Journal of Constructional Steel Research 128 (2017) 380–396

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Cost efficiency analyses of steel frameworks for economical design of

multi-storey buildings
Oğuzhan Hasançebi
Middle East Technical University, Department of Civil Engineering, Ankara, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: In the present study cost efficiencies of various steel frameworks are investigated for economical design of multi-
Received 8 February 2016 storey buildings. A total of thirteen steel frames that incorporate various types of beam-column connection and
Received in revised form 27 August 2016 bracing configuration are considered for detailed and comparative cost analyses. The three multi-storey buildings
Accepted 14 September 2016
consisting of 10, 20 and 30 floors are stiffened according to each of the thirteen steel frameworks to yield thirty-
Available online xxxx
nine test frames for numerical applications. First design optimizations are carried out using an evolution strategy
Keywords:
(ES) integrated parallel optimization algorithm to minimize the total member weight in each test frame. An ex-
Cost efficiency analysis tensive cost analysis is then carried out on the optimized design of each test frame to calculate its estimated con-
Multi-storey steel buildings struction cost using a cost model that itemizes costs of all production stages including material, manufacturing,
Steel frameworks erection and transportation. Cost-efficient frameworks are identified for the three steel buildings by comparing
Beam-column connections estimated costs of the test frames. Furthermore, the variations in cost efficiencies of the steel frameworks versus
Bracing configurations the storey number (or building height) are scrutinized. The results collected are utilized to reach certain recom-
Structural optimization mendations regarding the selection of economically feasible frames for design of multi-storey steel buildings.
Parallel evolution strategy
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction connection and bracing configuration. In Memari and Madhkhan [1],

The design of tall steel buildings is usually governed by lateral loads
especially as the height-to-width ratio of a structural system increases.
Hence, steel buildings must be designed and detailed to have sufficient
rigidity and stability to resist lateral loads safely. One way to provide lat-
eral stability in such structures is to have moment-resisting (rigid) type
connections between beams and columns. In such rigidly connected
systems the lateral loads are resisted through flexural stiffness of
beams and columns. Another way is to stiffen a structural framework
with a full-bracing system that behaves like a vertical truss throughout
the height of the building to transmit lateral forces to the ground. In
general, a bracing system can be arranged in a variety of different topo-
logical configurations depending on structural and architectural re-
quirements, and the most common ones used in practice are cross (X),
diagonal, Z, V and eccentric V-bracings, etc. It should be noted that at
times when all the beam-column connections are of moment-free
type (i.e. pin connections) in a steel frame, an integrated bracing system
is necessary to provide lateral stability of the structure. A bracing system
can also be used to stiffen rigidly connected frameworks to increase the
lateral stability of such structures.
Several studies in the literature have investigated design weight ef-
ficiencies of steel frames featuring various types of beam-column
E-mail address: oguzhan@metu.edu.tr.
optimum design weights of various two-dimensional braced and
unbraced steel frames were examined under gravity and lateral seismic
forces. The optimization problem was formulated as the minimum
weight design of a frame subject to a number of behavioural constraints
including combined bending and axial stress, shear stress, compression
buckling, tension slenderness and drift ratio according to AISC-ASD
(American Institute of Steel Construction – Allowable Stress Design)
[2] specifications. Several rigid and pin-jointed planar frames stiffened
with a bracing system were sized optimally using a feasible directions
optimization method. Amongst various bracing configurations consid-
ered in their studies, the frame with a V-bracing yielded the lightest de-
sign weight, whereas the one with an X-bracing led to the heaviest
design weight. Kameshki and Saka [3] investigated efficiencies of X, V
and Z-bracings in pin-jointed frames as well as rigidly connected frames
without any bracing system. They employed a genetic algorithm for the
optimum sizing designs of the planar frames according to BS 5950 [4]
specifications. Considering the design optimization of a 3-bay 15-storey
frame, it was demonstrated that the X-bracing system yielded the
lightest design weight for the frame. Further, it was concluded that in
rigidly connected frames as well as in pin-jointed frames with V or Z-
bracings, inter-storey drift constraints were dominant criteria in the
design process. Other studies in the literature have attempted to find
optimal distribution of bracing members in steel frames. In Liang et al.
[5], the optimum topology design of bracing systems was searched for
planar steel frames using a performance based design optimization

http://dx.doi.org/10.1016/j.jcsr.2016.09.002
0143-974X/© 2016 Elsevier Ltd. All rights reserved.
 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396 381

method. In their approach a continuum design domain was used to stiff- the test frames, connection designs, and in-depth cost analyses are pre-
en a multi-storey building and inefficient materials were removed grad- sented in the sixth section. Finally, the concluding remarks are outlined
ually from this domain until a performance based index of the bracing in the last section of the paper.
system was maximized. Recently, Kaveh and Farhoodi [6] studied the
problem of layout optimization for X-bracing of steel frames according 2. Formulation of design optimization problem
to specifications of IBC2006 [7].
The literature survey reveals that efficiencies of various steel frame- For a steel frame consisting of Nm members that are collected in Nd
works have been so far investigated by comparing the design weights of design groups (variables), the optimum design problem according to
the resulting structures when different framing systems are adopted. In AISC-ASD [11] specifications yields the following discrete programming
these studies, a steel frame has been modeled as a planar structure and problem, if the design groups are selected from standard sections.
its design weight is associated only with the total member weight in the The objective is to find a vector of integer values I (Eq. (1))
frame, i.e. connection weights are not included. On the other hand, it is representing the sequence numbers of steel sections assigned to Nd
common to anyone that the minimum design weight of a steel frame member groups
only ensures the least material cost for members, yet cannot guarantee

the lowest construction cost on the whole. In particular, connection de- I T ¼ I 1 ; I2 ; :::; INd ð1Þ
signs may appreciably affect the manufacturing cost of a steel frame
since the fabrication cost of joints can be in excess of 30% of the total fab- to minimize the weight (W) of the frame
rication cost of a structure [8]. Therefore, it is extremely important that
design efficiencies of different steel frameworks are evaluated based on X
Nd X
Nt
W¼ ρi A i Lj ð2Þ
construction costs of the resulting structures, rather than design
i¼1 j¼1
weights only. In this regard, employing a precise and realistic cost
model is required to determine estimated costs of the generated de-
where Ai and ρi are the length and unit weight of a steel section adopted
signs. Pavlovcic et al. [9] performed a cost function analysis in design op-
for the member group i respectively, Nt is the total number of members
timization of steel frames, and developed a cost model that include all
in group i, and Lj is the length of the member j which belongs to the
essential fabrication and erection activities of steel frames. Unlike
group i.
usual practice in structural optimization where a simple cost function
The members subjected to a combination of axial compression and
is developed by multiplying some geometrical properties by suitable
flexural stress must be sized to meet the following stress constraints:
weights representing cost coefficients, their cost function itemizes all
stages of production including welding, cutting, drilling, surface prepa- 2 3
ration, assembly, flange aligning, painting as well as steel and bolting 6 7
f 6fa C mx f bx C my f by 7
material costs, transportation, and erection, etc. if a N0:15; 6
6 Fa þ ! þ ! 7
7−1:0≤0 ð3Þ
Fa 4 f f 5
The present study is concerned with investigating cost efficiencies of 1− 0a F bx 1− 0a F by
various steel frameworks for economical design of multi-storey build- F ex F ey
ings. In this context a total of thirteen steel frames that incorporate var-
 
ious types of beam-column connection (i.e. rigid or pin) and bracing fa f f by
configuration (i.e. X, Z, V, eccentric V-bracings) are considered for de- þ bx þ −1:0≤0 ð4Þ
0:60F y F bx F by
tailed and comparative cost analyses. The numerical examples are per-
formed using three multi-storey buildings consisting of 10, 20 and 30  
fa f f f by
floors. It is assumed that the buildings are subjected to gravity loads as if ≤0:15; a þ bx þ −1:0≤0 ð5Þ
Fa F a F bx F by
well as lateral wind loads according to ASCE 7-05 [10] code of practice.
Each building is modeled as a space frame stiffened according to each of
If the flexural member is under tension, then the following formula
the thirteen steel frameworks, resulting in thirty-nine test frames for
is used instead:
numerical applications. Each test frame is first sized using standard
hot rolled sections to attain minimum weight design of its members  
fa f f by
subject to stress, stability and displacement limitations in accordance þ bx þ −1:0≤0 ð6Þ
with AISC-ASD [11] specifications. The design optimization is performed 0:60F y F bx F by
using an evolution strategy (ES) integrated parallel optimization algo-
rithm developed earlier by the author [12] for optimizing very large
steel structures, especially high-rise buildings, in a timely manner. An In Eqs. (3)–(6), Fy is the yield stress of steel, and fa = (P/A) represents
extensive cost analysis is then carried out on the optimized design of the computed axial stress, where A is the cross-sectional area of the
each test frame using the cost model developed by Pavlovcic et al. [9] member. The computed flexural stresses due to bending of the member
to accurately estimate its construction cost. Cost-efficient frameworks about its major (x) and minor (y) principal axes are denoted by fbx and
are identified for the three steel buildings by comparing the estimated fby respectively. Fex′ and Fey′ denote the Euler stresses about principal axes
costs of the test frames. In addition, the variations in cost efficiencies of the member that are divided by a safety factor of 23/12. It should be
of the steel frameworks versus the storey number (or building height) noted unlike tension members for which the safety factor is given as
are examined. Based on the numerical investigations, certain recom- 5/3, the AISC-ASD [11] employs a higher safety factor for compression
mendations are reached regarding the selection of economically feasible members under Euler buckling. The reason for this is to account for
frames for design of multi-storey steel buildings. The following sections the P-delta magnification effect. Fa stands for the allowable axial stress
of the paper are organized as follows. The second section presents opti- under axial compression force alone, and is calculated depending on
mum design formulations of space steel frames according to AISC-ASD elastic or inelastic bucking failure mode of the member using Formulas
[11]. The ES integrated parallel optimization algorithm is briefly 1.5-1 and 1.5-2 given in AISC-ASD [11]. The allowable bending compres-
overviewed in the third section. The thirteen steel frameworks incorpo- sive stresses about major and minor axes are designated by Fbx and Fby,
rating various types of beam-column connection and bracing configura- which are computed using the Formulas 1.5-6a or 1.5-6b and 1.5-7
tion are introduced in the fourth section. The fifth section summarizes given in AISC-ASD [11]. It is important to note that while calculating al-
the cost model utilized in the study for estimating construction costs lowable bending stresses, a newer formulation (Eq. (7)) of the moment
of the steel frames. The numerical examples, sizing optimizations of gradient coefficient cb given in ANSI/AISC 360-05 [13] is employed in
382 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396

the study to account for the effect of moment gradient on lateral tor- bfb
sional buckling resistance of the members,

12:5M max B1
cb ¼ Rm ≤3:0 ð7Þ tf
2:5M max þ 3M A þ 4MB þ 3MC
dc
where Mmax, MA, MB and MC are the absolute values of maximum, quar-
ter-point, midpoint, and three-quarter point moments along the
bfc b'fb
unbraced length of the member respectively, and Rm is a coefficient
which is =1.0 for doubly symmetric sections. Cmx and Cmy are the re- B2
duction factors, introduced to counterbalance overestimation of the sec-
ond-order moments by the amplification factor (1− fa/F′). e For unbraced
frame members, they are taken as 0.85. For braced frame members
without transverse loading between their ends, they are calculated
from Cm = 0.6 − 0.4(M1/M2), where M1/M2 is the ratio of smaller end
moment to the larger end moment. For braced frame members having
transverse loading between their ends, they are determined from the
formula Cm = 1 + ψ(fa/Fe′) based on a rational approximate analysis
outlined in AISC-ASD [11] Commentary-H1, where ψ is a parameter
that considers maximum deflection and maximum moment in the
member.
For the computation of allowable compression and Euler stresses,
the effective length factors (K) are required. For beam and bracing
members, K is taken equal to unity. For column members, alignment
charts furnished in AISC-ASD [11] can be utilized. In this study, however,
the effective length factors of columns in braced and unbraced steel
frames are calculated from the following approximate formulas devel-
oped by Dumonteil [14], which are accurate to within about −1.0 and
+2.0% of the exact results [15]:
For unbraced members:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:6GA GB þ 4ðGA þ GB Þ þ 7:5
K¼ ð8Þ
GA þ GB þ 7:5

For braced members:

Fig. 1. Beam-column geometric constraints.

3GA GB þ 1:4ðGA þ GB Þ þ 0:64

K¼
3GA GB þ 2:0ðGA þ GB Þ þ 1:28
where GA and GB refer to stiffness ratio or relative stiffness of a column at
its two ends.
It is also required that the computed shear stresses (fv) in members
are smaller than the allowable shear stresses(Fv), as formulated in Eq.
(10).
f v ≤ F v ¼ 0:40C v F y
In Eq. (10), Cv is referred to as web shear coefficient. It is taken equal
to Cv =1.0 for rolled W-shaped members with h/tw ≤2.24E/Fy, where h is
the clear distance between flanges, E is the elasticity modulus and tw is
the thickness of web. For all other symmetric shapes, Cv is calculated
from Formulas G2-3, G2-4 and G2-5 in ANSI/AISC 360-05 [13].
Apart from stress constraints, slenderness limitations are also im-
posed on all members such that maximum slenderness ratio (λ = KL/
r) is limited to 300 and 200 for tension and compression members re-
spectively. The displacement constraints are imposed such that the
maximum lateral displacements are restricted to be less than H/400,
and upper limit of storey drift is set to be h/400, where H is the total
height of the frame building and h is the height of a storey.
Finally, we consider geometric constraints between beams and col-
umns framing into each other at a common joint for practicality of an
optimum solution generated. For the two beams B1 and B2 and the col-
umn shown in Fig. 1, one can write the following geometric constraints:
bfb
−1:0≤0
bfc
ð9Þ
0
bfb
  −1:0≤0 ð12Þ
dc −2t f
where bfb, bfb′ and bfc are the flange width of the beam B1, the beam B2
and the column respectively, dc is the depth of the column, and tf is
the flange width of the column. Eq. (11) simply ensures that the flange
width of the beam B1 remains smaller than that of the column. On the
ð10Þ other hand, Eq. (12) enables that flange width of the beam B2 remains
smaller than clear distance between the flanges of the column(dc − 2tf).
3. A parallel evolution strategy algorithm
Numerous methods have been developed by the researchers in the
past few decades for solving optimization problems. Amongst them a
group of techniques called nature-inspired metaheuristics have re-
ceived a tremendous attention from all disciplines in science and engi-
neering. These techniques use nature as a source of inspiration for
their algorithmic models and implementations. Apparently, a sound
reputation of metaheuristics in structural design optimization is owing
to their ease of application, robust performances, independency on gra-
dient information, and capability of handling both continuous and dis-
crete design variables. The state-of-the-art reviews of metaheuristics
in the realm of structural design optimization are outlined in some ex-
cellent review articles, such as Saka [16], Lamberti and Pappalettere
[17], Saka and Doğan [18], etc.
In the present study the above-formulated optimization problem is
ð11Þ solved using a metaheuristic approach, namely a discrete evolution
strategy (ES) integrated parallel optimization algorithm developed
 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396 383

earlier by the author [12]. A comprehensive comparison of this algo-
rithm with other metaheuristics in Hasançebi et al. [19] indicates the ef-
ficiency of the former in the context of design optimization of steel
frames. The ES algorithm employs an iterative procedure that aims to
improve (evolve) a population of designs (individuals) over a selected
number of generations. In each iteration of this algorithm, several evo-
lutionary operators including recombination, a geometric distribution
based mutation, and selection are applied in sequence to generate indi-
viduals of the next population. For each individual generated in the
course of the optimization, a structural analysis needs to be carried
out to obtain force and deformation responses of the corresponding de-
sign. Accordingly, the response calculations of generated individuals
may be computationally massive especially for large scale problems,
reaching as much as 98% of the total computing time in a serial execu-
tion of the algorithm. On the other hand, the discrete ES algorithm is
very suitable for parallel implementation since evaluation of the design
population can be conducted independently and concurrently using a
cluster of computers rather than a single computer. To this end, the ES
algorithm is parallelized based on a master-slave type parallelization
model of the optimization procedure. In this parallelization model,

Fig. 2. The thirteen steel frames investigated for comparative cost analyses of steel buildings.
384 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396

while a master processor executes evolutionary operators (recombina- integrated bracing system, and (iii) pin-jointed frames with an integrat-
tion, mutation and selection) and provides overall coordination, com- ed bracing system. In a fourth group a special steel connection design
munication and synchronization within the network, multitask loops application in some countries is considered, in which steel frames con-
such as structural analyses of generated designs are distributed to a sist of rigid connections at interior joints and pin connections at exterior
number of slave processors. The interested reader is referred to ones with an integrated bracing system. A total of thirteen frameworks
Hasançebi et al. [12] for complete details of the discrete ES algorithm (forms) featuring different types of beam-column connection and brac-
as well as the subtle characteristics of its parallelization and multi- ing configuration are produced from these four groups for detailed and
tasking environment. comparative cost analyses. They are displayed in Fig. 2 in conjunction
with the design of 10-storey building, and their structural features are
4. Steel frameworks for lateral stability summarized in Table 1 for the purpose of clarity. Accordingly, the
forms investigated here include a rigidly connected frame (form A;
The rigidity and stability of structural systems to lateral forces is a group 1), four rigidly connected frames with X, Z, V and eccentric V-
major concern in the design of tall buildings since high wind forces bracings (forms B, C, D, E; group 2), three pin-jointed frames with X, Z
acting on the facades of the buildings generally cause significant and V and eccentric V-bracings (forms F, G, H, I; group 3), and four
overturning moments and lateral drifts. Hence, the steel buildings are frames having rigid connections at interior joints and pin connections
integrated with various framing systems to form structures having suf- at exterior ones with X, Z, V and eccentric V-bracings (forms J, K, L, M;
ficient lateral stiffness and drift index within acceptable limits. One group 4).
framing system is a rigid frame, which consist of moment-resisting
(rigid) beam-column connections. In such a frame the lateral loads are 5. A cost analysis of steel frames
resisted through flexural stiffness of beams and columns. An alternative
way is to stiffen a steel frame with a bracing system in the form of a ver- The minimum weight design of a steel building ensures the least
tical truss (placed throughout the height of a building) that resists later- usage of material, yet may not truly reflect its overall construction
al loads by a cantilever truss action. This way the flexural action cost. Therefore, the efficiency of a steel frame adopted for multi-storey
exhibited by a rigid frame is eliminated and the horizontal shear is pri- buildings should rely on its cost, rather than its design weight. In this
marily resisted by axial stiffness of columns and bracing members in the context it is essential to employ a precise and realistic cost model that
vertical truss. There are other structural systems, such as framed tube, accurately estimates the overall cost of a steel frame. In the study the
trussed tube, bundled tube, etc., which might be adopted for economical cost model developed by Pavlovcic et al. [9] is implemented for cost ef-
design of mega-structures [20], which are out of the scope of this paper. ficiency analyses of the steel frames generated according to different
In general, a number of possibilities exist for topological configura- forms. This cost model itemizes all stages of production for steel frames,
tions of bracing members, and the selection is governed by structural in which total cost of a structure (Cst) can be formulated as summation
and architectural requirements. The most typical configuration is an of five essential items; namely material cost of elements(Cel), material
X-bracing, where lateral forces induce tension on one diagonal and and manufacturing cost of joints(Cj), transportation cost (Ct), erection
compression on the other. However, the contribution of compression cost (Cer)and extra costs (Cex).
diagonal to the lateral stiffness is usually ignored, and it is assumed
that only one (tension) diagonal remains effective alternately depend- C st ¼ C el þ C j þ C t þ C er þ C ex ð13Þ
ing on the direction of lateral loads. Another configuration is a diagonal
bracing, where a single bracing member is used in each panel instead of
two. Accordingly, in this configuration the diagonals must be designed All the terms in Eq. (13) are described in sufficient detail in the fol-
to resist both tension and compression since lateral loads can act in lowing subsections. For more details, the reader is referred to Pavlovcic
both directions. The problem associated with this type of bracing system et al. [9]. It is worth mentioning that the costs in the present study are
is that the frame is less stiff in the direction inducing compression on expressed in Euro currency, whereas the unit of mass is kilogram (kg).
bracing members. A modified form of diagonal bracing is Z-bracing,
where a single bracing member is used in each panel, yet the inclina- 5.1. Cost of elements
tions of bracing members in successive stories are opposite. Hence, for
a certain direction of loading the bracing members in successive stories The cost of elements (Cel) depends on the total mass of elements in a
are in opposite (tension, compression) forces. This leads to non-uniform steel frame, and is computed from Eq. (14),
lateral stiffness on the storey basis such that the stories with compres-
sion braces will be weaker than those with tension braces. Unlike for- C el ¼ ks mel ð14Þ
mers, a V-bracing does not run across the wall and is used if some
openings are required in braced panels. In this bracing system the V-
Table 1
braces are connected at mid-spans of the beams leading to reduced flex-
A summary of the thirteen steel frames investigated for cost-efficient design of multi-sto-
ural forces for these members. When further spacing is required than rey buildings.
provided by V-bracing, one can alternatively use eccentric V-bracing
Frame Form Type of connection Bracing
in which the V-braces do not meet concentrically on the beams. Instead
group configuration
some eccentricity is introduced between the ends of the V-braces at the
expense of having more spacing, introducing additional flexure and I A Rigid No bracing
II B Rigid X
shear into framing beams. C Rigid Z
The bracing systems are essential to provide lateral stability of steel D Rigid V
frames that consist of moment-free connections (i.e. pin-joints) only. E Rigid Eccentric V
However, they can also be incorporated into rigidly connected frames III F Pin X
G Pin Z
to produce mixed-type structural systems that absorb the horizontal
H Pin V
shear by developing flexural and cantilever truss actions together. Ac- I Pin Eccentric V
cordingly, steel frameworks are classified into four groups here based IV J Rigid at interior joints and pin at exterior joints X
on the type of beam-column connection and whether the frame is stiff- K Rigid at interior joints and pin at exterior joints Z
ened by an integrated bracing system or not. The three of these groups L Rigid at interior joints and pin at exterior joints V
M Rigid at interior joints and pin at exterior joints Eccentric V
are (i) rigidly connected frames, (ii) rigidly connected frames with an
 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396 385

where ks =0.40 is the material cost factor for steel members in Euro/kg,
and mel is the total element mass in kg.
5.2. Cost of joints
The cost estimation for joints is relatively more complicated, and in-
cludes the sum of material and manufacturing costs of all joints in a steel
frame, as expressed in Eqs. (15)–(17).
for hole forming in Euro/min, nh is the number of holes, tp is the thick-
ness of drilled plate in cm, tpe is the additional drilling path in cm, Vh is
the drilling speed in cm/min, which is calculated from
2
V h ¼ 0:763d −5:720d þ 20:96 ð27Þ
where d is the hole diameter in cm, and finally Thne = 11.9 is the addi-
tional time for preparation in min.

C j ¼ C jm þ C jn ð15Þ 5.3. Cost of transportation

C jm ¼ C wm þ C bm þ C sm þ C pm ð16Þ The transportation cost (Ct) is estimated in proportion to the total

mass of the structure using Eq. (28),
C jn ¼ C wn þ C hn ð17Þ
C t ¼ kt ms ð28Þ
In Eqs. (15)–(17), Cjm and Cjn are the total material and manufactur-
ing costs of joints respectively; Cwm, Cbm, Csmand Cpm are the total costs of where kt = 0.022 is the cost factor for transportation in Euro/kg, and ms
materials used for welding, bolting, stiffeners, and end plates respective- is the total mass of the structure in kg.
ly; and finally Cwn and Chn are the total manufacturing costs of welding
and hole forming respectively. In the following sub-sections the formu- 5.4. Cost of erection
lations are presented to calculate the above-mentioned cost quantities.
Similar to the cost of transportation, the cost of erection (Cer) is also
5.2.1. Material cost of joints formulated as a function of structural mass using Eq. (29),
The material cost of joints is obtained through Eq. (16) by calculating
C er ¼ ker T er ms ð29Þ
the cost quantities Cwm, Cbm, Csm, and Cpm from Eqs. (18)–(21).

C wm ¼ kwm M w Lw ð18Þ where ker = 120.2 is the cost factor for erection in Euro/hour, and Ter =
0.0014 is the labor-hour per unit quantity of steel in hour/kg.
C bm ¼ kbm nb ð19Þ
5.5. Extra costs
C sm ¼ ks msm ð20Þ
There are other costs (Cex), such as painting, flange aligning, surface
C pm ¼ ks mpm ð21Þ preparation, cutting and welding of the elements, etc., which should
also be taken into account while estimating the final cost of a steel
In Eqs. (18)–(21), kwm = 1.40 is the material cost factor for welding construction project. Detailed formulations are also presented for these
in Euro/kg, Lw is the total length of weld in meter (m), Mw is the mass cost quantities in Pavlovcic et al. [9]. However, for the sake of simplicity
of weld material in kg/m, which is calculated from Eq. (22), the extra costs are also treated as a function of total structural mass in
this study based on the results of the numerical examples discussed in
Mw ¼ 1:33a2w þ 0:19aw −0:02 ð22Þ [9], and it is set to 0.184 Euro/kg for each steel frame investigated.

where aw is the weld size in centimeter (cm), nb is the total number of 6. Numerical examples
bolts, kbm is the material cost factor for bolts in Euro/bolt, which is calcu-
lated from The numerical examples are carried out using three multi-storey
2
steel buildings consisting of 10, 20 and 30 floors. All the buildings
kbm ¼ 3:076d −7:373d þ 4:62 ð23Þ have an identical plan view shown in Fig. 3, where each storey is
3.66 m (12 ft) high. Each building is modeled as a space frame stiffened
where d is the bolt diameter in cm, msm is the total mass of stiffeners,
according to each of the thirteen steel frameworks, resulting in thirty-
and mpm is the total mass of end plates.

5.2.2. Manufacturing cost of joints

The manufacturing cost of joints is obtained through Eq. (17) by
calculating the cost quantities Cwn and Chn from the following two
equations.

C wn ¼ kwn ð f wn þ T wn Lw þ T wne Þ ð24Þ

   
C hn ¼ khn nh t p þ t pe =V h f wn þ T hne ð25Þ

In Eqs. (24) and (25), kwn =0.12 is the manufacturing cost factor for
welding in Euro/min, fwn = 1.40 is a welding factor that increases
welding arch time in case of short welds or positional welding, Twn is
the operation time per unit length of weld in min/m, which is calculated
from

T wn ¼ 17:26a2w þ 2:9aw þ 1:82 ð26Þ

where aw is the weld size in cm, Lw is the weld length in m, Twne =0.3 is
any additional welding time in min, khn =0.32 is the manufacturing cost Fig. 3. Typical plan view of the 10, 20 and 30-storey steel buildings.
386 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396

Table 2 Table 4
Gravity loading for roof and floor beams of all the test frames. Wind loads calculated for the test frames of the 20-storey building.

Beam type Outer span beams Inner span beams Floor z (m) Kz x-Direction y-Direction

(kN/m) (kN/m) Distributed Distributed Distributed Distributed

windward leeward windward leeward
Long span floor beams 9.79 19.59
force (kN/m) force (kN/m) force (kN/m) force (kN/m)
Short span floor beams 8.04 16.07
Long span roof beams 6.75 13.50 1 3.7 0.57 2.33 2.83 2.33 3.21
Short span roof beams 5.54 11.07 2 7.3 0.66 2.66 2.83 2.66 3.21
3 11.0 0.74 2.99 2.83 2.99 3.21
4 14.6 0.80 3.24 2.83 3.24 3.21
5 18.3 0.85 3.46 2.83 3.46 3.21
nine test frames in all. Each test frame is first sized for minimum weight 6 21.9 0.90 3.64 2.83 3.64 3.21
design of its members using standard sections in AISC. Accordingly, the 7 25.6 0.94 3.81 2.83 3.81 3.21
beams and columns are selected from 297 and 171 economical wide- 8 29.3 0.98 3.95 2.83 3.95 3.21
flange (W) sections respectively, whereas the bracing members are 9 32.9 1.01 4.09 2.83 4.09 3.21
10 36.6 1.04 4.22 2.83 4.22 3.21
adopted from 147 economical W-sections. The combined stress, stabil-
11 40.2 1.07 4.33 2.83 4.33 3.21
ity, displacement and geometric constraints are imposed according to 12 43.9 1.10 4.44 2.83 4.44 3.21
the provisions of AISC-ASD [11], as discussed in Section 2. For all the 13 47.5 1.12 4.54 2.83 4.54 3.21
test frames, the steel modulus of elasticity is set to (E) = 29,000 ksi 14 51.2 1.15 4.64 2.83 4.64 3.21
(203,893.6 MPa) and the yield stress is taken as (Fy) = 36 ksi 15 54.9 1.17 4.73 2.83 4.73 3.21
16 58.5 1.19 4.82 2.83 4.82 3.21
(253.1 MPa). 17 62.2 1.21 4.91 2.83 4.91 3.21
18 65.8 1.23 4.99 2.83 4.99 3.21
6.1. Member grouping 19 69.5 1.25 5.06 2.83 5.06 3.21
20 73.2 1.27 2.57 1.41 2.57 1.61
For sizing optimization of each test frame the members are collected
in a number of groups, considering symmetry of the structure as well as
fabrication requirements of the members. The member grouping is con- floors are tabulated in Table 2, and they are applied as uniformly distrib-
ducted such that corner columns (CC) are grouped together as having uted loads on the beams.
the same section over two adjacent floors, as are interior columns (IC), On the other hand, wind loads acting on the frames are computed
outer columns on long facades (OCl), outer columns on short facades using the following equation in ASCE 7-05 [10]:
(OCs), inner beams (IB), outer beams (OB) and bracing members
(BM). It should be noted that such grouping of columns are carried
 
out considering different tributary areas for columns under gravity pw ¼ 0:613K z K zt K d V 2 I GC p ð30Þ
loads (dead, live and snow loads). The details of the member grouping
are depicted in Fig. 3 on a plan view of a steel frame. This leads to 30,
60 and 90 independent size variables for the test frames of 10, 20 and
Table 5
30-storey buildings stiffened according to the form A respectively, Wind loads calculated for the test frames of the 30-storey building.
whereas the corresponding numbers of size variables are 35, 70 and
Floor z (m) Kz x-Direction y-Direction
105 for the test frames of all other forms due to presence of bracing
members in the latters. Distributed Distributed Distributed Distributed
windward leeward windward leeward
force (kN/m) force (kN/m) force (kN/m) force (kN/m)
6.2. Design loads
1 3.7 0.57 2.33 3.17 2.33 3.61
2 7.3 0.66 2.66 3.17 2.66 3.61
For design purpose, all the frames are subjected to various gravity
3 11.0 0.74 2.99 3.17 2.99 3.61
loads as well as lateral wind forces. The gravity loads acting on floor 4 14.6 0.80 3.24 3.17 3.24 3.61
slabs include dead, live, and snow loads. All floors excluding the roof 5 18.3 0.85 3.46 3.17 3.46 3.61
are subjected to a dead load of 2.88 kN/m2 (60.13 lb./ft2) and a live 6 21.9 0.90 3.64 3.17 3.64 3.61
load of 2.39 kN/m2 (50 lb./ft2). The roof is assumed to be subjected to 7 25.6 0.94 3.81 3.17 3.81 3.61
8 29.3 0.98 3.95 3.17 3.95 3.61
a dead load of 2.88 kN/m2 (60.13 lb./ft2) plus a snow load of 0.75 kN/ 9 32.9 1.01 4.09 3.17 4.09 3.61
m2 (15.75 lb./ft2). The resulting gravity loads on the beams of roof and 10 36.6 1.04 4.22 3.17 4.22 3.61
11 40.2 1.07 4.33 3.17 4.33 3.61
12 43.9 1.10 4.44 3.17 4.44 3.61
Table 3 13 47.5 1.12 4.54 3.17 4.54 3.61
Wind loads calculated for the test frames of the 10-storey building. 14 51.2 1.15 4.64 3.17 4.64 3.61
15 54.9 1.17 4.73 3.17 4.73 3.61
Floor z (m) Kz x-Direction y-Direction 16 58.5 1.19 4.82 3.17 4.82 3.61
17 62.2 1.21 4.91 3.17 4.91 3.61
Distributed Distributed Distributed Distributed
18 65.8 1.23 4.99 3.17 4.99 3.61
windward leeward windward leeward
19 69.5 1.25 5.06 3.17 5.06 3.61
force (kN/m) force (kN/m) force (kN/m) force (kN/m)
20 73.2 1.27 5.14 3.17 5.14 3.61
1 3.7 0.57 2.33 2.32 2.33 2.63 21 76.8 1.29 5.21 3.17 5.21 3.61
2 7.3 0.66 2.66 2.32 2.66 2.63 22 80.5 1.30 5.28 3.17 5.28 3.61
3 11.0 0.74 2.99 2.32 2.99 2.63 23 84.1 1.32 5.35 3.17 5.35 3.61
4 14.6 0.80 3.24 2.32 3.24 2.63 24 87.8 1.34 5.41 3.17 5.41 3.61
5 18.3 0.85 3.46 2.32 3.46 2.63 25 91.4 1.35 5.48 3.17 5.48 3.61
6 21.9 0.90 3.64 2.32 3.64 2.63 26 95.1 1.37 5.54 3.17 5.54 3.61
7 25.6 0.94 3.81 2.32 3.81 2.63 27 98.8 1.38 5.60 3.17 5.60 3.61
8 29.3 0.98 3.95 2.32 3.95 2.63 28 102.4 1.40 5.66 3.17 5.66 3.61
9 32.9 1.01 4.09 2.32 4.09 2.63 29 106.1 1.41 5.71 3.17 5.71 3.61
10 36.6 1.04 2.11 1.16 2.11 1.32 30 109.7 1.42 2.88 1.59 2.88 1.80
 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396 387

Table 6 Table 8
The optimized member weights of the test frames for the 10, 20 and 30-storey buildings. The optimized designs of the 20-storey building for selected forms A, B, F and J.

Frame group Form Optimized member design weights (kg) Stories Member group Ready sections

10–Storey 20-Storey 30-Storey A20 B20 F20 J20

I A 149,103 573,637 1,442,439 (Group 1) (Group 2) (Group 3) (Group 4)

II B 111,408 447,835 1,197,546 1–2 CCa W40X215 W16X89 W12X45 W18X86
C 107,941 430,327 1,180,520 OCla W40X297 W21X147 W33X201 W40X192
D 110,377 451,132 1,189,864 OCsa W30X235 W40X244 W36X720 W44X248
E 115,222 462,026 1,243,245 ICa W40X436 W40X531 W36X848 W36X393
III F 129,207 678,491 3,060,820 OBa W30X90 W8X21 W8X28 W10X22
G 121,059 630,559 2,844,586 IBa W30X90 W14X30 W14X43 W12X35
H 132,174 681,461 3,074,219 BRa None W10X22 W8X31 W8X40
I 135,736 702,303 3,171,121 3–4 CCa W40X199 W24X94 W8X40 W10X77
IV J 109,242 487,153 1,447,472 OCla W40X298 W14X176 W36X256 W14X211
K 103,598 452,588 1,407,455 OCsa W40X244 W40X221 W40X593 W44X248
L 107,208 486,772 1,447,045 ICa W30X292 W40X436 W36X848 W40X298
M 109,997 502,992 1,495,717 OBa W27X84 W8X21 W8X28 W10X22
IBa W30X99 W21X44 W14X43 W16X36
BRa None W10X22 W8X31 W8X31
where pw is the design wind pressure in kN/m2, Kz is the velocity expo- 5–6 CCa W27X194 W14X109 W10X39 W14X90
sure coefficient, Kzt is the topographic factor, Kd is the wind direction OCla W40X268 W14X132 W33X152 W21X166
factor, V is the basic wind speed, G is the gust factor, and Cp is the exter- OCsa W12X190 W30X132 W40X436 W36X194
ICa W36X260 W44X285 W36X848 W40X277
nal pressure coefficient. Assuming that the frames are located in a flat
OBa W30X90 W12X26 W8X28 W12X26
terrain with a basic wind speed of V = 56m/s (125 mph) and exposure IBa W30X90 W14X34 W14X43 W40X149
category B, the following values are used for these parameters: Kzt = BRa None W8X35 W8X48 W6X20
1.0, Kd = 0.85, I = 1.0, G = 0.85, Cp = 0.8 on windward face, Cp = − 7–8 CCa W24X162 W14X109 W8X31 W27X84
0.44 and −0.5 on leeward face for winds along x and y-directions re- OCla W40X268 W33X141 W30X211 W12X136
OCsa W24X192 W12X106 W40X480 W36X194
spectively. The wind loads calculated for the test frames of 10, 20 and ICa W36X245 W27X307 W36X848 W44X248
30-storey buildings are documented in Tables 3, 4 and 5 respectively, OBa W24X62 W14X30 W8X28 W18X40
IBa W27X84 W18X50 W14X43 W16X36
BRa None W6X15 W10X39 W6X20
Table 7 9–10 CCa W14X120 W14X176 W10X33 W21X73
The optimized designs of the 10-storey building for selected forms A, B, F and J. OCla W40X215 W14X145 W30X191 W30X108
Stories Member group Ready sections OCsa W40X221 W36X135 W33X387 W27X146
ICa W40X264 W24X131 W40X480 W40X167
A10 B10 F10 J10 OBa W21X62 W40X149 W8X28 W12X26
IBa W33X118 W16X40 W14X43 W24X55
(Group 1) (Group 2) (Group 3) (Group 4)
BRa None W6X20 W10X39 W12X30
1–2 CCa W24X68 W16X36 W8X28 W8X24 11–12 CCa W40X192 W12X53 W8X28 W18X55
OCla W33X141 W12X65 W8X28 W10X22 OCla W40X192 W14X145 W33X130 W18X86
OCsa W40X221 W24X94 W8X28 W10X22 OCsa W40X192 W33X130 W40X244 W21X93
ICa W40X192 W12X120 W8X24 W10X22 ICa W40X215 W36X170 W33X619 W30X99
OBa W14X30 W8X18 W8X24 W10X22 OBa W14X34 W10X22 W8X28 W24X55
IBa W18X46 W18X35 W12X58 W12X53 IBa W27X84 W24X62 W14X43 W24X55
BRa None W5X19 W10X45 W12X45 BRa None W6X15 W8X31 W8X31
3–4 CCa W12X65 W10X33 W10X33 W8X31 13–14 CCa W14X99 W14X90 W8X28 W12X58
OCla W14X109 W10X54 W12X26 W8X24 OCla W40X199 W14X120 W30X99 W14X74
OCsa W14X109 W12X72 W8X24 W10X22 OCsa W40X277 W18X130 W24X207 W30X90
ICa W21X166 W30X90 W12X72 W10X60 ICa W14X145 W30X173 W40X298 W30X90
OBa W24X55 W8X18 W12X53 W10X45 OBa W30X90 W18X35 W8X28 W24X68
IBa W18X46 W12X26 W10X39 W8X31 IBa W24X68 W24X55 W14X43 W24X62
BRa None W6X15 W12X26 W8X24 BRa None W6X9 W8X31 W5X16
5–6 CCa W18X50 W16X31 W8X24 W10X22 15–16 CCa W12X53 W8X40 W8X28 W8X31
OCla W14X90 W10X49 W14X145 W12X152 OCla W14X90 W14X159 W10X88 W16X57
OCsa W18X97 W10X54 W12X96 W14X99 OCsa W24X117 W21X57 W36X150 W30X99
ICa W14X109 W12X58 W12X65 W14X68 ICa W27X161 W24X84 W27X307 W30X173
OBa W18X35 W8X21 W21X73 W24X55 OBa W18X35 W12X30 W8X28 W21X50
IBa W16X36 W12X26 W8X31 W8X28 IBa W21X50 W21X50 W14X43 W18X40
BRa None W6X15 W8X28 W10X22 BRa None W6X15 W5X19 W6X9
7–8 CCa W12X40 W16X26 W8X28 W10X22 17–18 CCa W12X58 W10X60 W8X28 W8X28
OCla W18X76 W10X49 W8X28 W10X22 OCla W12X72 W10X45 W10X33 W21X50
OCsa W18X76 W10X49 W10X30 W10X22 OCsa W24X104 W18X76 W21X101 W24X55
ICa W12X79 W14X38 W10X30 W10X22 ICa W24X104 W24X62 W33X118 W36X160
OBa W14X30 W8X21 W14X43 W14X34 OBa W14X34 W10X22 W12X30 W21X44
IBa W14X30 W12X26 W14X43 W14X34 IBa W18X35 W14X30 W14X43 W36X135
BRa None W6X9 W14X43 W14X34 BRa None W6X9 W10X22 W6X9
9–10 CCa W12X26 W12X26 W14X43 W12X30 19–20 CCa W10X33 W12X30 W12X30 W8X28
OCla W8X35 W8X31 W16X45 W12X30 OCla W10X49 W8X35 W12X30 W8X28
OCsa W10X49 W8X40 W6X15 W6X15 OCsa W18X76 W16X26 W12X30 W21X50
ICa W18X55 W8X28 W6X15 W6X15 ICa W14X61 W8X31 W14X48 W24X62
OBa W12X26 W8X18 W6X15 W6X15 OBa W8X21 W8X21 W12X30 W21X50
IBa W12X26 W12X30 W6X9 W6X9 IBa W12X26 W16X36 W14X43 W16X40
BRa None W6X9 W6X9 W6X9 BRa None W6X9 W5X16 W6X9
Weight (kg) 149,103 111,408 129,207 109,242 Weight (kg) 573,637 447,835 678,491 487,153
a a
CC: Corner Columns, IC: Inner Columns, OCl: Outer Columns on long facades, OCs: CC: Corner Columns, IC: Inner Columns, OCl: Outer Columns on long facades, OCs:
Outer Columns on short facades IB: Inner Beams, OB: Outer Beams, BR: Bracing Members. Outer Columns on short facades IB: Inner Beams, OB: Outer Beams, BR: Bracing Members.
388 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396

Table 9
The optimized designs of the 30-storey building for selected forms A, B, F and J.

Stories Member group Ready sections

A30 B30 F30 J30

(Group 1) (Group 2) (Group 3) (Group 4)

a
1–2 CC W36X848 W33X424 W12X65 W40X531
OCla W14X550 W40X328 WLARGE2 W40X655
OCsa W40X324 W36X359 WLARGE2 W40X328
ICa W40X397 W40X655 WLARGE2 W40X593
OBa W36X135 W10X22 W8X28 W12X26
IBa W30X90 W14X30 W14X43 W24X55
BRa None W12X45 W12X65 W12X26
3–4 CCa W36X588 W30X477 W10X60 W14X370
OCla W27X539 W36X328 WLARGE1 W36X848
OCsa W40X277 W36X260 WLARGE2 W44X198
ICa W33X567 W33X387 WLARGE2 W40X362
OBa W36X135 W12X26 W8X28 W12X26
IBa W30X108 W16X36 W14X43 W14X34
BRa None W8X35 W8X18 W14X74
5–6 CCa W14X550 W36X393 W10X54 W36X650
OCla W36X527 W27X368 WLARGE1 W36X848
OCsa W40X277 W27X235 WLARGE2 W44X198
ICa W33X387 W33X318 WLARGE2 W40X397
OBa W40X149 W30X90 W8X28 W44X198
IBa W24X94 W14X43 W14X43 W44X224
BRa None W8X31 W14X82 W6X20
7–8 CCa W33X567 W33X468 W12X58 W14X311
OCla W27X494 W30X477 W36X848 W14X370
OCsa W40X436 W40X235 WLARGE2 W36X182
ICa W40X324 W33X354 WLARGE2 W44X285
OBa W30X90 W27X94 W8X28 W44X198
IBa W30X108 W30X116 W14X43 W24X62
BRa None W6X20 W8X35 W10X49
9–10 CCa W30X581 W40X298 W10X49 W27X407
OCla W40X531 W14X455 W40X655 W40X362
OCsa W36X393 W40X277 WLARGE1 W40X436
ICa W40X249 W44X285 WLARGE2 W40X397
OBa W33X118 W21X50 W8X28 W21X50
IBa W40X149 W44X224 W16X45 W40X167
BRa None W10X45 W14X61 W6X15
11–12 CCa W30X433 W14X455 W10X45 W36X300
OCla W36X485 W12X336 W36X328 W36X588
OCsa W40X221 W36X194 WLARGE2 W40X167
ICa W36X260 W40X264 WLARGE2 W44X248
OBa W33X152 W30X90 W8X28 W44X248
IBa W30X99 W24X55 W14X43 W44X198
BRa None W8X24 W10X54 W6X20
13–14 CCa W36X328 W14X426 W8X40 W36X245
OCla W33X424 W21X275 W40X244 W21X364
OCsa W30X211 W33X141 WLARGE1 W44X248
ICa W40X249 W36X182 WLARGE2 W44X248
OBa W40X167 W24X55 W8X28 W24X55
IBa W36X160 W21X50 W14X43 W44X224
BRa None W8X48 W12X45 W6X20
15–16 CCa W14X370 W14X398 W10X39 W30X211
OCla W14X500 W33X263 W40X328 W40X264
OCsa W40X397 W36X160 W30X581 W40X192
ICa W40X362 W40X149 W36X848 W40X362
OBa W24X68 W33X118 W8X28 W44X248
IBa W40X149 W18X65 W14X43 W44X224
BRa None W6X20 W10X45 W6X15
17–18 CCa W33X291 W40X328 W8X31 W18X175
OCla W40X268 W14X500 W24X408 W18X175
OCsa W40X244 W14X257 W36X798 W27X146
ICa W40X249 W44X248 WLARGE1 W40X397
OBa W33X118 W40X149 W12X30 W10X22
IBa W40X149 W40X167 W14X43 W21X44
BRa None W6X9 W10X60 W10X60
19–20 CCa W40X235 W40X221 W8X31 W36X182
OCla W40X244 W14X370 W40X268 W14X159
OCsa W33X130 W21X182 W40X655 W30X108
ICa W40X328 W40X298 W40X480 W27X146
OBa W30X90 W24X55 W8X28 W24X55
IBa W44X198 W40X149 W14X43 W16X36
BRa None W8X13 W14X43 W16X67
21–22 CCa W40X328 W36X150 W8X28 W36X170
OCla W14X145 W40X244 W33X354 W27X178
 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396 389

Table 9 (continued)

Stories Member group Ready sections

A30 B30 F30 J30

(Group 1) (Group 2) (Group 3) (Group 4)

a
OCs W21X111 W40X192 W44X248 W40X167
ICa W36X588 W40X192 W33X468 W40X249
OBa W27X84 W40X149 W8X28 W44X198
IBa W21X50 W24X62 W14X43 W33X118
BRa None W6X25 W12X58 W4X13
23–24 CCa W21X147 W14X90 W8X28 W24X104
OCla W14X176 W12X106 W36X230 W12X106
OCsa W40X149 W40X167 W40X297 W24X104
ICa W40X192 W40X199 W40X298 W21X73
OBa W24X62 W21X44 W8X28 W44X198
IBa W30X116 W18X35 W18X46 W16X36
BRa None W4X13 W10X45 W8X24
25–26 CCa W21X122 W12X53 W8X28 W8X28
OCla W40X192 W14X90 W12X230 W14X82
OCsa W24X104 W27X84 W30X211 W44X224
ICa W44X198 W18X50 W40X277 W30X90
OBa W44X248 W21X44 W12X30 W21X44
IBa W21X44 W18X35 W14X43 W30X90
BRa None W5X19 W12X40 W4X13
27–28 CCa W10X60 W14X109 W8X24 W12X26
OCla W14X90 W12X72 W30X99 W14X43
OCsa W14X82 W16X77 W33X141 W33X118
ICa W27X114 W16X40 W27X194 W36X150
OBa W21X68 W18X40 W10X30 W12X26
IBa W21X44 W14X34 W14X43 W36X150
BRa None W6X9 W10X39 W8X13
29–30 CCa W12X87 W12X53 W14X30 W8X28
OCla W8X35 W14X145 W14X38 W8X24
OCsa W12X40 W14X38 W8X28 W8X24
ICa W24X84 W24X62 W14X48 W30X90
OBa W16X40 W18X40 W8X28 W8X24
IBa W16X40 W24X55 W16X45 W16X36
BRa None W6X9 W8X24 W6X9
Weight (kg) 1,442,439 1,197,546 3,060,820 1,447,472
a
CC: Corner Columns, IC: Inner Columns, OCl: Outer Columns on long facades, OCs: Outer Columns on short facades IB: Inner Beams, OB: Outer Beams, BR: Bracing Members.

and they are applied as uniformly distributed lateral loads on the exter-
nal beams at every floor level on windward and leeward faces of the
frames.
The gravitational and wind loads are combined under two load com-
binations in order to determine the design loads. In the first loading con-
dition, the gravity loads are applied with wind loads acting along x-axis
(i.e. 1.0GL + 1.0WL-x), whereas in the second one they are combined
with wind forces acting along y-axis (i.e. 1.0GL + 1.0WL-y).

6.3. Sizing optimization of steel frames

The sizing optimizations of the thirty-nine test frames are performed

using the ES integrated parallel optimization algorithm mentioned in
Section 3, where each frame is sized for minimum weight design of its
members by selecting sections from the standard W-section list in
AISC. The discrete ES technique employed has a stochastic nature such
that it locates different solutions in independent executions of the algo-
rithm. Hence, each frame is independently designed five times by run-
ning a (10, 50)-ES algorithm over 1000 generations, resulting in
50,000 function evaluations (structural analyses) at each run. The best
(minimum weight) solution obtained out of the five runs is considered
as the optimized sizing design of a test frame. The further implementa-
tion details of the design optimization process and the whole set of
parameter values chosen can be found in Hasançebi et al. [12].
The optimized member weights of the thirty-nine test frames are
presented overall in Table 6. The optimized designs of 10, 20 and 30-
storey buildings for some selected forms (namely A, B, F and J) are tab-
ulated in Tables 7, 8 and 9, where section designations attained in the Fig. 4. A typical rigid beam-column connection model.
390 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396

optimized designs of these frames are documented for all the member
groups.

6.4. Connection designs

The connection designs are performed next, where the beam-col-

umn connection models shown in Figs. 4 and 5 are used as a basis for es-
timating average connection weights and costs in the optimized designs
of the test frames. Fig. 4 illustrates a bolted end-plate type connection,
which is regarded as a typical moment-resisting joint in steel frames.
In this connection model the stiffeners are welded along three sides
using fillet welds and the beam is attached to the end-plate with a com-
plete joint penetration groove weld. On the other hand, shown in Fig. 5
is a double angle beam-column connection, which is one of the most
commonly used shear connections in steel frames idealizing a typical
pin behaviour.
The optimized designs of the test frames corresponding to the forms
A, B, F and J are used only for the connection designs in 10, 20 and 30-
storey buildings. Accordingly, one form is selected to represent each
frame group (namely, A for group 1; B for group 2; F for group 3; J for
group 4), and it is reasonably assumed that connection weight and
cost analyses performed on a representative form will also give good es-
timates for other forms belonging to the same group. A total of nine
sample joints are selected from predetermined locations in each repre-
sentative frame for detailed design calculations. These locations corre-
spond to the corner, side and inner joints depicted in Fig. 6 on the
Fig. 5. A typical pin-jointed beam-column connection.
first, middle and top storey levels of a steel frame. The beam-column
connections at these joints are designed using a spreadsheet program
(specifically developed for this study), where a number of design

Fig. 6. Selection of joints for detailed cost analyses.

 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396 391

Table 10
The connection weights and costs for the frame groups in the design of 10-storey building.

Frame group Average connection weight (kg) Average connection cost (Euro) The total weight of The total cost of
connections (kg) connections (Euro)
Inner joint Side joint Corner joint Inner joint Side joint Corner joint

I 54.81 92.66 127.17 71.79 118.91 161.14 14,692 18,830

II 52.00 84.46 120.57 68.11 108.38 152.79 13,660 17,506
III 5.47 7.95 10.17 11.71 15.95 20.22 1261 2553
IV 5.35 43.54 124.46 11.25 58.84 156.51 8676 11,418

Table 11
The connection weights and costs for the frame groups in the design of 20-storey building.

Frame group Average connection weight (kg) Average connection cost (Euro) The total weight of The total cost of
connections (kg) connections (Euro)
Inner joint Side joint Corner joint Inner joint Side joint Corner joint

I 121.62 182.95 267.46 175.19 253.71 363.38 60,398 83,679

II 109.93 177.28 245.17 159.88 248.42 326.50 56,773 78,657
III 12.21 17.73 22.10 23.82 34.95 47.12 5581 11,268
IV 11.02 99.67 248.63 21.05 144.37 331.46 36,719 51,299

parameters, such as thickness of stiffeners, size of bolts, weld thickness,
etc. are determined optimally to provide the minimum strength for the
connections under the computed joint forces. The connection weights
and costs calculated for a representative frame are averaged separately
for corner, side and inner joints to obtain average connection weights
and costs of its respective frame group in a particular building. The aver-
age connection weights and costs calculated for all the frame groups in
the designs of 10, 20 and 30-storey buildings are displayed in Tables 10,
11 and 12 respectively. The total weight and cost of connections corre-
sponding to each frame group in these buildings are also provided in the
respective tables.
6.5. Discussion on design weights of steel frames
The final design weights of the test frames for the 10, 20 and 30-sto-
rey buildings are obtained by adding their corresponding member and
connection weights, as performed in Table 13. The frame weights ap-
pear to be in the range of [112,274, 163,795] (kg) for the 10-storey
building, [487,100, 707,885] (kg) for the 20-storey building, and
[1,331,982, 3,191,009] (kg) for the 30-storey building. The group aver-
age weights, representing the arithmetic mean of all the frame weights
belonging to a particular group, are also shown in Table 13 to use as in-
dices for comparing the design weight efficiencies of different groups. In
Figs. 7, 8 and 9, the normalized design weights are displayed for the test
frames of 10, 20 and 30-storey buildings respectively, where the design
weight of each frame is proportioned to that of the rigidly connected
frame (form A) in order to visualize the design weight efficiencies of dif-
ferent forms with respect to one another.
Accordingly, in the design of 10-storey building the group 4 frames
lead to minimum design weights on average, followed by group 2,
group 3 and group 1 frames in the order of design weight efficiency.
On the other hand, the group 2 frames turn out to be the most
weight-efficient systems in the designs of 20 and 30-storey buildings,
followed by group 4, group 1 and group 3 frames. Overall, these results
reveal that groups 2 and 4 frames, which take advantage of flexural ac-
tion through full or sectional rigid beam-column connections in their

Table 12
The connection weights and costs for the frame groups in the design of 30-storey building.

Frame group Average connection weight (kg) Average connection cost (Euro) The total weight of The total cost of
connections (kg) connections (Euro)
Inner joint Side joint Corner joint Inner joint Side joint Corner joint

I 213.33 333.94 441.06 328.64 484.02 641.28 158,673 232,557

II 207.79 315.61 423.17 323.41 462.02 602.44 151,461 221,987
III 28.59 41.70 53.75 71.79 102.69 134.87 19,889 49,445
IV 22.52 182.02 446.35 44.26 270.38 645.53 99,950 147,666

Table 13
The final design weights of the test frames for the 10, 20 and 30-storey buildings.

Frame group Form Optimum design weights (kg) Group average weights (kg)

10–Storey 20-Storey 30-Storey 10–Storey 20-Storey 30-Storey

I A 163,795 634,035 1,601,112 163,795 634,035 1,601,112

II B 125,068 504,608 1,349,008 124,897 504,603 1,354,255
C 121,601 487,100 1,331,982
D 124,037 507,905 1,341,325
E 128,882 518,799 1,394,706
III F 130,468 684,072 3,080,708 130,805 678,785 3,057,575
G 122,320 636,140 2,864,475
H 133,435 687,043 3,094,108
I 136,998 707,885 3,191,009
IV J 117,918 523,872 1,547,422 116,187 519,096 1,549,373
K 112,274 489,307 1,507,406
L 115,884 523,492 1,546,995
M 118,673 539,712 1,595,667
392 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396

Fig. 7. Normalized design weights of the structural forms for the 10-storey building.

structural forms, and which are also stiffened by an integrated bracing
system produce lesser design weights, as compared to groups 1 and 3
frames where lateral loads are resisted by only one frame action, i.e. ei-
ther flexural or cantilever truss action. Another observation is that as the
storey or building height increases, an integrated bracing system alone
becomes less effective in comparison to the rigid frame behaviour.
This observation is justified by the fact that as group 3 frames produce
lighter design weights than does group 1 frame for the 10-storey build-
ing, the average design weight of the formers exceeds that of the latter
for the 20-storey building, and even for the 30-storey building the de-
signs produced by group 3 frames become almost twice as heavy as
those of the group 1 frame. In fact, the relatively heavy designs pro-
duced by the group 3 frames for the 30-storey building can be attributed
to inadequate lateral stiffness of the resulting structures, in which case
the member sections are increased tremendously to satisfy the displace-
ment and drift constraints.
Regarding the individual performances of the structural forms, it is
seen that the Z- bracing consistently outperforms X, V and eccentric V
configurations regardless of frame group type in all the three buildings.
The relative performances of X and V-bracings appear to be alternating
such that sometimes V-bracing produces lesser design weights than
does X-bracing, and vice versa. However in all the test cases investigated
here the designs produced by these two bracings configurations differ
very little from each other in terms of the final design weights attained.
Accordingly, the design weight efficiencies of these two bracings are
found very comparable with each other. Finally, as anticipated the ec-
centric V-bracing gives rise to steel frameworks that are heavier than
those of X and V-bracings in the order of 2–4%.
6.6. Cost analyses of steel frames
The overall construction costs of the test frames are computed using
the cost formulas presented in Section 5 to evaluate the cost efficiencies
of the frames. The detailed cost calculations of the test frames are pre-
sented in Tables 14, 15 and 16 for the 10, 20 and 30-storey buildings re-
spectively. The group average costs, representing the arithmetic mean
of all the frame costs belonging to a particular group, are also presented
in these tables for comparing the cost efficiencies of different groups. In
addition, in Figs. 10, 11 and 12, the normalized construction costs are
displayed for the frames of 10, 20 and 30-storey buildings respectively,
where the cost of each steel frame is proportioned to that of the rigidly
connected frame (form A) in order to visualize the cost efficiencies of
different forms with respect to one another.

Fig. 8. Normalized design weights of the structural forms for the 20-storey building.
 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396 393

Fig. 9. Normalized design weights of the structural forms for the 30-storey building.

For the design of 10-storey building, the costs of the steel frames ap-
pear to be in the range of [94,879, 139,776] (Euro), indicating that the
cost difference between different forms can be in excess of 30% in refer-
ence to the most costly one. The group 4 frames come out to be the most
cost-efficient systems, whereas the group 1 frame leads to the most ex-
pensive one. There is a certain level of correlation between the weights
and costs in this sense, since these two groups also lead to the lowest
and highest design weights for the building at hand. However, the re-
sults also indicate that group 3 frames lead to more economical designs
than do group 2 frames, even though they have higher design weights
than the latters. Especially, the frame that incorporates form G ranks
the second amongst thirteen structural forms tested, outperforming
even the three forms (J, L and M) of the group 4 frames. Overall, it is de-
duced that groups 3 and 4 frames are economically the most feasible
systems for the design of 10-storey building.
For the design of 20-storey building, the costs of the frames appear
to be in the range of [415,472, 550,440] (Euro), indicating that the
cost difference between different forms can be in excess of 25% in

Table 14
Cost analyses for the test frames of the 10-storey building.

Frame Form Element mass Joint Element cost Joint cost Transport cost Erection cost Extra costs Total cost Group average cost
group (kg) mass (kg) (Euro) (Euro) (Euro) (Euro) (Euro) (Euro) (Euro)

I A10 149,103 14,692 59,641 18,830 3603 27,563 30,138 139,776 139,776
II B10 111,408 13,660 44,563 17,506 2752 21,047 23,013 108,880 108,748
C10 107,941 13,660 43,176 17,506 2675 20,463 22,375 106,195
D10 110,377 13,660 44,151 17,506 2729 20,873 22,823 108,082
E10 115,222 13,660 46,089 17,506 2835 21,688 23,714 111,833
III F10 129,207 1261 51,683 2553 2870 21,955 24,006 103,068 103,329
G10 121,059 1261 48,424 2553 2691 20,584 22,507 96,759
H10 132,174 1261 52,870 2553 2936 22,455 24,552 105,365
I10 135,736 1261 54,294 2553 3014 23,054 25,208 108,123
IV J10 109,242 8676 43,697 11,418 2594 19,843 21,697 99,249 97,909
K10 103,598 8676 41,439 11,418 2470 18,893 20,658 94,879
L10 107,208 8676 42,883 11,418 2549 19,501 21,323 97,674
M10 109,997 8676 43,999 11,418 2611 19,970 21,836 99,833

Table 15
Cost analyses for the test frames of the 20-storey building.

Frame Form Element mass Joint mass Element cost Joint cost Transport cost Erection cost Extra costs Total cost Group average cost
group (kg) (kg) (Euro) (Euro) (Euro) (Euro) (Euro) (Euro) (Euro)

I A20 573,637 60,398 229,455 83,679 13,949 106,695 116,662 550,440 550,440
II B20 447,835 56,773 179,134 78,657 11,101 84,915 92,848 446,655 446,652
C20 430,327 56,773 172,131 78,657 10,716 81,969 89,626 433,099
D20 451,132 56,773 180,453 78,657 11,174 85,470 93,455 449,208
E20 462,026 56,773 184,810 78,657 11,414 87,303 95,459 457,643
III F20 678,491 5581 271,396 11,268 15,050 115,116 125,869 538,699 534,605
G20 630,559 5581 252,223 11,268 13,995 107,050 117,050 501,586
H20 681,461 5581 272,585 11,268 15,115 115,616 126,416 540,999
I20 702,303 5581 280,921 11,268 15,573 119,123 130,251 557,136
IV J20 487,153 36,719 194,861 51,299 11,525 88,157 96,392 442,235 438,537
K20 452,588 36,719 181,035 51,299 10,765 82,341 90,033 415,472
L20 486,772 36,719 194,709 51,299 11,517 88,093 96,322 441,941
M20 502,992 36,719 201,197 51,299 11,874 90,823 99,307 454,499
394 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396

Table 16
Cost analyses for the test frames of the 30-storey building.

Frame Form Element mass Joint mass Element cost Joint cost Transport cost Erection cost Extra costs Total cost Group average cost
group (kg) (kg) (Euro) (Euro) (Euro) (Euro) (Euro) (Euro) (Euro)

I A30 1,442,439 158,673 576,976 232,557 35,224 269,435 294,605 1,408,797 1,408,797
II B30 1,197,546 151,461 479,018 221,987 29,678 227,011 248,217 1,205,912 1,209,976
C30 1,180,520 151,461 472,208 221,987 29,304 224,146 245,085 1,192,730
D30 1,189,864 151,461 475,946 221,987 29,509 225,718 246,804 1,199,964
E30 1,243,245 151,461 497,298 221,987 30,684 234,701 256,626 1,241,296
III F30 3,060,820 19,889 1,224,328 39,105 67,776 518,422 566,850 2,416,481 2,398,569
G30 2,844,586 19,889 1,137,835 39,105 63,018 482,034 527,063 2,249,056
H30 3,074,219 19,889 1,229,688 39,105 68,070 520,676 569,316 2,426,856
I30 3,171,121 19,889 1,268,448 39,105 70,202 536,983 587,146 2,501,885
IV J30 1,447,472 99,950 578,989 147,666 34,043 260,400 284,726 1,305,824 1,307,334
K30 1,407,455 99,950 562,982 147,666 33,163 253,666 277,363 1,274,840
L30 1,447,045 99,950 578,818 147,666 34,034 260,328 284,647 1,305,494
M30 1,495,717 99,950 598,287 147,666 35,105 268,519 293,603 1,343,179

reference to the most costly one. Similar to the 10-storey building, the Z-bracing in groups 2 and 4 frames, have shown to be the most econom-
group 4 frames come out to be the most cost-efficient systems again ical forms out of the thirteen investigated here. The fact that the costs
for the 20-storey building, whereas the group 1 frame leads to the and weights are not always correlated in steel frames is also verified
most expensive one. The structural forms C and K, both consisting of here; in that the group 4 frames slightly outperform the group 2 frames

Fig. 10. Normalized costs of the structural forms for the 10-storey building.

Fig. 11. Normalized costs of the structural forms for the 20-storey building.
 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396
Fig. 12. Normalized costs of the structural forms for the 30-storey building.
even though the formers have higher design weights than the latters.
Likewise, even though the group 3 frames are approximately 7% heavier
on average than the group 1 frame, they are 3% cheaper than the latter.
To sum up, the groups 2 and 4 frames are economically the most feasible
systems for the design of 20-storey building, and the groups 1 and 3
frames cost relatively higher as compared to these groups.
For the design of 30-storey building, the costs of the frames appear
to be in the range of [1,192,730, 2,501,885] (Euro), indicating that the
cost difference between different forms can be in excess of 50% in refer-
ence to the most costly one. This time the group 2 frames lead to the
lowest construction costs on the whole, outperforming all steel frames
in other groups. The group 3 frames have shown to be the least cost-ef-
ficient systems for this building, indicating that an integrated bracing
system alone becomes less effective as compared to the rigid frame be-
haviour as the storey or building height increases. Overall, it is deduced
that the group 2 frames are economically feasible systems for the design
of 30-storey building, and the use of group 3 frames should be avoided
due to inadequate stiffness of the resulting structures to lateral forces.
7. Concluding remarks
This study has addressed cost efficiencies of various steel frames for
economical design of multi-storey buildings. A total of thirteen steel
frameworks that incorporate various beam-column connections (rigid
or pin) and bracing configurations (X, Z, V and eccentric V) are consid-
ered for detailed and comparative cost analyses. The steel frames are
classified into four groups as (i) rigidly connected frame (form A),
(ii) rigidly connected frames with an integrated bracing system
(forms B, C, D, E), (iii) pin-jointed frames with an integrated bracing
system (forms F, G, H, I), and (iv) frames having rigid connections at
interior joints and pin connections at exterior ones with an integrated
bracing system (forms J, K, L, M).
The numerical examples are performed using three multi-storey
buildings with 10, 20 and 30 floors in an effort to observe variations in
the cost efficiencies of the steel frames when the height of a building
changes. Accordingly, incorporation of thirteen structural forms into
the three buildings has resulted in thirty-nine test frames for numerical
applications. The design optimization is first implemented using an ES
integrated parallel optimization algorithm to minimize the total mem-
ber weight in each test frame. An extensive cost analysis is then carried
out on the optimized design of each frame to calculate its estimated con-
struction cost using a cost model that itemizes costs of all production
stages including material, manufacturing, erection and transportation.
The final design weights of the test frames, incorporating member and
395
connection weights together, are reproduced in Table 13, and also are
presented as normalized design weights in Figs. 7, 8 and 9. Similarly, de-
tailed cost calculations of the test frames are carried out in Tables 14, 15
and 16, and are also shown as normalized construction costs in Figs. 10,
11 and 12.
In summary, groups 2 and 4 frames lead to the lightest structural
systems regardless of storey number of building height since they take
advantage of both a flexural action through full or sectional rigid
beam-column connections in their structural forms, and a cantilever
truss action through an integrated bracing system. On the other hand,
the groups 3 and 4 frames offer economically more advantageous
systems for the design of 10-storey building. Relatively high costs of
connections in group 2 frames make them economically less attractive
for the design of short buildings. However, when the building height in-
creases, the advantage of low connection costs of pin-joints is eventual-
ly offset by low lateral stiffness in group 3 frames, resulting in heavy
design weights and high construction costs for this group. As a result,
groups 2 and 4 frames appear to be most cost-efficient systems for the
design of 20-storey building. As the number of storey increases further,
the significance of rigid frame behaviour becomes much more pro-
nounced, making group 2 frames the only choice for an economical
design of 30-storey building.
References
[1] A.M. Memari, M. Madhkan, Optimal design of steel frames subject to gravity and
seismic codes' prescribed lateral forces, Struct. Optim. 18 (1999) 56–66.
[2] AISC-ASD, Manual of Steel Construction-allowable Stress Design, 8th ed., 1983 (Chi-
cago, IL, USA).
[3] E.S. Kameshki, M.P. Saka, Genetic algorithm based optimum bracing design of non-
swaying tall plane frames, J. Constr. Steel Res. 57 (2001) 1081–1097.
[4] BS5950, Steelwork Design British Standards, British Standards Institution, London,
UK, 2000.
[5] Q.Q. Liang, Y.M. Xie, G.P. Steven, Optimal topology design of bracing systems for
multi-storey steel frames, J. Struct. Eng. ASCE 126 (7) (2000) 823–829.
[6] A. Kaveh, N. Farhoodi, Layout optimization for X-bracing of planar steel frames, Int. J.
Civ. Eng. 8 (3) (2010) 256–275.
[7] International Building Code, International Building Council, Washington, DC, 2006.
[8] N.B. Hadj Ali, M. Sellami, A.-F. Cutting-Decelle, J.-C. Mangin, Multi-stage production
cost optimization of semi-rigid steel frames using genetic algorithms, Eng. Struct. 31
(11) (2009) 2766–2778.
[9] L. Pavlovcic, A. Krajnc, D. Beg, Cost function analysis in the structural optimization of
steel frames, Struct. Multidiscip. Optim. 28 (2004) 286–295.
[10] ASCE 7-05, Minimum Design Loads for Building and Other Structures, American So-
ciety of Civil Engineering, 2005.
[11] AISC-ASD, Manual of Steel Construction-allowable Stress Design, 9th ed., 1989 (Chi-
cago, IL, USA).
[12] O. Hasançebi, T. Bahçecioğlu, Ö. Kurç, M.P. Saka, Optimum design of high-rise steel
buildings using an evolution strategy integrated parallel algorithm, Comput. Struct.
89 (21–22) (2011) 2037–2051.
396 O. Hasançebi / Journal of Constructional Steel Research 128 (2017) 380–396
[13] ANSI/AISC 360-05, Specification for Structural Steel Buildings, AISC, Chicago, IL, USA,
2005.
[14] P. Dumonteil, Simple equations for effective length factors, Eng. J. AISC 29 (3) (1992)
111–115.
[15] J. Hellesland, Review and Evaluation of Effective Length Formulas, University of Oslo,
1994 (Report No. 94-2).
[16] M.P. Saka, Optimum design of steel frames using stochastic search techniques based
on natural phenomena: a review, in: B.H.V. Topping (Ed.), Civil Engineering Com-
putations: Tools and Techniques, Saxe-Coburg Publications, Stirlingshire, UK 2007,
pp. 105–147.
[17] L. Lamberti, C. Pappalettere, Metaheuristic design optimization of skeletal struc-
tures: a review, in: T. BHV (Ed.), Computational Technology Reviews, vol. 4, 2011,
pp. 1–32.
[18] M.P. Saka, E. Doğan, Recent developments in metaheuristic algorithms: a
review, in: T. BHV (Ed.), Computational Technology Reviews, vol. 5, 2012,
pp. 31–78.
[19] O. Hasançebi, S. Çarbaş, E. Doğan, F. Erdal, M.P. Saka, Comparison of non-determin-
istic search techniques in the optimum design of real size steel frames, Comput.
Struct. 88 (17–18) (2010) 1033–1048.
[20] B.S. Taranath, Wind and Earthquake Resistant Buildings, CRC Press, 2005.