1. 정형조, 세종대학교 토목환경공학과 조교수 최강민, 한국과학기술원 건설 및 환경공학과 박사과정 지한록, 한국과학기술원 건설 및 환경공학과 석사과정 고만기, 공주대학교 토목환경공학과 교수 이인원, 한국과학기술원 건설 및 환경공학과 교수 2005 년 한국강구조학회 학술발표회 장소 : 서울산업대학교 일시 : 2005 년 6 월 4 일 2005 년 한국강구조학회 학술발표회 장소 : 서울산업대학교 일시 : 2005 년 6 월 4 일 MR 댐퍼가 설치된 스마트 면진 구조물의 반능동 진동제어 알고리즘 비교 평가

2. Dynamics & Smart Structures Lab., Sejong Univ., Korea 2 Contents Introduction Benchmark Base Isolated Building MR Damper-based Control Systems Control Algorithms Numerical Simulation Results C onclusions

3. Dynamics & Smart Structures Lab., Sejong Univ., Korea 3 Introduction Base isolation systems, such as elastomeric, friction, and lead-rubber bearing systems, have been accepted as an effective means for seismic protection of building structures. Base isolation systems can reduce inter-story drifts and floor accelerations, whereas base displacements in those systems may be increased.  expensive loss of space for seismic gap Hybrid-type base isolation systems employing additional active control devices have been studied to limit base drift.

4. Dynamics & Smart Structures Lab., Sejong Univ., Korea 4 Because of its adaptability and reliability, an MR damper- based hybrid-type base isolation system could solve the large base drift problem of the passive-type base isolation system. To systematically compare the effectiveness of control systems for base isolated buildings, the benchmark study developed by Narasimhan et al. (2003, 2004) based on input from the ASCE structural control committee. In the benchmark problem, three different kinds of base isolation systems, such as linear elastomeric with low damping, frictional systems, and bilinear or nonlinear elastomeric systems, are considered.

5. Dynamics & Smart Structures Lab., Sejong Univ., Korea 5 To verify the effectiveness of the MR damper-based control systems considering some control algorithms, such as modified clipped-optimal control, maximum energy dissipation, modulated homogeneous friction algorithms, fuzzy control and neural network-based control, for seismic protection of base isolation system. Objective of This Study

6. Dynamics & Smart Structures Lab., Sejong Univ., Korea 6 Benchmark Structure an eight-story base isolated steel-braced framed building length: 82.4m width: 54.3m similar to existing buildings in LA, California Benchmark Base Isolated Building

7. Dynamics & Smart Structures Lab., Sejong Univ., Korea 7 Linear elastomeric isolation system: 92 low damping elastomeric bearings Base Isolation Systems Considered

8. Dynamics & Smart Structures Lab., Sejong Univ., Korea 8

9. 9 Supplemental active or semiactive control devices: 16 active or semiactive control devices at the isolation level (8 in X- and 8 in Y-direction)

10. Dynamics & Smart Structures Lab., Sejong Univ., Korea 10 Control Diagram for MR Damper-based System MR Damper-based Control System Controller (Control Algorithm) MR Damper Base Isolated Building Structure

11. Dynamics & Smart Structures Lab., Sejong Univ., Korea 11 MR Damper Model - The damper is modeled using a spring, a dash pot and hysteretic element in parallel. MR damper model Force displacement relationship of MR damper

12. Dynamics & Smart Structures Lab., Sejong Univ., Korea 12 Original Clipped-Optimal Control Algorithm Control Algorithms Optimal Control (LQG) MR Damper Base Isolated Building Structure Clipped Algorithm (0 or V max ) fcfc f MR v=0 v=V max

13. Dynamics & Smart Structures Lab., Sejong Univ., Korea 13 Modified Clipped-Optimal Control Algorithms Optimal Control (LQG) MR Damper Base Isolated Building Structure Clipped Algorithm (V c ) Modified version proposed by Yoshida and Dyke (2004) fcfc f MR v=0 v=  f c f c - f MR =0

14. Dynamics & Smart Structures Lab., Sejong Univ., Korea 14 Optimal Control (LQG) MR Damper Base Isolated Building Structure Clipped Algorithm (V c ) Another modified version proposed in this study fcfc f MR v=0 v=  f c f c - f MR =0 v=V max f c -  f MR =0

15. Dynamics & Smart Structures Lab., Sejong Univ., Korea 15 Maximum Energy Dissipation Algorithm This algorithm considers a Lyapunov function that represents the relative energy in the structure (Jansen and Dyke, 2000). MR Damper Base Isolated Building Structure Clipped Algorithm (0 or V max )

16. Dynamics & Smart Structures Lab., Sejong Univ., Korea 16 Modulated Homogeneous Friction Algorithm This algorithm originally developed for use with variable friction devices was modified for MR dampers (Jansen and Dyke, 2000). MR Damper Base Isolated Building Structure Clipped Algorithm (0 or V max ), in which, the most recent local extrema in the deformation of the MR damper

17. Dynamics & Smart Structures Lab., Sejong Univ., Korea 17 Fuzzy Control Base Isolated Building Structure Fuzzy Controller MR Damper Design of fuzzy controller - Fuzzy input : base displacement and base velocity ( ) - Fuzzy output : desired command voltage ( ) - Membership functions NLNSZEPSPL -YY0 NEZEPO X0 -X Input Output

18. Dynamics & Smart Structures Lab., Sejong Univ., Korea 18 Neural Network-based Control Active neuro-controller (Kim et al. 2000, 2001) - New training algorithm using cost function - Sensitivity evaluation algorithm Structure of neural network - Input layer: 5 nodes (base displacement/acceleration, ground acceleration acceleration at 8 th floor, acceleration at device location) - Hidden layer: 8 nodes - Output layer: 1 node (desired control force) Neural Network Controller MR Damper Base Isolated Building Structure Clipped Algorithm (0 or V max )

19. Dynamics & Smart Structures Lab., Sejong Univ., Korea 19 Numerical Simulation Results Control Algorithms Considered - Original modified clipped-optimal (OCO) - Modified clipped-optimal by Yoshida and Dyke (MCO-1) - Modified clipped-optimal proposed herein (MCO-2) - Maximum energy dissipation (MED) - Modulated homogeneous friction (MHF) - Fuzzy control (FC) - Neural network-based control (NNC) Base Isolation Systems Considered - Linear elastomeric isolation system

20. Dynamics & Smart Structures Lab., Sejong Univ., Korea 20 Evaluation Criteria (Narasimhan et al., 2003) - J 1 : normalized peak base shear - J 2 : normalized peak structure shear - J 3 : normalized peak base displ. or isolator deformation - J 4 : normalized peak inter-story drift - J 5 : normalized peak absolute floor acceleration - J 6 : normalized peak force generated by all control devices - J 7 : normalized RMS base displacement - J 8 : normalized RMS absolute floor acceleration - J 9 : normalized total energy absorbed by all control devices

21. Dynamics & Smart Structures Lab., Sejong Univ., Korea 21 Earthquakes Used (Narasimhan et al., 2003) Both the fault-normal (FN) and the fault-parallel (FP) components of - Newhall record in Northridge earthquake - Sylmar record in Northridge earthquake - El Centro record in Imperial Valley earthquake - Rinaldi record in Northridge earthquake - Kobe record in Hogoken Nanbu earthquake - Jiji068 record in Jiji earthquake - Erzinkan record in Erzinkan earthquake

22. Dynamics & Smart Structures Lab., Sejong Univ., Korea 22

23. Dynamics & Smart Structures Lab., Sejong Univ., Korea 23 Time history responses (in the fuzzy control case under the El Centro record)

24. Dynamics & Smart Structures Lab., Sejong Univ., Korea 24 EarthquakesJ1J1 J2J2 J3J3 J4J4 J5J5 Sylmar OCO0.900.910.730.871.16 MCO-10.950.960.900.931.00 MCO-20.930.940.880.900.95 MED0.97 0.920.930.96 MHF0.93 0.920.940.97 FC0.96 0.920.940.97 NNC0.830.850.760.750.87 El Centro OCO1.251.240.541.261.61 MCO-10.950.930.780.830.85 MCO-20.940.930.640.840.87 MED0.890.860.640.762.21 MHF0.950.930.670.820.86 FC0.950.920.620.830.90 NNC0.74 0.360.661.55 Control performance of several control algorithms

25. Dynamics & Smart Structures Lab., Sejong Univ., Korea 25 EarthquakesJ1J1 J2J2 J3J3 J4J4 J5J5 Kobe OCO1.041.030.521.001.63 MCO-10.86 0.780.890.97 MCO-20.820.810.690.800.97 MED0.880.860.690.861.54 MHF0.830.820.760.830.97 FC0.880.870.630.831.02 NNC0.700.720.460.721.58 Jiji OCO0.84 0.650.860.87 MCO-10.92 0.840.93 MCO-20.920.910.830.910.93 MED0.96 0.920.95 MHF0.93 0.900.930.94 FC0.91 0.860.910.93 NNC0.90 0.800.920.95 Control performance of several control algorithms

26. Dynamics & Smart Structures Lab., Sejong Univ., Korea 26 Conclusions Some control algorithms, such as MCOs, MED, MHF, FC and NNC, are considered to verify the effectiveness of MR damper-based control systems for seismic protection of a base isolated building. Performance of MCO-2 is slightly better than that of other clipped-optimal control algorithms (i.e., OCO, MCO-1). The neural network-based control could be considered as one promising candidate for the linear benchmark base isolated systems.

27. Dynamics & Smart Structures Lab., Sejong Univ., Korea 27 Thank you for your attention!