Problem 1 ySP (s) + k = 1.6 − 5 (s + 1)(2s + 1) y(s) The closed loop transfer function can be written as follows: 8 y(s) 8 89 (s + 1)(2s + 1) G CL (s) = = = 2 = 2 8 ySP (s) 1 + 2s + 3s + 9 2 9s + 1 3s + 1 (s + 1)(2s + 1) (S1.1) Thus, comparing (S1.1) to the standard form of 2nd order systems, one obtains: τ2 = 2 9 ⇒ τ = 2 3 = 0.471 (S1.2) 2ξτ = 1 2 ⇒ ξ = 2 4 = 0.354 (S1.3) k = 8 9 = 0.889 (S1.4) Since ξ is smaller than 1, the system is underdamped. a) From textbook, we know that: ⎛ −πξ ⎞ max dev final value y max − y(∞) ⎟ = 0.305 = OS = exp ⎜ = ⎜ 1 − ξ2 ⎟ final value y(∞) ⎝ ⎠ (S1.5) The output of the system for a step change of magnitude 0.1 in the set point is: y(s) = 0.1 89 2 s 2 9s + 1 3s + 1 (S1.6) Applying the final value theorem, yields: y(∞) = lim sy(s) = lim s →0 s →0 0.8 9 0.8 = = 0.0889 2 9s + 1 3s + 1 9 2 (S1.7) Hence, the maximum value of the response is: y max = y(∞)(1 + OS) = 0.116 (S1.8) b) For a servo problem, the offset is given by: offset = new set point − y(∞) = 0.1 − 0.0899 = 0.0111 (S1.9) c) From the textbook, the period of the oscillation is: Τ = 2π ω where ω = 1 − ξ 2 τ (S1.10) Therefore, the value of the period of the oscillation Τ is: Τ = 3.17 min (S1.11) Problem 2 h (ft) The resistance of a liquid to a hydrostatic pressure can be defined as the rate of changing of the liquid level due to the change of the output low rate. Thus: dh R= dq R (S2.1) q (ft3/min) Assuming linear resistances, R1 and R2 can be directly evaluated by plotting h (ft) versus q (ft3/min) and estimating the slope of the straight line. Hence, one obtains: R1 = R 2 = 0.5 (S2.2) Since the cross sections are equal to 2, the time constant of the two tanks are both equal to τ1 = A1R1 = 1 (min) and τ2 = A 2 R 2 = 1 (min) , while the static gains of two tanks are equal to k1 = R1 = 0.5 (min ft 2 ) and k 2 = R 2 R1 = 1 (min ft 2 ) . Thus, the transfer functions of the two tanks are the following: G1 (s) = 0.5 s +1 ⇒ G 2 (s) = 1 s +1 (S2.3) Employing a proportional controller, the corresponding transfer function is: G s (s) = k c (S2.4) Plotting the change in pressure to the valve versus the change in flow provides the transfer function for the final control element: G f (s) = dP 0.1 = = 0.1 dq 1 (S2.5) With no lag in the measuring device dynamics, the corresponding transfer function is: G m (s) = 1 (S2.6) Therefore, the block diagram is the following: a) controller ySP (s) valve kc q(s) 0.1 tank 1 tank 2 0.5 s +1 1 s +1 y(s) Measuring device 1 b) The close loop transfer function is: 0.05 k c 0.05 k c 2 1 + 0.05 k c y(s) (s + 1) G CL (s) = = = 1 2 ySP (s) 1 + 0.05 k c s2 + s +1 2 1 + 0.05 k c 1 + 0.05 k c (s + 1) (S2.7) Comparing the closed loop transfer function to the standard form of a 2nd order system transfer function one obtains: τ = 1 1 + 0.05 k c (S2.8) ξ = 1 1 + 0.05 k c (S2.9) k = 0.05 k c (1 + 0.05 k c ) (S2.10) For a critically damped 2nd order system, ξ=1 and therefore: k c(CD) = 0 (S2.11) Therefore, the critical damping cannot occur. c) For interacting tanks, the transfer function between the output of the second tank and the input of the first tank is given by: G(s) = y(s) R2 0.5 = = 2 2 q(s) τ1τ2s + (τ1 + τ2 + A1R 2 )s + 1 s + 3s + 1 Thus, the close loop transfer function becomes: (S2.12) 0.05 k c 0.05 k c 2 1 + 0.05 k c y(s) G CL (s) = = s + 3s + 1 = 0.05 k 1 3 ySP (s) 1 + c s2 + s +1 2 1 + 0.05 k c s + 3s + 1 1 + 0.05 k c (S2.13) Comparing the closed loop transfer function to the standard form of a 2nd order system transfer function one obtains: τ = 1 1 + 0.05 k c (S2.14) 3 2 1 + 0.05 k c (S2.15) ξ= k = 0.05 k c (1 + 0.05 k c ) (S2.16) For a critically damped 2nd order system, ξ=1 and therefore: k c(CD) = 25 psi/ft (S2.17) Hence, for interacting capacities the critical damping does occur. d) Assuming kc=1.5kc(CD), one obtains: k c = 1.5k c(CD) = 37.5 psi/ft (S2.18) Thus, the natural period of the oscillations, damping factor and gain becomes: τ = 0.590 (min) (S2.19) ξ = 0.885 (S2.20) k = 0.652 (ft) (S2.21) For a step change of 1/12 ft in the se point, ysp(s) is equal to 1/(s12). Therefore, one obtains: y(s) = 0.652 0.054 ySP (s) = 2 0.348s + 1.043s + 1 (0.348s + 1.043s + 1)s (S2.22) 2 Since ξ is smaller than 1, from textbook we have: ⎧⎪ ⎡ ⎛ 1 − ξ2 1 −ξt τ1 −1 2 t y(t) = 0.054 ⎨1 − e sin ⎢ 1 − ξ + tg ⎜ ⎜ τ1 ⎢⎣ 1 − ξ2 ⎪⎩ ⎝ ξ = 0.054 − 0.116e −1.5t sin(0.8t + 28o ) ⎞ ⎤ ⎫⎪ ⎟⎥ ⎬ = ⎟ ⎠ ⎥⎦ ⎪⎭ (S2.23) Problem 3: a) The block diagram is the following: mA mA mA Controller mA m3/min psgi Transducer Delay mA Valve Kg/m3 Process Composition Transmitter Data Kg/m3 Line Kg/m3 b) Controller (pure PID) Transfer function is: Gc ( s) = K c (1 + 1 + τ D s) τIs Transducer ~ p ( s) p v ( s) − 3 15 − 3 12 3 = = = = Transfer Function is: GT ( s) = ~v p ( s) p ( s) − 4 20 − 4 16 4 Control Valve Linearizing qA around an operative point: q A = q A0 + 0.03 ln(20)20 ( pv 0 −3) / 12 ( p v − pv 0 ) Choosing qA0=0.17 and pV0=3, gives: q A = q A0 + 0.0025ln(20)(p v − p v0 ) q (s) q (s) − 0.17 Transfer Function is: G v (s) = A = A = 0.0025ln(20) p v (s) p v (s) − 3 Also, considering delay: G v (s) = 0.0025ln(20)e − s Process dc V = q A c A + q F c F − (q A + q F )c . Note q A + q F ≈ q F . dt dc Hence V + qF c = q Ac A + qF cF . dt At steady state: q Fcs = q As c A + q Fc Fs Defining c~ = c − c s , q~A = q A − q As , c F = c F − c Fs yields: dc dc or V + q Fc = c A q A + q Fc F τp + c = k p q A + k d c F where τP = V / q F , dt dt K P = c A / q F and K d = 1 . Taking Laplace transform gives: kd kP c(s) = q A (s) + c F (s) = G P (s)q A (s) + G D (s)c F (s) τPs + 1 τPs + 1 Therefore, process can be expanded as illustrated in figure. Kg/m3 GD(s) Trasmission Line Transfer function is: G L (s) = e − t Ls , m3/min Kg/m3 GP(s) where t L = ⎡⎣ 20π(0.5) 2 ⎤⎦ / 4q F Composition Transmitter Data c~ ( s ) c m ( s ) − 4 20 − 4 16 Transfer Function is: GCTA ( s ) = ~m = = = c ( s) c( s) 200 200