Feedback Control of a Water Heater

Transcription

1 Feedback Control of a Water Heater

2 Introduction v A simple feeback control system involving a stirred tank, temperature measurement, controller and manipulated heater is shown in Figure T 0, F Measurement T R, F Water Tank T R Q T R Controller T RSet Heater Figure. Feedback control of a simple continuous water heater.

3 Model Ø The energy balance for the tank is dt dt R F = ( T 0 -TR ) V Q + VrCp Ø Where Q is the heat input from the controlled heater. A proportional-integral feedback controller can be modelled as Q Kp = Q + Kp + ò 0 ε εdt t Ø Where the control error is given by ε = TRSet -T R I

4 Model v Nomenclature Symbols C p Specific heat Kcal/(Kg C) f Frequency of oscillations l/h F Flow rate m 3 /h K P Controller constant Kcal/C K D Controller constant for differential term (Kcal h)/c Q Heat input Kcal/h T Temperature C t I Controller cinstant h V Reactor volume m 3 є Error C ρ Density Kg /m 3 Indices M Refers to measured value R Refers to reactor Set Refers to setpoint 0 Refers to inlet

5 Program Ø Initial condition Variables Unit Value Inlet temperature C Set point C 80 Reactor Volume m Inlet flow rate m 3 /hr 10 Density Kg/m 3 1 Specific heat capacity Kcal/Kg C 1 Initial Heat duty Kcal/hr 6 Ø Initial condition parameters Variables Symbols Unit Value Controller constant K P Kcal/C 10 First order lag τ Q hr 1 τ S hr 2 Time constant τ I hr 5 Error ε C 0

6 Program v M-file

7 Program v M-file

8 Exercises 1 vdisconnect the controller (set K P =0), allow the temperature to reach steady state, and measure the temperature response to changes in water flow. Water flow rate (5 m 3 /hr)

9 Exercises 1 vdisconnect the controller (set K P =0), allow the temperature to reach steady state, and measure the temperature response to changes in water flow. Water flow rate (15 m 3 /hr)

10 Exercises 2 vrepeat exercise 1 for a change in inlet temperature Inlet temperature (15 C)

11 Exercises 2 vrepeat exercise 1 for a change in inlet temperature Inlet temperature (25 C)

12 Exercises 3 vmeasure the controlled response to a step change in F with proportional rtional control only (set t Ivery high). Notice the offset error є,, and change t I to a low value. Dose the offset disappear? Proportional control only

13 Exercises 3 vmeasure the controlled response to a step change in F with proportional rtional control only (set t Ivery high). Notice the offset error є,, and change t I to a low value. Dose the offset disappear? Step change in F with proportional control ( F: 20 à 25 m 3 /hr, after 80 hr)

14 Exercises 3 vmeasure the controlled response to a step change in F with proportional rtional control only (set t Ivery high). Notice the offset error є,, and change t I to a low value. Dose the offset disappear? Step change in F with proportional control ( F: 20 à 15 m 3 /hr, after 80 hr)

15 Exercises 3 vmeasure the controlled response to a step change in F with proportional rtional control only (set t Ivery high). Notice the offset error є,, and change t I to a low value. Dose the offset disappear? Proportional-integral control ( F: 20 m 3 /hr)

16 Exercises 3 vmeasure the controlled response to a step change in F with proportional rtional control only (set t Ivery high). Notice the offset error є,, and change t I to a low value. Dose the offset disappear? Step change in F with proportional-integral control ( F: 20 à 25 m 3 /hr, after 80 hr)

17 Exercises 3 vmeasure the controlled response to a step change in F with proportional rtional control only (set t Ivery high). Notice the offset error є,, and change t I to a low value. Dose the offset disappear? Step change in F with proportional-integral control ( F: 20 à 15 m 3 /hr, after 80 hr)

18 Exercises 4 vdisconnect the controller and by forcing the system with a stepwise change in water flow rate, tobtain I the process reaction curve as explained in Ch. t I 7. Analyze this curve to obtain the parameters for the Ziegler-Nichols Method. Use Table 7.1 to obtain the best controller setting for P and d PI control. Try these out in a simulation. Water flow change (process reaction curve) S = Slope A = =

19 Exercises 4 vdisconnect the controller and by forcing the system with a stepwise change in water flow rate, tobtain I the process reaction curve as explained in Ch. t I 7. Analyze this curve to obtain the parameters for the Ziegler-Nichols Method. Use Table 7.1 to obtain the best controller setting for P and d PI control. Try these out in a simulation. Table. Controller settings based on process responses (Zigler-Nichols) Cont roller Kp t I t D P 1/T L S PI 0.9/T L S 3.33T L PID 1.2/T L S 2T / 2 L T L

20 Exercises 4 P controller

21 Exercises 4 PI controller

22 Exercises 4 PID controller

23 Exercises 5 vproceed as in Example 4, but use the Cohen-Coon Coon settings from Table 7.1 A=0.5 B=0.03 T L =2.5 S= K = A B = 0.06 = S B t = 5.63 Table. Controller settings based on process responses (Cohen-Coon) Cont roller Kp t I t D P PI PID t KT t KT L t KT L L (1+ TL ) 3t TL (0.9 + ) 12t 4 ( 3 TL + ) 12t T T L L T T T 8T L L L L / t / t / t / t T L T L /t

24 Exercises 5 P controller

25 Exercises 5 PI controller

26 Exercises 5 PID controller

27 Exercises 6 vas explained in Ch. 7, try to establish the controller settings by trial and error. Trial and error method Optimal parameter Kp t I t D = 50 = 8 = 2

28 Exercises 7 vwith proportional control only, increase K to K P P0, until oscillations in the response occur. Use this frequency, f 0, to set the controller according to the Ultimate Gain Method (K =0.45 K P P0, t =f 0 /1.2), where f I 0 is the frequency of the oscillations at K =K P P0 (see Ch. 7). Measure the response to a change in T 0 =K P0 Table. Controller settings based on process responses (Ultimate Gain) Cont roller Kp t I t D P 0.5K P0 PI 0.45K P0 f 0 /1.2 PID 0.6K P0 f 0 / 2 / 8 f 0

29 Exercises 7 vwith proportional control only, increase K to K P P0, until oscillations in the response occur. Use this frequency, f 0, to set the controller according to the Ultimate Gain Method (K =0.45 K P P0, t =f 0 /1.2), where f I 0 is the frequency of the oscillations at K =K P P0 (see Ch. 7). Measure the response to a change in T 0 =K P0 increase K P to K P0

30 Exercises 7 vwith proportional control only, increase K to K P P0, until oscillations in the response occur. Use this frequency, f 0, to set the controller according to the Ultimate Gain Method (K =0.45 K P P0, t =f 0 /1.2), where f I 0 is the frequency of the oscillations at K =K P P0 (see Ch. 7). Measure the response to a change in T 0 =K P0 Ultimate Gain method Kp = 45 t I = 4.17 f 0 = 4

31 Exercises 7 vwith proportional control only, increase K to K P P0, until oscillations in the response occur. Use this frequency, f 0, to set the controller according to the Ultimate Gain Method (K =0.45 K P P0, t =f 0 /1.2), where f I 0 is the frequency of the oscillations at K =K P P0 (see Ch. 7). Measure the response to a change in T 0 =K P0 T 0 (20 à 15) with PI controller (Ultimate Gain method)

32 Exercises 7 vwith proportional control only, increase K to K P P0, until oscillations in the response occur. Use this frequency, f 0, to set the controller according to the Ultimate Gain Method (K =0.45 K P P0, t =f 0 /1.2), where f I 0 is the frequency of the oscillations at K =K P P0 (see Ch. 7). Measure the response to a change in T 0 =K P0 T 0 (20 à 25) with PI controller (Ultimate Gain method)

33 Exercises 8 vadjust the controller by trial and error to obtain the best results to a change in set point. Set point change (80 ºC à 90 ºC )

34 Exercises 8 vadjust the controller by trial and error to obtain the best results to a change in set point. Set point change (80 ºC à 70 ºC )

35 Exercises 9 vinclude a loss term in the energy balance. Trial and error method (Q loss = 0)

36 Exercises 9 vinclude a loss term in the energy balance. Trial and error method (Q loss = 10%)

37 Exercises 9 vinclude a loss term in the energy balance. Trial and error method (Q loss = 20%)

38 Exercises 10 v Change the controller to give PID-control by adding a differential term,

39 Exercises 11 v Include a first order time lag in the temperature measurement using u the equations : dt dt m = T R - t m T m ε = TRSet -T m

40 Conclusions Ø A simple feedback control system involving a stirred tank, temperature measurement, controller and manipulated heater is shown. Ø The Cohen-coon method were also compared with the predicted values by using the controllers of P, PI and PID. Among them, the P controller gave the best results. Ø The best results of Ziegler-Nichols method follows in the order of PID > PI > P controller. Ø Trial and error method of controllers can be adjusted by changing the values of gain K P, reset time and derivative time.