CM 331 Process Control, Spring 217 Instructor: Dr. om Co Lecture 21 (Back to Process Control opics ) General Control Configurations and Schemes. a) Basic Single-Input/Single-Output (SISO) Feedback Figure 1. Basic Feedback Control Configuration. CM331 Lecture 21 Page 1 Remarks: 1. Process is often not considered for modification via controller design. Nonetheless, process design can often simplify the control issues. In Figure 1, we are assuming the process block already includes the actuator. 2. he set-point can be fixed (e.g. constant or step changes) or scheduled (e.g. with segments of ramp up/down ) 3. he most common approach in using feedback information is to obtain an error signal by taking the difference between measured value with the set-point as e = y setpoint y measured Afterwards, a control decision u is obtained by using the (present and past) error information using a chosen control algorithm (e.g. on-off, PID, etc.) which is then sent to the actuator (e.g. valve). (More details later) 4. here are two (often complementary) objectives of feedback control: a. Regulation (Disturbance compensation) to keep the process working at the desired point despite changes in the disturbance loads. b. Set-point tracking move the process to new desired positions fast but within other specifications and constraints (overshoots, oscillations, etc.) CM331 Lecture 21 Page 2
uning Methods: 1. Cohen-Coon a. Identify parameters of FOPD (first order plus time delay) step response K p = y ss / u ss (process gain), τ del (delay time) and τ (process time constant) CM331 Lecture 21 Page 3 b. Evaluate control parameters based on desired modes Let r = τ delay, τ K c τ I τ d P PI 1 K p r (1 + r 3 ) 1 K p r (.9 + r 12 ) τ delay 3 + 3r 9 + 2r PID 1 K p r (4 3 + r 4 ) τ delay 32 + 6r 13 + 8r τ delay 4 11 + 2r CM331 Lecture 21 Page 4
2. Autotune/Ziegler-Nichols/yreus-Luyben (see e.g. http://www.chem.mtu.edu/~tbco/cm416/zn.html and http://www.chem.mtu.edu/~tbco/cm416/atune.html ) 3. Via Simulation ypical Control Optimization Performance Criteria: - IAE (integrated absolute error) = e - ISE (integrated squared error) = e 2 dt - IAE (integrated time absolute error) = t e dt dt - ISE (integrated time squared error) = te 2 - IWSE (integrated weighted squared error = (e 2 + αu 2 ) dt dt CM331 Lecture 21 Page 5 b) Feedforward-Feedback Figure 2. Feedforward-feedback Control Configuration. CM331 Lecture 21 Page 6
Remarks: 1. Often used when detection of disturbance effects are too slow (i.e. lags and delays). 2. A feedforward control signal u feedforward is evaluated based on a predictive model of the disturbance effect on the process. his is then added to the control signal u feedback based on feedback error information. he sum u = u feedforward + u feedback is the signal sent to the actuator in the process. CM331 Lecture 21 Page 7 Example: Consider the following transfer function block diagram G p = Figure 2. Feedforward-Feedback. 5 (2s + 1) 2 ; G 2 d = 2s + 1 ; G.4s +.1 c = s (when G m = then no feedforward control) he equivalent transfer function equation is given by y = ( G cg p ) y 1 + G c G set + ( G d G m G p ) d p 1 + G c G p CM331 Lecture 21 Page 8
Q: How does one construct G m? A: We want (G d G m G p ) to get to zero as fast the dynamics allow. hus, the ideal case is (G m ) ideal = G d 2(2s + 1)2 = G p 5(2s + 1) Problem is that the order of the numerator is now higher than order in denominator. A compromise is to introduce extra terms such that the order of numerator is at least equal or less than order of denominator, e.g. add (s + 1) to denominator. hus, (G m ) real = 2(2s + 1) 2 5(2s + 1)(s + 1) = 8s2 + 8s + 2 1s 2 + 15s + 5 Figure 3 shows the comparison of responses with G m = (G m ) real and that with no feedforward for the case where y set = 1 and d = 1. CM331 Lecture 21 Page 9 Figure 3. Comparison of Feedback with Feedforward and without Feedforward control. CM331 Lecture 21 Page 1
c) Cascade Control Figure 3. Cascade Control Configuration. CM331 Lecture 21 Page 11 Remarks: 1. Consider cascade control when a. Inner loop is around three times faster than outer loop. b. Inner loop contains some nonlinear dynamics (e.g. stiction, hysteresis ) 2. radeoff issues a. Hardware: More sensors and controllers b. Software: added complexity and tuning issues CM331 Lecture 21 Page 12
d) Split-range Figure 4. Split range control configuration. CM331 Lecture 21 Page 13 Remarks: 1. If control signals ranges need to be implemented via different mechanisms (e.g. heating vs. cooling), a map is needed to specify the movements of each actuators (e.g. valves). Example: Figure 5. Example of a split range map. 2. Different maps may be needed at different operation modes. CM331 Lecture 21 Page 14
e) Ratio Figure 4. One implementation of ratio control Remarks: 1. Often used in blending (e.g. air/fuel) 2. Also used in distillation control (e.g. reflux/feed) ( If feed is considered as load disturbance, will have feedforward control aspect) CM331 Lecture 21 Page 15 f) Override Figure 5. Override control configuration. CM331 Lecture 21 Page 16
Remarks: 1. Often used for safety: a. Low or high selector decided based on safety consideration b. One of the controlled outputs chosen based on direct or indirect constraints 2. If the safety loop is on-off, then acts as safety interlock (hard override). 3. If the primary loop (one that addresses main operational objective) contains reset term (e.g. PI or PID), then anti-reset windup should be included when this controller was overrode. CM331 Lecture 21 Page 17