You are on page 1of 9

SEISMIC HAZARD MITIGATION OF MULTISTORIED BUILDINGS USING SEMI-ACTIVE MR DAMPER

Arnav Anuj Kasar, Anand Paul, S. D. Bharti 1, M. K. Shrimali 2


Department of Civil Engineering, Malaviya National Institute of Technology Jaipur, Jaipur 302017, Rajasthan

ABSTRACT
Earthquakes are the most devastating catastrophic activity. Earthquakes, affect the structures, thereby causing serious threat to life and propertied housed in the structure. The semi-active control has emerged as a very attractive proposition of structural control in past decade. In this study the effectiveness of Magneto-Rheological (MR) damper, which is one of the most effective semi-active control device has been studied. Then, an Artificial Neural Network (ANN) has been developed to predict the response of the structure considering the parameters of two different ground motions and the parameters of the structure. A simplified lumped mass model was subjected to unidirectional excitation by real earthquake ground motions and the seismic response of building model was obtained by solving governing equations of motion using Newmarks stepby-step method. Further, Artificial Neural Networks (ANN) was developed and trained to obtain the responses of the structure, when subjected to a unidirectional ground motion. It was found that the ANN is quite effective in predicting the result.

INTRODUCTION In past devastating earthquakes around the world have underscored the tremendous importance of understanding the way in which civil engineering structurerespond to natural events of earthquakes, strong winds etc. The strong earthquakes in last two decades includes Loma Prieta, San Francisco Bay (1994), Northridge, California (1994), Kobe, Japan (1995), Kocaeli, Turkey (1999), Chi-Chi, Taiwan (1999), Bhuj, India (2001) and Haiti, (2010), were destructive and caused large scale damage to civil structures and loss of human lives and properties. It is of utmost concern to protect structures from significant damage/failure, under such catastrophic natural events. Semi-active control systems have emerged as very attractive proposition for seismic response reduction of structures. These devices are characterized by their ability to dynamically change their properties, without adding energy to the controlled system. Semi-active control strategies are dissipative in nature, inherently stable, and require a
1 2

Associate Professor Associate Professor

little energy to operate (Spencer and Sain, 1997). Studies have shown that appropriately implemented semi-active damping system perform significantly better than passive devices and have the potential to achieve, or even surpass, the performance of fully active system, thus allowing for the possibility of effective response reduction during a wide array of dynamic loading conditions (Spencer and Sain, 1997). Magnetorheological (MR) dampers MR damper typically consists of a hydraulic cylinder containing micron-sized, magnetically polarizable particles suspended within a fluid (usually oil). MR fluid behavior is controlled by subjecting the fluid to a magnetic field in the absence of a magnetic field, the MR fluid flows freely while in the presence of a magnetic field the fluid behaves as a semi-solid. Structural Control using Neural Networks Artificial Neural Networks is a relatively new technique which is a generalization of mathematical model of biological nervous system.In recent years, Artificial Neural Networks (ANN) has shown wide applications in the field of Structural Engineering. The learning capability of Neural Networks along with their nonlinearity has appealed the structural engineers, as with such properties Neural Networks can be a promising tool for the control of Structures. REVIEW OF LITERATURE For the last three decades, the reduction of structural response caused by dynamic effect has become a subject of intensive research. Many structural control concepts have been evolved for this purpose and quite a few of them have been implemented in practice. Dyke et al. (1996) proposed a clipped-optimal control strategy based on acceleration feedback for controlling MR dampers to reduce the structural responses due to seismic loads. The effectiveness of proposed control algorithm and the usefulness of MR dampers for structural response reduction were demonstrated through a numerical example of three story model structure. Spencer et al. (1997) proposed a model to predict the dynamic behavior of MR damper, referred as phenomenological model for MR damper, that can effectively portrays the behavior of a typical MR damper overcoming the drawbacks of previous models. Xu et al. (2000) suggested two optimal displacement control strategies for semi-active control of earthquake excited structures using ER or MR dampers and compared those with the optimal force control strategy. The work on the example building showed that the installation of smart dampers with proper parameters and control strategy significantly reduced the seismic response of the building structure and the performance

was better than that of the common brace or passive device.Jansen and Dyke (2000) evaluated several semi-active control algorithms for applications in a structural control system using multiple MR dampers. Based on the results, three control algorithms, namely Lyapunov controller algorithm, the clipped optimal algorithm and the modulated homogeneous friction algorithm were found to be most suited for use with MR dampers in a multi-input control system. Yang et al. (2002) designed and developed a 20 ton MR damper capable of providing semi-active damping for structural applications. The authors developed and used a mechanical model based on Bouc-Wen hysteresis model to dynamic response analysis of the MR damper.Kurata et al. (2000) studied the effectiveness of the semi-active structural control technique in high rise buildings. It was shown that the semi active structural control technique was effective in achieving high performance.Sahasrabudhe et al. (2005) investigated the performance of smart sliding isolation system that is, sliding isolation combined with semi-active MR dampers, experimentally and analytical under near-field ground motion. Authors found that the MR damper reduces bearing displacement, maintaining the isolation forces less than the passive high-damping case. SCOPE AND OBJECTIVES OF THE STUDY Magneto-rheological (MR) damper has emerged as one of the most attractive semiactive devices that are capable of generating the magnitude of force necessary for fullscale application to civil engineering structures. Moreover the device is operated with very low power and offers highly reliable operation and its operation is relatively insensitive to temperature fluctuation or impurities in the fluid. The effectiveness of MR damper as seismic response control device is investigated taking numerical example of a 25-story realistic RC building modeled as linear shear building subjected to unidirectional excitation and the performance of MR damper with an un-damped system was compared. Structural Model The building is idealized as a linear shear type building with lateral degrees of freedom at floor levels. The system is assumed to remain in linear elastic state and hence does not yield under excitation. It is considered that the system is subjected to unidirectional excitation and spatial variation of ground motion and any effect due to soil structure interaction is neglected. The lateral resistance of the buildings is assumed to be so large that it does not affect the dampers performance adversely. The structure model of the building with MR damper is shown in Figure 1 (Left). The governing equations of motion of the multi degree of freedom (MDOF) system are written as:

[ M ]{u}+[C ]{u}+[ K ]{u}= [ D ]{ f m } - [ M ][ r ]{ug }


where, M, C, and K are the mass, damping and stiffness matrices of the system, respectively; f m is the MR damper force vector, respectively; D is the damper location matrix; u is the displacement vector with respect to the ground; r is a influence coefficient vector; and ug is the earthquake ground acceleration.
mn ffn mnffn-1 m ff4 m ff3 m2 ff2 m1 ff1 k4 k3 kn kn-1

xd k0 c1 c0

F Bouc-Wen

k2 k1

k1 Figure 1: (Left) Structural Model of the building with MR Damper (Right) Modified Bouc-Wen Model of MR Damper (Spencer et al 1997)

Computation of MR Damper Force In order to compute response of the MDOF system with MR damper, an appropriate control algorithm and modeling of dynamic behavior of MR damper is essential.The governing Eq.(1) can be expressed in the state-space form as below

u {Z } = [ A]{Z }+[ B ]{ f }+[ E ]{ }


m/f g

where, Z is the state vector, A is the system matrix; B and E are the distribution matrix of the control force and the excitation, respectively. In this study, Lyapunovs direct approach is employed for the control algorithm. Leitmann (1994) applied Lyapunovs direct approach for the design of semi-active controller. In this approach, a Lyapunov function is chosen of the form

L ({Z }) =

1 {Z } 2

2 p

where

{Z } p

For predicting the MR damper force accurately there are several models which have been used. (Wen, 1976; Stanway et al. 1987; Spencer et al.1997). In this study, modified Bouc-Wen model (Spencer et al. 1997) is used (Fig. 1 (Right)). The equation governing the force predicted by this model is

f m = x + k1 (ud x0 ) c1
Where, the evolutionary variable z is given by

z = vd x ( z ) z

( n 1)

(vd x) z + Am (vd x)

Solution Procedure Input voltage to MR damper is given by the filter equation and damper force is obtained from Eq.(4). Thereafter, the governing equations of motion Eq.(1) are solved using Newmarks step-by step method assuming linear variation of acceleration over a small time interval t, the Newmarks algorithm used. Neural Network The next step was to generate Artificial Neural Networks which were then trained using a fraction of results generated by the Lyapunov Controller. Once trained, the Neural Networks were used to predict the responses of the structure, using the remaining Data Set, and the result generated was then compared with the results generated using the Lyapunov control strategy. A plot, comparing the results obtained was then prepared, showing the effectiveness of Neural Network as a structural control tool. (Figure 5) NUMERICAL STUDY A lumped mass structural model of a twenty five story RC frame is considered with each floor mass and stiffness as 130 ton and 150000kN/m, respectively and MR dampers

rigidly connected between floors. The building is subjected to unidirectional excitation for which four real earthquake ground motions are considered. The displacement and acceleration response spectra for the four considered ground motion for 5% of critical damping are shown in (Figures 2). The response parameters of interest for the study are: peak top floor displacement, peak normalized story shear force (V x /W), and peak normalized base shear (B x /W). Here, story shear force (V x ) and base shear (B x ) are normalized with the weight of the building (W). Response Control with MR Damper The effectiveness of MR damper is compared with the response of undamped system. Under Controlledstrategythe damper behaves as semi-active device and thecommand voltage is governed by the control law. The parameters of MR damper (Yang et al. 2002) have been suitably scaled up to suit the damper deformation. The performance of MR is investigated by providing dampers at all floors. This parametric study is conducted with command voltage of 6V.A comparison of uncontrolled and controlled time response of top floor displacement, acceleration and normalized base shear is shown in (Figures 3). The figures indicate effectiveness of MR dampers for seismic response mitigation. It is observed that in overall analysis, Controlled strategy is quite effective in reducing the displacement response.
60

80

Spectral Displacement (cm)

Spectral Displacement (cm)

60

40

40

20

20

0 0 3.0 2.5 1 2 3 4 5

0 0 3.0 2.5 1 2 3 4 5

Pseudo Acceleration (g)


0 1 2 3 4 5

Pseudo Acceleration (g)

2.0 1.5 1.0 0.5 0.0 Period (second)

2.0 1.5 1.0 0.5 0.0 0 1 2 3 4 5 Period (second)

Fi 1 R

K b N00E(1995) D

ti 5 %

Figure 2: Response Spectrum, Damping Ratio: 5% (Left) Kobe N00E (1995)(Right) Northridge N00E (1994)

60 40 20 0 -20 -40 -60 0 2 15 30 45 60 75 90

Tf_disp (cm)

Tf_disp (cm)

Uncontrolled Controlled

150 75 0 -75 -150 -5 2


Tf_acc (g)

Uncontrolled Controlled

105 120 135 150

5 10 15 20 25 30 35 40 45 50 55 60 65

T (sec) Uncontrolled Controlled

T (sec)

Tf_acc (g)

1 0 -1 -2 0 20 40 60 80 100 120 140

1 0 -1 -2 -5 0

Uncontrolled Controlled

160

5 10 15 20 25 30 35 40 45 50 55 60 65

T (sec)
Base Shear (kN)

0.3 0.2 0.1 0.0 -0.1 -0.2 0 20 40 60 80 100 120 140

Uncontrolled Controlled

Base Shear (kN)

0.4 0.2 0.0 -0.2 -0.4 -5 0

T (sec) Uncontrolled Controlled

160

5 10 15 20 25 30 35 40 45 50 55 60 65

T (sec)

T (sec)

Figure 3 : Time based responses of the structure. Ground Motion: Kobe (left) Northridge (Right)

Training of the Neural Network The development and training of Neural Net has been carried out in Palisade Decision Making tool. The results obtained from analysis of the model, along with the ground motion parameters (i.e. Time and Ground Acceleration) and the Structural parameters (i.e. Mass, Stiffness and Damping Ratio of the structure) were used to train a network, to develop the Neural Controller. Only a few values from the available data set were supplied to train the network, and the trained network is then tested on the remaining values in the data set. This procedure has been carried out for both the ground motions for controlled as well as uncontrolled conditions.
Conventional Uncontrolled Neural Uncontrolled
Tf_disp (cm)

60
Tf_disp (cm)

40 20 0 -20 -40 0 60 40
Tf_disp (cm)

Time (s)

20

40

150 120 90 60 30 0 -30 -60 -90 -120 -150 0 90 72 54 36 18 0 -18 -36 -54 -72 -90 0 10 20 30 40 20 40

Conventional Uncontrolled Neural Uncontrolled

60

Time (sec)

Conventional Controlled Neural Controlled


Tf_disp (cm)

Conventional Controlled Neural Controlled

20 0 -20 -40 0 10 20 30 40 50

50

60

Time (s)

Time (sec)

Figure 4 : Comparison of Results, Ground Motion: Kobe (Left) Northridge (Right)

RESULTS AND DISCUSSIONS The effectiveness of semi-active dampers (MR) in response reduction for individual building frame has been investigated. It is observed from the results of numerical study that the dampers are quite effective in response control for ground motion. The damper parameters must be appropriately chosen in order to attain best control performance. Following precise conclusions were made from the study; there exists an optimum value of command voltage for overall best control performance for a wide range of ground motion; there exists an optimum value of parameter R f for overall best control performance for a wide range of ground motion. Another important interpretation from the study was that the Artificial Neural Networks has great potential to be used for structural control. If trained properly, they simulate nearly the same results as generated by the proper conventional design approach Figure 4. REFERENCES 1. Jangid, R. S., and Datta, T. K. (1992). Seismic behaviour of torsionally coupled base isolated structure. European Earthquake Engineering (Italy) 6, 2-13. Soong, T.T., and Spencer Jr, B.F. (2002). Supplemental energy dissipation: state-of-the-art and state-of-the-practice. Engineering Structures, 24, 243-259. Shrimali, M. K., and Jangid, R. S. (2002). Non-linear seismic response of base-isolated liquid storage tanks to bi-directional excitation. Nuclear Engineering and Design, 217, 1-20. Spencer Jr, B. F., and Nagarajaiah, S. (2003). State of the art of structural control. Journal of structural Engineering,ASCE, 129(7), 845-856. Spencer Jr, B. F., Dyke, S. J., Sain, M. K., and Carlson, J. D. (1997). Phenomenological model for magnetorheological dampers. Journal of Engineering Mechanics, ASCE, 123(3), 230-238 Ivers, D. E., and Miller L. R. (1991). Semi-active suspension technology: An evolutionary view. Advance Automotive Technologies, ASME, 40, 327-346. Dyke, S. J., Spencer Jr, B. F., Sain, M. K., and Carlson, J. D. (1996a). Modeling and control of magnetorheological dampers for seismic response reduction. Smart Materials and Structures, 5, 565-575. Dyke, S. J., Spencer Jr, B. F., Quast, P., Sain, M. K., KaspariJr, D. C., and Soong, T. T. (1996b). Acceleration feedback control of MDOF structures. Journal of Engineering Mechanics, ASCE, 122(9), 907-918. Dyke. S. J., and Spencer Jr, B. F. (1996). Seismic response control using multiple MR dampers. Proceedings of the 2nd International Workshop on Structural Control.

2. 3.

4. 5.

6. 7.

8.

9.

10. Dyke, S. J., and Spencer Jr, B. F. (1997). A comparison of semi-active control strategies for the MR damper. Proceedings of the IASTED International Conference, on Intelligent Information Systems, the Bahamas, 580-584. 11. Xu, Y. L., Qu, W. L., and Ko, J. M. (2000). Seismic response control of frame structures using magnetorheological/electrorheological dampers. Earthquake Engineering and Structural Dynamics, 29, 557-575. 12. Jansen, L. M., and Dyke, S. J. (2000). Semi-active control strategies for MR dampers: comparative study. Journal of Engineering Mechanics, ASCE, 126(8), 795-803. 13. Yang, G., Spencer Jr, B. F., Carlson, J. D., and Sain, M. K. (2002). Largescale MR fluid dampers: modeling and dynamic performance considerations. Engineering Structures, 24 , 309-323. 14. Yang, G., Spencer Jr, B. F., Carlson, J. D., and Sain, M. K. (2002). Largescale MR fluid dampers: modeling and dynamic performance considerations. Engineering Structures, 24 , 309-323. 15. Sahasrabudhe, S. S., and Nagarajaiah, S. (2005). Semi-active control of sliding isolated bridges using MR dampers: an experimental numerical study. Earthquake Engineering and Structural Dynamics, 34, 965-983.

You might also like