This article appeared in a journal published by Elsevier.

The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elseviers archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy

Journal of Constructional Steel Research 74 (2012) 1725

Contents lists available at SciVerse ScienceDirect

Journal of Constructional Steel Research

Multi-objective seismic design method for ensuring beam-hinging mechanism in steel frames
Se Woon Choi, Hyo Seon Park
Dept. of Architectural Engineering, Yonsei University, 134 Shinchon-dong, Seoul, Korea 120-749

a r t i c l e

i n f o

a b s t r a c t
Previous research efforts have shown that the column-beam exural strength ratios of joints in moment resisting steel frames should be higher than 1.0 or even 2.0 for a beam-hinging collapse mechanism. However, it has been pointed out that, in order to prevent a weak story mechanism in a structure, it is not practical to use a specic single value as a limit for the column-beam exural strength ratio for all joints of a structure. Therefore, an optimal design technique is needed to determine the column-beam exural strength ratios for joints in a structure. In this paper, a multi-objective seismic design method for ensuring beam-hinging mechanism in steel moment resisting frame structures is presented and applied to optimal seismic design of well-known steel moment frames. In addition to the constraint for ensuring beam-hinging mechanism, the relationship between the structural cost and the energy dissipation capacity of structures is provided by considering the two conicting objective functions. In order to select the best design among the candidate designs, as a guide for structural engineers, a simple rule is presented in the form of dissipated energy density dened by the ratio of the energy dissipation capacity to the structural weight. 2012 Published by Elsevier Ltd.

Article history: Received 2 October 2011 Accepted 28 January 2012 Available online 13 March 2012 Keywords: Optimization Steel moment resisting frame Seismic design Column-to-beam strength ratio Inter-story drift

1. Introduction Seismic design criteria currently used in several countries such as KBC 2005 [1] and ANSI/AISC 341-05 [2] allow the plastic deformation of structures caused by seismic loads. Seismic designs enable plastic behavior to dissipate seismic energy and use relatively smaller seismic loads since a structure designed to maintain its elastic state during an earthquake is far from being optimum or efcient design. Moreover, for a severe earthquake, ductile behavior of a structure can reduce the repair cost and human injuries by preventing a sudden collapse of the structure. A moment resisting frame is a simple seismic-resistant system consisting only of beams and columns with rigid beam-to-column joints that secure exural rigidity. It features advantages such as excellent ductility and variety of architectural planning and, thus, is being widely used in several countries. As it has rigid beam-tocolumn joints, a moment frame experiences lateral deformations through exural behavior of columns and beams when an earthquake occurs. Due to such behavioral characteristics, exural rigidity and strength of members are important factors in determining the overall behavior of a frame structure [3]. Depending on the design of exural rigidity and strength of the member, damage to the structure can be alleviated by forming plastic hinges only at beams as illustrated in Fig. 1(a), or plastic hinges can be
Corresponding author at: Dept. of Architectural Eng., Yonsei University, 134 Shinchon-dong, Seodaemun-ku, Seoul 120-749, Rep. of Korea. Tel.: + 82 2 2123 2794; fax: + 82 2 365 4668. E-mail address: hspark@yonsei.ac.kr (H.S. Park). 0143-974X/$ see front matter 2012 Published by Elsevier Ltd. doi:10.1016/j.jcsr.2012.01.012

formed at the both ends of all columns supporting a specic story which induces a sudden failure as described in Fig. 1(b). Dooley and Bracci [4], Medina and Krawinkler [5], and Nakashima and Sawaizumi [6] discuss why the column-beam exural strength ratios of joints should be higher than 1.0 or even 2.0 for a beam-hinging collapse mechanism. In ANSI/AISC 341-05 [2], the strong column-weak beam condition is specied to reduce the possibility of column failures that lead to a sudden collapse of the overall structure. However, Kuntz and Browning [7] pointed out that, in order to prevent a weak story mechanism in a structure, it is not practical to use a specic single value as a limit for the column-beam exural strength ratio for all joints of a structure. Consequently, an optimal design technique is needed to determine the column-beam exural strength ratios for joints in a structure. Also, Moghaddam et al. [8] and Xu et al. [9] showed that an even distribution of inter-story drifts can enhance ductility of a structure and reduce seismic damage. For such reasons, several researchers have proposed techniques that can evenly distribute the inter-story drifts of a structure caused by an earthquake as given in Fig. 1(a). Leelataviwat et al. [10] proposed a new design technique that can evenly distribute the inter-story drifts by dissipating the seismic energy through the plastic-hinge collapse mechanism of beams. Xu et al. [9] suggested a performance-based optimization seismic design technique that minimizes the coefcient of variation of inter-story drifts and the weight of a structure using an optimality criteriabased algorithm. However, design alternatives obtained from the methods for minimizing the magnitudes or variations of inter-story drifts cannot guarantee to improve the initial stiffness and the energy dissipation capacity of a structure.

Author's personal copy

18 S.W. Choi, H.S. Park / Journal of Constructional Steel Research 74 (2012) 1725

Vu Vy
Base shear force

Bilinear idealization Pushover curve

Area of pushover curve =Dissipated energy =Elastic energy + Inelastic energy =0.5* y*Vy + 0.5*(Vu +Vy ) *( u- y) Ductility = /

Roof lateral displacement

Fig. 2. A bilinear idealization of a pushover curve for a calculation of elastic and inelastic dissipated energy of a structure. Fig. 1. Types of collapse mechanisms depending on the formation of hinges at a moment frame. (a) Beam-hinge collapse mechanism. (b) Column-hinge collapse mechanism.

2.1. Objective functions This study uses two objective functions: i) minimizing the structural weight of the building in order to consider the structural cost and ii) maximizing the energy dissipation capacity calculated from a non-linear static analysis. In addition to the most widely used objective function in structural optimization, the second objective function is adopted to improve the initial stiffness and ductility of the structure since the feasible solutions obtained from the method minimizing only inter-story drift ratios (Xu et al. [9] and Leelataviwat et al. [10]) cannot guarantee to improve the initial stiffness and the energy dissipation capacity of the structure. The rst and second objective functions are dened as in Eqs. (1) and (2): Minimize f 1
m X i 1

In this paper, a multi-objective seismic design method for ensuring beam-hinging mechanism in steel moment resisting frame structures is presented and applied to seismic design of steel moment resisting frame structures. In the multi-objective formulation, a set of feasible designs are obtained by considering two conicting objective functions: the structural weight and the energy dissipation capacity. The formulation can provide the relationship between the structural cost and the energy dissipation capacity of a structure for structural engineers. Constraints on the designs include member strength, inter-story drift, plastic hinge formation at joints of columns, and cross-sectional dimensions of columns. Rather than producing a single optimized design with minimal material cost or maximum energy dissipation capacity, the formulation provide a group of feasible designs with beam-hinging mechanism. Structural engineers then compare the feasible designs and select the most preferred one. When facing a group of feasible designs, structural engineers need to have a judgmental decision rule. In this paper, a simple rule to select the best design among the candidate designs is presented in the form of dissipated energy density (DED) dened by the ratio of the energy dissipation capacity to the structural weight. The proposed method is applied to optimal design of the 3-story and 9-story steel moment frames used by Xu et al. [9], Hasan et al. [11], and Gupta and Krawinkler [12]. 2. Formulation of the structural optimization problem In this paper, the purpose of the optimal seismic design is to nd a design with the minimized structural cost as well as the maximized seismic performance. Since the optimization problem needs two conicting objective functions, in general, a set of optimal solutions are found instead of a single optimal solution. It is because making one objective good inevitably makes the other bad. For instance, when only the weight of a system, which is mainly used in the area of structural optimization, is minimized, the initial costs can go down. This, however, is likely to cause deterioration of seismic performance of the structural system. Therefore, it is necessary to approach the optimization issues from a multi-objective optimization standpoint. Multi-objective genetic algorithms [13] can nd multiple Pareto-optimal solutions in a single run where classical single-objective optimization methods require multiple runs to nd Pareto-optimal solutions. For such reasons, in this paper, one of the Pareto-based approaches [13], a non-dominated sorting genetic algorithm-II (NSGA-II) [14] is used to nd a set of optimal designs.

i A i li

1 2

Minimize f 2 1=Areapushover

where i is the mass density of the ith member, Ai is the cross-sectional area of the ith member, li is the length of the ith member, and m is the total number of members. As shown in Fig. 2, Areapushover, which is the area below the roof lateral displacement-base shear force graph (pushover curve) obtained from a non-linear static analysis, is the same as total dissipated energy of a structure. By assuming a bilinear idealization of a pushover curve [15], Areapushover in Fig. 2 consists of elastic energy and inelastic energy. 2.2. Constraints Constraints on the designs considered in this paper can be classied into 6 types. The constraints on member strengths of columns and beams are given in Eqs. (3) and (4): c1 Rc =Dc 1:0
i i

i 1 to Nc

Table 1 Load values used in the analysis. Load name Floor dead load for weight calculations Floor dead load for mass calculations Roof dead load excluding penthouses Penthouse dead load Reduced live load Load value 4596.51 Pa 4117.70 Pa 3974.06 Pa 5554.11 Pa 957.61 Pa

Author's personal copy

S.W. Choi, H.S. Park / Journal of Constructional Steel Research 74 (2012) 1725 19

9 3 6

9 6

9 6

9 3

7 1 4

7 4

7 4

7 1

4@9.14m : Design variable

Fig. 3. Elevation of the 3-story example.

c2 Rb =Db 1:0

i 1 to Nb

i i is the required strength of the ith column member, Dc is the where Rc design strength of the ith column member, Nc is the number of columns, i i Rb is the required strength of the ith beam member, Db is the design strength of the ith beam member, and Nb is the number of beams. The constraint on inter-story drift is given in Eq. (5):

i+1 i where Ac and Ac are the cross-sectional areas of the verticallycontinuous column members on the i + 1th and ith stories, respectively. Among constraints generally considered in seismic designs of moment frames, it is known that the constraint on the inter-story drift is more dominant than other strength conditions [16,17]. Eq. (8) is the constraint on the structural weight considered in the rst objective function:

c3 =a 1:0

c6 W design =W lim 1:0

3@3.96m

where is the maximum inter-story drift ratio considering the drift amplication coefcient and a is the allowable inter-story drift ratio. For ensuring beam-hinging mechanism in steel moment resisting frame structures, Eq. (6) is used to avoid formation of plastic hinges at joints of columns:   c4 1:0 Nh;c = 2 Njoint 1:0 6

where Wdesign is the structural weight of a design and Wlim is the limitation on the structural weight. It is known that the two objective functions considered in this study are proportional in general. If there is no constraint on the structural weight, the structural weight may increase impractically as the energy dissipation capacity grows. 3. Examples The multi-objective seismic design method for ensuring beamhinging mechanism in steel moment resisting frame structures presented in this paper is applied to seismic design of two examples used by Xu et al. [9], Hasan et al. [11], and Gupta and Krawinkler [12]. For comparison of seismic performance of the example structures, the sections for columns and beams shown in Hasan et al. [11] and Gupta and Krawinkler [12] are dened as initial design in this paper.
4000 Initial design Feasible solution 3500 3000 2500 2000 1500 1000 250

c5 Ac =Ac 1:0

i1

i 1 to Nstory 1

Table 2 List of columns and beams used in the 3-story example. Column W14 370 W14 342 W14 311 W14 283 W14 257 W14 233 W14 211 W14 193 W14 176 W14 159 W14 145 W14 132 W14 120 W14 109 W14 99 W14 90 Beam W24 229 W24 207 W24 192 W24 176 W24 162 W24 146 W24 131 W24 117 W24 104 W24 103 W24 94 W24 84 W24 76 W24 68 W24 62 W24 55

Total dissipated energy (kNm)

where Nh,c is the number of plastic hinges occurred at joints of columns and Njoint is the number of joints considered. With this constraint, the structure is designed not to allow formation of plastic hinges at joints of columns excluding the beam-column joints of the top story. The joints at columns in the top story are not considered in the constraint since formation of plastic hinges at the columns of the top story will have little effect on the plastic behavior of the overall structure. For that reason, ANSI/AISC 341-05 [2] allows the exception of top story joints from the strong column-weak beam condition. Eq. (7) is the constraint on the cross-sectional areas of verticallycontinuous columns so that the cross-sectional area of the lower column is larger than or the same as that of the upper column:

300

350

400

450

500

550

Structural weight (kN)

Fig. 4. Distribution of objective functions for both the initial design and the feasible solutions (3-story example).

Author's personal copy

20 S.W. Choi, H.S. Park / Journal of Constructional Steel Research 74 (2012) 1725

750 35 Constraint c1 Constraint c3 Constraint c5 Constraint c2 Constraint c4 Constraint c6

Initial design Feasible solution

Dissipated energy density (DED)

0.8-1.0

No. of feasible solutions

30 25 20 15 10 5 0 0-0.2

700

650

600

550

500 250 300 350 400 450 500 550

0.2-0.4

0.4-0.6

0.6-0.8

Constraint value
Fig. 5. Frequency distribution of constraint ratios in the feasible solutions (3-story example).

Structural weight (kN)

Fig. 7. Distribution of DED and structural weight of the structure for both the feasible solutions and the initial design (3-story example).

It is assumed that the structure is located in LA, US where the soil grade is D (stiff soil) and the importance factor of the structure is 1.0. The steels used for beams and columns are A36 and A572 Grade 50, respectively. ASCE 7-05 [18] and ANSI/AISC 360-05 [19] are used to determine load combinations, inter-story drift conditions, and member strength design conditions. Three types of load combinations considered in this paper are given in Eqs. (9)(11): 1:4D 1:2D 1:6L 1:2D 1:0E 1:0L 9 10 11

elasticBeamColumn and nonlinearBeamColumn elements of OpenSees [20] for static elastic and inelastic analysis, respectively. Beam-to-column connections in the examples are assumed to be fully restrained and panel zone deformation is not considered. The yield strengths of beams and columns are 248.21 MPa (36ksi) and 344.74 MPa (50ksi), respectively. Eq. (12) is used to check the formation of a plastic hinge at an end of column members:
i

FyZ EI
i i

12

where D is dead load, L is live load, and E is earthquake load. The dead and live loads used in this study are given in Table 1. The earthquake load is calculated based on the equivalent lateral force calculation process suggested in ASCE 7-05 [18]. As to the lateral load distribution used in the non-linear static analysis, a ratio of seismic loads for each story used in static elastic analysis is applied to control the lateral displacement at the top of the building. The pushover analysis is done with OpenSees [20]. In the pushover analysis, the lateral load is gradually increased until the lateral displacement at the top of the building reaches 5% of overall structural height, which is corresponding to the collapse prevention level suggested in FEMA 356 [15]. Structural members in the examples are modeled using
1.0 Initial design

i , Z i, E i, and I i are the plastic curvature, yield strength, cross where, i, Fy sectional plastic modulus, young's modulus, and second moment of area of the ith column member, respectively. Plastic hinges are assumed to occur at the both ends of each member. For both examples, the optimal solutions are obtained after 400 generations with a population size of 40. Thus, the whole optimization process for each of examples with two different objective functions requires 32,000 times of tness function evaluations.

3.1. Example of 3-story moment frame The 3-story, 4-bay steel moment frame is shown in Fig. 3 (Xu et al. [9], Hasan et al. [11], and Gupta and Krawinkler [12]). As identied in Fig. 3, 27 members of the structure are linked to 9 design variables: 6 for columns and 3 for beams. As given in Table 2, the sets for available sections for columns and beams consist of 16 commercially available W14 sections and W24 sections, respectively. A total of 35 feasible solutions out of the population size of 40 are obtained from the multi-objective seismic design problem in Eqs. (1)(8). The relationship between the structural weight and the energy dissipation capacity for both the feasible solutions and the initial design is shown in Fig. 4. As depicted in Fig. 4, the energy dissipation capacity of feasible solutions is almost directly proportional to the
Table 3 List of columns and beams for the selected feasible solution (3-story example).

Value of inter-story drift constraint

Feasible solution 0.9

0.8

0.7

0.6

Column Design variable Section type W14 233 W14 233 W14 90 W14 311 W14 311 W14 176

Beam Design variable 7 8 9 Section type W24 207 W24 192 W24 84

0.5 250

300

350

400

450

500

550

Structural weight (kN)

Fig. 6. Distribution of structural weight and inter-story drift constraint ratio for both the initial design and the feasible solutions (3-story example).

1 2 3 4 5 6

Author's personal copy

S.W. Choi, H.S. Park / Journal of Constructional Steel Research 74 (2012) 1725 21

7000 6000 5000 4000 3000 2000 1000 0 0 10 20

L.S. level (2.5%) (29.13, 6291.92) L.S. level (2.5%) (26.47, 4893.91)

C.P. level (5.0%) (58.85, 6342.36) C.P. level (5.0%) (56.16, 4943.73)

Table 4 Minimum and maximum exural strength ratios of the initial design and the selected feasible solution (3-story example). Initial design Minimum exural strength ratio 2.00 Maximum exural strength ratio 3.55 Selected feasible solution Minimum exural strength ratio 1.14 Maximum exural strength ratio 1.99

Base shear (kN)

I.O. level (0.7%) (7.26, 3485.36) I.O. level (0.7%) (7.04, 3382.77)

Initial design Selected feasible solution 30 40 50 60

Roof lateral displacement (cm)

Fig. 8. Pushover curves of the initial design and the selected feasible solution (3-story example).

structural weight. In other words, putting more structural weight tends to increase the energy dissipation capacity of a structure. Also, it is notable that the energy capacity of the optimized solution with the same structural weight of the initial design is increased by about 8.3%. The distribution of constraint ratios for the feasible solutions is illustrated using a frequency table, as given in Fig. 5. For this example, the plastic hinge constraint c4 in Eq. (6) and column continuity constraint c5 in Eq. (7) turn out to be active constraints with values in the range of 0.8 to 1.0. The other dominant constraint is the interstory drift constraint c3 in Eq. (5) with values in the range of 0.6 to 0.8. Fig. 6 shows the relationship between structural weight and inter-story drift constraint ratio for both the feasible solutions and the initial design. It is observed that the value of inter-story drift constraint ratio decreases with an increase in the weight of the structure. For this example, it took 2.46 hours to complete the whole optimization process on Intel Core i7-2600 processor of 3.40 GHz. Among the 35 feasible solutions with beam-hinging mechanism from the multi-objective optimization in this paper, a simple rule is developed to select the best design. Dissipated energy density (DED) dened by the ratio of the energy dissipation capacity to the structural weight in Eq. (13) is used to select the best design. DED Total dissipated energy of a structure Weight of a structure 13

Fig. 7 shows the relationship between DED and structural weight of the structure for both the feasible solutions and the initial design. The feasible solution in Table 3 with the largest value of DED is selected as the best design. In other words, the design has the greatest energy dissipation capacity per unit structural weight. In the nonlinear static analysis with the target lateral displacement of 0.60 m at the top of the building equal to 5% of overall structural height, the push-over curves obtained from 468 steps with the displacement increment of 1.27 mm (0.05 in.) for both the selected design and the initial design are illustrated in Fig. 8. The roof lateral displacements at the maximum inter-story drift ratios of 0.7%, 2.5%, and 5.0% corresponding to the three performance levels of immediate occupancy (I.O.), life safety (L.S.), and collapse prevention (C.P.) [15] are compared in Fig. 8. It is notable, as illustrated in Fig. 8, that the yield strength of the selected design appears to be greater than that of the initial design even though the initial stiffness of the deigns are very similar to each other. The distribution of inter-story drift ratios at the three performance levels of I.O., L.S., and C.P. for the selected design is more uniform than that of the initial design as shown in Fig. 9. For both the initial and selected designs, the column-beam exural strength ratios of joints were calculated using Eq. (14):     F y Z = F y Z
c

14

where (FyZ)c is the plastic exural strength of the column at a joint and (FyZ)b is the plastic exural strength of the beam at a joint. As given in Table 4, for both designs, the ratios are greater than 1.0, which is in agreement with the suggestion in ANSI/AISC 341-05 [2]. This is also in conformity with the studies of Dooley and Bracci [4], Medina and Krawinkler [5], and Nakashima and Sawaizumi [6] in concluding that the exural strength ratio of each joint should be larger than 1.0 in order to induce a beam-hinge collapse mechanism. The relationship between the maximum value and the standard deviation of column-beam exural strength ratios of joints for all 35 feasible solutions is shown in Fig. 10. As shown in Fig. 10, the standard
2.5

Selected feasible solution 3

Standard deviation of column-to-beam strength ratio

Initial design

2.0

1.5

Story

1.0

0.5

1
I.O. level (0.7%) L.S. level (2.5%) C.P. level (5.0%)

0.0 1 2 3 4 5 6 7

0.00

0.01

0.02

0.03

0.04

0.05

Maximum column-to-beam strength ratio

Fig. 10. The relationship between the maximum value and the standard deviation of column-beam exural strength ratios of joints in all 35 feasible solutions (3-story example).

Inter-story drift ratio

Fig. 9. Distribution of inter-story drift ratios for the initial design and the selected feasible solution (3-story example).

Author's personal copy

22 S.W. Choi, H.S. Park / Journal of Constructional Steel Research 74 (2012) 1725

19 5 10

19 10

19 10

19 10

19 5

18 5 10

18 10

18 10

18 10

18 5

17 4 9

17 9

17 9

17 9

17 4

16 4 9

16 9

16 9

16 9

16 4

15 3 8

15 8

15 8

15 8

15 3

14 3 8

14 8

14 8

14 8

14 3

13 2 7

13 7

13 7

13 7

13 2

12 2 7

12 7

12 7

12 7

12 2

11 1 6

11 6

11 6

11 6

11 1

5@9.14m :Design variable

Fig. 11. Elevation of the 9-story example.

deviation is clearly proportional to the maximum value of columnbeam exural strength ratios of joints. 3.2. Example of 9-story moment frame

A total of 7 feasible solutions out of the population size of 40 are obtained from the multi-objective seismic design problem in Eqs. (1)(8). The relationship between the structural weight and
13000 Initial design Feasible solution 12000

Total dissipated energy (kNm)

The 9-story, 5-bay steel moment frame is shown in Fig. 11 (Xu et al. [9], Hasan et al. [11], and Gupta and Krawinkler [12]). As identied in Fig. 11, 99 members of the structure are linked to 19 design variables: 10 for columns and 9 for beams. As given in Table 5, the sets for available sections for columns and beams consist of 16 commercially available W14 sections and W36 sections, respectively.
Table 5 List of columns and beams used in the 9-story example. Column W14 665 W14 605 W14 550 W14 500 W14 455 W14 426 W14 398 W14 370 W14 342 W14 311 W14 283 W14 257 W14 233 W14 211 W14 193 W14 176 Beam W36 210 W36 194 W36 182 W36 170 W36 160 W36 150 W36 135 W36 130 W36 118 W36 108 W36 99 W36 94 W36 84 W36 68 W36 62 W36 55

11000

10000

9000

8000 1800 1900 2000 2100 2200

Structural weight (kN)

Fig. 12. Distribution of objective functions for both the initial design and the feasible solutions (9-story example).

5.48m

8@3.96m

Author's personal copy

S.W. Choi, H.S. Park / Journal of Constructional Steel Research 74 (2012) 1725 23

10 9 8

Constraint c1 Constraint c3 Constraint c5

Constraint c2 Constraint c4 Constraint c6

Table 6 List of columns and beams for the selected feasible solution (9-story example). Column Design variable 1 2 3 4 5 6 7 8 9 10 Section type W14 605 W14 550 W14 500 W14 370 W14 233 W14 500 W14 455 W14 426 W14 398 W14 176 Beam Design variable 11 12 13 14 15 16 17 18 19 Section type W36 182 W36 170 W36 210 W36 182 W36 150 W36 182 W30 108 W27 94 W24 68

No. of feasible solution

7 6 5 4 3 2 1 0 0-0.2 0.2-0.4 0.4-0.6 0.6-0.8 0.8-1.0

Constraint value
Fig. 13. Frequency distribution of constraint ratios for the feasible solutions (9-story example).

the energy dissipation capacity for both the feasible solutions and the initial design is shown in Fig. 12. As shown in Fig. 12, the energy dissipation capacity of the example is not clearly proportional but, by

and large, the increase of structural weight led to the increase of energy dissipation capacity. The distribution of constraint ratios for the feasible solutions is illustrated using a frequency table, as given in Fig. 13. As is the case with the 3-story example, it is conrmed that the plastic hinge constraint c4 and column continuity constraint c5 are active constraints with values in the range of 0.8 to 1.0. In addition, for this example, the inter-story drift condition is also active constraint with values in

1.1

9000 8000 7000 I.O. level (0.7%)

L.S. level (2.5%) (81.79, 8433.45) L.S. level (2.5%) (64.89, 6445.19)

C.P. level (5.0%) (171.58, 8601.09) C.P. level (5.0%) (146.69, 6809.84)

Value of inter-story drift constraint

1.0

Base shear (kN)

0.9

(19.43, 5125.39)

6000 5000 4000 3000 2000 1000 0 0

I.O. level (0.7%) (23.37, 4993.37)

0.8

0.7 Initial design 0.6 Feasible solution

Initial design Selected feasible solution

0.5 1800 1900 2000 2100 2200

50

100

150

200

Structural weight (kN)

Fig. 14. Distribution of structural weight and inter-story drift constraint ratio for both the initial design and the feasible solutions (9-story example).

Roof lateral displacement (cm)

Fig. 16. Pushover curves of the initial design and the selected feasible solution (9-story example).

Initial design 600 Initial design Feasible solution 8 560 540 6 500 480 460 2 440 420 1800 1900 2000 2100 2200 0 0.00
I.O. level (0.7%) L.S. level (2.5%) C.P. level (5.0%)

10

Selected feasible solution

Dissipated energy density (DED)

580

Story
4

520

0.01

0.02

0.03

0.04

0.05

Structural weight (kN)

Fig. 15. Distribution of DED and structural weight of the structures for both the feasible solutions and the initial design (9-story example).

Inter-story drift ratio

Fig. 17. Distribution of inter-story drift ratios for the initial design and the selected feasible solution (9-story example).

Author's personal copy

24 S.W. Choi, H.S. Park / Journal of Constructional Steel Research 74 (2012) 1725

Table 7 Energy dissipation capacities, DEDs, and column-beam exural strength ratios for both 7 feasible solutions and initial design for 9-story frame example. Structural Energy dissipation capacity (kNm) weight (kN) Elastic Initial design 1863.29 Feasible 1837.51 solutions 1915.16 2103.80 2115.56 2194.45 2211.65 2218.41 Inelastic Total 8588.92 8734.09 8363.78 9563.43 10815.69 10876.99 11152.47 13058.49 460.95 475.32 436.71 454.58 511.24 495.66 504.26 588.64 DED Column-tobeam strength ratios Min. 1.47 1.47 1.63 1.44 1.54 1.33 1.33 1.50 Max. 4.91 4.91 4.91 6.51 5.05 11.64 11.34 4.85

4. Conclusions A multi-objective seismic design method for ensuring beamhinging mechanism in steel moment resisting frame structures is presented and applied to seismic design of steel moment resisting frame structures. In addition to the constraint for ensuring beam-hinging mechanism, the relationship between the structural cost and the energy dissipation capacity of structures is provided by considering the two conicting objective functions. Rather than producing a single optimized design with minimal material cost or maximum energy dissipation capacity, the formulation provides a group of feasible designs with beam-hinging mechanism. In order to select the best design among the candidate designs, as a guide for structural engineers, a simple rule is presented in the form of dissipated energy density (DED) dened by the ratio of the energy dissipation capacity to the structural weight. For the example structures, the selected design based on DED shows an improvement in the initial stiffness and yield strengths. Moreover, it is found that the distribution of inter-story drift ratios for the selected design is more uniform than that for the initial designs. Also, it is notable that column-beam exural strength ratios of joints in the selected optimum designs with beam-hinge collapse mechanisms are in the range of 1.14 to 4.85, which is larger than the suggested ratios in ANSI/AISC 341-05 [2]. From the relationship between the maximum value and the standard deviation of column-beam exural strength ratios of joints, it is found that the standard deviation of columnbeam exural strength ratios of joints is proportional to the maximum value. For nal check of the selected design, a nonlinear dynamic analysis may be necessary to consider more accurate dynamic responses of a structure. Acknowledgements This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (No. 2011-0018360). References
[1] AIK. Korea building code and commentary (KBC 2005). Architectural Institute of Korea; 2005. [2] AISC. Seismic provisions for structural steel buildings (ANSI/AISC 341-05). Chicago (IL): American Institute of Steel Construction; 2005. [3] Bruneau M, Uang CM, Whittaker A. Ductile design of steel structures. McGraw-Hill; 1997. Chapter 8. [4] Dooley KL, Bracci JM. Seismic evaluation of column-to-beam strength ratios in reinforced concrete frames. ACI Struct J 2001;98(6):83451. [5] Medina RA, Krawinkler H. Strength demand issues relevant for the seismic design of moment-resisting frames. Earthq Spectra 2005;21(2):41539. [6] Nakashima M, Sawaizumi S. Column-to-beam strength ratio required for ensuring beam-collapse mechanisms in earthquake responses of steel moment frames. Proceedings of the 12th World Conference on Earthquake Engineering. New Zealand Society for Earthquake Engineering; 2000. [7] Kunts GL, Browning J. Reduction of column yielding during earthquakes for reinforced concrete fames. ACI Struct J 2003;100(5):57380. [8] Moghaddam H, Hajirasouliha I, Doostan A. Optimum seismic design of concentrically braced steel frames: concepts and design procedures. J Constr Steel Res 2005;61(2):15166. [9] Xu L, Gong Y, Grierson DE. Seismic design optimization of steel building frameworks. J Struct Eng 2006;132(2):27786. [10] Leelataviwat S, Goel SC, Stojadinovic B. Energy-based seismic design of structures using yield mechanism and target drift. J Struct Eng 2002;128(8):104654. [11] Hasan R, Xu L, Grierson DE. Push-over analysis for performance-based seismic design. Comput Struct 2002;80(31):248393. [12] Gupta A, Krawinkler H. Seismic demands for performance evaluation of steel moment resisting frame structures. The John A. Blume Earthquake Engineering Center, Report No. 132. Stanford, California: Dept. of Civil Engineering, Stanford University; 1999. [13] Deb K. Multi-objective optimization using evolutionary algorithms. John Wiley & Sons; 2001. Chapter 2. [14] Deb K, Pratap A, Agarwal S, Meyarivan T. A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE Trans Evol Comput 2002;6(2):18297. [15] FEMA. Prestandard and commentary for the seismic rehabilitation of buildings (FEMA 356). Washington: Federal Emergency Management Agent; 2000.

941.86 7647.06 955.25 7778.85 907.87 7455.91 1115.34 8448.09 1054.13 9761.56 1176.82 9700.18 1210.78 9941.69 1297.70 11760.79

the range of 0.8 to 1.0. For this reason, as shown in Fig. 14, there is no distinct difference in the values of inter-story drift constraint for the feasible solutions relation with an increase in the weight of the structure. For this example, it took 16.67 hours to complete the whole optimization process on Intel Core i7-2600 processor of 3.40 GHz. Fig. 15 shows the relationship between DED and structural weight of the structure for both the feasible solutions and the initial design. The feasible solution in Table 6 with the largest value of DED is selected as the best design. In the push-over curves for both the selected design and the initial design illustrated in Fig. 16, roof lateral displacements at the three performance levels of I.O., L.S., and C.P. are compared. In the nonlinear static analysis with the target lateral displacement of 1.86 m at the top of the building equal to 5% of overall structural height, the push-over curves are obtained from 1464 steps with the displacement increment of 1.27 mm (0.05 in.). The selected design appears to have higher initial stiffness and yield strength than those for the initial design. Moreover, it is found that the distribution of inter-story drift ratios for the selected design is more uniform than that for the initial design as shown in Fig. 17. For both of the initial design and the selected feasible solution, energy dissipation capacities, DEDs, and column-beam exural strength ratios of joints are calculated and listed in Table 7. The relationship between the maximum value and the standard deviation of columnbeam exural strength ratios of joints for all 7 feasible solutions is shown in Fig. 18. As shown in Fig. 18, the standard deviation is proportional to the maximum value of column-beam exural strength ratios of joints.

2.6

Standard deviation of column-to-beam strength ratio

2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 4 5 6 7 8 9 10 11 12

Maximum column-to-beam strength ratio

Fig. 18. The relationship between the maximum value and the standard deviation of column-beam exural strength ratios of joints in all 7 feasible solutions (9-story example).

Author's personal copy

S.W. Choi, H.S. Park / Journal of Constructional Steel Research 74 (2012) 1725 [16] Foutch DA, Yun S. Modeling of steel moment frames for seismic loads. J Constr Steel Res 2002;58:52964. [17] Liu M, Burns SA, Wen YK. Multiobjective optimization for performance-based seismic design of steel moment frame structures. Earthq Eng Struct Dyn 2005;34:289306. 25

[18] ASCE. Minimum design loads for buildings and other structures (ASCE 7-05). New York (NY): American Society of Civil Engineer; 2006. [19] AISC. Specication for structural steel buildings (ANSI/AISC 360-05). Chicago (IL): American Institute of Steel Construction; 2005. [20] Mazzoni S, Mckenna F, Fenves GL. OpenSees command language manual; 2006.