INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING

Int. J. Numer. Meth. Engng 2002; 55:1451–1462 (DOI: 10.1002/nme.549)

Life-cycle cost optimization of steel structures

Kamal C. Sarma1; 2 and Hojjat Adeli2; ∗; †

1 Qwest Communications International Inc.; Dublin; OH 43016; U.S.A.
2 Department of Civil and Environmental Engineering and Geodetic Science; The Ohio State University;
470 Hitchcock Hall; 2070 Neil Avenue; Columbus; OH 43210; U.S.A.

SUMMARY
This article addresses the life-cycle cost optimization of steel structures. The main factors in
uencing
the life-cycle cost of a structure are delineated and their eects on various cost functions are discussed.
A four-criteria optimization model is presented for the life-cycle cost optimization of steel structures.
These criteria are (i) select discrete commercially available sections with the lowest cost, (ii) select
commercially available sections with the lightest weight, (iii) select the minimum number of dierent
types of commercially available sections, and (iv) select commercially available sections with the min-
imum total perimeter length. The last criterion models a representative type of cost incurred over the
life of the structure, that is, preventative maintenance in the form of periodic painting of an exposed
steel structure to avoid corrosion. The life-cycle cost optimization model is based on fuzzy logic with
the goal of formalizing the life-cycle design process but with some input from the design engineer
through introduction of weighting coecients re
ecting the relative importance of various criteria.
The model is applied to a large steel structure with over 3300 members. Copyright ? 2002 John
Wiley & Sons, Ltd.

KEY WORDS: cost optimization; fuzzy logic; life-cycle design; multi-criteria optimization; optimization;
steel structures; structural optimization

INTRODUCTION

Life-cycle cost is the total cost of a structure during its lifetime. The life-cycle cost of a
structure includes the initial costs, including costs of design and construction, plus costs of
operation (utilities), maintenance (including repair), and eventually dismantling or demolishing
the structure over the lifetime of the structure. If the initial cost of a structure is low, but
the utilities and maintenance costs are high the structure may not be considered the most
economical design. From an economic point of view the ideal goal of cost optimization of
structures should be minimizing the total life-cycle cost.

∗ Correspondence to: Hojjat Adeli, Department of Civil and Environmental Engineering and Geodetic Science;
The Ohio State University; 470 Hitchcock Hall; 2070 Neil Avenue; Columbus; Ohio 43210, U.S.A.
† E-mail: adeli.1@osu.edu

Received 19 March 2001

Revised 19 December 2001
Copyright ? 2002 John Wiley & Sons, Ltd. Accepted 28 January 2002
1452 K. C. SARMA AND H. ADELI

Only a small fraction of the papers published in the area of structural optimization deal
with cost optimization of structures; the rest deal with minimum weight design [1–4]. For
a chronological review of papers on the cost optimization of reinforced concrete and steel
structures published in archival journals, the reader should refer to Sarma and
Adeli [5; 6].
Recently, the authors developed a new methodology for cost optimization of three-dimen-
sional steel frame structures [7; 8] employing multicriteria optimization [1], fuzzy logic [9],
and evolutionary computing approaches [10]. In particular they present a fuzzy discrete multi-
criteria cost optimization model for the design of space steel structures subjected to the ac-
tual constraints of AISC allowable stress design (ASD) [11] and load and resistance fac-
tor design (LRFD) [12] specications by considering three design criteria: (i) minimum
material cost, (ii) minimum weight, and (iii) minimum number of dierent section types.
The computational model begins with a continuous-variable minimum weight solution with
a pre-emptive constraint violation strategy as a preliminary solution [7] for the nal dis-
crete optimum cost solution. In that work, only the initial cost of the structure was
considered.
The three major costs during the life of a structure are the initial design=build cost, the
operating cost, and the maintenance cost. The initial design=build cost includes the costs of
mechanical services, electrical services, nishing, interior and exterior decorations, external
and internal facilities, and landscaping. The share of design and construction costs of the
super- and sub-structures can be less than 50% of the total initial cost of the structure [13].
The operating cost of a structure includes the costs of care-taking and cleaning, energy
(gas and electricity), water and sewerage, insurance, mortgage interests, and security and
management. The maintenance cost includes inspection, painting, repair, and replacement costs.
The maintenance cost varies with the use, utility, and importance of the structure. Considering
a discount rate (time value of money) of 2% above in
ation, a 50-year old building usually
has an operating cost 0.8–1.3 times the initial cost [13].
The design life (or economic life) of most structures is often in the range 30–40 years.
But, the anticipated life (or actual life expectancy) of a structure is much longer; in the
US, it may be 60–85 years or even more [13]. The anticipated life of a structure is an im-
portant factor in the life-cycle design. Factors in
uencing the anticipated life of a structure
include obsolescence, natural or man-made catastrophe, and inadequate and out-of-fashion
facilities.
The life expectancy as well as repair and maintenance also depend on the type of material
used in the structure. Concrete may lose its strength with the passing of time. Frequent
changes in weather, cracks, shrinkage, and corrosion of reinforcements may reduce the life
of a concrete structure. In these structures even if repair is done, the initial strength may not
be achieved. Similarly, in steel structures, the joints are the most vulnerable points. In an
exposed steel structure the joints may accumulate dirt and debris, and if proper maintenance
is not done this may lead to corrosion or rustling leading to failure. In this respect riveted
or bolted joins are inferior to welded joints. Figure 1 presents a pictorial view of the general
eect of repair on the strength of a structure with the passing of time.
In this article, extending the work of Sarma and Adeli [8], a computational model is pre-
sented for life-cycle cost optimization of structures with inclusion of an additional parameter:
the perimeter of member cross-sections, which determines the maintenance cost of exposed
steel structures like sports utilities, bridges, towers, etc.

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
 LIFE-CYCLE COST OPTIMIZATION OF STEEL STRUCTURES 1453

Initial Strength

Major Repairs
Minimum Acceptable Strength

Strength

Life Cycle of the Structure

Time in Years

Figure 1. A pictorial view of the eect of repair on the strength of a structure with the passing of time.

LIFE-CYCLE COST OF A STEEL STRUCTURE AND

THE PRIMARY CONTRIBUTING FACTORS

The life-cycle cost of a steel structure can be considered as the sum of seven dierent cost
components or functions:

(1) initial cost which may include the following costs as dened in Reference [8]
(a) cost of planning and design
(b) material cost of structural members such as beams, columns, and bracings
(c) fabrication cost including the material costs of connection elements
(d) cost of transporting rolled sections to the fabrication shop and transporting the
fabricated pieces to the construction eld
(e) receiving, handling, and storage costs of rolled sections and fabricated pieces
(f) erection cost
(g) cost of operation of tools and machinery on the construction site
(h) cost of preparing the project site including the cost of the foundation,
(2) maintenance cost such as painting of exposed members of a steel structure
(3) inspection cost to prevent a potentially major damage to the structure
(4) repair cost
(5) operating cost required for proper functional use of the structure such as heating and
electricity
(6) probable failure cost, based on an acceptable probability of failure
(7) dismantling or demolishing costs.

Figure 2 presents dierent cost functions and their relationships in contributing to the
life-cycle cost. We identify eleven main factors that in
uence the life-cycle cost of a structure

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
1454 K. C. SARMA AND H. ADELI

Life-cycle
Cost

1 2 3 4
Initial Cost Painting and Repair Cost
Inspection
Maintenance
Cost
Cost

k
5 6 Importance of
7 the Structure
Operating Failure Cost Dismantling
Costs Cost j
Use of the
Structure
c f i
Weight of Geographic
Rolled Sections Discount
Location rate
b e h
Number of Number of Life Span of
Section Types Connections the Structure
a d g
Cost of Rolled Perimeter of Maintenance
Sections sections Policy

Cost Component Main Factors Influencing the Life-cycle Cost

Contribution to the Life-cycle Cost Influence on the Cost Component

Figure 2. Dierent cost functions and their relationships in contributing to the life cycle cost.

signicantly. These are:

(a) cost of the rolled sections used for initial construction of the structure
(b) number of dierent section types used in the structure
(c) weight of rolled sections used in the structure
(d) perimeter of rolled sectioned in the structure
(e) number of connections
(f) geographic location of the project site
(g) maintenance policy of the structure
(h) anticipated life of the structure
(i) discount rate of the currency
(j) use of the structure,
(k) importance of the structure.

The in
uence of these factors on dierent cost functions is shown in Figure 2 with single
lines.
Out of the aforementioned 11 factors, factors (a)–(f) are responsible for reducing the
cost function 1 [8]. Factor (b) also in
uences the cost function 7. Factor (d) is responsi-
ble for reducing the cost function 2 as the cost of painting on an exposed steel structure

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
 LIFE-CYCLE COST OPTIMIZATION OF STEEL STRUCTURES 1455

for preventing possible corrosion depends primarily on the perimeter of the members where
paint is applied. Factor (e) in
uences the cost function 1 in the form of fabrication, erec-
tion, connection materials, and labor costs [8]. This factor also in
uences the cost functions
2–4, because connections are usually the most vulnerable points for failure and have to be
inspected periodically. Furthermore, painting of connections is usually time consuming and
consequently costs more than the rest of the structure.
Factor (f) in
uences the cost functions 1–6. Apart from in
uencing the initial cost [8],
this factor in
uences the painting cost because the site may be near to the seacoast or in
an industrially polluted area where corrosion is high. Geographic location also in
uences
the maintenance and repair costs as a place with abundance of skilled and unskilled labour
force costs less than a place where labour force is scarce and expensive. Maintenance and
repair costs of structures in a dicult terrain are often expensive. Geographic location also
in
uences the operating cost of a structure such as heating and air-conditioning. Furthermore,
the probable failure cost also depends on the geographic location. For example, the probabil-
ity of failure of a structure in an active earthquake zone increases signicantly. Factor (g)
in
uences the cost functions 2–6. A poor maintenance policy often leads to early failure.
A conservative maintenance policy, on the contrary, may result in excessive costs. Factors
(h) and (i) in
uence all the cost functions, except the initial cost function 1 to be discussed
further in the next section. Factor (k) in
uences the cost functions 2–6. For very important
structures like a nuclear power plant the maintenance and probable failure costs are high, as
such structures are designed with a low probability of failure.
In this work, we consider only the rst four factors, as the structural designer has no control
over the other seven factors to reduce the life-cycle cost of a structure. In this article, ex-
tending the authors’ recent work [8], a four-criteria cost optimization model is presented for
life-cycle cost optimization of steel structures. These criteria are (i) select discrete com-
mercially available sections with the lowest cost, (ii) select discrete commercially available
sections with the lightest weight, (iii) select minimum number of dierent types of discrete
commercially available sections, and (iv) select discrete commercially available sections with
minimum total perimeter length.

FORMULATION OF TOTAL LIFE-CYCLE COST

The life-cycle cost of a structure can be dened as follows:

1
1
1
C Lifecycle = C Initial + yn1 C Maintenance + yn2 C Inspection + C Repair
(1 + i) (1 + i) (1 + i)yn3

1 1 1
+ yn4 C Operating + yn5 C Failure + C Dismantle (1)
(1 + i) (1 + i) (1 + i)yn6

where C Lifecycle , C Initial , C Maintenance , C Inspection , C Repair , C Operating , C Failure , and C Dismantle are the
total life cycle, initial, and annual maintenance, inspection, repair, operating, probable failure,
and dismantling costs of a steel structure, respectively; i is the discount rate
of money; yn1 ,
yn2 , yn3 , yn4 , yn5 , and yn6 are the years when the associated costs incur; denotes the
summation of all the costs of the same category during the life of the structure. Equation (1)
is based on the concept of single present worth [14]. The cost terms in the right-hand side

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
1456 K. C. SARMA AND H. ADELI

of Equation (1) are the costs in the year they actually occur. The 1=(1 + i) n factor is used to
convert the cost into its present value discounted by the discount rate of i for the yn period.
The discount rate depends on the prevailing interest rate and the depreciation of the currency
or in
ation rate. This rate is not a constant term and may vary over the life of the structure.
A discount rate of 2 or 3% above in
ation is considered an appropriate value [13]. Unfor-
tunately, actual cost data needed in the life cycle optimization of a structure are virtually
non-existent in the literature. The scant information available is based on insucient statisti-
cal data or assumptions. Therefore, it is currently not feasible to optimize the life-cycle cost
of a structure using Equation (1) due to the lack of statistically meaningful cost data.
Wilson et al. [15] discuss the development of a decision support system for analysing
the life-cycle costs of alternative bridge deck designs. A few researchers have presented
probabilistic models for life cycle cost optimization of structures [16; 17]. The use of such
probabilistic models is also limited due to the lack of statistical data. The authors have
developed a life-cycle cost optimization model based on fuzzy logic [9] with the goal of
formalizing the life-cycle design process but with some input from the design engineer. In
this article, the model is described by considering only one representative type of cost incurred
over the life of the structure. That is, preventive maintenance in the form of periodic painting
of an exposed steel structure to avoid corrosion or rustling. Thus, in this article only two
kinds of costs are considered: initial cost and the maintenance cost in the form of painting.

FUZZY DISCRETE MULTICRITERIA LIFE-CYCLE COST OPTIMIZATION

Following Sarma and Adeli [8], we dene four fuzzy functions: material cost of the struc-
ture C̃(ỹ), weight of the structure W̃ (ỹ), number of dierent section types T̃ (ỹ) and the
total cross-sectional perimeter of all the sections P̃(ỹ) in terms of the fuzzy discrete variables
(commercially available W shapes), ỹ. These are fuzzy functions (identied by the sign ∼
above them) because the variables ỹ are treated as fuzzy variables. The objective of the four-
criteria optimization model is to minimize the functions, C̃(ỹ), W̃ (ỹ), T̃ (ỹ) and P̃(ỹ). Out of
these four fuzzy functions C̃(ỹ), W̃ (ỹ) and P̃(ỹ) can be expressed explicitly in terms of the
fuzzy variables ỹ. The expression for C̃(ỹ) and W̃ (ỹ) are presented by Sarma and Adeli [8].
The expression for P̃(ỹ) is as follows:


Nt
P̃(ỹ)= li ỹpi ; p̃pi ∈ Spi (2)
i=1

where ỹpi is the perimeter of the discrete standard shape with the maximum membership
function corresponding to the minimum perimeter criterion; li is the total length of members
linked to variable xi (representing the cross-sectional area of the ith group of members linked
together in the continuous variable fuzzy GA optimization stage); Spi is a set of fuzzy discrete
candidate standard shapes for the design variable xi corresponding to the minimum perimeter
criterion; Nt is the number of initial section types (equal to the number of design variables in
the rst continuous variable stage of the optimization). The following approximate equation
is used for the perimeter of wide
ange shapes:
ỹpi =4bf i − 2twi + 2di (3)
where bf i , twi , and di , are the
ange width, web thickness, and the total depth of the section i.

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
 LIFE-CYCLE COST OPTIMIZATION OF STEEL STRUCTURES 1457

µp

pj
0 pjmin pjmax

Figure 3. Membership function for the minimum perimeter criterion.

In the fuzzy discrete multi-criteria cost optimization model, for each continuous variable
xi (i =1; 2; : : : ; Nt ), Ni commercially available sections (ỹij ) are selected such that, ỹij ¿xi ,
(i =1; 2; : : : ; Nt and j = 1; 2; : : : ; Ni ). The sign ∼ in ỹij indicates that it has four fuzzy attributes
corresponding to the four fuzzy objective functions. All the Ni shapes selected are graded with
membership functions for each one of the four criteria. The expressions for the membership
functions for minimum cost, minimum weight, and minimum section type are presented in
Reference [8]. Denoting the cross-sectional perimeter of the jth candidate section by pj , and
the maximum and the minimum values within the Ni candidate shapes by pjmax and pjmin , a
linear membership function for the minimum perimeter criterion is dened in the following
form (Figure 3):
pj − pjmin

Pj =1 − ; j =1; 2; : : : ; Ni (4)
pjmax − pjmin

The four fuzzy sets of candidate shapes corresponding to the four aforementioned criteria
for design variable xi are dened as SCj , S Wj , S Tj and SPj . To nd the multicriteria opti-
mum solution, the max–min procedure of Bellman and Zadeh [18] is used. The solution is
obtained by forming a fuzzy set, SDj , dened as the intersection of the four fuzzy sets as
follows:

SDj (ỹij )=SCj (ỹij ) ∩ S Wj (ỹij ) ∩ STj (ỹij ) ∩ SPj (ỹij ); j =1; 2; : : : ; Ni (5)

The membership of this intersection fuzzy set SDj is set equal to the minimum values of the
membership functions of the four sets SCi , S Wi , S Ti , and S Pi :

Dj (ỹij )= min{
C (ỹij );
W (ỹij );
T (ỹij );
P (ỹij )}; j =1; 2; : : : ; Ni (6)

where
C (ỹij );
W (ỹij );
T (ỹij ), and
P (ỹij ) are the fuzzy membership functions for the four
criteria: minimizing material cost, minimizing weight, minimizing the number of section types,
and minimizing the total perimeter of sections, respectively. The discrete commercially avail-
able section corresponding to the maximum membership function,
Di(best) , in the Ni selected

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
1458 K. C. SARMA AND H. ADELI

candidate shapes represents the best section [19].

Di(best) = max [
Dj (ỹij )]; i =1; 2; : : : ; Nt (7)

j=1; 2; :::; Ni

This step is known as defuzzication. This concept is further explained through an example
in Reference [8]. Thus, by using the max–min operation of fuzzy membership functions
(Equations (6) and (7)), the best commercially available discrete shapes are obtained for all
the design variables. It has been proved mathematically that this solution is Pareto optimal [8].
In obtaining the intersection fuzzy set S Dj (ỹij ) (Equation (4)), it is assumed that all the four
criteria are of equal importance. However, one of them may be of greater importance than
others. In such a case, S Dj (ỹij ) may be expressed as a convex combination of the four criteria
with weighting coecients re
ecting their relative importance. In this work four weighting
coecients w C , w W , w T , and wP are used for minimizing the material cost, sectional weight,
number of section types, and total perimeter of sections, respectively. There is an inverse
relationship between the weighting coecients and their impact on the multicriteria cost op-
timization (to increase the importance of a criterion its weighting coecient is reduced). The
modied membership functions are dened as:

Ci = w C
Ci


Wi = w W
Wi
(8)


Ti = w T
Ti


Pi = wP
Pi

where the sum of the weighting coecients is equal to 1:

wC + w W + w T + w P =1 (9)

EXAMPLE

The fuzzy multicriteria discrete life-cycle cost optimization model created in this research has
been implemented in the C language using the IRIX operating system on the SGI Origin 2000
supercomputer at the Ohio Supercomputer Center.
The life-cycle cost optimization model presented in this research has been applied to the
36-storey irregular moment resisting steel space frame structure with setbacks and cross-
bracings shown in Figures 4 and 5. Details of this example and the loading and displacement
constraints are presented in Reference [2] and Sarma and Adeli [8] and will not be repeated
here. The 3328 members of the structure are divided into 186 groups of initial design variables
as described in Reference [8]. Adeli and Park [2] present a minimum weight solution for the
same example. For columns and axially loaded members A572 Grade 50: W6, W8, W10,
W12 and W14 shapes are used. Similarly, for beams W16, W18, W21, W24, W27, W30,
W33, W36 and W40 shapes from the same grade of steel are used. The design constants
used are modulus of elasticity, E =198:91 GPa (29 000 ksi), specic weight =76:97 kN=m3

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
 LIFE-CYCLE COST OPTIMIZATION OF STEEL STRUCTURES 1459

12x3.56 m
= 42.67 m
3

12x3.56 m
= 42.67 m
2

12x3.56 m
= 42.67 m
1

12x4.57 m = 54.86 m

6x4.57 m = 27.43 m

Figure 4. 36-storey steel space moment-resisting frame structure.

(490:0 lb=ft3 ), and yield stress, Fy = 344:75 MPa (50 ksi). Cost data are obtained from Nucor
[20–22] as explained in Reference [8].
The design engineer has to make a decision on the relative values of the importance-
weighting coecients for the four membership functions. In this work the following ve
cases are studied:

Case A wC =0:001; w W =0:333; w T =0:333; wP =0:333 (Ni =6 for minimum material cost
design).
Case B wC =0:333; w W =0:001; w T =0:333; wP =0:333 (Ni =6 for minimum weight design).
Case C wC =0:333; w W =0:333; w T =0:333; wP =0:001 (Ni =6; for minimum perimeter de-
sign)
Case D wC =0:333; w W =0:333; w T =0:001; wP =0:333 (Ni =15; for minimum section type
design).
Case E wC =0:25; w W =0:25; w T =0:25; wP =0:25 (Ni =15; for equal preference for the four
criteria).

The optimum design solutions obtained by the fuzzy multicriteria discrete optimization
model based on the ASD and LRFD codes for Cases A–E are presented in Tables I and II.
A comparison of the cost values in the two tables leads to the conclusion that optimum design
costs based on the LRFD code are consistently less than the corresponding costs based on
the ASD code. Compared with the minimum material cost design (Case A), the material cost

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
1460 K. C. SARMA AND H. ADELI

1 2 column type

beam type
1
6 x 4.57 m cross bracing
2
(6 x 15 ft) column
3
beam

4 x 4.57 m (4 x 15 ft)
Plan for stories 1 to 12 (Section 3)

1
1 2
3

6 x 4.57 m 3
(6 x 15 ft) 4 2

8 x 4.57 m (8 x 15 ft)
Plan for stories 1 to 12 (Section 2)

1 2

6 x 4.57 m 3 3 y
(6 x 15 ft) 2
1

x
12 x 4.57 m (12 x 15 ft)

Plan for stories 1 to 12 (Section 1)

Figure 5. Floor plan for the 36-storey steel space moment-resisting frame structure.

of the minimum perimeter design (Case C) is 5.2% more using the ASD code (5.9% more
using the LRFD code). But, the total number of section types in Case C is reduced to 38 (36
for LRFD code) from 50 (51 for LRFD code) for Case A, resulting in reduced fabrication,
erection, storing, and handling cost.
Case E with equal preference for all the criteria appears to yield a good compromise in
terms of all the criteria. However, the designer can in
uence the life-cycle optimum design by
using dierent weighting factors based on his=her experience and considering the prevailing
economic and local conditions. The methodology presented in this work provides a logical
way for the designer to consider the best design for the life cycle of the structure. The method-
ology can be extended to include additional parameters relevant to the life cycle design of a
structure.

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
 LIFE-CYCLE COST OPTIMIZATION OF STEEL STRUCTURES 1461

Table I. Optimum design solutions obtained by the fuzzy discrete multicriteria cost optimization
model based on the ASD code for the 36-storey frame example.
Material cost Weight of Total no. of Total sectional
Type of study of the structure the structure section types perimeters
Case A for $6 543 527 15 516:0 kN 50 1 115 768:6 m
Ni = 6 (3488.3 kips) (43 927 897:4 in:)

Case B for $6 733 198 15 410:1 kN 58 1 160 980:7 m

Ni = 6 (3464.5 kips) (45 707 902:5 in:)

Case C for $6 885 985 15 852:2 kN 38 1 079 128:2 m

Ni = 6 (3563.9 kips) (42 485 360:3 in:)

Case D for $7 750 952 18 142:1 kN 20 1 138 301:4 m

Ni = 15 (4078.7 kips) (44 815 017:1 in:)

Case E for $7 364 961 17 216:0 kN 24 1 111 154:1 m

Ni = 15 (3870.5 kips) (43 746 225:5 in:)

Table II. Optimum design solutions obtained by the fuzzy discrete multicriteria cost optimization
model based on the LRFD code for the 36-storey space frame example.
Material cost Weight of Total no. of Total sectional
Type of study of the Structure the structure section types perimeters
Case A for $6 133 422 14 538:3 kN 51 1 107 473:5 m
Ni = 6 (3268.5 kips) (43 601 319:2 in:)

Case B for $6 265 238 14 470:2 kN 60 1 125 928:5 m

Ni = 6 (3253.2 kips) (44 327 894:1 in:)

Case C for $6 497 895 14 961:7 kN 36 1 064 846:2 m

Ni = 6 (3363.7 kips) (41 923 077:1 in:)

Case D for $7 504 254 17 475:3 kN 23 1 141 289:7 m

Ni = 15 (3928.8 kips) (44 932 664:1 in:)

Case E for $6 810 375 15 983:4 kN 25 1 114 614:8 m

Ni = 15 (3593.4 kips) (43 882 471:0 in:)

ACKNOWLEDGEMENT
Computing time for this research was provided by the Ohio Supercomputer Center.

REFERENCES
1. Adeli H (ed.). Advances in Design Optimization. Chapman & Hall: London, 1994.
2. Adeli H, Park HS. Neurocomputing for Design Automation. CRC Press: Boca Raton, FL, 1998.
3. Adeli H, Kumar S. Distributed Computer-Aided Engineering for Analysis, Design, and Visualization. CRC
Press: Boca Raton, FL, 1999.

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462
1462 K. C. SARMA AND H. ADELI

4. Adeli H, Soegiarso R. High-Performance Computing in Structural Engineering. CRC Press: Boca Raton, FL,
1999.
5. Sarma KC, Adeli H. Cost optimization of concrete structures. Journal of Structural Engineering (ASCE) 1998;
124(5):570 –578.
6. Sarma KC, Adeli H. Cost optimization of steel structures. Engineering Optimization 2000; 32(6):777– 802.
7. Sarma KC, Adeli H. Fuzzy genetic algorithm for optimization of steel structures. Journal of Structural
Engineering (ASCE) 2000; 126(5):596 – 604.
8. Sarma KC, Adeli H. Fuzzy discrete multicriteria cost optimization of steel structures. Journal of Structural
Engineering (ASCE) 2000; 126(11):1339 –1347.
9. Zadeh LA. Fuzzy sets. Information and Control 1965; 8(3):338–353.
10. Adeli H, Hung SL. Machine Learning—Neural Networks, Genetic Algorithms, and Fuzzy Sets. Wiley:
New York, 1995.
11. AISC. Manual of Steel Construction—Allowable Stress Design (9th edn, 2nd revision). American Institute of
Steel Construction: Chicago, IL, 1995.
12. AISC. Manual of Steel Construction—Load and Resistance Factor Design (2nd edn, 2nd revision), vol. I.
American Institute of Steel Construction: Chicago, IL, 1998.
13. Tietz SB. Lifecycle costing and whole-life design. The Structural Engineer 1987; 65A(1):10 –11.
14. Kirk SJ, Dell’isola AJ. Life Cycle Costing for Design Professionals (2nd edn). McGraw-Hill, Inc.: New York,
1995.
15. Wilson JL, Wagaman SJ, Veshosky DA, Shi CG, Adury P, Beidleman CR. Life-cycle engineering of bridges.
Microcomputers in Civil Engineering 1997; 12:445– 452.
16. Frangopol DM, Lin K-Y, Estes AC. Life-cycle cost design of deteriorating structures. Journal of Structural
Engineering (ASCE) 1997; 123(10):1390 –1401.
17. Koskisto OJ, Ellingwood BR. Reliability-based optimization of plant precast concrete structures. Journal of
Structural Engineering (ASCE) 1997; 123(3):298–304.
18. Bellman RE, Zadeh LA. Decision-making in a fuzzy environment. Management Science 1970; 17:B141–164.
19. Klir GJ, Folger TA. Fuzzy Sets, Uncertainty, and Information. Prentice-Hall: Englewood Clis, NJ, 1988.
20. Nucor. Nucor Steel Price List, E
ective date: 11 January 1999. Nucor Steel, a Division of Nucor Corporation,
Jewel, Texas, 1999.
21. Nucor. Nucor-Yamato Price List, 15 February 1999. Nucor-Yamato Steel Company: Blytheville, Arkansas,
1999.
22. Nucor. Nucor Steel, Berkley, Start up Rolling Forecast Wide Flange Beams, 24 March 1999. Nucor Steel
Company: Berkley, 1999.

Copyright ? 2002 John Wiley & Sons, Ltd. Int. J. Numer. Meth. Engng 2002; 55:1451–1462