You are on page 1of 9

Systemic Analysis

http://www.intersections.ro

Analysis of Active Seismic Response for Controlled Structures Using Stochastic Models
Fideliu Păuleţ-Crăiniceanu
Dept. of Struct. Mechanics, “Gh. Asachi” Tech. Univ., Iasi, 700050, Romania, fideliu@ce.tuiasi.ro

Summary
This paper investigates the seismic response of active controlled structures using a stochastic approach. The study is at a starting point for the author’s direction of research and it is intended to focus on principles and methodologies for simple structures and simple control strategies. The research is envisioning validating stochastic domain strategies for controlled structures that have been already validated in time and frequency domains. In a relative recent past the author has been studied structural active control. Especially energy base optimal active control was used as the strategy of control. This strategy was used for large structures (bridges and buildings). Variants for full state and reduced order controllers were tested. Other variants were used for taking into account the noise and time lag. Overlapping and disjoint distributed controllers have been in the views, too. Time-history and frequency analysis were done. For showing the validity of the methodology, internationally proposed benchmark structures and conditions have been employed. A stochastic approach for the seismic response of structures with control is proposed. The seismic action is described as a system with a white noise input and a seismic-like time-history response. A non-negative envelope function is modeling the input stationary random process. Application of the methodology on very simple structures is in the views in order to begin a longer term investigation in the field. The main results show the differences between the controlled and the non-controlled cases. From the stochastic point of view, results show that the controlled structures behave very well. It is confirmed that this type of analysis is very useful in judging a control methodology together with other analysis as time-history response and/or frequency response analysis. KEYWORDS: stochastic response, passive control, seismic action, structural model.

ISSN 1582-3024

Article No.4, Intersections/Intersecţii, Vol.5, 2008, No.4, “Systematic Analysis”

40

ro Fideliu Păuleţ-Crăiniceanu 1.Systemic Analysis http://www. Variants for full state and reduced order controllers were tested. In this case. “Systemic Analysis” 41 . the frequency response principle is shown in Figure 1.e. The study is at a starting point for the author’s direction of research and it is intended to focus on principles and methodologies for simple structures and simple control strategies. This paper investigates the seismic response of active controlled structures using a stochastic approach. too.4.5. FREQUENCY RESPONSE The frequency response can be seen as the degree of amplification of the input or the ratio between the output and the input of a structure under harmonic excitation. Vol. This strategy was used for large structures (bridges and buildings). The frequency response principle Graphically. Especially energy base optimal active control was used as the strategy of control. 2008. Overlapping and disjoint distributed controllers have been in the views. No. 2. a complex harmonic function with amplitude equal to unity and with a circular frequency ω . Other variants were used for taking into account the noise and time lag.intersections.4. ISSN 1582-3024 Article No. Intersections/Intersecţii. internationally proposed benchmark structures and conditions have been employed. For showing the validity of the methodology. the excitation is usually written: f (t ) = e iwt (1) i. e iωt SYSTEM G(iω )e iωt Figure 1. Time-history and frequency analysis were done. INTRODUCTION Structural active control is seen as a direction of development for mitigating the effects of actions as natural disasters against civil engineering constructions. The research is envisioning validating stochastic domain strategies for controlled structures that have been already validated in time and frequency domains. The author has been studied this broad field in a relative recent past [1-5].

for this kind of action. It is obvious that. MODELING THE SEISMIC ACTION Artificial accelerograms are extremely useful in structural analysis because of the possibility to have a clearer image of the action and of the response of a structure under the earthquake excitation.5 1 Period (sec) 1. in Figure 4. 50 45 40 35 Accel. Acceleration amplification 3. 2008. Figure 5 shows a general concept in time-history analysis using generated accelerograms.Systemic Analysis http://www.ro Analysis of Active Seismic Response for Controlled Structures Using Stochastic Models ua c2 k2 c1 k1 m2 m1 & 2 && x2 x x2 & 1 && x1 x x1 Figure 3. ampl. the passive control is very effective. “Systemic Analysis” 42 .5. ISSN 1582-3024 Article No. 30 25 20 15 10 5 0 0 0.4. Active Tuned Mass Damperg && x As an example. Vol. Intersections/Intersecţii.4.intersections. No. the effect of the TMD over the acceleration response in frequency domain for the problem shown Figure 3 is presented.5 2 Figure 4.

8 Envelope Function 0.1 5 10 15 Time (s) 20 25 30 0 0 5 10 Time (s) 15 20 Figure 6. “Systemic Analysis” 43 .3 0.2 0 0 0.Systemic Analysis http://www.4 0. Time-history response using generated accelerogram In mathematical terms. [6.c (t -t 2 ) ⎪ ⎩e t<0 0 ≤ t ≤ t1 t1 ≤ t ≤ t2 t > t2 (4) ISSN 1582-3024 Article No. Vol.6 0. 2008. Envelope function Based on statistical observations there were proposed many formulae for the envelope function.4.25 and b=0.7]. One example is expressed by Equation (3): ψ (t ) = e.4.2 0. Intersections/Intersecţii. as it is given in Equation (2).9 0. Another very used envelope function is that from Equation (4): ⎧ 0 ⎪ ⎛ t ⎞2 ⎪ ⎪⎜ ⎟ ⎟ ψ (t ) = ⎨ ⎜ ⎝ t1 ⎠ ⎪ 1 ⎪ .5.5 0.at .at .6 0. No.ro Fideliu Păuleţ-Crăiniceanu White Noise Ground Model (Kalman filter) Simulated Accelerogram Structural Model Time-history Response Figure 5. 1 1 Envelope Function 0.e-bt max(e. [6]: & &g (t ) = ψ (t ) & &(t ) x x (2) &(t ) is a where ψ (t ) is a deterministic nonnegative envelope function and & x stationary random process.e-bt ) (3) Letting a=.5. this function is drawn in Figure 6.intersections.7 0.4 0. Envelope function Figure 7. the earthquake acting the structure is modeled as a nonstationary random process.8 0.

26. psifc. w &&(t ) Figure 8. for i=1:np if t(i)<t1.intersections.csp. from Equation (2).4. it is very practical and gives good results in applications to structural control problems.np). w ωg = ςg = kg mg cg is the ground circular frequency.tf. Intersections/Intersecţii. [7]. The equation describing this model is (5): 2 2 & &(t ) + 2ς g ω g x & (t ) + ω g & (t ) + ω g x x(t ) = 2ς g ω g w w(t ) (5) where: & &(t ) is the absolute acceleration of the mass. Kanai-Tajimi model for the ground &(t ) . “Systemic Analysis” 44 . c=0. 2m g ω g ISSN 1582-3024 Article No. this function is plotted in Figure 7. t2=13. end end end Acceleration output. Even if this model is one of the simplest models. psi(i)=1. No. psi(i)=(t(i)/t1)^2. is shown below: function [psi. x &&(t ) is the acceleration of the base (white noise). is the ground damping coefficient.4. Vol.t2. & x widely accepted the Kanai-Tajimi model for the ground. else if t(i)<t2.Systemic Analysis http://www. else psi(i)=exp(-csp*(t(i)-t2)). x &&(t ) mg kg cg Broad band acceleration input.ro Analysis of Active Seismic Response for Controlled Structures Using Stochastic Models and. it is In order to generate the stationary random process.tf.t]=psifc(t1. for t1=3. a MATLAB function.5. For computing this. 2008. shown in Figure 8. t=linspace(0.np).

“Systemic Analysis” 45 .65 . can be also seen as a Kalman filter and it can be easily employed using a state equation (please see the next section).ω 2 ) 2 + (2ς gω gω ) 2 ⎛ 2 1 + 4ς g 2 Φ (ω ) = ω2 2 ωg 2 2 ⎞ 2 ω ⎜1 − ω ⎟ + 4ς g 2 2 ⎟ ⎜ ωg ωg ⎝ ⎠ S (7) with S being the intensity of the white noise.19): ISSN 1582-3024 Article No.4 1. Function from Equation (7) Figure 9 shows the power spectral density function described by Equation (7).2 Φ 1 0.5. 4. 1. S is actually the power spectral density of the white noise. No.4 0.Systemic Analysis http://www. Note that the model shown in this section. when S = 1m 2 /sec 4 and ς g = 0. 2008. Intersections/Intersecţii. Vol.8 0.2 0 0 20 40 60 80 100 120 (ω/ωg)2 Figure 9.4.4. Equation (5).6 0.intersections.ω + 2iς gω gω (6) and the power spectral density is shown by Equation (7): 4 + ( 2ς g ω gω ) 2 ωg S= 2 (ω g .6 1. STOCHASTIC RESPONSE The Equations (2) and (5) can be transformed in a state form (3.8 1.ro Fideliu Păuleţ-Crăiniceanu For the system (5) the frequency response function is (6) 2 ωg + 2iς g ω g ω H (ω ) = 2 2 ω g .

B g = ⎢ ⎥.ω ⎤ ⎡0 ⎤ 2 . Vol. [9.10]. Be = ⎢ ⎥ .ro Analysis of Active Seismic Response for Controlled Structures Using Stochastic Models & 1 (t ) = A g x1 (t ) + B g w(t ) ⎧x ⎨ && ⎩ x g (t ) = ψ (t )C g x1 (t ) ⎧ x(t ) . Intersections/Intersecţii. supposing the structure’s equation of motion is (9) & s (t ) = Ax s (t ) + B& & g (t ) x x (9) then the whole system (earthquake acceleration generation plus structure) will be written in a new state equation (10) & (t ) = A e (t )y (t ) + B e w(t ) y (10) ⎧ x s (t ) ⎫ because a new state vector y (t ) = ⎨ ⎬ have been introduced and. ⎩ x g (t )⎭ ⎡ A ψ (t )BC g ⎤ ⎡O⎤ A e (t ) = ⎢ . therefore. The covariance.2ς g ω g ] Above. Also it should be observed that the system matrix is time dependent. the state equation for the action was written. and C g = . ⎥ Ag ⎦ ⎣O ⎣B g ⎦ Looking to the equation (10).intersections. Now. “Systemic Analysis” 46 . Stochastic response principle Using the Equation (10).ω g ⎥ . matrix of the response is the solution of the Riccati equation (11): & (t ) = A (t )R (t ) + R (t ) A ′ (t ) + B WB ′ R e e e e (11) ISSN 1582-3024 Article No.w & (t ) ⎭ ⎩x ⎡ 0 Ag = ⎢ 2 ⎣.2ς g ω g ⎦ 1 ⎣ ⎦ 1 (8) [ . No.4.Systemic Analysis http://www. one can compute the stochastic response shown in the Figure 10.w(t )⎫ where: x 1 (t ) = ⎨ ⎬ is the state vector and & (t ) .5. 2008. time dependent. White Noise Intensity Riccati Covariance Equation Time-history Stochastic Response Ground Model (Kalman filter) Structural Model Figure 10.4. it is seen that the input is the white noise.

Vol. The matrix R(t) is defined as follows. b=0.m y )′] = E[y (t )y′(t )] (12) where E[.4. Equation (12).25. EXAMPLE For the same example of a TMD from Figure 3. 5. No.2 . From the stochastic point of view. 6.4. and S=9cm2/sec4 was chosen. “Systemic Analysis” 47 .5. Intersections/Intersecţii. It is confirmed that this type of analysis is very useful in judging ISSN 1582-3024 Article No.5. Displacement covariance Figure 11 presents a comparison between the covariance of the displacement in the case with TMD and in the case in which the TMD is not employed. the ground coefficients ω g = 6. the envelope function (3) with a=0.ro Fideliu Păuleţ-Crăiniceanu where the prime sign means the transpose of a matrix and W=2πS.28 . A non-negative envelope function is modeling the input stationary random process. ς g = 0.m y )(y (t ) .Systemic Analysis http://www.intersections. The seismic action is described as a system with a white noise input and a seismic-like time-history response. 2008.] means expectation and my=E[y(t)]=0 is the mean value of y. results show that the controlled structures behave very well. The effectiveness of the passive control is seen again. 15 Expectation of x*x (cm*cm) 10 5 0 0 5 10 15 Time (s) 20 25 30 Figure 11. CONCLUSIONS A stochastic approach for the seismic response of structures with control is proposed. [11]: R (t ) = E[(y (t ) .

IEEE Trans CST. ICAPV2000. F.intersections. No. A. pp. The main results show the differences between the controlled and the non-controlled cases.: A Statistical Method of Determining the Maximum Response of A Building Structure During an Earthquake. Y. Rodellar. Iaşi. Bryson. Proceedings of the International Conference on Advanced Problems in Vibration Theory and Applications. Păuleţ-Crăiniceanu. Memoriile Secţiilor Ştiinţifice. pp.. ISSN 10636536. Xi’an.. Estimation and Control. Washington DC. Proceedings of the Second World Conference on Earthquake Engineering. Yokohama.K. Bucureşti. Bakule. Vol. 1960 9. June 19-22. 2ECSC.. Masri S. vol. ENPC. Journal of Engineering Mechanics Division. Wolfe R. Application of the methodology on very simple structures is in the views in order to begin a longer term investigation in the field. Y.C. No. Caffrey J. Vol. Shinozuka. H.: An Experimental Study of an Adaptive Momentum Exchange Device for Structural Control Applications.663-669 5. September. 1979 11. Ho. Vol. F.13.: Probabilistic Theory of Structural Dynamics. J. A.4. Editura Societăţii Academice „Matei-Teiu Botez”. Liu. EM4. 1st International Conference on Motion and Vibration Control. F.A.: Response of MDOF Systems to Nonstationary Random Excitation. Tom XXI/1998. Sato. F. July 3-6. EM1.5.M. Champs-sur-Marne. Rossell: Overlapping Reliable Control for a Cable-Stayed Bridge Benchmark. Science Press. Second European Conference on Structural Control. L. 1965 7. pp. 105. McGraw-Hill. Păuleţ-Crăiniceanu..197-215 4.: Applied Optimal Control: Optimization.: Active versus passive control of seismic response for civil engineering structures. M. Editura Academiei Române.: Hybrid Versus Tuned or Active Mass Dampers for Response Control of Tall Buildings. 1975 10. 2000. Intersections/Intersecţii. 2001. 2000 3.4. Paulet-Crainiceanu. 2008.: Structural Active Control For Buildings And Bridges In Seismic Areas.: Simulation of Nonstationary Random Processes. IEEE Transactions on Control Systems Technology. 1992 8. July 2005. Tajimi. 2003 6. M. Gasparini. D. 1967 ISSN 1582-3024 Article No. References 1. J. Păuleţ-Crăiniceanu. Păuleţ-Crăiniceanu. Kaneko. II. 147-151 2. Journal of Engineering Mechanics Division.Systemic Analysis http://www.. Y.. China.4.E.jr. Beijing.ro Analysis of Active Seismic Response for Controlled Structures Using Stochastic Models a control methodology together with other analysis as time-history response and/or frequency response analysis. F. Seria IV. Hemisphere Publishing Corporation.. 2000. Japan. France. Antiseismic Protection through Structural Active Control. “Systemic Analysis” 48 . Mita.