Stochastic Simulation of Ground Motion Components for a Specified Design Scenario Sanaz Rezaeian Armen Der Kiureghian (PI) University of California, Berkeley Sponsor: State of California through Transportation Systems Research Program of Pacific Earthquake Engineering Research (PEER) Outline: Motivation Ground motion model Extend to simulate multiple components o Principal axes of ground motions o High correlations between model parameters Example Conclusion Motivation: In seismic hazard analysis, development of design ground motions is a crucial step. High levels of intensity Expected structural behavior: Nonlinear Approach: Response-history dynamic analysis Requires: Ground motion time-series Motivation: In seismic hazard analysis, development of design ground motions is a crucial step. High levels of intensity Expected structural behavior: Nonlinear Approach: Response-history dynamic analysis Requires: Ground motion time-series Difficulties come from scarcity of previously recorded motions. Controversies come from methods of selecting and modifying real records. Alternative: Use simulated time-series in conjunction or in the place of real records. Motivation: In seismic hazard analysis, development of design ground motions is a crucial step. High levels of intensity Expected structural behavior: Nonlinear Approach: Response-history dynamic analysis Requires: Ground motion time-series Our Goal: Earthquake and site characteristics Suite of simulated time-series (F, M, Rrup, Vs30) Site Controlling Fault F: Faulting mechanism M: Moment magnitude … VS30: Shear wave velocity of top 30m R: Closest distance to ruptured area Motivation: In seismic hazard analysis, development of design ground motions is a crucial step. High levels of intensity Expected structural behavior: Nonlinear Approach: Response-history dynamic analysis Requires: Ground motion time-series Our Goal: Earthquake and site characteristics Suite of simulated time-series (F, M, Rrup, Vs30) Site Controlling Fault F: Faulting mechanism M: Moment magnitude … VS30: Shear wave velocity of top 30m R: Closest distance to ruptured area For 2D/3D structural analysis, need ground motion components. Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Acceleration time-series Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Temporal non-stationarity: Variation of intensity in time Spectral non-stationarity: Variation of frequency content in time Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Source of stochasticity Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Impulse response function (IRF) corresponding to pseudo-acceleration response of a SDOF linear oscillator Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Duhamel’s integral (superposition of filter responses to a sequence of statistically independent pulses with the time of application τ) Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter 0 time tn Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Unit-variance process with spectral non-stationarity Normalization by standard deviation Fully non-stationary process x(t ) Time modulating filter Simulated ground acceleration z(t ) High-pass filter Non-zero residuals! Over estimates response spectrum at long periods! Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Critically damped oscillator z 2 c z c2 z x(t) Ground Motion Model: [Rezaeian and Der Kiureghian, 2008] 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) White-noise w(t ) Linear filter with time-varying parameters Filtered white-noise Normalization by standard deviation Fully non-stationary process x(t ) Unit-variance process with spectral non-stationarity Time modulating filter Simulated ground acceleration z(t ) High-pass filter Model Parameters: 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) Model Parameters: 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) Modulating function parameters: Ia 2g t 2 0n (accel (t )) dt : Arias intensity D595 : Effective duration, between 5% to 95% Ia tmid : Time at the middle of strong shaking, at 45% Ia Filter parameters: ωmid : Frequency at the middle of strong shaking ω' : Rate of change of frequency over time : Damping ratio Model Parameters: 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) Modulating function parameters: Ia 2g t 2 0n (accel (t )) dt : Arias intensity D595 : Effective duration, between 5% to 95% Ia tmid : Time at the middle of strong shaking, at 45% Ia Filter parameters: ωmid : Frequency at the middle of strong shaking ω' : Rate of change of frequency over time : Damping ratio Model Parameters: 1 t x(t ) q(t, α) h[t , λ ( )]w( )d f (t ) Modulating function parameters: Ia 2g t 2 0n (accel (t )) dt Filter parameters: ωmid : Frequency at the middle of strong shaking : Arias intensity D595 : Effective duration, between 5% to 95% Ia : Time at the middle of strong shaking, at 45% Ia : Rate of change of frequency over time : Damping ratio Model parameters are identified for many recorded motions to develop predictive equations in terms of F, M, R, VS30 Acceleration, g tmid ω' 0.15 0 Recorded -0.25 0 Time, sec 40 Match statistical characteristics Representing: • Intensity • Frequency • Bandwidth Identify model parameters Ia, tmid, D5-95 ωmid, ω’ , ζ Two Horizontal Components: Component 1: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h Component 2: 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h Two Horizontal Components: Component 1: Component 2: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h w1(τ) and w2(τ) are statistically independent if along the principal axes. source of stochasticity Two Horizontal Components: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h Component 1: 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h Component 2: source of stochasticity w1(τ) and w2(τ) are statistically independent if along the principal axes. Principal Axes of Ground Motion: A set of orthogonal axes along which the components are uncorrelated. Minor Expected Epicenter Site Major Horizontal Plane Intermediate Penzien and Watabe (1975) Two Horizontal Components: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h Component 1: 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h Component 2: source of stochasticity w1(τ) and w2(τ) are statistically independent if along the principal axes. Principal Axes of Ground Motion: A set of orthogonal axes along which the components are uncorrelated. Rotate recorded motions in the database. ρ a ,a ≠0 ρ a ,a =0 1 1,θ a2,θ θ 2 2,θ Site a1,θ a1 a2 Horizontal Plane Rotating Recorded Motions: Correlation Coefficient Between The Two Components Northridge earthquake recorded at Mt. Wilson Station 0.5 (55,0) 0 (0,-0.42) -0.5 0 10 20 30 40 50 60 70 80 90 Rotation Angle, degrees 0.3 0.2 As-Recorded Component 1 0.1 Acceleration, g 0.3 0.2 0 -0.1 -0.2 -0.3 0 -0.1 -0.2 -0.3 0.3 0.2 0.3 0.2 As-Recorded Component 2 0.1 0 -0.1 -0.2 -0.3 5 10 15 20 Time, s 25 30 35 Principal Component 2 0.1 0 -0.1 -0.2 -0.3 0 Principal Component 1 0.1 40 0 5 10 15 20 Time, s 25 30 35 40 Two Horizontal Components: Component 1: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h Component 2: 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h Two Horizontal Components: Component 1: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h Component 2: 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h Predictive equations: R rup M β 4 ln Vs30 β3 ln -1 [ F p ( p)] β0 β1 (F) β 2 750 m/s 25 km 7.0 M β3 1[ F p ( p)] β0 β1 (F) β 2 7 . 0 R rup β4 25 km Vs30 750 m/s η ε if p I a,maj , I a,int η ε if p D5-95 , tmid , ωmid , ω' , ζ f Two Horizontal Components: Component 1: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h Component 2: 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h Predictive equations: R rup M β 4 ln Vs30 β3 ln -1 [ F p ( p)] β0 β1 (F) β 2 750 m/s 25 km 7.0 M β3 1[ F p ( p)] β0 β1 (F) β 2 7 . 0 Model parameter p transformed to the standard normal space R rup β4 25 km Vs30 750 m/s η ε if p I a,maj , I a,int η ε if p D5-95 , tmid , ωmid , ω' , ζ f Two Horizontal Components: Component 1: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h Component 2: 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h Predictive equations: R rup M β 4 ln Vs30 β3 ln -1 [ F p ( p)] β0 β1 (F) β 2 750 m/s 25 km 7.0 M β3 1[ F p ( p)] β0 β1 (F) β 2 7 . 0 R rup β4 25 km Vs30 750 m/s Predicted mean conditioned on earthquake and site characteristics η ε if p I a,maj , I a,int η ε if p D5-95 , tmid , ωmid , ω' , ζ f Independent normally-distributed errors Two Horizontal Components: Component 1: 1 t x1 (t ) q(t, α1 ) h [ t τ, λ ( )] w ( τ ) dτ 1 1 σ ( t ) h Component 2: 1 t x 2 (t ) q(t, α 2 ) h[t τ,λ 2 ( )]w2 ( τ )dτ σ ( t ) h Predictive equations. High correlations expected between parameters of the two components. Two Horizontal Components: Correlation Matrix: Major Component (larger Arias intensity) I a, tf D595, tf tmid , tf ωmid , tf −0.38 −0.04 −0.21 −0.25 −0.06 +0.92 −0.30 1 +0.68 −0.07 −0.21 −0.26 −0.31 1 −0.24 −0.22 −0.26 1 −0.19 1 ω'tf ω'tf ζ tf −0.03 −0.13 +0.09 +0.02 +0.89 +0.68 −0.17 −0.11 −0.17 +0.04 +0.65 +0.96 −0.30 −0.24 −0.21 +0.28 −0.13 −0.15 −0.29 +0.94 −0.10 +0.29 I a, tf D595, tf tmid , tf ωmid , tf −0.06 +0.19 −0.21 −0.22 −0.10 +0.52 −0.13 ω'tf 1 −0.01 −0.23 −0.29 +0.32 −0.02 +0.75 ζ tf 1 −0.31 +0.01 −0.08 +0.07 −0.005 1 +0.69 −0.20 −0.18 −0.17 1 −0.34 −0.24 −0.22 1 −0.19 +0.28 I a, tf D595, tf tmid , tf ωmid , tf 1 −0.05 ω'tf I a, tf D595, tf tmid , tf 1 ζ tf Intermediate Component Symmetric ωmid , tf ζ tf Major Component 1 Intermediate Component (smaller Arias intensity) Two Horizontal Components: Correlation Matrix: Major Component (larger Arias intensity) I a, tf D595, tf tmid , tf ωmid , tf −0.38 −0.04 −0.21 −0.25 −0.06 +0.92 −0.30 1 +0.68 −0.07 −0.21 −0.26 −0.31 1 −0.24 −0.22 −0.26 1 −0.19 1 ω'tf ω'tf ζ tf −0.03 −0.13 +0.09 +0.02 +0.89 +0.68 −0.17 −0.11 −0.17 +0.04 +0.65 +0.96 −0.30 −0.24 −0.21 +0.28 −0.13 −0.15 −0.29 +0.94 −0.10 +0.29 I a, tf D595, tf tmid , tf ωmid , tf −0.06 +0.19 −0.21 −0.22 −0.10 +0.52 −0.13 ω'tf 1 −0.01 −0.23 −0.29 +0.32 −0.02 +0.75 ζ tf 1 −0.31 +0.01 −0.08 +0.07 −0.005 1 +0.69 −0.20 −0.18 −0.17 1 −0.34 −0.24 −0.22 1 −0.19 +0.28 I a, tf D595, tf tmid , tf ωmid , tf 1 −0.05 ω'tf I a, tf D595, tf tmid , tf 1 ζ tf Intermediate Component Symmetric ωmid , tf ζ tf Major Component 1 Intermediate Component (smaller Arias intensity) Two Horizontal Components: Correlation Matrix: Major Component (larger Arias intensity) I a, tf D595, tf tmid , tf ωmid , tf −0.38 −0.04 −0.21 −0.25 −0.06 +0.92 −0.30 1 +0.68 −0.07 −0.21 −0.26 −0.31 1 −0.24 −0.22 −0.26 1 −0.19 1 ω'tf ω'tf ζ tf −0.03 −0.13 +0.09 +0.02 +0.89 +0.68 −0.17 −0.11 −0.17 +0.04 +0.65 +0.96 −0.30 −0.24 −0.21 +0.28 −0.13 −0.15 −0.29 +0.94 −0.10 +0.29 I a, tf D595, tf tmid , tf ωmid , tf −0.06 +0.19 −0.21 −0.22 −0.10 +0.52 −0.13 ω'tf 1 −0.01 −0.23 −0.29 +0.32 −0.02 +0.75 ζ tf 1 −0.31 +0.01 −0.08 +0.07 −0.005 1 +0.69 −0.20 −0.18 −0.17 1 −0.34 −0.24 −0.22 1 −0.19 +0.28 I a, tf D595, tf tmid , tf ωmid , tf 1 −0.05 ω'tf I a, tf D595, tf tmid , tf 1 ζ tf Intermediate Component Symmetric ωmid , tf ζ tf Major Component 1 Intermediate Component (smaller Arias intensity) Two Horizontal Components: Correlation Matrix: Major Component (larger Arias intensity) I a, tf D595, tf tmid , tf ωmid , tf −0.38 −0.04 −0.21 −0.25 −0.06 +0.92 −0.30 1 +0.68 −0.07 −0.21 −0.26 −0.31 1 −0.24 −0.22 −0.26 1 −0.19 1 ω'tf ω'tf ζ tf −0.03 −0.13 +0.09 +0.02 +0.89 +0.68 −0.17 −0.11 −0.17 +0.04 +0.65 +0.96 −0.30 −0.24 −0.21 +0.28 −0.13 −0.15 −0.29 +0.94 −0.10 +0.29 I a, tf D595, tf tmid , tf ωmid , tf −0.06 +0.19 −0.21 −0.22 −0.10 +0.52 −0.13 ω'tf 1 −0.01 −0.23 −0.29 +0.32 −0.02 +0.75 ζ tf 1 −0.31 +0.01 −0.08 +0.07 −0.005 1 +0.69 −0.20 −0.18 −0.17 1 −0.34 −0.24 −0.22 1 −0.19 +0.28 I a, tf D595, tf tmid , tf ωmid , tf 1 −0.05 ω'tf I a, tf D595, tf tmid , tf 1 ζ tf Intermediate Component Symmetric ωmid , tf ζ tf Major Component 1 Intermediate Component (smaller Arias intensity) Example: Design scenario: F = 1 (Reverse) , M = 7.35 , R =14 km , VS30 = 660 m/s Example: Design scenario: F = 1 (Reverse) , M = 7.35 , R =14 km , VS30 = 660 m/s Major Component Recorded Intermediate Component Ia D5-95 tmid ωmid/(2π) ω’/(2π) s.g s s Hz Hz/s 16.7 17.3 27.2 18.3 10.1 17.1 3.9 8.1 –0.08 –0.12 –0.03 0.0165 0.0147 Simulated 0.0099 3.2 ζf 0.12 0.42 0.20 Ia D5-95 tmid ωmid/(2π) ω’/(2π) s.g s s Hz Hz/s 0.0135 0.0047 0.0034 17.0 21.0 24.8 17.8 10.7 16.9 4.1 8.6 3.7 –0.02 –0.18 –0.13 ζf 0.11 0.50 0.35 Example: Design scenario: F = 1 (Reverse) , M = 7.35 , R =14 km , VS30 = 660 m/s Major Component Recorded Intermediate Component Ia D5-95 tmid ωmid/(2π) ω’/(2π) s.g s s Hz Hz/s 16.7 17.3 27.2 18.3 10.1 17.1 3.9 8.1 –0.08 –0.12 –0.03 0.0165 0.0147 Simulated 0.0099 3.2 Ia D5-95 tmid ωmid/(2π) ω’/(2π) s.g s s Hz Hz/s 0.0135 0.0047 0.0034 17.0 21.0 24.8 17.8 10.7 16.9 4.1 8.6 3.7 –0.02 –0.18 –0.13 ζf 0.12 0.42 0.20 0.1 Recorded Recorded Simulated Simulated Simulated Simulated ζf 0.11 0.50 0.35 0 Acceleration, g -0.1 0.1 0 -0.1 0.05 0 -0.05 0 20 40 Time, s 60 80 0 20 40 60 Time, s 80 Example: Design scenario: F = 1 (Reverse) , M = 7.35 , R =14 km , VS30 = 660 m/s Major Component Recorded Intermediate Component Ia D5-95 tmid ωmid/(2π) ω’/(2π) s.g s s Hz Hz/s 16.7 17.3 27.2 18.3 10.1 17.1 3.9 8.1 –0.08 –0.12 –0.03 0.0165 0.0147 Simulated 0.0099 3.2 0.01 Ia D5-95 tmid ωmid/(2π) ω’/(2π) s.g s s Hz Hz/s 0.0135 0.0047 0.0034 17.0 21.0 24.8 17.8 10.7 16.9 4.1 8.6 3.7 –0.02 –0.18 –0.13 ζf 0.12 0.42 0.20 Recorded Recorded Simulated Simulated Simulated Simulated ζf 0.11 0.50 0.35 0 Velocity, m/s -0.01 0.05 0 -0.05 0.05 0 -0.05 0 20 40 Time, s 60 80 0 20 40 60 Time, s 80 Example: Design scenario: F = 1 (Reverse) , M = 7.35 , R =14 km , VS30 = 660 m/s Major Component Recorded Intermediate Component Ia D5-95 tmid ωmid/(2π) ω’/(2π) s.g s s Hz Hz/s 16.7 17.3 27.2 18.3 10.1 17.1 3.9 8.1 –0.08 –0.12 –0.03 0.0165 0.0147 Simulated 0.0099 3.2 0.02 Ia D5-95 tmid ωmid/(2π) ω’/(2π) s.g s s Hz Hz/s 0.0135 0.0047 0.0034 17.0 21.0 24.8 17.8 10.7 16.9 4.1 8.6 3.7 –0.02 –0.18 –0.13 ζf 0.12 0.42 0.20 Recorded Recorded Simulated Simulated Simulated Simulated ζf 0.11 0.50 0.35 Displacement, m 0 -0.02 0.05 0 -0.05 0.05 0 -0.05 0 20 40 Time, s 60 80 0 20 40 60 Time, s 80 Conclusion: Developed a stochastic model for earthquake ground motion components Created a database of principal ground motion components Identified model parameters for the records in the database predictive equations for model parameters in terms of F , M , R , VS30 Identified correlation coefficients between model parameters of the components For given F , M , R , VS30 , correlated model parameters are randomly simulated and used along with statistically independent white-noise processes to generate a pair of horizontal ground motion components in the directions of principal axes. The proposed methods can be easily extended to simulate the vertical component. Related Publications: 1. Rezaeian S, Der Kiureghian A. "A stochastic ground motion model with separable temporal and spectral nonstationarities”, Earthquake Engineering and Structural Dynamics, 2008, Vol. 37, pp. 1565-1584. 2. Rezaeian S, Der Kiureghian A. "Simulation of synthetic ground motions for specified earthquake and site characteristics”, Earthquake Engineering and Structural Dynamics, 2010, Vol. 39, pp. 1155-1180. 3. Rezaeian S, Der Kiureghian A. "Simulation of orthogonal horizontal ground motion components for specified earthquake and site characteristics”, Submitted to Earthquake Engineering and Structural Dynamics. Thank You This project was made possible with support from: State of California through Transportation Systems Research Program of Pacific Earthquake Engineering Research Center (PEER TSRP). 44 Ground Motion Model: Advantages Extra Small number of parameters that have physical meaning and can be easily identified by matching with features of a given accelerogram Completely separable temporal and spectral nonstationary characteristics, which facilitates identification and interpretation of the parameters No need for sophisticated processing of the target accelerogram, e.g. Fourier analysis or estimation of evolutionary PSD Simple simulation of sample functions, requiring little more than generation of standard normal random variables k xˆ (t ) q(t ) si (t ) ui i 1 for where where tk t tk 1 ti i t i 0,...,n ui ~ N(0,1) Form of the model facilitates nonlinear random vibration analysis (e.g., by using TELM).