Sei sulla pagina 1di 6

This article is an author-created prior version for The Fourth China-Japan-Korea Joint Symposium on

Optimization of Structural and Mechanical Systems, Kunming, pp.89-94, Nov. 6-9, 2006, China

A SIMPLE ESTIMATION OF FABRICATION COST AND MINIMUM COST


DESIGN FOR STEEL FRAMES
Kiichiro Sawada1* , Hitoshi Shimizu2 , Akira Matsuo1, Takaichi Sasaki1, Takashi Yasui3 and Atsushi Namba3
1

Social and Environmental Engineering, Hiroshima University, 1-4-1, Kagamiyama, Higashi-hiroshima, Japan
* Corresponding author:kich@hiroshima-u.ac.jp
2
Takenaka Corporation Hiroshima office, 10-10, Hashimoto-cho, Hiroshima,Japan
3

Minami Kogyo,10883-40 ,Hara, Hachihonmatsu ,Higashi-hiroshima , Japan

Abstract
In this study, the steel fabrication cost functions are first shown. Next, the minimum cost frames are designed by the
optimization method and the presented cost functions and are compared with the minimum weight frames. The problem
solved here is to determine the cross section of each member, which minimizes the sum of the fabrication cost and
material cost under the constraints of the aseismic design in Japan. It includes the constraints on the member stress,
story drift and collapse load for the building subjected to the vertical load and the horizontal load. This problem is
solved by Genetic Algorithm based on the ranking selection.
Keywords: Building Structures, Fabrication Cost, Minimum Cost Design

1 Introduction
There are a lot of studies on structural optimization of buildings. Many of these studies solve the minimum weight
design problems [1,2]. At the initial stage of the structural steel design in Japan, the fabrication cost of the building is
often predicted based on the structural weight. However, structural weight cant necessarily predict the fabrication cost
exactly. One of the reasons is because the fabrication cost for steel members depends upon the complexity of the
connections rather than the structural weight.
We have already shown that the fabrication cost of steel structures depends upon the number of parts and the jointed
sectional area from the questionnaires on fabrication time to fabricating workers [3]. In addition, we have derived the
fabrication cost functions from the estimated cost data of the fabricating company and the questionnaires [3].
In this study, the fabrication cost functions are first shown. Next, the minimum cost frames are designed by Genetic
Algorithm based on the ranking selection [4,5], and are compared with the minimum weight frames.

2 The Typical Beam-to-Column Connection in Japan


Figure1 shows the typical H-beam-to-RHS-column connections in Japan. In the fabricating company, the beam to
column connection is first welded to two through diaphragms by full-penetration welds as shown in Fig.2. In this paper,
the beam to column connection consisting of these three parts as shown in Fig.2 is named the connection box. The
connection box is welded to the flanges of the bracket by full-penetration welds and to the web of the bracket by fillet
welds as shown in Fig.3. Finally, the columns are welded to the connection box by full-penetration welds as shown in
Fig.4. The bracket is connected to the beam by high strength bolts in the field. This study deals with the buildings
having the beam-to-column connection shown in Fig.1.

3 Fabrication Cost Functions


In this study, the following function is presented to predict the steel fabrication cost.
CF=CP+CB+CW+CI
(1)
where CF represents the steel fabrication cost, CP represents the preparatory process cost, CB represents the assembly
cost, CW represents the welding cost, and CI represents the information cost such as making shop drawings.

Connection box

Bracket
Figure 1. Typical Beam
to column connection

Figure 2.

Figure 3.

Figure 4.
The preparatory process is comprised of marking, drilling, blasting and flange bevels of diaphragms, beams and
brackets. It has been observed from the questionnaires [3] that the preparatory process cost depends on the number of
parts such as diaphragms, girders and brackets rather than structural weight. The following function is proposed to
estimate the preparatory process cost, CP.
nj

CP KP ( NPD i PB NP B )

(2)

i 1

where NPDi represents the number of diaphragms for beam to column connection i, nj represents the number of beam to
column connections, NPB represents the number of beams and brackets, PB and KP represent the coefficients to
evaluate the preparatory process cost.
It has also been observed from the questionnaires [3] that the assembly cost depends on the number of parts rather
than structural weight. The following fucntion to estimate the assembly cost, CB is proposed.
nj

CB KB ( NB0 i BC NBC )

(3)

i 1

where NB0i represents the number of parts in connection boxes and brackets for beam to column connection i, NBC
represents the number of columns, BC and KB represent the coefficients to evaluate the assembly cost.
It has been observed from the questionnaires [3] that the welding cost depends on sum of jointed sectional areas. The
following function is proposed to estimate the welding cost, CW.
nbb
nj

(4)
CW KW AD i ABB i
i 1
i 1

,where ADi represents the jointed sectional area between the column and the diaphragm for beam to column connection i,
nbb represents the number of brackets, KW represent the coefficient to evaluate the welding cost.
The information cost represents the cost for shop drawings and full scaling. Since the information cost depends on the
number of sheets of shop drawings, the following function based on the number of columns and beams is proposed.
CI KIc NIc KIb NIb KIgW
(5)
,where NIc represents the number of shop fabricated column trees, NIb represents the number of beams having different
cross sectional size, W represents the total structural weight of the frame, KIc, KIb, KIg represent the coefficient to
evaluate the information cost.
Computational examples of NPDi in Eq.(2), NB0i in Eq.(3) and ADi in Eq.(4) for beam to column connection i
NPDi in Eq.(2), NB0i in Eq.(3) and ADi in Eq.(4) for beam to column connection i are computed as follows.
For a standard beam to column connection shown in Fig.5,
NPDi=2, NB0i=7, ADi=4Aci
Aci: the cross sectional area of the column adjoining to connection i

For a beam to column connection including an internal diaphragm shown in Fig.6,


NPDi=3, NB0i=8, ADi=5Aci
For a beam to column connection having different upper and lower column depths shown in Fig.7,
NPDi=2, NB0i=10, ADi=4Aci+4HbiTci
Hbi: the cross sectional depth of the beam adjoining to connection i
Tci: the thickness of the lower column adjoining to connection i

through
diaphragm

beam

internal
diaphragm

column

Fig.5

Fig.6

Fig.7

4 Cost Coefficients
The values of PB and BC in Eqs.(2) and (3) were computed from questionnaires on fabrication time as follows [3].
PB=15, BC=3
The values of KP, KB, KW, KIb, KIc, KIg in Eqs.(2),(3),(4) and (5) were computed from the least square
approximation based on estimated cost data of the fabricating company as follows [3].
KP=196 yen, KB=1150yen, KW=16.4yen/cm2, KIb=3550yen, KIc=12200yen, KIg= 999yen/t

6 The Minimum Weight Design Problem and The Minimum Cost Design Problem
The minimum weight design problem of steel structural frames as shown in Fig.8 can be formulated as follows.

Find Didc (idc 1,..., NDC ), Tic (ic 1,..., NC ),


H idb (idb 1,..., NDB ), Bib , Twib , Tf ib (ib 1,..., NB )
M

which minimize W Ai Li

(1)

i 1

subjected to

gS j
gDk

Nj
Aj f N j

k / Hk

1 / 200
g P P 1

Mj
Z j fM

1 ( j 1,2..., NM )
j

(k 1,2...., NF )

(2a-c)

where W,,Ai, and Li denote respectively the structural weight, weight per unit volume of steel, the cross-sectional area,
and member length; NM, NF denote respectively the number of members and the number of stories. Nj, Mj, fNj, fMj,k,
and Hk denote respectively the axial force, bending moment, allowable stress for the axial force and bending moment,
interstory drift of story k, and height of story k. p is the collapse load factor. The design variables, Didc, Tic, Hidb, Bib,

Twib, Tfib are chosen from the list of standard section sizes [1].
The abovementioned constraints are based on the Japanese building standard law [4] and the Japanese Design Standard
for Steel Structures [5]. In this study, the stiffness matrix method is used for elastic constraints, and compact
procedure [6], one of limit analysis methods, is used for plastic constraints.

Beam

Column
Column section
Fig.8(A) Steel structural frame

f
Beam section

Fig.8(B) Column section and beam section

The minimum cost design problem of steel frames can be formulated as follows.
Find Dic , Tic (ic 1,..., nc ), Bib , H ib , Twib , Tf ib
(ib 1,..., nb)
which minimize
nj
nd

C CF KS Ai Li V Di
i 1
i 1

(3)

subjected to Eq.(2a ) (2c)


where CF represents the fabrication cost shown in Eq.(1), KS represents steel material cost per unit weight, VDi
represents the total volume of diaphragm in beam to column connection i, nj represents the number of beam to column
connections.

7 Design Examples
The minimum weight design and the minimum cost design are executed for the center plane frame of five-story
building shown in Fig.9. Youngs modulus E, the yield stress F, and the steel weight per unit volume are specified as
follows. E= 2.06105(N/mm2), F=235(N/mm2), =76.93(N/cm3)
The design load for the elastic constraints, Eq.(2a) and (2b) is shown in Fig.10. The design horizontal load for the
plastic constraints, Eq.(2c), is twice as much as the one shown in Fig.10. The design variables of the members are also
shown in Fig.10. Genetic Algorithm (GA) is applied for both the minimum weight design and the minimum cost design.
GA control parameters are population of 100, crossover probability of 1.0 and mutation probability of 0.01. The
following computational results are the solutions having the minimum objective function for 10 solutions obtained by
using a different initial random number each time. Table 1(A),(B) show the minimum cost solutions for KS=40000yen/t
and KS=80000yen/t. Table 1(C) shows the minimum weight solution. The total cost C of the minimum cost solution for
KS=40000yen/t and KS=80000yen/t is less than that of the minimum weight solution, respectively. Figure 11 shows the
fabrication cost and material cost for the minimum weight frame and the minimum cost frame. It is observed from this
figure that the minimum cost frame has lower fabrication cost than the minimum weight frame, while both frames have
almost same material cost. Figure12(A),(B) show the cross sectional area diagram of the minimum cost solution for
KS=40000yen/t and the minimum weight solution. The numerical values in these figures represent cross sectional depth.
It is observed from these figures that the minimum cost frames consist of several unified member depths, while most of
the member depths of the minimum weight frames are different.

94.1kN

3.5m

343kN

3.5m

211kN

3.5m
3.5m
4m
8m
6m
8m
8m

6m

165kN

141kN 165kN

188kN

94.1kN

H-H9B9Tw9Tf9
H-H10B10Tw10Tf10 H-H9B9Tw9Tf9
-D2T2
-D2T2
-D2T2
H-H7B7Tw7Tf7
-D2T2

164kN

H-H5B5Tw5Tf5
-D2T2

125kN

H-H3B3Tw3Tf3
-D1T1

88.6kN

H-H1B1Tw1Tf1

H-H8B8Tw8Tf8
-D2T2

H-H7B7Tw7Tf7
-D2T2

H-H6B6Tw6Tf6
-D2T2

H-H5B5Tw5Tf5
-D2T2
H-H3B3Tw3Tf3

H-H4B4Tw4Tf4
-D1T1

-D1T1

H-H2B2Tw2Tf2
-D1T1

-D1T1

H-H1B1Tw1Tf1
-D1T1

8m

Fig.10 Five-story plane frame

Fig.9 Five-story building


Table 1(A) The minimum cost design
(KS=40000yen/t)

Table 1(B) The minimum cost design


(KS=80000yen/t)
2

Sectional size (mm)


Column
5F
4F
3F
2F
1F
Outer girder
RF
5F
4F
3F
2F
Inner girder
RF
5F
4F
3F
2F
Structural Weight
Total Cost

188kN

Sectional size (mm)

A (cm )

-40016
-40016
-40016
-50016
-50016

234.8
234.8
234.8
298.8
298.8

H-500250916
H-500200922
H-500250916
H-7002501225
H-5002001219

123.6
130.5
123.6
205.8
132.9

H-500200912
H-500200912
H-5002001222
H-7002501222
H-5002001222
26.88 (t)
2,288,100yen

92.29
92.29
144.2
191.5
144.2

Column
5F
4F
3F
2F
1F
Outer girder
RF
5F
4F
3F
2F
Inner girder
RF
5F
4F
3F
2F
Structural Weight
Total Cost

A (cm2)

-40016
-40016
-40016
-45016
-45016

234.8
234.8
234.8
266.8
266.8

H-400200916
H-500200916
H-500250922
H-6002501222
H-7002001225

98.57
107.6
152.6
178.2
180.8

H-400200912
H-500250919
H-500250916
H-6002501219
H-7002001222
26.26 (t)
3,347,200yen

83.29
138.0
123.6
163.9
169.5

Table 1(C) The minimum weight design


Sectional size (mm)
Column
5F
4F
3F
2F
1F
Outer girder
RF
5F
4F
3F
2F
Inner girder
RF
5F
4F
3F
2F
Structural Weight
Total Cost

A (cm2)

-40016
-40016
-40016
-45016
-45016

234.8
234.8
234.8
266.8
266.8

H-400200916
H-400200919
H-500200919
H-7002001225
H-7003001222

98.57
110.0
119.0
180.8
213.5

H-400200919
110.0
H-500200922
130.5
H-7003001219 196.2
H-400200912
83.29
H-400200912
83.29
25.94 (t)
2,340,000yen (KS=40,000yen/t)
3,450,000yen (KS=80,000yen/t)

fabrication cost
material cost

minimum weight
design
minimum cost
design

Cost
(yen)
Fig.11(A) Costs for the minimum weight frame and
the minimum cost frame (KS=40000yen/t)
0.00E+00 5.00E+05 1.00E+06 1.50E+06 2.00E+06 2.50E+06

fabrication cost
material cost

minimum weight
design
minimum cost
design

Cost
(yen)
Fig.11(B) Costs for the minimum weight frame and
the minimum cost frame (KS=80000yen/t)
0.00E+00

1.00E+06

2.00E+06

3.00E+06

4.00E+06

400

500

500

500

500

400

400

400

500

500

500

400

500

400

500

500

500

500

700

500

700

700

700

700

400

700

500

500

500

700

400

700

400

450

Fig.12(A) Cross sectional area diagram for the


minimum cost frame (KS=40000yen/t)

Fig.12(B) Cross sectional area diagram for


the minimum weight frame

8 Conclusions
In this study, the steel fabrication cost functions have been first shown. Next, the minimum cost frames have been
designed by Genetic Algorithm based on the ranking selection, and compared with the minimum weight frames. The
following remarks have been obtained in computational results.
(1) The total cost C of the minimum cost solution is less than that of the minimum weight solution under the same
condition. The minimum cost frame has lower fabrication cost than the minimum weight frame, while both frames have
almost same material cost.
(2) The minimum cost frames consist of several unified member depths, while most of member depths of the minimum
weight frames are different.

References
1.
2.
3.
4.
5.
6.
7.
8.

Sawada K., Matsuo A. : An Exact Algorithm and Approximate Algorithms for Discrete Optimization of Steel
Building Frames, CJK-OSM3, 663-668, 2004.10
Uetani, K., Tsuji, M. and Takewaki, I., Application of an optimum design method to practical building frames with
viscous dampers and hysteretic dampers, Engineering Structures 25, 579-592, 2003
Shimizu H., Sawada K., Matsuo A., Sasaki T. and Namba T. : A Study on Fabrication Cost Estimation for Steel
Frames, Journal of constructional steel, Vol.14, 2006.11(In Japanese)
Jenkins W.M. Plane frame optimum design environment based on genetic algorithm, ASCE,
Vol.118(11),pp3103-3112,1992
Ohsaki, M., Genetic Algorithm for Topology Optimization of Trusses, Computers & Structures, Vol. 57. No. 2. pp.
219-225. 1995.1
The Ministry of Construction of Japan, The Building Standard Law of Japan, 1994(In Japanese)
Architectural Institute of Japan, Design Standard for Steel Structures, 2002(In Japanese)
Livsley, R.K., A Compact FORTRAN Sequence for Limit Analysis, Int. J. of Num. Meth. Eng. Vol.5, No.3,
pp.446-449, 1973

Potrebbero piacerti anche