12/11/2021

Control System Technology 1 Controller Design Control System Technology 2 Controller Design

Objectives
To give an entry-level ability to discuss controller design, tune a
controller, complete a computer-aided Bode design of a controller,
and design simple compensation networks
After completing this chapter, you will be able to
[Link] the ultimate cycle method for tuning a controller
[Link] the process reaction method for tuning a controller
16. Controller Design [Link] self-tuning adaptive controllers
[Link] a computer-aided Bode design of a 𝑃𝐼𝐷 controller
[Link] compensation networks that will improve the response
of the control system

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 3 Controller Design Control System Technology 4 Controller Design

§[Link] §[Link] Ultimate Cycle Method

Controller design consists of the selection of the control modes The ultimate cycle method uses controller adjustment formulas
and control mode settings and/or compensation networks that will peak 1 period to determine the controller settings
result in a stable system that meets the control objectives 𝐴 peak 2 The formulas require
The control objectives may specify some or all of the following 𝐴/4
• minimum controller gain 𝐺𝑢
characteristics • period of the oscillation 𝑃𝑢
[Link] accuracy and speed of response of the measuring transmitter The quarter amplitude decay response is
a common control system design criterion
[Link] residual error allowed by the controller after a load change Control Mode Original Method Modified Method
[Link] response of the control system to a step change in load/setpoint. Propotional control
𝐾𝑃 = 0.5𝐺𝑢
Adjust the gain to obtain quarter-amplitude
(P) decay response to a step change in setpoint
Quarter amplitude decay or critical damping may be specified Proportional Integral 𝐾𝑃 = 0.45𝐺𝑢
1
𝐾𝐼 = (𝑚𝑖𝑛−1 )
control 2.0 𝑃𝑢
[Link] maximum frequency limit (PI)
𝐾𝐼 = (𝑚𝑖𝑛−1 ) Adjust the gain to obtain quarter-amplitude
𝑃𝑢 decay response to a step change in setpoint
[Link] response time, the rise time, and the settling time of the 1.5
𝐾𝑃 = 0.6𝐺𝑢 𝐾𝐼 = (𝑚𝑖𝑛−1 )
control system Proportional Integral 1.2 𝑃𝑢
Derivative control 𝐾𝐼 = (𝑚𝑖𝑛−1 ) 𝑃𝑢
𝑃 𝐾𝐷 = (𝑚𝑖𝑛−1 )
[Link] cost: hardware, maintenance, operating, reject (PID)
𝑢
𝑃𝑢
6
Adjust the gain to obtain quarter-amplitude
−1
product, environmental, … 𝐾𝐷 =
8
(𝑚𝑖𝑛 )
decay response to a step change in setpoint

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 5 Controller Design Control System Technology 6 Controller Design

§[Link] Ultimate Cycle Method §[Link] Ultimate Cycle Method

𝐾𝑃 𝐾𝑃
𝑅 𝐸 1 𝐶 𝑅 𝐸 1 𝐶
𝐾𝐼 Process 𝐾𝐼 Process
𝑠 𝑠

𝐾𝐷 𝑠 𝐾𝐷 𝑠

Closed-loop feedback control using a PID controller Closed-loop feedback control using a PID controller

Control Mode rise time overshoot settling time Steady-state error

Control Mode Original Method Modified Method
Propotional control Adjust the gain to obtain quarter-amplitude Propotional control
𝐾𝑃 = 0.5𝐺𝑢 decrease increase small change decrease
(P) decay response to a step change in setpoint (P)
1 Proportional Integral
Proportional Integral 𝐾𝑃 = 0.45𝐺𝑢 𝐾𝐼 = (𝑚𝑖𝑛−1 ) decrease increase increase decrease
2.0 𝑃𝑢 control (PI)
control
𝐾𝐼 = (𝑚𝑖𝑛−1 ) Adjust the gain to obtain quarter-amplitude Proportional Integral
(PI) 𝑃𝑢 small change decrease decrease nochange
decay response to a step change in setpoint Derivative control (PID)
1.5
𝐾𝑃 = 0.6𝐺𝑢 𝐾𝐼 = (𝑚𝑖𝑛−1 ) The general effects of each controller parameter (𝐾𝐼 , 𝐾𝑃 , 𝐾𝐷 ) on a closed-loop system
Proportional Integral 1.2 𝑃𝑢
Derivative control 𝐾𝐼 = (𝑚𝑖𝑛−1 ) 𝑃𝑢
𝑃𝑢 𝐾𝐷 = (𝑚𝑖𝑛−1 )
(PID) 6
𝑃𝑢 Adjust the gain to obtain quarter-amplitude
𝐾𝐷 = (𝑚𝑖𝑛−1 )
8 decay response to a step change in setpoint

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

1
 12/11/2021

Control System Technology 7 Controller Design Control System Technology 8 Controller Design

§[Link] Ultimate Cycle Method §[Link] Process Reaction Method

- Ex.16.1 A process control system is tested at start-up. The Assumption: the process can be approximated by a first-order
derivative mode is turned OFF, and the integral mode is set at lag plus dead-time model
the lowest setting. The gain is gradually increased until the The formulas obtained from an open-loop step response test
controlled variable starts to oscillate. The gain setting is 2.2 and
- Operate the process on manual control (open loop) until the
the period of oscillation is 12𝑚𝑖𝑛. Use the original ultimate cycle measured variable remains constant
method to determine the PID controller settings
- Study the step response
Solution 𝐶𝑚
final value
From the problem, the gain setting 𝐺𝑢 = 2.2 and the period of tangent to steepest

measured value, %
portion of curve
oscillation 𝑃𝑢 = 12𝑚𝑖𝑛
effective delay
∆𝐶𝑚
Then 𝐾𝑃 = 0.6𝐺𝑢 = 0.6 × 2.2 = 1.32 in minutes
2.0 2.0
𝐾𝐼 = = = 0.167𝑚𝑖𝑛−1 𝐿 ∆𝑡
𝑃𝑢 12 original value
time, minute 𝑡
𝑃𝑢 12 step input output response
𝐾𝐷 = = = 1.5𝑚𝑖𝑛
8 8 A first-order lag plus dead-time model of the process is derived from the graph with the
Proportional Integral Derivative Controller: 𝐾𝑃 = 0.6𝐺𝑢 , 𝐾𝐼 = 2.0/𝑃𝑢, 𝐾𝐷 = 𝑃𝑢 /8 dead time equal to 𝐿 and the first-order time constant equal to ∆𝑡
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 9 Controller Design Control System Technology 10 Controller Design

§[Link] Process Reaction Method §[Link] Process Reaction Method

- Process Reaction Method - Ex.16.2 During start-up, the control valve of a process control
𝐿 = effective delay, 𝑚𝑖𝑛 system was maintained constant until the controlled variable
∆𝐶𝑚 stopped changing and reached a steady value. Then the
𝑁= %/𝑚𝑖𝑛 control valve position was changed by ∆𝑃 = 10%. A response
∆𝑡
∆𝑃 = change in the manipulating element, % graph was obtained: 𝐿 = 5𝑚𝑖𝑛 , ∆𝑡 = 10𝑚𝑖𝑛 , 𝐶𝑚 0 = 40% ,
[Link] control (𝑃) 𝐶𝑚 ∞ = 48%. Use the process reaction method to determine
∆𝑃 the settings of a three-mode controller
𝐾𝑃 = Sol. 𝐶𝑚 ∆𝐶𝑚 48 − 40
𝑁𝐿 final value
tangent to steepest
𝑁= = = 0.8%/𝑚𝑖𝑛
∆𝑡 10
measured value, %

[Link] Integral control (𝑃𝐼) portion of curve

∆𝑃 10
∆𝑃 0.3 effective delay → 𝐾𝑃 = 1.2 = 1.2 =3
𝐾𝑃 = 0.9 , 𝐾𝐼 = 𝑚𝑖𝑛 −1 in minutes
∆𝐶𝑚 𝑁𝐿 0.8 × 5
𝑁𝐿 𝐿 0.5 0.5
𝐾𝐼 = = = 0.1𝑚𝑖𝑛 −1
[Link] Integral Derivative control (𝑃𝐼𝐷) 𝐿
𝐿 5
∆𝑃 0.5 original value
∆𝑡 𝑡
𝐾 = 0.5𝐿 = 0.5 × 5 = 2.5𝑚𝑖𝑛
𝐾𝑃 = 1.2 , 𝐾𝐼 = 𝑚𝑖𝑛 −1 , 𝐾𝐷 = 0.5𝐿𝑚𝑖𝑛 time, minute 𝐷
𝑁𝐿 𝐿
PID Controller: 𝑁 = ∆𝐶𝑚/∆𝑡, 𝐾𝑃 = 1.2∆𝑃/(𝑁𝐿), 𝐾𝐼 = 0.5/𝐿, 𝐾𝐷 = 0.5𝐿
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 11 Controller Design Control System Technology 12 Controller Design

§[Link]-Tuning Adaptive Controllers §[Link]-Tuning Adaptive Controllers

- A major problem in designing or tuning process control systems - Objective of self-tuning adaptive controllers: to maintain the
is that the process models are often very complex, difficult to desired control criteria when the load/production rate changes
obtain, and inaccurate - Adaptive controllers use a conventional 𝑃𝐼𝐷 algorithm within an
- The control engineer has three choices in handling the control outer loop that handles the adaptive algorithm. The outer loop
of a process whose model changes with changes in production observes the control system and uses these observations to
rate (or other reasons) determine the values of 𝐾𝑃 , 𝐾𝐼 , and 𝐾𝐷 that will achieve
• to tune the controller for the worst case and accept a sluggish optimum performance
response for other conditions - Many different techniques are used to adapt a controller to
• to change the controller mode settings every time the process changes in the process. Adaptive controllers fall into one of
model changes three categories
• to design a controller that will change itself every time the • use programmed adjustment of the controller gain
process changes • use a model of the process to determine the 𝑃𝐼𝐷 values
→ The self-tuning adaptive controller is an implementation of • use pattern recognition to determine the 𝑃𝐼𝐷 values
the third choice
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

2
 12/11/2021

Control System Technology 13 Controller Design Control System Technology 14 Controller Design

§[Link]-Tuning Adaptive Controllers §[Link]-Tuning Adaptive Controllers

Programmed Adaptive Controller Model-Based Adaptive Controllers
Programmed adaptive controller automatically adjusts the proportional Model-based adaptive controller uses an internal model of the
gain 𝑃 as a function of a number of process-related variables process to determine the optimum PID values
- P may be programmed as a function of the controlled variable, the Adaptive algorithm uses an identification/self-learning mode to
setpoint, the error, the controller output, or a remote input variable determine the model of the process: step changes are
- Adaptive algorithm can be set for independent gain adjustment introduced above and below the setpoint and observes the
using any combination of the process-related variables reaction until sufficient data have been obtained to establish a
- The controller could process reaction curve similar to the figure → process model is
final value
𝐶𝑚 obtained from the process
• adjust the controller gain according to the production rate to tangent to steepest

measured value, %
portion of curve reaction curve → 𝑃𝐼𝐷 values
maintain the quarter amplitude decay criteria for all are determined
effective delay
production rates, also ∆𝐶𝑚
in minutes Some algorithms also provide
• make an independent adjustment in the gain based on the 𝐿 a continuous update: process
value of the setpoint original value model is continuously updated,
∆𝑡 𝑡
time, minute then 𝑃𝐼𝐷 values accordingly
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 15 Controller Design Control System Technology 16 Controller Design

§[Link]-Tuning Adaptive Controllers §[Link]-Aided PID Controller Design

Pattern-Recognition Adaptive Controllers peak 1 period Proportional Response
A pattern-recognition adaptive controller 𝐴 peak 2 The proportional controller reduces the rise time, increases the
𝐴/4 overshoot, and reduces the steady-state error
uses a graph of error versus time similar to
the figure to obtain the optimum PID values Integral Response
The adaptive algorithm constantly examines the response of the An integral control has the effect of eliminating the steady-state
control system to naturally occurring disturbances caused by error, but it may make the transient response worse
changes in load or setpoint → search for peaks in the magnitude
Derivative Response
of the error to find two or three peaks → calculate the time
between peaks, the magnitudes of the peaks → determine the Derivative control “applies the brakes,” slowing the controlled
amplitude decay ratio, the period of the oscillation → compute variable just before it reaches its destination
the optimum 𝑃𝐼𝐷 values Most practical control systems use very small derivative time,
because the derivative response is highly sensitive to noise in
the process variable signal

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 17 Controller Design Control System Technology 18 Controller Design

§[Link]-Aided PID Controller Design §[Link]-Aided PID Controller Design

Ex.: Mass-Spring-Damper System 𝑋(𝑠) 1
𝐺 𝑠 = =
𝐾
𝑥(𝑡)
𝑀 𝑥ሷ + 𝑓𝑣 𝑥ሶ + 𝐾𝑥 = 𝑓 𝐹(𝑠) 𝑠 2 + 10𝑠 + 20
𝑓𝑣 𝑀
𝑓(𝑡) 𝑋(𝑠) 1 Define a 𝑃𝐼𝐷 controller in MATLAB using a TF model directly
→𝐺 𝑠 = =
𝐹(𝑠) 𝑀𝑠 2 + 𝑓𝑣 𝑠 + 𝐾 Kp = 1; Ki = 1; Kd = 1;
s = tf('s');
using the numerical values 𝑚 = 1𝑘𝑔, 𝑓𝑣 = 10𝑁𝑠/𝑚, 𝐾 = 20𝑁/𝑚,
C = Kp + Ki/s + Kd*s
𝐹 = 1𝑁, the TF
𝑋(𝑠) 1
𝐺 𝑠 = = Alternatively, MATLAB's pid object to generate an equivalent
𝐹(𝑠) 𝑠 2 + 10𝑠 + 20
continuous-time controller
The goal of the 𝑃𝐼𝐷 control problem is to show how each of the
C = pid(Kp,Ki,Kd)
terms, 𝐾𝑃 , 𝐾𝐼 , and 𝐾𝐷 , contributes to obtaining the common goals
- Fast rise time
Open-Loop Step Response
- Minimal overshoot
s = tf('s');
- Zero steady-state error P = 1/(s^2 + 10*s + 20);
step(P)
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

3
 12/11/2021

Control System Technology 19 Controller Design Control System Technology 20 Controller Design

§[Link]-Aided PID Controller Design §[Link]-Aided PID Controller Design

Proportional Control 𝑅 𝐸 𝐶
𝑋 1 𝐾𝑃 reduces 𝑡𝑟 , increases %𝑂𝑆, 𝐾𝑃 𝐺(𝑠)
𝐺= =
𝐹 𝑠 2 + 10𝑠 + 20 and reduces 𝑒𝑠𝑠
Open-Loop Step Response 𝐾𝑃 𝐺(𝑠) 𝐾𝑝
𝑇 𝑠 = =
s = tf('s'); 1 + 𝐾𝑃 𝐺(𝑠) 𝑠 2 + 10𝑠 + (20 + 𝐾𝑃 )
P = 1/(s^2 + 10*s + 20);
step(P) Kp = [100, 200, 300]; t = 0:0.01:1.5;
for i=1:3
- The final value of the output to a unit step input 𝑐𝑠𝑠 = 0.05 → C = pid(Kp(i)); T = feedback(C*P,1);
large steady-state error 𝑒𝑠𝑠 = 1 − 0.05 = 0.95 step(T,t);
hold on;
- The rise time 𝑡𝑟 = 0.884𝑠, and the settling time 𝑡𝑠 = 1.59𝑠
end
→Let's design a controller that will reduce 𝑡𝑟 , 𝑡𝑠 , and eliminate the
→ the proportional controller 𝐾𝑃 reduced both 𝑡𝑟
steady-state error 𝑒𝑠𝑠 = 0 and 𝑒𝑠𝑠 , increased %𝑂𝑆, and decreased 𝑡𝑠 by
a small amount

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 21 Controller Design Control System Technology 22 Controller Design

§[Link]-Aided PID Controller Design §[Link]-Aided PID Controller Design

Proportional-Derivative Control 𝑅 𝐸 𝐾𝑃 𝐶 Proportional-Integral Control 𝑅 𝐸 𝐾𝑃 𝐶
𝐺(𝑠) 𝐺(𝑠)
Addition of 𝐾𝐷 tends to reduce 𝐾𝐷 𝑠
Addition of 𝐾𝐼 tends to decrease 𝑡𝑟 , 𝐾𝐼
1
𝑠
both %𝑂𝑆 and 𝑡𝑠 increase %𝑂𝑆, 𝑡𝑠 , and reduces 𝑒𝑠𝑠
(𝐾𝑃 + 𝐾𝐷 𝑠)𝐺(𝑠) 𝐾𝑃 + 𝐾𝐷 𝑠 (𝐾𝑃 + 𝐾𝐼 /𝑠)𝐺(𝑠) 𝐾𝑃 𝑠 + 𝐾𝐼
𝑇 𝑠 = = 𝑇 𝑠 = =
1 + (𝐾𝑃 + 𝐾𝐷 𝑠)𝐺(𝑠) 𝑠 2 + (10 + 𝐾𝐷 )𝑠 + (20 + 𝐾𝑃 ) 1 + (𝐾𝑃 + 𝐾𝐼 /𝑠)𝐺(𝑠) 𝑠 3 + 10𝑠 2 + (20 + 𝐾𝑃 )𝑠 + 𝐾𝐼

Kp = 300; Kd=[1,10,100]; t = 0:0.01:1.5; Kp = 30; Ki=[70,90,110]; t = 0:0.01:1.5;

for i=1:3 for i=1:3
C = pid(Kp,0,Kd(i)); T = feedback(C*P,1) C = pid(Kp,Ki(i)); T = feedback(C*P,1)
step(T,t) step(T,t)
hold on; hold on;
end end
→ the addition of the derivative term 𝐾𝐷 reduced → the integral controller 𝐾𝐼 also reduces 𝑡𝑟 and
both %𝑂𝑆 and 𝑡𝑠 , and had a negligible effect increases %𝑂𝑆 as 𝐾𝑃 does (double effect)
on 𝑡𝑟 and 𝑒𝑠𝑠 the integral controller eliminated the steady-
state error 𝑒𝑠𝑠
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 23 Controller Design Control System Technology 24 Controller Design

§[Link]-Aided PID Controller Design §[Link]-Aided PID Controller Design

𝐾𝑃
Proportional-Integral-Derivative 𝑅 𝐸 1 𝐶 Note for Designing a PID Controller
𝐶(𝑠) 𝐾𝐼
𝑠
𝐺(𝑠)
- Set all gains to zero, obtain an open-loop response and
𝑇 𝑠 =
𝑅(𝑠) 𝐾𝐷 𝑠 determine what needs to be improved
𝐾𝐷 𝑠 2 + 𝐾𝑃 𝑠 + 𝐾𝐼 - Increase the 𝐾𝑃 gain until the response to a step input is
= 3
𝑠 + (10 + 𝐾𝐷 )𝑠 2 + (20 + 𝐾𝑃 )𝑠 + 𝐾𝐼 steady oscillation, to improve the rise time 𝑡𝑠
- Increase the 𝐾𝐷 gain until the oscillations go away, to reduce
Kp = 350; Ki=300; Kd=50; the overshoot %𝑂𝑆
t = 0:0.01:1.5;
- Repeat steps 2 and 3 until increasing the 𝐾𝐷 gain does not
C = pid(Kp,Ki,Kd);
stop the oscillations
T = feedback(C*P,1)
step(T,t) - Set 𝐾𝑃 and 𝐾𝐷 to the last stable values
- Increase the 𝐾𝐼 gain to reduce the steady-state error 𝑒𝑠𝑠 until
→ the closed-loop system with no overshoot 𝑒𝑠𝑠 → 0
%𝑂𝑆, fast rise time 𝑡𝑟 , and no steady-state
error 𝑒𝑠𝑠
Keep in mind that do not need to implement all three controllers
(𝐾𝑃 , 𝐾𝐼 , and 𝐾𝐷 ) into a single system, if not necessary
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

4
 12/11/2021

Control System Technology 25 Controller Design Control System Technology 26 Controller Design

§[Link]-Aided PID Controller Design §[Link]-Aided PID Controller Design

Automatic PID Tuning P Tuning
- MATLAB provides tools for automatically choosing optimal 𝑃𝐼𝐷 s = tf('s');
gains which makes the trial and error process described above P = 1/(s^2 + 10*s + 20);
unnecessary. The tuning algorithm can be accessed directly pidTuner(P,'p')

using pidtune or through a nice graphical user interface (GUI)

using pidTuner
- The MATLAB automated tuning algorithm chooses PID gains to
balance performance (response time, bandwidth) and
robustness (stability margins). By default, the algorithm designs
for a 60-degree phase margin

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 27 Controller Design Control System Technology 28 Controller Design

§[Link]-Aided PID Controller Design §[Link] Design of A Three-Loop Control System

PID Tuning The process blends and heats a mixture of water and syrup, and
s = tf('s'); has three control loops: concentration, temperature, and level
Kp = 350; Ki=300; Kd=50; C = Kp + Ki/s + Kd*s
syrup level
P = 1/(s^2 + 10*s + 20); controller
pidTuner(P,C)
level
water sensor

heating fluid concentration

controller

blending and
concentration
heating tank
analyzer
(mixer not shown)

𝐿2
temperature
𝐿
controller

product to bottling line

Temperature
measuring transmitter

The blending and heating process

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 29 Controller Design Control System Technology 30 Controller Design

§[Link] Design of A Three-Loop Control System §[Link] Design of A Three-Loop Control System
Concentration control loop: manipulate the syrup input flow rate to Level control loop: manipulate the water input flow rate to maintain
maintain the desired concentration of syrup in the finished product the level of liquid in the blending tank
syrup level syrup level
controller controller

level level
water sensor water sensor

heating fluid concentration heating fluid concentration

controller controller

blending and blending and

concentration concentration
heating tank heating tank
analyzer analyzer
(mixer not shown) (mixer not shown)
𝐿2

𝐿2

temperature temperature
𝐿 𝐿
controller controller

product to bottling line product to bottling line

Temperature Temperature
measuring transmitter measuring transmitter

The blending and heating process The blending and heating process

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

5
 12/11/2021

Control System Technology 31 Controller Design Control System Technology 32 Controller Design

§[Link] Design of A Three-Loop Control System §[Link] Design of A Three-Loop Control System
Temperature control loop: manipulate the flow rate of the heating Blending and Heating Process Specifications
fluid to maintain the temperature of the liquid in the blending tank [Link] [Link] Measuring
Production rate: 2.5 × 10−4 𝑚3/𝑠 Transmitter
Composition: 900 water + 10% syrup Model: 1𝑠𝑡 order lag + dead time
syrup level
controller Density, 𝜌: 1005𝑘𝑔/𝑚3 Time constant, 𝜏1 : 200𝑠
Viscosity, 𝜇: 0.1𝑃𝑎. 𝑠 Dead-time delay, 𝑡𝑑1: 15𝑠
level Specific heat, 𝐻ℎ : 4170𝐽/𝑘𝑔 0𝐾 Gain: 1 (%output/%input)
water sensor
[Link] Tank [Link] Measuring Transmitter
heating fluid concentration Diameter, 𝐷: 0.75𝑚 Model: 1𝑠𝑡 order lag + dead time
controller
Height: 1𝑚 Time constant, 𝜏2 : 2𝑠
blending and
Operating level, ℎ: 0.75𝑚 Dead-time delay, 𝑡𝑑2: 0.5𝑠
concentration
heating tank
analyzer
Product temp.: 600 𝐶 Gain: 1 (%output/%input)
(mixer not shown)
Heating fluid temp.: 1000 𝐶
Film coef., ℎ𝑓 = ℎ𝑜 : 1349𝑊/𝑚2 0𝐾
𝐿2

temperature
[Link] Measuring
𝐿
controller Outlet pipe diameter: 0.0351𝑚 Transmitter
Distance from outlet Model: Over-damped 2𝑛𝑑 order lag + dead time
product to bottling line
Temperature to temp. probe: 0.26𝑚 Time constant, 𝜏3 : 50𝑠
measuring transmitter 𝜏4 : 240𝑠
Wall thickness: 0.01𝑚
The blending and heating process Wall material: steel Gain: 1 (%output/%input)

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 33 Controller Design Control System Technology 34 Controller Design

§[Link] Design of A Three-Loop Control System §[Link] Design of A Three-Loop Control System
Blending and Heating Process Specifications - Ex.16.4 Design a PID controller for the concentration control
[Link] Valves [Link] Measuring loop in the blending (water and syrup) and heating process
Model: underdamped 2𝑛𝑑order lag Model (𝜏5 ): 1𝑠𝑡 order lag
Water valve, 𝜔01 : 10.2𝑟𝑎𝑑/𝑠 Gain (𝐺): 1 (%output/%input) Solution
ζ1: 0.75
Gain: 5 (%output/%input) [Link] Process
[Link] the concentration process time constant, 𝜏5
Syrup valve, 𝜔02 : 2.4𝑟𝑎𝑑/𝑠 Model: Integral (nonregulating) Liquid volume 𝑉 = area × operating height
ζ2 : 0.90 𝐹𝑆𝑖𝑛 : 0.01𝑚3 /𝑠
Gain: 31.25 (%output/%input) 𝐹𝑆𝑜𝑢𝑡 : 1𝑚 𝜋𝐷2 ℎ 𝜋 × 0.752 × 0.75
21.6𝑟𝑎𝑑/𝑠
= = = 0.3313𝑚3
Heating valve,𝜔03 : 4 4
ζ3 : 0.8 [Link] Process −4 3
Gain: 8 (%output/%input) Model (𝜏6 ,𝑡𝑑3): 1𝑠𝑡 order lag Liquid flow rate 𝑄 = 2.5 × 10 𝑚 /𝑠
Gain (𝐺): 1 (%output/%input) 𝑉 0.3313
𝜏5 = = = 1325𝑠
𝑄 2.5 × 10−4
[Link] the syrup control valve TF coefficients
2ζ2 2 × 0.9 1 1
𝐵 1 = = = 0.75 𝐵 2 = 2 = = 0.1736
𝜔02 2.4 𝜔02 2.42

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 35 Controller Design Control System Technology 36 Controller Design

§[Link] Design of A Three-Loop Control System §[Link] Design of A Three-Loop Control System
[Link] the overall TF of the measuring transmitter, - Ex.16.5 Design a PID controller for the temperature control
process, and control valve loop in the blending and heating process
𝑒 −15𝑠 1 31.25 Solution
𝑇𝐹 = × ×
1 + 200𝑠 1 + 1325𝑠 1 + 0.75𝑠 + 0.1736𝑠 2 [Link] the thermal process time constant, 𝜏6
[Link]: This is the same TF as the blending system in Ex.16.3. Thermal resistance (from the program THERMRES)
The following summary was obtained from a run of DESIGN ℎ1 = ℎ0 = 1349𝑊/𝑚2 0𝐾
(see Ex.16.3)
𝜋𝐷2 𝐷 0.75
Design Summary 𝐴= + 𝜋𝐷ℎ = 𝜋𝐷 = 𝜋 × 0.75 × = 2.209𝑚2
Proportional gain: 1.85 4 4+ℎ 4 + 0.75
Integral action rate: 2000𝑠 −1 𝑥 = 0.01𝑚
Derivative action time constant: 106𝑠
Derivative limiter: 0.100 𝐾 = 45𝑊/𝑚 0𝐾
M (−180): −9.3𝑑𝐵 From THEMERES: 𝑅𝑇 = 7.718 × 10−4 0𝐾/𝑊
ANGLE (ODB): −1400
Gain margin: 9.3𝑑𝐵
Phase margin: 400
Frequency limit: 3 × 10−2𝑟𝑎𝑑/𝑠

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

6
 12/11/2021

Control System Technology 37 Controller Design Control System Technology 38 Controller Design

§[Link] Design of A Three-Loop Control System §[Link] Design of A Three-Loop Control System
Thermal capacitance [Link] the overall TF of the measuring transmitter,
𝐶 = 𝑚𝐻ℎ process, and control valve
𝜌ℎ𝜋𝐷2 1005 × 0.75 × 𝜋 × 0.752 1
𝑚= = = 333𝑘𝑔 𝑇𝐹 = ×
4 4 1 + 290𝑠 + 12.000𝑠 2
𝐶 = 333 × 4170 = 1.389 × 106 𝐽/ 0𝐾 𝑒 −𝑠 8
×
𝜏6 = 𝑅𝐶 = 7.71 × 10−4 × 1.389 × 106 = 1070𝑠 1 + 1070𝑠 1 + 0.0741𝑠 + 0.00214𝑠 2
[Link] the thermal dead-time lag, 𝑡𝑑3 [Link] following summary was obtained from a run of DESIGN
Design Summary
𝑄 2.5 × 10−4 Proportional gain: 2.33
Outlet flow rate = 𝑈 = = = 0.258𝑚/𝑠 0.001𝑠 −1
𝐴 0.03512 𝜋/4 Integral action rate:
Derivative action time constant: 197𝑠
distance 0.26 Derivative limiter: 0.100
Dead time lag = 𝑡𝑑3 = = = 1𝑠
velocity 0.258 M (−180): −12.7𝑑𝐵
ANGLE (ODB): −1400
c. Determine the temperature measuring transmitter TF Gain margin: 12.7𝑑𝐵
2ζ3 2 × 0.8 1 1 Phase margin: 400
𝐵 1 = = = 0.0741 𝐵 2 = 2 = = 0.00214 Frequency limit: 1.6 × 10−2𝑟𝑎𝑑/𝑠
𝜔03 21.6 𝜔03 21.62
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 39 Controller Design Control System Technology 40 Controller Design

§[Link] Design of A Three-Loop Control System §[Link] Design of A Three-Loop Control System
- Ex.16.6 Design a PID controller for the level control loop in the [Link] the overall TF of the measuring transmitter,
blending and heating process process, and control valve
Solution 𝑒 −0.5𝑠 1 5
𝑇𝐹 = × ×
[Link] the level process integral action time constant 1 + 2𝑠 442𝑠 1 + 0.147𝑠 + 0.00961𝑠 2
𝜋𝐷2 𝜋 × 0.752 [Link] following summary was obtained from a run of DESIGN
𝐴= = = 0.4418 Design Summary
4 4 Proportional gain: 74.4
𝐹𝑆𝑜𝑢𝑡 1 Integral action rate: 1𝑠 −1
𝑇1 = 𝐴 = 0.4418 = 442𝑠 Derivative action time constant: 2.43𝑠
𝐹𝑆𝑖𝑛 0.001
Derivative limiter: 0.100
[Link] the water control valve TF coefficients M (−180): −6.0𝑑𝐵
2ζ1 2 × 0.75 ANGLE (ODB): −134.80
𝐵 1 = = = 0.147 Gain margin: 6.0𝑑𝐵
𝜔01 10.2 Phase margin: 450
1 1 Frequency limit: 1.3𝑟𝑎𝑑/𝑠
𝐵 2 = 2 = = 0.00961
𝜔01 10.22

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 41 Controller Design Control System Technology 42 Controller Design

§[Link] System Compensation §[Link] System Compensation

- Use for servo control systems: position, velocity, force control of - Designer begins the servo controller design with an analysis of
mechanical loads, … a simple loop closure. If a loop gain adjustment
- The component transfer functions in a servo loop are well • satisfies the performance objectives → designer simply
defined and seldom involve dead-time delays provides the means of establishing the necessary loop gain.
- The design of a servomechanism This may involve adjusting the gain of one or more
components in the loop, or it may involve the addition of an
• Establish the performance objectives: static accuracy,
amplifier in the loop
response time, overshoot, stability criteria, …
• Selects the system components: a servo actuator, a power • does not satisfy the performance objectives → designer must
supply, a drive amplifier, and a feedback transducer, … then modify the open-loop frequency response of the system
+change/modify components in the loop
• Use component transfer functions to construct the open-loop
Bode diagram or root-locus plot of the system +insert additional components into the loop → frequency
compensation or simply compensation
- The derivative and integral modes of a 𝑃𝐼𝐷 controller are
examples of compensation
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

7
 12/11/2021

Control System Technology 43 Controller Design Control System Technology 44 Controller Design

§[Link] System Compensation §[Link] System Compensation

- The Bode diagrams of the following four simple compensation - The Bode diagrams of the following four simple compensation
networks networks
40 40 0 0 40 40 0 0
Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

gain gain gain gain gain gain gain gain
Gain (dB)

Gain (dB)
Gain (dB)

Gain (dB)

Gain (dB)

Gain (dB)

Gain (dB)

Gain (dB)
20 20 −20 −20 20 20 −20 −20
0 0 −40 −40 0 0 −40 −40
0 90 0 0 0 90 0 0
phase phase phase phase phase phase
phase phase
−90 0 −90 −90 −90 0 −90 −90
0.1 1 10 1 1 0.1 1 10 1 1 0.1 1 10 1 1 0.1 1 10 1 1
𝑇1 𝑇1 𝑇1 𝑇𝑙𝑒𝑎𝑑 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑒𝑎𝑑 𝑇1 𝑇1 𝑇1 𝑇𝑙𝑒𝑎𝑑 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑒𝑎𝑑

An integration-lead A lead-lag A lag compensation A lag-lead An integration-lead A lead-lag A lag compensation A lag-lead
compensation networkcompensation network network compensation network compensation networkcompensation network network compensation network

Straight-line Bode diagrams of four compensation networks. Observe Straight-line Bode diagrams of four compensation networks. Observe
the similarity between network (a) and the integral mode, between network (b) the similarity between network (a) and the integral mode, between network (b)
and the derivative mode, and between network (c) and a low-pass filter and the derivative mode, and between network (c) and a low-pass filter

• Integration-lead network compensation • Lead-lag network compensation 𝑇𝐿𝑒𝑎𝑑 > 𝑇𝐿𝑎𝑔

𝑉𝑜𝑢𝑡 1 + 𝑇1 𝑠 𝑉𝑜𝑢𝑡 1 + 𝑇𝐿𝑒𝑎𝑑 𝑠
= (16.11) = (16.12)
𝑉𝑖𝑛 𝑇1 𝑠 𝑉𝑖𝑛 1 + 𝑇𝐿𝑎𝑔 𝑠
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 45 Controller Design Control System Technology 46 Controller Design

§[Link] System Compensation §[Link] System Compensation

- The Bode diagrams of the following four simple compensation - The Bode diagrams of the following four simple compensation
networks networks
40 40 0 0 40 40 0 0
Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

Phase Angle (°)

gain gain gain gain gain gain gain gain
Gain (dB)

Gain (dB)
Gain (dB)

Gain (dB)

Gain (dB)

Gain (dB)

Gain (dB)

Gain (dB)
20 20 −20 −20 20 20 −20 −20
0 0 −40 −40 0 0 −40 −40
0 90 0 0 0 90 0 0
phase phase phase phase phase phase
phase phase
−90 0 −90 −90 −90 0 −90 −90
0.1 1 10 1 1 0.1 1 10 1 1 0.1 1 10 1 1 0.1 1 10 1 1
𝑇1 𝑇1 𝑇1 𝑇𝑙𝑒𝑎𝑑 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑒𝑎𝑑 𝑇1 𝑇1 𝑇1 𝑇𝑙𝑒𝑎𝑑 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑎𝑔 𝑇𝑙𝑒𝑎𝑑

An integration-lead A lead-lag A lag compensation A lag-lead An integration-lead A lead-lag A lag compensation A lag-lead
compensation networkcompensation network network compensation network compensation networkcompensation network network compensation network

Straight-line Bode diagrams of four compensation networks. Observe Straight-line Bode diagrams of four compensation networks. Observe
the similarity between network (a) and the integral mode, between network (b) the similarity between network (a) and the integral mode, between network (b)
and the derivative mode, and between network (c) and a low-pass filter and the derivative mode, and between network (c) and a low-pass filter

• Lag network compensation • Lead-lag network compensation 𝑇𝐿𝑒𝑎𝑑 < 𝑇𝐿𝑎𝑔

𝑉𝑜𝑢𝑡 1 𝑉𝑜𝑢𝑡 1 + 𝑇𝐿𝑒𝑎𝑑 𝑠
= (16.13) = (16.14)
𝑉𝑖𝑛 1 + 𝑇𝐿𝑎𝑔 𝑠 𝑉𝑖𝑛 1 + 𝑇𝐿𝑎𝑔 𝑠
HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien

Control System Technology 47 Controller Design Control System Technology 48 Controller Design

§[Link] System Compensation §[Link]

- integration-lead compensation ≈ integral controller
1 Section 16.2: 16.01, 16.02
𝑉𝑜𝑢𝑡 1 + 𝑇1 𝑠 𝑉𝑜𝑢𝑡 1 + 𝐼 𝑠 𝐼 + 𝑠
= → = = 𝐿 (16.15) Section 16.3: 16.03
𝑉𝑖𝑛 𝑇1 𝑠 𝑉𝑖𝑛 1 𝑠
𝑠
𝐼 Section 16.5: 16.07, 16.10
- lead-lag compensation ≈ derivative controller
𝑉𝑜𝑢𝑡 1 + 𝑇𝐿𝑒𝑎𝑑 𝑠 𝑉𝑜𝑢𝑡 1 + 𝐾𝐷 𝑠
= → = , 0 < 𝛼 < 1(16.16) Group working: 5 students
𝑉𝑖𝑛 1 + 𝑇𝐿𝑎𝑔 𝑠 𝑉𝑖𝑛 1 + 𝛼𝐾𝐷 𝑠
Dead time: 2 weeks after class
- lag compensation ≈ low-pass filter
→ used to remove undesirable high-frequency components of a signal
- lag-lead compensation → used to cancel dominant low-frequency
lead components in much the same way that lead-lag
compensation is used to cancel dominant low frequency lag
components

HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien HCM City Univ. of Technology, Faculty of Mechanical Engineering Nguyen Tan Tien