System Response LAPLACE DOMAIN TRANSFER FUNCTION Y (S) = H(S) • F(S ) y (t) = h(t) ⊗ f (t) RESPONSE MODELS Y ( jω) = H( jω) • F( jω) FRF IMPULSE t y ( t ) = ∫ h (τ)f ( t − τ)dτ 0 TIME DOMAIN Peter Avitabile Mechanical Engineering Department University of Massachusetts Lowell FREQUENCY DOMAIN 22.451 Dynamic Systems – System Response 1 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – First-order System τy& + y = f ( t ) τ > 0 I.C. y(0) = y o Free Response No forcing function – Only I.C. Laplace Transform yo τy o τ[sY (s) − y o ] + Y(s) = 0 ⇒ Y (s) = = τs + 1 s + 1 Inverse Laplace y( t ) = y o e τ ( τ) − t start at yo and exponentially ∞ decay to zero as t Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 2 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – First-order System Forced Response – Step Response f ( t ) = A if t > 0 otherwise = 0 Laplace Transform A τy o s + A τ[sY (s) − y o ] + Y (s) = ⇒ Y (s) = s s(τs + 1) break up using partial fraction expansion A y os + c1 c2 τ = + Y (s) = 1 s s+1 s s + τ τ 22.451 Dynamic Systems – System Response 3 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – First-order System where c1 = sY (s) s = 0 τy o s + A = τs + 1 1 c 2 = s + Y (s) s = − 1 τ τ s=0 =A A y os + τ = s s=− 1 τ = yo − A A 1 + (y o − A ) ∴ Y (s) = 1 s s+ τ 22.451 Dynamic Systems – System Response 4 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – First-order System Inverse Laplace −t y STEP ( t ) = A + ( y o − A)e τ −t if y o = 0 → y STEP ( t ) = A1 − e τ y(τ) = 0.632A y(2τ) = 0.865A y(3τ) = 0.950A y(4τ) = 0.982A y(5τ) = 0.995A Scan Fig. 7.2 p. 337 Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 5 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – First-order System Forced Response – Ramp Response f ( t ) = At if t > 0 otherwise = 0 Laplace Transform A τ[sY (s) − y o ] + Y (s) = 2 s A + y os 2 B 2 B1 C τ = 2 + + Y (s) = 1 s s+1 2 s s s + τ τ where B = A; B = − τA; C = y + τA 2 1 o 22.451 Dynamic Systems – System Response 6 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – First-order System Inverse Laplace y RAMP ( t ) = At − Aτ + Aτe −t τ Note that the SS response is At − Aτ. After the transient portion decays, the response will track the ramp but with an error of Aτ. Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 7 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – Second-order System &x& + 2ζω n x& + ωn 2 x = f ( t ) I.C. x o ; x& o Free Response – No forcing function – Only I.C. [ Laplace Transform 2 ] 2 & s X (s) − sx (0) − x (0) + 2ζω n [sX (s) − x (0)] + ωn X (s) = 0 ( s + 2ζω )x X(s) = • − x0 2 s 2 + 2ζωn s + ωn n 0 Solution form depends on the poles of the characteristic equation s 2 + 2ζω n s + ωn 2 = 0 ∴ s1, 2 = −ζω n ± (ζω n )2 − ωn 2 = −ζω n ± ωn ζ 2 − 1 22.451 Dynamic Systems – System Response 8 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – Second-order System Three cases of damping •Case 1 – Overdamped (ζ>1) •Case 2 – Critically damped (ζ=1) •Case 3 – Underdamped (0<ζ<1) •Case 4 – Undamped (ζ=0) Scan Fig. 7.5 p. 343 Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 9 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – Second-order System Case 1 – Overdamped (ζ>1) – Two Real Roots (s + 2ζω n )x 0 + v 0 (s + 2ζω n )x 0 + v 0 X(s) = 2 = 2 (s − s1 )(s − s 2 ) s + 2ζω n s + ωn 2 s1 = −ζω n + ωn ζ − 1; s 2 = −ζω n − ωn ζ 2 − 1 Partial Fraction Expansion A1 A2 X(s) = + s − s1 s − s 2 22.451 Dynamic Systems – System Response ωn ζ + ζ 2 − 1 x 0 + v 0 A1 = 2 2ω n ζ − 1 − ωn ζ − ζ 2 − 1 x 0 + v 0 A2 = 2 2ω n ζ − 1 10 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – Second-order System Case 1 – Overdamped (ζ>1) – Two Real Roots ω ζ + ζ 2 − 1 x + v 0 0 n −ω x(t) = e 2 2ω n ζ − 1 n (ζ − ω ζ − ζ 2 − 1 x + v 0 0 n −ω e − 2 2ω n ζ − 1 22.451 Dynamic Systems – System Response 11 ) ζ 2 −1 t n (ζ + ) ζ 2 −1 t Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – Second-order System Case 2 – Critically damped (ζ=1) – Two Real Repeated Roots s1, 2 = −ωn ( two real repeated roots) (s + 2ωn )x 0 + v 0 (s + ωn )x 0 + ωn x 0 + v 0 = X(s) = 2 2 (s + ωn )2 s + 2ω n s + ω n x0 ωn x 0 + v 0 X(s) = + s + ωn (s + ωn )2 Inverse Laplace x(t) = x 0e −ω n t −ω t ( ) + ωn x 0 + v 0 te 22.451 Dynamic Systems – System Response n 12 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – Second-order System Case 3 – Underdamped (0<ζ<1) – Complex Conjugate Pair Define: σ = ζω n ωd = ω n 1 − ζ 2 Then s 2 + 2ζω n s + ωn 2 = (s + ζω n )2 − ζ 2 ωn 2 + ωn 2 = (s + σ ) + ωd 2 2 Then (s + 2ζω n )x 0 + v 0 (s + 2σ )x 0 + x 0 + v 0 X(s) = 2 = 2 (s + σ )2 + ωd 2 s + 2ζω n s + ωn 22.451 Dynamic Systems – System Response 13 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Transient Response – Second-order System which can be rearranged into (s + σ )x 0 σx 0 + v 0 X(s) = + 2 2 2 2 (s + σ ) + ωd (s + σ ) + ωd (s + σ )x 0 (σx 0 + v 0 ) ωd = + 2 2 ωd (s + σ ) + ωd (s + σ )2 + ωd 2 Inverse Laplace x(t) = x 0e OR =e − σt σx 0 + v 0 − σt cos ωd t + e sin ωd t ωd − σt σx 0 + v 0 sin ωd t x 0 cos ωd t + ωd 22.451 Dynamic Systems – System Response 14 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Impulse &x& + 2ζω n x& + ωn 2 x = f ( t ) = δ( t ) with I.C. x o ; x& o Laplace Transform [s X(s) − sx(0) − x& (0)]+ 2ζω [sX(s) − x(0)] + ω 2 n Grouping terms X(s) = F(s) 2 s + 2ζω n s + ωn 2 + n 2 X (s) = F(s) (s + 2ζω n )x o + v o s 2 + 2ζω n s + ωn 2 If I.C. are zero, then the result is the system transfer function X (s) 1 H(s) = = 2 F(s) s + 2ζω n s + ωn 2 The inverse Laplace of this yields the impulse response of the system. (Note that I.C. are zero) 22.451 Dynamic Systems – System Response 15 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Impulse Case 1 – Overdamped x(t) = 1 2ω n ζ 2 − 1 e ( ) − ω n ζ − ζ 2 −1 t − 1 2ω n ζ 2 − 1 ω ζ + ζ 2 − 1 x + v 0 0 n −ω e + 2 2 ω ζ −1 n ω ζ − ζ 2 − 1 x + v 0 0 n −ω e − 2 2 ω ζ −1 n 22.451 Dynamic Systems – System Response 16 n (ζ − e ( ) − ω n ζ + ζ 2 −1 t ) ζ 2 −1 t I.C. n (ζ + ) ζ 2 −1 t Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Impulse Case 2 – Critically Damped x ( t ) = te − ω t + x o e − ω t + (ωn x o + v o )te − ω n n n t I.C. 22.451 Dynamic Systems – System Response 17 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Impulse Case 3 – Underdamped (σx o + v o ) 1 − σt − σt x(t) = e sin ωd t + e x o cos ωd t + sin ωd t ωd ωd I.C. Case 4 – Undamped vo 1 x(t) = sin ωn t + x o cos ωn t + sin ωn t ωn ωn 22.451 Dynamic Systems – System Response 18 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Impulse Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 19 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Impulse Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 20 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Step &x& + 2ζω n x& + ωn 2 x = f ( t ) = u ( t ) with I.C. x o ; x& o Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 21 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Step Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 22 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Step Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 23 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Forced Response – Unit Step Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 24 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Frequency Response Function The FRF can be obtained from the Fourier Transform of Input-Output Time Response (and is commonly done so in practice) The FRF can also be obtained from the evaluation of the system transfer function at s=jω. For a 1st order system 1 1 H(s) = ⇒ H (s) s = jω = H( jω) = τs + 1 1 + jωτ For a 2nd order system H(s) = 1 s 2 + 2ζω n s + ωn 2 ⇒ H(s) s = jω = H( jω) = 22.451 Dynamic Systems – System Response 1 − ω2 + 2ζω n jω + ωn 2 25 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Frequency Response Function The FRF for a mechanical system H(s) = 1 2 ms + cs + k ⇒ H( jω) = 1 − mω2 + jcω + k This is normally presented in a LOG-MAG and PHASE plot called a BODE DIAGRAM. 22.451 Dynamic Systems – System Response 26 Dr. Peter Avitabile Modal Analysis & Controls Laboratory FRF – Bode Diagram – 1st Order Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 27 Dr. Peter Avitabile Modal Analysis & Controls Laboratory FRF – Bode Diagram – 2nd Order Source: Dynamic Systems – Vu & Esfandiari 22.451 Dynamic Systems – System Response 28 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Fourier Series A periodic function is a function that satisfies the relationship where f (t ) = f (t + T ) T is the period of the function The fundamental angular frequency is defined by 2π ωo = (rad/sec) T 22.451 Dynamic Systems – System Response 29 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Fourier Series Consider a family of waveforms f1 (t ) = 1 f 2 (t ) = sin (2πt ) 1 f 3 (t ) = sin (6πt ) 3 1 f 4 (t ) = sin (10πt ) 5 1 f 4 (t ) = sin (14πt ) 7 Summed as N g (t ) = ∑ f n (t ) 1< N ≤ 5 n =1 22.451 Dynamic Systems – System Response 30 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Fourier Series The summed waveforms can be plotted in Matlab to approximate a square wave 22.451 Dynamic Systems – System Response 31 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Fourier Series () A Fourier Series representation of a real periodic function f t is based upon the summation of harmonically related sinusoidal components. If the period is T , then the harmonics are sinusoids with frequencies that are integer multiples of ωo ; This is written as f n (t ) = a n cos(nωo t ) + b n sin (nωo t ) = A n sin (nωo t + φn ) Note: These two are equivalent – one is written with sine and cosine terms – the other is written as a sin with amplitude and phase 22.451 Dynamic Systems – System Response 32 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Fourier Series The Fourier Series representation of an arbitrary periodic waveform f t is an infinite sum of harmonically related sinusoids and is commonly written as () ∞ 1 f (t ) = a 0 + ∑ [a n cos(nωo t ) + b n sin (nωo t )] 2 n =1 ∞ 1 = a o + ∑ [A n sin (nωo t + φn )] 2 n =1 Or ( ) ( a n jnω t b n jnω t − jnω t f (t ) = e +e + e + e − jnω t 2j 2 1 1 jnω t = (a n − jb n )e + (a n + jb n )e − jnω t 2 2 o o o ∞ [ = ∑ Fn e jnω t −∞ o o o ) o ] 22.451 Dynamic Systems – System Response 33 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Fourier Coefficients The derivation of the expression for computing the coefficients in a Fourier Series is beyond the scope of this course. Without proof, the coefficients Fn are 1 Fn = T t1 +T − jnω t ( ) f t e dt ∫ o t1 which is evaluated over any periodic segment or in sinusoidal form 1 an = T 1 bn = T t1 +T ∫ f (t )cos(nωo t )dt t1 t1 +T ∫ f (t )sin(nωo t )dt t1 22.451 Dynamic Systems – System Response 34 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Consider a linear SDOF system with a steady state periodic input (assume that the initial conditions and transients have decayed to zero) The input excitation can be described by a Fourier Series representation as ∞ f (t ) = ∑ Fn e jnω t o −∞ ∞ 1 = a o + ∑ A n sin (nωo t + φn ) 2 n =1 22.451 Dynamic Systems – System Response 35 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs The system output response is related to the input excitation through the system frequency response function as Y( jω) = H( jω) ⋅ F( jω) The nth real harmonic input component of the Fourier Series f n (t ) = A n sin (nωo t + φ) Generates an output sinusoidal component y n (t ) with a magnitude and phase determined by the nth component of H( jω) y n (t ) = H( jnωo ) A n sin[nωo + φn + ∠H( jnωo )] 22.451 Dynamic Systems – System Response 36 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs From superposition, the total output y n (t ) is the sum of all the n components as y(t ) = ∞ ∑ y n (t ) n =0 ∞ 1 = a o H( jo ) + ∑ A n H( jnωo ) sin[nωo t + φn + ∠H( jnωo )] 2 n =1 Note: This is also a Fourier Series with the same fundamental and harmonic frequencies as the input In complex form, this is written as y n (t ) = H( jnωo )Fn e jnω t o The complete output Fourier Series is y(t ) = n =∞ ∑ y n (t ) = n = −∞ n =∞ jnω t ( ) H jn ω F e ∑ o n o n = −∞ 22.451 Dynamic Systems – System Response 37 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Consider a simple mass spring dashpot system that is under steady state sinusoidal excitation (but not a ωn of a SDOF system Time Domain h (t ) f (t ) = sin (ωt ) System FFT h ( jω) Frequency Domain f ( jω) 22.451 Dynamic Systems – System Response y(t ) = A sin (ωt + φ) 38 IFT y ( jω) Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Now consider excitation at the natural frequency, ωn ,of the SDOF system 22.451 Dynamic Systems – System Response 39 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Now consider a periodic input which is not sinusoidal. However, the Fourier Series representation is nothing more than the sum of different amplitude sine waves. 22.451 Dynamic Systems – System Response 40 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.7 The first-order electric network shown in Fig. 15.7 is excited with the sawtooth function. Find an expression for the series representing the output Vout(t). 22.451 Dynamic Systems – System Response 41 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.7 (cont.) The electric network has a transfer function 1 H(s ) = RCs + 1 and therefore has a frequency response function 1 H ( j ω) = (ωRC )2 + 1 ∠H( jω) = tan −1 (− ωRC ) the input function u(t) may be represented by the Fourier Series 2(− 1)n +1 sin (nωo t ) u (t ) = ∑ nπ n =1 ∞ 22.451 Dynamic Systems – System Response 42 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.7 (cont.) At the output the series representation is 2(− 1)n +1 y ( t ) = ∑ | H ( jω ) | sin[nωo t + ∠H( jnωo )] nπ n =1 ∞ n +1 ( ) 2 1 − sin nω t + tan −1 (− nω RC ) y(t ) = ∑ o o nπ (nω RC )2 + 1 n =1 o [ ∞ ] As an example consider the response if the period of the input is chosen to be T = πRC so that ω = 2 ,then o RC ∞ 2(− 1)n +1 2n sin t + tan −1 (− 2n ) 2 RC n =1 nπ (2n ) + 1 y (t ) = ∑ 22.451 Dynamic Systems – System Response 43 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.7 (cont.) Figure 15.8 shows the computer response found by summing the first 100 terms of the Fourier Series. T = πRC 22.451 Dynamic Systems – System Response ωo = 44 2 RC Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.8 A cart shown in Figure 15.9a, with a mass m=1.0 kg is supported on low friction bearings that exhibit a viscous drag B = 0.2 N-s/m and is coupled through a spring with stiffness K=25 N/m to a velocity source with a magnitude of 10 m/s but which switches direction every π seconds as shown in Figure 15.9b. Vin (t ) = 10 m / s 0≤t<π − 10 m / s π ≤ t < 2π The task is to find the resulting velocity of the mass v m (t ) 22.451 Dynamic Systems – System Response 45 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.8 (cont.) The input Ω(t ) has a period of T = 2π and a fundamental frequency of ωo = 2π / T = 1 rad/s. The Fourier Series for the input contains only odd harmonics: 20 ∞ 1 u (t ) = ∑ sin[(2n − 1)ωo t ] π n =1 2n − 1 = 20 1 1 ( ) ( ) ( ) sin ω t + sin 3 ω t + sin 5 ω t + ... o o o π 3 5 22.451 Dynamic Systems – System Response 46 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.8 (cont.) The frequency response of the system is H ( jω ) = 25 25 − ω2 + j0.2ω ( ) which when evaluated at the harmonic frequencies of the input nωo = n radians/s is H( jnωo ) = 25 25 − n 2 + j0.2n [( ) 22.451 Dynamic Systems – System Response 47 ] Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.8 (cont.) The following table summarizes the first five odd spectral components at the system input and output. 22.451 Dynamic Systems – System Response 48 Dr. Peter Avitabile Modal Analysis & Controls Laboratory Response of Linear System Due to Periodic Inputs Example 15.8 (cont.) Figure 15-10a shows the computed frequency response magnitude for the system and the relative gains and phase shifts (rad) associated with the first five terms in the series. 22.451 Dynamic Systems – System Response 49 Dr. Peter Avitabile Modal Analysis & Controls Laboratory 22.451 Dynamic Systems – System Response 50 Dr. Peter Avitabile Modal Analysis & Controls Laboratory