi, Tribe-charged system search for global optimization, algorithm with circular region perturbation for global optimization problems,
Appl. Math. Model. 93 (2021) 115–133, http://dx.doi.org/10.1016/j.apm.2020. Appl. Intell. 50 (3) (2020) 951–975.
12.007. [47] A. Askarzadeh, M. Montazeri, L.D.S. Coelho, A modified harmony search algo-
[22] S. Talatahari, M. Azizi, An extensive review of charged system search algorithm rithm applied to capacitor placement of radial distribution networks considering
for engineering optimization applications, in: S. Carbas, A. Toktas, D. Ustun voltage stability index, Int. J. Bio-Inspired Comput. 13 (3) (2019) 189–198.
(Eds.), Nature-Inspired Metaheuristic Algorithms for Engineering Optimization [48] K. Ji, W. Wang, X. Wang, Z. Wei, J. Yang, H. Qin, Optimal installation of
Applications, Springer Singapore, Singapore, 2021, pp. 309–334. distributed generators based on an enhanced harmony search algorithm, J. Phys.
[23] M. Azizi, Designing fuzzy controllers for frame structures based on ground Conf. Ser. (2019) 022006, IOP Publishing.
motion prediction using grasshopper optimization algorithm: A case study of [49] B. Keshtegar, P. Hao, Y. Wang, Y. Li, Optimum design of aircraft panels based
Tabriz, Iran, in: S. Carbas, A. Toktas, D. Ustun (Eds.), Nature-Inspired Metaheuris- on adaptive dynamic harmony search, Thin-Walled Struct. 118 (2017) 37–45,
tic Algorithms for Engineering Optimization Applications, Springer Singapore,
http://dx.doi.org/10.1016/j.tws.2017.05.004.
Singapore, 2021, pp. 153–180.
[50] H. Ouyang, W. Wu, C. Zhang, S. Li, D. Zou, G. Liu, Improved harmony search
[24] M. Azizi, U. Aickelin, H.A. Khorshidi, M. Baghalzadeh Shishehgarkhaneh, Shape
with general iteration models for engineering design optimization problems, Soft
and size optimization of truss structures by chaos game optimization considering
Comput. 23 (20) (2019) 10225–10260.
frequency constraints, J. Adv. Res. (2022) http://dx.doi.org/10.1016/j.jare.2022.
[51] Y.Y. Moon, Z.W. Geem, G.-T. Han, Vanishing point detection for self-driving
01.002.
car using harmony search algorithm, Swarm Evol. Comput. 41 (2018) 111–119,
[25] M. Farshchin, M. Maniat, C.V. Camp, S. Pezeshk, School based optimization
http://dx.doi.org/10.1016/j.swevo.2018.02.007.
algorithm for design of steel frames, Eng. Struct. 171 (2018) 326–335.
[52] J. Yi, X. Li, C.-H. Chu, L. Gao, Parallel chaotic local search enhanced harmony
[26] N. Khodadadi, M. Azizi, S. Talatahari, P. Sareh, Multi-objective crystal structure
search algorithm for engineering design optimization, J. Intell. Manuf. 30 (1)
algorithm (MOCryStAl): Introduction and performance evaluation, IEEE Access
(2019) 405–428, http://dx.doi.org/10.1007/s10845-016-1255-5.
9 (2021) 117795-117812, http://dx.doi.org/10.1109/ACCESS.2021.3106487.
[53] B. Keshtegar, T. Ozbakkaloglu, A. Gholampour, Modeling the behavior of FRP-
[27] A. Kaveh, A. BolandGerami, Optimal design of large-scale space steel frames
confined concrete using dynamic harmony search algorithm, Eng. Comput. 33
using cascade enhanced colliding body optimization, Struct. Multidiscip. Optim.
(3) (2017) 415–430, http://dx.doi.org/10.1007/s00366-016-0481-y.
55 (1) (2017) 237–256.
[54] M. Hasanipanah, B. Keshtegar, D.-K. Thai, N.-T. Troung, An ANN-adaptive
[28] S. Talatahari, A.H. Gandomi, X.-S. Yang, S. Deb, Optimum design of frame
dynamical harmony search algorithm to approximate the flyrock resulting from
structures using the eagle strategy with differential evolution, Eng. Struct. 91
blasting, Eng. Comput. 38 (2) (2022) 1257–1269, http://dx.doi.org/10.1007/
(2015) 16–25, http://dx.doi.org/10.1016/j.engstruct.2015.02.026.
s00366-020-01105-9.
[29] M.R. Maheri, H. Shokrian, M. Narimani, An enhanced honey bee mating
[55] J. Yi, L. Gao, X. Li, C.A. Shoemaker, C. Lu, An on-line variable-fidelity
optimization algorithm for design of side sway steel frames, Adv. Eng. Softw.
109 (2017) 62–72. surrogate-assisted harmony search algorithm with multi-level screening strategy
for expensive engineering design optimization, Knowl.-Based Syst. 170 (2019)
[30] A. Kaveh, M.I. Ghazaan, Enhanced whale optimization algorithm for sizing
optimization of skeletal structures, Mech. Based Des. Struct. Mach. 45 (3) (2017) 1–19.
345–362. [56] R. Sheikholeslami, A.C. Zecchin, F. Zheng, S. Talatahari, A hybrid cuckoo–
[31] S. Kazemzadeh Azad, O. Hasançebi, S. Kazemzadeh Azad, Upper bound strategy harmony search algorithm for optimal design of water distribution systems, J.
for metaheuristic based design optimization of steel frames, Adv. Eng. Softw. 57 Hydroinform. 18 (3) (2016) 544–563.
(2013) 19–32. [57] B. Keshtegar, P. Hao, Y. Wang, Q. Hu, An adaptive response surface method
[32] S. Kazemzadeh Azad, O. Hasançebi, S. Kazemzadeh Azad, Computationally and Gaussian global-best harmony search algorithm for optimization of aircraft
efficient optimum design of large scale steel frames, Int. J. Optim. Civil Eng. stiffened panels, Appl. Soft Comput. 66 (2018) 196–207, http://dx.doi.org/10.
4 (2014). 1016/j.asoc.2018.02.020.
[33] O. Hasançebi, Cost efficiency analyses of steel frameworks for economical design [58] H. Ouyang, G. Ma, G. Liu, Z. Li, X. Zhong, Hybrid teaching-learning based
of multi-storey buildings, J. Construct. Steel Res. 128 (2017) 380–396. optimization with harmony search for engineering optimization problems, in:
[34] C. Tort, S. Şahin, O. Hasançebi, Optimum design of steel lattice transmission line 2017 36th Chinese Control Conference (CCC), IEEE, 2017, pp. 2714–2717.
towers using simulated annealing and PLS-TOWER, Comput. Struct. 179 (2017) [59] B. Keshtegar, S. Etedali, Nonlinear mathematical modeling and optimum design
75–94. of tuned mass dampers using adaptive dynamic harmony search algorithm,
[35] T. Kundu, H. Garg, A hybrid ITLHHO algorithm for numerical and engineering Struct. Control Health Monit. 25 (7) (2018) e2163, http://dx.doi.org/10.1002/
optimization problems, Int. J. Intell. Syst. http://dx.doi.org/10.1002/int.22707, stc.2163.
n/a(n/a). [60] S. Gholizadeh, A. Barzegar, Shape optimization of structures for frequency
[36] A. Kaveh, Y. Vazirinia, An upgraded Sine cosine algorithm for tower crane constraints by sequential harmony search algorithm, Eng. Optim. 45 (6) (2013)
selection and layout problem, Periodica Polytechn. Civil Eng. 64 (2) (2020) 627–646.
325–343, http://dx.doi.org/10.3311/PPci.15363. [61] M. Jaberipour, E. Khorram, A new harmony search algorithm for solving
[37] I. Brajević, M. Tuba, An upgraded artificial bee colony (ABC) algorithm for mixed–discrete engineering optimization problems, Eng. Optim. 43 (5) (2011)
constrained optimization problems, J. Intell. Manuf. 24 (4) (2013) 729–740, 507–523.
http://dx.doi.org/10.1007/s10845-011-0621-6. [62] A. ANSI, ANSI/AISC 360-10, in: Specification for Structural Steel Buildings,
[38] I. Brajević, J. Ignjatović, An upgraded firefly algorithm with feasibility-based American Institute of Steel Construction, Inc., Chicago, IL, 2010.
rules for constrained engineering optimization problems, J. Intell. Manuf. 30 (6) [63] K.S. Lee, Z.W. Geem, A new structural optimization method based on the
(2019) 2545–2574, http://dx.doi.org/10.1007/s10845-018-1419-6. harmony search algorithm, Comput. Struct. 82 (9–10) (2004) 781–798.
[39] V.K. Pathak, A.K. Srivastava, A novel upgraded bat algorithm based on cuckoo [64] ASCE, Minimum Design Loads for Buildings and Other Structures, American
search and sugeno inertia weight for large scale and constrained engineering Society of Civil Engineers, 2013.
design optimization problems, Eng. Comput. (2020) http://dx.doi.org/10.1007/ [65] S. Arora, S. Singh, Butterfly optimization algorithm: a novel approach for global
s00366-020-01127-3. optimization, Soft Comput. 23 (3) (2019) 715–734.

19
M. Azizi, S. Talatahari, M. Basiri et al. Decision Analytics Journal 3 (2022) 100067

[66] A.A. Heidari, S. Mirjalili, H. Faris, I. Aljarah, M. Mafarja, H. Chen, Harris hawks [68] V. Muthiah-Nakarajan, M.M. Noel, Galactic swarm optimization: A new global
optimization: Algorithm and applications, Future Gener. Comput. Syst. 97 (2019) optimization metaheuristic inspired by galactic motion, Appl. Soft Comput. 38
849–872, http://dx.doi.org/10.1016/j.future.2019.02.028. (2016) 771–787.
[67] S. Mirjalili, S.M. Mirjalili, A. Hatamlou, Multi-verse optimizer: a nature-inspired [69] A.H. Gandomi, A.H. Alavi, Krill herd: a new bio-inspired optimiza-
algorithm for global optimization, Neural Comput. Appl. 27 (2) (2016) 495–513, tion algorithm, Commun. Nonlinear Sci. Numer. Simul. 17 (12) (2012)
http://dx.doi.org/10.1007/s00521-015-1870-7. 4831–4845.

20