Process control

Course Objectives
➢ What is the process control?
➢ Basis of process control
➢ Basic hardware and instrumentation
➢ Process dynamics (modeling)
➢ Analysis of dynamic systems
(Source: Petro Times)
➢ Design of feedback controllers (PID)
➢ Various control strategies (Feedforward, IMC, cascade)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
References

Books
1. Dale E. Seborg, Thomas F. Edgar, Duncan A. Mellichamp, Process
Dynamics and Control, John Wiley & Sons, 2011.
2. Donald R. Coughanowr, Process System Analysis and Control,
McGraw-Hill International Editions, 2009.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Assessments
▪ Exercises at classes and homework: 20%
▪ Midterm examination: 30%
▪ Final examination: 50%

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
A. Introduction

Input Output
Raw Product
Plant

Requirements:
1. Safety: 2bar, 2-136 °C
2. Production specifications:
10000 kg/h, 99.5%
3. Environment: CO2, SO2…
4. Operational constraint:
250-514 °C
5. Economic: Qmin
Source:https://doi.org/10.14483/22487638.11499
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
There are three issue:
1. Suppressing the influence external disturbance:

Objective:
t = tr
f i , ti

Sensor h,t
t
Start up -> stability
tr f o , to

; If f st = cont Then
e
u fi change t
FCV

Controller

steam f st ti change ; If f st = cont Then t

Controller

t  tr
e > 0 more steam
e = tr − t e < 0 reduce steam
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
2. Ensuring stability
t
 =  o t is disturbed
 =  o +  t  tr
tr  No need control
 =0

tr   =  o t is disturbed

 =  o +  t  tr
f i , ti
need control
h,t
3. Optimize the performance Sensor A->B->C
t
max imize = cos t of purcha sin g A + tr f o , to
e
( revenue from the sale of product B  − cos t of steam)dt
tr

u

FCV
Controller
o

Q(steam)->min steam f st

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and Instrumentation Diagram (P&ID)
• The system is too complex to describe in text.
• We must use standard symbols.
• Piping and instrumentation (P&I) drawings provide projects.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
We must use standard symbols. Why?
if you use symbols in a non-standard fashion, be sure to do it consistently
so your readers understand your meaning for that symbol each time they
see it

Source: https://instrumentationandcontroltoday.blogspot.com

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and instrumentation (P&I) drawings

- No line: The instrument is

located in the field near the
process & operator.
– Solid line: The instrument is
located in a control room
(accessible to the operator).
– Dotted line: The instrument is
not directly accessible.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and instrumentation (P&I) drawings
Instruments that are
mounted on process plant
(i.e sensor that mounted on
pipeline or process
equipments.

Field mounted
on pipeline

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and instrumentation (P&I) drawings

Instruments that are

mounted behind a
control panel board

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and instrumentation (P&I) drawings

Instruments that
are mounted on
control board

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and instrumentation (P&I) drawings
Instruments that are functioned in Distributed
Control System (DCS)

A distributed control system (DCS) refers to

a control system usually of a manufacturing
system, process or any kind of dynamic
system, in which the controller elements are
not central in location (like the brain) but are
distributed throughout the system with each
component sub-system controlled by one or
https://tuoitre.vn/nha-may-
loc-dau-dung-quat-lap-dat-
more controllers. The entire system of
he-thong-camera-quan- controllers is connected by networks for
sat-bao-ve-294741.htm communication and monitoring
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Piping and instrumentation (P&I) drawings
FC Flow Controller PT Pressure Transmitter
FE Flow Element PTD Pressure Transducer
FI Flow Indicator
FT Flow Transmitter LC Level Controller
FS Flow Switch LG Level Gauge
FIC Flow Indicating Controller LR Level Recorder
FCV Flow Control Valve LT Level Transmitter
FRC Flow Recording Controller LS Level Switch
LIC Level Indicating Controller
PC Pressure Controller LCV Level Control Valve
PG Pressure Gauge LRC Level Recording Controller
PI Pressure Indicator
PR Pressure Recorder TE Temperature Element

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and instrumentation (P&I) drawings

PS Pressure Switch TI Temperature Indicator

PIC Pressure Indicating Controller TR Temperature Recorder
PCV Pressure Control Valve TS Temperature Switch
PRC Pressure Recording Controller TC Temperature Controller
PDI Pressure Differential Indicator TT Temperature Transmitter
PDR Pressure Differential Recorder
PDS Pressure Differential Switch
PDT Pressure Differential Transmitter

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and instrumentation (P&I) drawings

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Piping and instrumentation (P&I) drawings
Signal Lines Symbology

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

Source: https://instrumentationandcontroltoday.blogspot.com

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

Source: https://instrumentationandcontroltoday.blogspot.com

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
We are going to divide the valve types
in five groups:

Source: https://instrumentationandcontroltoday.blogspot.com
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Valves and actuators can be represented with different drawings like the
instruments
With or without Positioned & Actuator: With or without handwheel:

With the fail position: With the status during normal operation:

Source: https://instrumentationandcontroltoday.blogspot.com
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Transducer and transmitter
The instruments usually have: the sensor (transducer) and the transmitter.
The sensor (transducer) interprets the measured variable and converts it
to other measurable energy
Example a resistance temperature detector
transducer (PT100) converts the
temperature in a resistance value.
This value is not usually sent
directly to the control cabinet,
instead it is converted in a head-
mounting transmitter. The
transmitter sends the final signal (4-
2mA) to the control cabinet
Source: https://instrumentationandcontroltoday.blogspot.com
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
- The head-mounting transmitter can be -Also, It can be indicated with the transducer
represented like the following drawings: or the thermowell.

- If the temperature transmitter isn’t head-

mounting, the transmitter can be installed in the
cabinet or in a junction box

Source: https://instrumentationandcontroltoday.blogspot.com
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
P&ID instrument codes

Example: Source: https://instrumentationandcontroltoday.blogspot.com

The “TAG” (name) is "20-FT-1982-A". This code will not be repeated for other plant devices.
-First part: “the Unit”. It indicates two digits "20". The instrument is installed in the unit “20”.
-Second part: "the instrument type”. It indicates two letters "FT". It is a flow transmitter.
- Third part: “the Loop Number”. It indicates "1982". Each control loop is named with a sequential number. A “Loop” is a
set of devices with a control logic link.
-Fourth part: “Suffix”. “A” indicates that there is a redundant device at least, as the "A" and "B" in the previous drawing

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
P&ID Title block

Source: https://instrumentationandcontroltoday.blogspot.com
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Example 1

With using these following symbols

LC
LC
Feed z
LV 200
V 200
LT

Complete control loop for LCV 201

LCV 201

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Example 2
PE Where PE is locally mounted on V-100

PRV 200 PT Where PT is locally mounted

PIC
PIC Where PIC is function in DCS
PE
Feed z
LV 200
V 200
PT

Draw control loop to show that PRV-100 will be activated to relief

pressure when the pressure in the V-100 is higher than desired value.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

Exercise 1
pHIC 1
pHE 1 pHT 1

pHIC 2
pHE 2 pHT 2

CV-102 TK-102
The diagram shows pH adjustment;
(base feed tank) part of waste water treatment process.
With using above symbols, draw
control loop where the process need
is:
The process shall maintained at
pH=6. When the process liquid states
below pH=6, CV-102 will be opened
to dosing NaOH to the tank TK-100.
When the process liquid states above
CV-101 pH=6, CV-101 will be operated to
TK-100 TK-101
dosing HCl.
(pH adjustment tank) (acid feed tank)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
B. Design Aspects of process control system
• Variable in process control systems.

• Select measurements (select controlled variable).

• Select manipulated variable.

• Select control configuration

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
B. Design Aspects of process control system
DV

MV Process system CV

Variable in process control

- Disturbance variable (DV)
- Manipulated variable (MV) Depend on objective
- Controlled variable (CV)
Example: f ,t i i

Objective: h, t
Sensor h,t
t
CV MV DV
tr f o , to
e h fi , ( f o ) f i , ti
u
FCV

Controller
t f st , fi , ti f i , ti
steam f st

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Depend on control objectives Select processing variable
and measured variable (or and measurements
unmeasured variable) - temperature (t)
Product(vapor)
Cooling water
- pressure (p)
- flow (f)
- level (l)
Product (Liquid)
- composition (c)
CIC CA

Feed z
f CV MV

f , ( )
Product (c%)
f st Actuator: Controller: Sensor: cr
Control PLC Composition
valve analyzer

Feedback control systems

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Product(vapor) Cooling water

CV MV
CIC Product (Liquid)

z
CA
f cr f , ( )
Product (c%)
Steam f st
Feed forward control systems
Product(vapor) Cooling water

CV MV
CIC Product (Liquid)

f , ( )
Computer TT
with t i
Measurements
and estimate
composition (c)
TT f
cr
Feed z Product (c%)

Steam f st
Measurement: ti to estimate composition
Inferential control systems
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Control configuration DVi

MVi Process system CVi

Depend on how many MV and CV we have in systems. We can

distinguish the control configuration as either single-input, single-output
(SISO, i = 1) or multiple-input, multiple output (MIMO, i>= 2)
DVi
DVi DVi
MVi CVi Controller MVi CVi
Process system Process system

CVi
Controller MVi Process system Controller Estimator

Feedback Control Feedforward Control Inferential Control

configuration configuration configuration
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
C. Modeling the dynamic and static behavior of process
•We (the researchers) need process model. Why?
•Classification of process model
•How to build process model?
fi fi
h??? if fi is from fi1 to fi2
f i2 - To understand the process control.

h f i1 - To train plant operating personnel.

fo
A  - To design the control strategy for new
process/ to design controller.
h
h2 -Optimize process operating condition:
Process model +economic information
h1 ->determine the most profitable
MV CV
Process  operating conditions.
0  0

We need process model and CV=f(MV,…) is a mathematical model

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Input change Process Effect on output variable
fi
h
fi
h2
f i2
h1
f i1
h 
 fo 0  0
A

To design the
How does the -How far?
Math models control strategy
process
help us answer -How fast for new process/
influence the
these questions! to design
response? “Shape”
controller
Classification of process model
- Theoretical model (TM): Principle conservation.
- Empirical model (EM): By fitting experimental data.
- Semi-empirical model: combination of TM+EM.
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
How to build process model?
Six (6) - step modelling procedure

1. Define Goals
We apply this procedure
2. Prepare information
• to many physical systems
3. Formulate the model • overall material balance

4. Determine the solution • component material balance

• energy balances
5. Analyze Results

6. Validate the model

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
f i , ti
1.Define Goals

Sensor h,t
t
tr f o , to
e
u

FCV
Controller

steam f st

liquid level → total mass in liquid

pressure → total moles in vapor
What variable?
temperature → energy balance
concentration → component mass
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
2. Prepare information
- Define system
MV CV
Process - State assumptions
- Collect data
fi

f i2
f i1

h
h2

h1

0  0

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
3. Formulate the model
fi
Objective: h = hr
Level (h) Total mass balance
m3 kg
h Rv
fo
fi , f o ( );  ( 3 )
s m

 Rate of mass   Rate of mass   Rate of mass 

 = − 
 accumulation   input   output 
d (V  )
= i f i − o f o
d
h dh h
if f o = A + = fi Differential equation
Rv d Rv (first order)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
4. The solution
- Initial condition h = ?????
Example: h = 0, fi = 0 at τ=0
- Some change to input variable
Example: fi = 2.5 at τ>=0
dh h dh h dh
A + = fi  A = fi − A = d
d Rv d Rv fi −
h
Rv
h
h −
dh  A h 
A = d  ln f i − = h = Rv f i  (1 − e ARv
h −1 0 )
0 fi −
0 Rv h =0
Rv Rv  h 
 f i − 
Rv
hn = hn −1 + ( )  
 A 
Or using numerical methods, example: Euler method  
n −1

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
5. Analyze Results
u(V) fi = 20 cm3 / min; Rv = 2,5; A = 218,34 cm 2
K u (cm3 /( phút.V ))
3
P F(
i cm / phút)
 ( kg / cm 3 )
h(cm) Process −
Rv fi h
A(cm ) 2
Fo (cm3 / phút)
h = Rv f i  (1 − e ARv
)
c. Trong trường hợp, Fo = Rv.h, Ku = 0.2. Ban đầu mức ở
ở trong bình h = 0 cm. Sau đó bơm được cấp nguồn với
cấp điện áp u =100V. Sau 10 phút mức ở trong bình là 20 fi
30 cm. Mức khi đạt trạng thái tĩnh là 50 cm:
0 
c1: Xác định hệ số cản trở của van (Rv)?. Bán kính của ARv1  ARv 2  ARv 3
bình (ống) chứa? h
Rv fi
c2: Vẽ đồ thị mức theo thời gian với 0< t <30 phút.
0 
c3: Tại thời điểm t = 30 phút điện áp cấp cho bơm là
120V; Vẽ đồ thị mức theo thời gian với 30< t < 50 phút.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Dynamic simulation using MATLAB
Apply the Laplace transforms to equation

dh h Laplace H (s) K p0
H (s)
A + = fi AsH ( s ) + = Fi Gp0 = =
d Rv Rv Fi ( s )  p 0 s + 1

2.5 Where,
Gp0 = ; Rv = 2,5; A = 218.34
545.45s + 1 Kp0 = Kv: Process gain.

2.5 τp0 = AKv: Effective process time

Gp0 = ; Rv = 2,5; A = 50 , time constant .
125s + 1
2.5
Gp0 = ; Rv = 2,5; A = 300
750 s + 1
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Dynamic simulation using MATLAB

Conclusion?????????
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
C1. Time delay in process dynamics and control
Time delay play an important role in process control system. They
commonly occur as a result of the transport time (θ) required for fluid (or
materials) to flow (or to transfer) through a pipe (or devices).
f ( )
The f d ( ) is the function f(t) delayed
by θ time unit. Therefor, f d (t ) and f(t)
 are related by
f d ( ) = f ( −  )1( −  )
f d ( )

Where, 1(τ-θ) is the unit step function


And, Laplace transformation of the f d (t )
 of time
Fd ( s ) = e − s F ( s )
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
C2. Time delay in process dynamics and control
Fi

H (s) 3
G p (s) = =
h Rv
Fo
Fi ( s ) 4 s + 1
Fi θ=2 dead time

H ( s ) 3e −2 s
G p (s) = =
h Rv
Fo
Fi ( s ) 4 s + 1

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
C3. Linearization of nonlinear models
▪ (𝑥𝑠 , 𝑢𝑠 ) are the steady-state of the system, we have:
dx
= f ( xs , u s ) = 0 (c3.1)
d
▪ The equation describing the system is differentiable at the
steady state
A linear approximation of this equation can be obtained by
using a Taylor series expansion and truncating after the first-
order terms
Given
 x = xs + x (c3.2)

u = u s + u
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
C3. Linearization of nonlinear models

dx f f
 f ( xs , u s ) + x + u (c3.3)
d x xs ,us u xs ,us

Example:
Determine the mathematic
fi ,ρ model of the system, given
the outlet flow rate:

h
fo ,ρ f o ( h) = Rv h (c3.4)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
C3. Linearization of nonlinear models
We have the differential equation:
dh
A = f i − f o (h) (c3.5)
d
Because of the nonlinearity of the right term, we could not apply
the Laplace transforms. Then using Taylor expansion at the
steady-state

f o ( h) = Rv h  f o  f o (hs ) + f o (hs )[h − hs ]

1 Rv (c3.6)
 f o ( h)  Rv hs + [ h − hs ]
2 hs

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
C3. Linearization of nonlinear models
Replace eq. (c3.6) into (c3.5)
dh 1 Rv
A = f i − Rv hs − [h − hs ] (c3.7)
d 2 hs
At the steady-state, we have:
f is = Rv h s (c3.8)

Add (c3.7) and (c3.8)

dh 1 Rv
A + f is = f i − [h − hs ]
d 2 hs (c3.8)
dh 1 Rv
 A + [ h − hs ] = f i − f is
d 2 hs
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
C3. Linearization of nonlinear models
Given h = h − hs , f 
= f i − f is
Eq. (c3.8) is rewritten
d ( h − hs ) 1 Rv
A + [ h − hs ] = f i − f is
d 2 hs
dh 1 Rv 
 A + h = f (c3.9)
d 2 hs
Apply the Laplace transforms to eq. (c3.9)
1 Rv 1 Rv
AsH ( s ) + H ( s ) = F ( s )  [ As + ]H ( s ) = F ( s ) (c3.10)
2 hs 2 hs

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
C3. Linearization of nonlinear models
The transfer function of linearized model
2 hs
H (s) 1 Rv K p0 (c3.11)
 = G p 0 (s) = = =
F (s) As +
1 Rv 2 A hs  p0s + 1
2 hs s +1
Rv

Where, 2 hs
K p0 =
Rv

2 A hs
 p0 =
Rv
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
C4. Concentration control
fi ; c Ai Objective: c = cr
Concentration (C)

V ; cA Component mass balance

f o ; c Ao Units:
m3  mg  Source:https://knightequip.
fi , f o ( ); c Ai , c A ; c Ao  3 ;0 − 1  com/cconcentration.html
s m 
 Rate of accumulation   Rate of input   Rate of generation /   Rate of output 
 =
    − 
 of component A   of component A   dissappearace of component A   of component A 
d (Vc A )
= fi c Ai  rAc AV − f o c Ao
d
dV dc A
cA +V = f i c Ai  rAc AV − f o c Ao
d d
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
dV
In level control = fi − f o
d
And c Ao = c A
dc A
cA ( fi − f o ) + V = f i c Ai  rAc AV − focA
d
dc A
V + cA fi = f i c Ai  rAc AV
d
dc A Case of rA ≈ 0
V + cA fi = f i c Ai  rAc AV
d
dc A
V + cA fi = f i c Ai
d
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Example Dòng lỏng của thứ nhất có lưu lượng khối lượng wi1 (kg/phút) và
thành phần là cA1 (nồng độ phần trăm 0-100%; 0-1). Dòng lỏng thứ
hai có lưu lượng khối lượng wi2 và thành phần là cA2. Dòng lỏng lấy
w i1 ; c A1 w i2 ; c A 2 ra khỏi bình có lưu lượng khối lượng là wo và thành phần là cAo, được
điều khiển bởi một van tuyến tính và wo = Rv.h1/2. Trong trường hợp
ứng dụng này, cA1 = cA2 = 0,5; wi1 và wi2 được duy trì là hằng số và là
120 kg/phút và 100 kg/phút. Bình chứa là hình trụ tròn có chiều cao
h  2,5m và đường kính 2m. Chất lỏng chảy vào bình là như nhau có
w o ; c Ao khối lượng riêng là 800 kg/m3. Hệ thống hoạt động tại một điểm tĩnh
A với h = 1,75m, cAo = 0,5. Với các thông số của hệ thống như trên, xác
định:
a. Chứng tỏ rằng, thành phần của sản phẩm lấy ra là
mg
wi1 , wi 2 ( ); c Ai , c A ; c Ao ( 0 − 1) dcAo
=
1
 wi1 (cA1 − wAo ) + wi2 (cA2 − wAo )
min d  Ah
b. Xác định lưu lượng lấy ra (wo) và hệ số cản trở của van lúc này?
c. Nếu cA1 thay đổi đột ngột từ 0.5 đến 0.6 (lưu lượng wi1, wi2 và cA2 không đổi). Lúc này, thành
phần của sản phẩm lấy ra khỏi bình (cAo) sẽ thay đổi. Giá trị cAo ở trạng thái xác lập (cuối cùng)
là bao nhiêu?. Thời gian cần thiết để cAo đạt được 99% so với giá trị cuối cùng?
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Solution
* mass balance
mg
wi1 , wi 2 ( ); c Ai , c A ; c Ao ( 0 − 1)
min
dm dh
= wi1 + wi 2 − w0  A = wi1 + wi 2 − w0 ; ( )
d d
* component balance

d (Vc A )
 = wi1c A1 + wi 2 c A 2 − wo c Ao
d
dV dc A
 cA + V = wi1c A1 + wi 2 c A 2 − wo c Ao
d d
dh dc A
 cA A + V = wi1c A1 + wi 2c A 2 − wo c Ao ; ()
d d
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Substituting for dh/dτ (*)

dc A
c A ( wi1 + wi 2 − w0 ) + V = wi1c A1 + wi 2c A 2 − wo c Ao
d
Note, CAo = CA

dc A 1
=  wi1 (cA1 −cA ) + wi 2 (cA2 − cA )
d  Ah
- How does the process influence the response ????????
- How to control CA ???????????

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Trong trường hợp ứng dụng này, cA1 = cA2 = 0,5; wi1 và wi2 được duy trì là hằng số và là 120
kg/phút và 100 kg/phút. Hệ thống hoạt động tại một điểm tĩnh với h = 1,75m, cAo = 0,5

0 = wi1 + wi 2 − w0  w0 = wi1 + wi 2 = 120 + 100 = 220kg / min

wo 220
Rv = = = 166.41
h 1.75

Nếu cA1 thay đổi đột ngột từ 0,5 đến 0,6 (lưu lượng wi1, wi2 và cA2 không đổi). Lúc này, thành
phần của sản phẩm lấy ra khỏi bình (cAo) sẽ thay đổi. Giá trị cAo ở trạng thái xác lập (cuối cùng)
là bao nhiêu?. Thời gian cần thiết để cAo đạt được 99% so với giá trị cuối cùng?

1
0 =  wi1 (cA1 −cA ) + wi 2 (cA2 − cA ) 
 Ah
wi1c A1 + wi 2 c A 2 120 x0.6 + 100 x0.5
cA = = = 0.555
wi1 + wi 2 120 + 100
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
dc A 1
= 120(0.6 −cA ) + 100(0.5 − cA )
d 800 x3.14 x 2 x1.75
dc A
= 0.0138 − 0.025c A
d
dc A
= d
0.0138 − 0.025c A
cAo2 
dc A 1
c 0.0138 − 0.025cA =  d  ln(0.0138 − 0.025c A ) c = 
cAo2

Ao1 0
−0.025 Ao1

with cAo1 = 0.5; cAo2 (99%c Ao ) = 0.555 − 0.555 x0.01 = 0.549

τ=116 min
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Dynamic simulation using MATLAB
dc A 1
=  wi1 (cA1 −cA ) + wi 2 (cA2 − cA )
d  Ah
dc A 1 1
+ ( wi1 + wi 2 ) c A =  wi1cA1 + wi 2cA2 
d  Ah  Ah
Because both inputs influence the dynamic of cA, it is necessary to
develop two transfer function for model
cA2 = 0
dc A 1 1
+ ( wi1 + wi 2 ) c A = wi1c A1
d  Ah  Ah
Laplace
C A (s) wi1
= G p 01 ( s ) =
C A1 ( s )  Ahs + ( wi1 + wi 2 )
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
cA1 = 0
dc A 1 1
+ ( wi1 + wi 2 ) c A = wi1c A 2
d  Ah  Ah
Laplace
C A (s) wi 2
= G p 02 ( s ) =
C A2 (s)  Ahs + ( wi1 + wi 2 )
The model of systems
120 100
C A ( s ) = G p 01 ( s )  CA1 + G p 02 ( s )  CA2 = CA1 + CA2
8792 s + 1 8792 s + 1

0.545 0.4545
C A (s) = CA1 + CA2
39.96 s + 1 39.39 s + 1
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
cA2 = 0 C A (s) 0.545
=
CA1 39.96 s + 1

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
C5. Temperature control

f i ,  , c p , ti
h,t
V , t , , cp

f o ,  , c p , to
steam
f st Source: https://www.pinterest.com
FCV
Q

m3 kg kJ
fi , f o ( );  ( 3 ); c p ( o ); t ( oC )
s m kg . C
 Rate of energy   Rate of generation 
 Rate of energy   Rate of energy   Rate of energy     
 = −   proviced / removed   / disappearance 
 accumulation   input   output     of energy 
 by heater / coolant   
d (mc p t )
= f i  c p ti − f o  c p t o  Q − (H r )rV
d
d (V  c p t )
= f i  c p ti − f o  c p t o  Q − (H r )rV
d
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
d (V  c p t )
= f i  c p ti − f o  c p t o  Q − (H r )rV
d
dV dt
 c pt +  c pV = f i  c p ti − f o  c p to  Q − (H r )rV ≈ 0
d d
Note : dt Q
V = (ti − t ) f i 
d cp 
to = t
 - How does the process influence the response ????????
 dV - How to control t ???????????
 = fi − f 0
 d
dt Q
V + f i t = f i ti 
d cp 

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Case of, electrical heated stirred tank with constant fo
f i ,  , c p , ti
h,t f o ,  , c p , to
V , t , , cp

Suppose that the metal heating element has a significant thermal

capacitance and that the electrical heating rate Q directly affects the
temperature of the element rather than the liquid contents
t Q
Ae
te P = he Ae (te − t ), W

Where, te is temperature of surface of heating element; he , Ae is the

product of the transfer coefficient and area available for hear transfer.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
So, dt he Ae (te − t )
V + f i t = f i ti + ()
d cp 
The transfer of the metal heating element
me ce dte
= Q − he Ae (te − t ) ()
d
dt
m m + c p  f i t + he Aet − f i ti c p 
(*) and V = te = d
cp he Ae
Substituting for te (**)

mme ce d 2t  me ce me ce m  dt me ce dti 1
+ + +  +t = + ti + Q

 fi he Ae d  he Ae  f i c p  f i
(***)
 d he Ae d  fic p
2

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
ti = 0

mme ce d 2t  me ce me ce m  dt 1
+ + +  +t = Q

 fi he Ae d  he Ae  f i c p  f i  d  fic p
2

Laplace

1
T (s)  fi c p
= = G1 ( s )
Q( s) mme ce 2  me ce me ce m 
s + + +  s + 1
 fi he Ae 
 he Ae  f i c p  f i 
 Process control
Q= 0

mme ce d 2t  me ce me ce m  dt me ce dti
+ + +  +t = + ti
 fi he Ae d  he Ae  f i c p  f i
2
 d he Ae d

Laplace

me ce
s +1
T (s) he Ae
= = G2 ( s )
Ti ( s ) mme ce 2  me ce me ce m 
s + + +  s + 1
 fi he Ae h A fc  fi
 e e i p 

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Summary
1 me ce
s +1
 fi c p he Ae
T (s) = Q( s) + Ti ( s )
mme ce 2  me ce me ce m  mme ce 2  me ce me ce m 
s + + +  s + 1 s + + +  s + 1
 fi he Ae h A fc  fi  f i he Ae h A fc  fi
 e e i p   e e i p 

Example
m mc mc 1
= 10 min; e e = 1, 0 min; e e = 1, 0 min; = 0, 05o C min/ kcal
 fi  fi c p he Ae  fi c p
Q = 5000kcal / min; ti = 100o C ;1kcal = 4,18855kJ
a. Calculate the nominal steady-state temperature?

b. Assume that the process is initially at steady-state determined in part (a). Calculate the
response, t, to a sudden change in Q from 5000 to 5400 kcal/min
mme ce
c. Suppose that it can be assumed that the term (  f h A ) is small relative to other terms in (***)
i e e

Calculate the response, t

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Solution
a.The steady-state form of (***) is
1
t s = ti + Q = 100 + 0, 05.5000 = 350o C
 fi c p

b and c.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
C6. Pressure control

https://uniduc.com/vi/blog/
(A.A. Kopyrin, A.I. Karelin, V.A. Karelin, 2006) may-chiet-rot-dang-ap
PYC
2
CV PT
p1 fi P 2

Pressure p2 , f 2
vA , p FCV
tank
2
p
vB V1
fo p1 , f1
V1

T1
C1

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Mathematical model of pressure control
Suppose the relation between pressure, volume, and temperature of air in
tank can be applied to ideal gas Laws
pv = nRt (C 6.1)

pv A = nRt (C 6.2)
dp dv A
vA +p =0 (t = const ) (C 6.3)
d d
Since v A + vB = C (const ) (C 6.4)
dv A dvB
+ =0 (C 6.5)
d d
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
dvB
In level control = fi − f o (C 6.6)
d
dp p
(C6.3), (C6.5) and (6.6), we have: = ( fi − f o ) (C6.7)
d v A
ps
At the steady-state, 0= ( fis − f os ) (C 6.8)
v As
d ( p − ps ) ps
(C6.7), (C6.8) we have:  ( fi − fis ) − ( f o − f os ) (C 6.9)
d v As
Given p  = p − ps , f i  = f i − f is , f o  = f o − f os (6.10)
dp ps
(C6.9), (6.10) we obtain   f i  − f o  (C 6.11)
d v As
Consider flow, fi , running through CV-valve, we obtain
fi = K v p1 − p (C 6.12)
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
− Kv
Differential (C6.12) df i = dp (C 6.13)
2 p1 − p
Or, − Kv
fi − f s = ( p − ps ) (C 6.14)
2 p1 − p

  − Kv
(C6.10) and (C6.14) fi = Kp , where : K = (C 6.15)
2 p1 − p

(C6.11) and (C6.16)

dp
1 + p = − K1 f o (C6.17)
d

1 − Kv v As
where : K1 = , K = ; 1 =
K 2 p1 − p Kps
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
D. Empirical model identification

Out-comes:
-Experimental design for model building
-Process reaction curve (graphical)
-Statistical parameter estimation

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

Start • Base case operating conditions

• Definition of perturbation
Experimental Design
• Measures
• Duration
Plant Experimentation

Determine Model Structure • Safely

• Small effect on product quality
Parameter Estimation • Small effect of profit

Diagnostic Evaluation
• What order, dead time, etc?
Model Verification

Complete

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Start

Experimental Design

Plant Experimentation

Determine Model Structure • Gain, time constant, dead time ...

Parameter Estimation
• Does the model fit the data used to
Diagnostic Evaluation evaluate the parameters?

Model Verification
• Does the model fit a new set of data
Complete
not used in parameter estimation.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

Start

Experimental Design • What is our goal?

Plant Experimentation We seek models good enough for

control design, controller tuning,
Determine Model Structure and process design.
Parameter Estimation • How do we know?

Diagnostic Evaluation

Model Verification

Complete

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
▪ Open-loop identification

Step input Process

Process output

▪ Closed-loop identification

Controller Process
Input Process
output

In this course, only open-loop identification is introduced. Profound

knowledge of system identification, students read other related books

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Open-loop identification -Process reaction curve - The simplest and
most often used method. Gives nice visual interpretation as well.

fi ; c Ai
1. Start at steady state
2. Single step to input V ; cA
f o ; c Ao
fi
3. Collect data until steady state
f i2
4. Perform calculations f i1

h
h2

h1

0  0

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
The transfer function of the first-order process
A(0, 63CV)
CV2
ΔCV
Kp =
ΔMV CV
τ p = τ 63% CV(s) Kp
CV1 G p (s) = =
MV2 MV(s) τ ps +1
ΔMV
MV1
τ 63% τ
time
τ0 τ1

∆𝐶𝑉
- Step 1: Determine 𝐾𝑝 =
∆𝑀𝑉

- Step 2:
+ Determine the point A at value 𝐶𝑉𝐴 = 0.632∆𝐶𝑉 on the graph
+ Determine the time corresponding with A on the time axis, 𝜏0
+ 𝝉𝒑 = 𝝉𝟔𝟑% = 𝝉𝟏 − 𝝉𝟎
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
The transfer function of first-order plus time delay models (FOPTD)
CV2
ΔCV 0.63CV
Kp = -θ s CV2
ΔMV
CV(s) K p e
p
CV
τ p = τ 63% a = maximum slope
CV1 G p (s) = = CV
MV(s) τ ps +1
MV2 ΔCV CV1
ΔMV Kp =
MV1 ΔMV θp
p τ ΔCV τp
time τp =
τ 63% a MV2
θ p = show in figure MV ΔMV
τ0 τ1 1
τ
∆𝐶𝑉 τ0 τ1 time
- Step 1: Determine 𝐾𝑝 = ∆𝑀𝑉
Process reaction curve - Method II
- Step 2: The two points method was proposed
CV2
+ From the step response curve we determine the times τ1 , τ2 that correspond to the 0, 63CV
28% and 63% the steady state value of the response, respectively.
0, 28CV CV
+ Determine the parameters by equations: τ p = 1,5( 63% -  28% )
ΔCV CV1
θ p =  63% -  p Kp =
ΔMV τ 28%
τ p = 1,5( 63% -  28% ) τ 63%
+ Note: Another two points can be used: 35% and 85%, then:
MV2
τ p = 0, 67( 85% -  35% ) θ p =  63% -  p ΔMV
MV1
θ p = 1,3 35% - 0, 29 85% τ 0 τ1 τ 2 time
τ

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Start Should we use this data?
CV2
Experimental Design Input should be close to a CVm
perfect step; this was basis of
equations. If not, cannot use
Plant Experimentation data for process reaction CV1

curve. good experimental design

MV2
Determine Model Signal/noise > 5 ΔMV
MV1
Structure τ
τ0 time
CVm
Parameter Estimation good experimental design
CV2
Not good experimental design CVm
MV2
Diagnostic Evaluation ΔMV
MV1
τ CV1
τ0 time
Model Verification
it
MVchecks for disturbances
The output must be “moved” 2
ΔMV
MV1
Complete enough!!! τ
τ0 time

Experiments were conducted in triplicate. All values were obtained in the average ± standard deviation (n = 3)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Process reaction curve - Method II
Start
CV2
0.63CVm

Experimental Design 0.28CVm CVm

-θps
ΔCVm CV1 CVm (s)K pe
Kp =
ΔMV τ 28% G p (s) = =
Plant Experimentation
τ p = 1,5( 63% -  28% ) τ 63% MV(s) τ ps +1
MV2
θ p =  63% -  p ΔMV
MV1
Determine Model τ
Structure τ 0 τ1 τ 2 time

55

Parameter Estimation 50

45

40

Diagnostic Evaluation 35
Temperature (oC)

30 45 oC,1.8 kW/m2

25 - Calculation Using statistical method

o Experiment
20
Model Verification 15

10

Complete 0
0 0.5 1 1.5 2 2.5
Time (s) 4
x 10

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Statistical analysis method
To evaluate the fitness of the mathematical model, these following
equations (D.1), (D.2), and (D.3) including, R2 (coefficient of
determination), RMSE (root mean square error), and MRE (mean
relative percentage error) were considered
σ𝑁
𝑖=1(𝑦𝑒𝑥𝑝,𝑖 −𝑦𝑝𝑟𝑒,𝑖 )
2
𝑅2 =1− σ𝑁 2 (D.1)
𝑖=1(𝑦𝑒𝑥𝑝,𝑖 −𝑦𝑎𝑣 )

1 𝑁
𝑅𝑀𝑆𝐸 = σ (𝑦 −𝑦𝑝𝑟𝑒,𝑖 )2 (D.2)
𝑁 𝑖=1 𝑒𝑥𝑝,𝑖

𝑦𝑒𝑥𝑝,𝑖 −𝑦𝑝𝑟𝑒,𝑖
100 σ𝑁
𝑖=1 𝑦𝑒𝑥𝑝,𝑖
𝑀𝑅𝐸 = (D.3)
𝑁

where, yexp,i, ypre,i, yav, and N are the measured data from the experiment, the predicted values,
the average experimental values, and the number of experiments, respectively
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Statistical analysis method
If the MRE value must be less than 10% then the modeling of the
process can be accepted.

If the R2 value more than 95% then the modeling of the process can be
accepted.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Example
A process consist of stirred tanks with input Q and output t. A step
change in Q is made from 82 W to 85 W, with output responses given in
table
Time t (oC) Time t (oC) Time t (oC)
(min) (min) (min)
0 10.00 8 17.44 16 17.96
1 12.40 9 17.60 17 17.97
f i ,  , c p , ti
2 13.82 10 17.50 18 17.98 h,t f o ,  , c p , to
V , t , , cp
3 15.16 11 17.86 19 17.80
4 15.89 12 17.89 20 17.99
Q
5 16.49 13 17.89 21 18.00
6 17.10 14 17.92 22 18.00
7 17.22 15 18.10

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Solution
CV2 0.63CVm
20.00 ΔCV
Kp =
18.00 ΔMV CV
16.00
τ p = τ 63%
CV1
14.00
MV2
12.00 ΔMV
MV1
10.00
τ 63% τ
time
t 8.00 τ0 τ1
6.00

4.00

2.00 CV(s) T(s) Kp

G p (s) = = =
0.00
0 5 10 15 20 25 MV(s) Q(s) τ ps +1
time (min)

CV 18 − 10 8 oC 2, 67 2, 67 3 2, 67
Kp = = = , G p (s) = T(s) = Q(s)  = 
MV 85 − 82 3 W 3s +1 3s +1 s 3s +1

0, 63CV = 5, 4 oC ;  p = 3, min

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
-1
 3 2, 67 
t( ) = t(0) + L T(s) = 10 + L  
-1

 s 3s +1 
= 10 + 3  2, 67 (1 − e − / 3 )
19.00 Response curve

18.00

17.00

t (exp) t(pre)
16.00

15.00

14.00

13.00

12.00

11.00

10.00 time (min)

0 5 10 15 20 25
𝑦𝑒𝑥𝑝,𝑖 −𝑦𝑝𝑟𝑒,𝑖
100 σ𝑁
𝑖=1 𝑦𝑒𝑥𝑝,𝑖 The modeling of the process
𝑀𝑅𝐸 = MRE=0.33% can be accepted
𝑁

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
E. Approximation for higher order systems
How to study higher-order transfer function models ??? example
K (−0,1s + 1)
G (s) = (E.1)
(5s + 1)(3s + 1)(0,5s + 1)
Answers: A approach for approximating higher-order transfer function
models with lower-order models (example 𝐺 ∗ ) that have similar dynamic
and steady-state characteristics. (Why?) The lower-order models are
more convenient for control system design and analysis
− s
Ke
G (s) = (E.2)

Method:  s +1
 2 2
s  3 3
s
- Taylor series expansion: e − s = 1 −  s + − + ...
2! 3!
 e − s  1 −  s
(E.3)
s
or e  1 +  s
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
- The 1/1 Padé approximation:

1− s
e − s  G1 ( s ) = 2 (E.4)

1+ s
2

- Skogestad’s “Half Rule” (2003): from higher – order transfer function

+ Determine the largest neglected time constant (𝜏𝑚 )
+ One-half of its (𝜏𝑚 ) value is added to a new time delay (𝜃)
+ Other half is added to the largest time constant (𝜏𝑟 )
+ Time constants (𝜏𝑖 ) that are smaller than the 𝜏𝑚 are approximated as
time delay using (E.3)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Application for (E.1)
- Using Taylor series expansion
K (−0,1s + 1) Ke − s
G (s) = 
(5s + 1)(3s + 1)(0,5s + 1)  s + 1

Time constant (5) is retained, applying the approximation in (E.2) gives

−0,1s + 1 = e −0,1s ;3s + 1 = e3 s ;0,5s + 1 = e0,5 s

K (−0,1s + 1) K .e −0,1s K .e −3,6 s  = 3, 6

G (s) =  =
(5s + 1)(3s + 1)(0,5s + 1) (5s + 1).e .e
3 s 0,5 s
(5s + 1)  =5
K (−0,1s + 1) Ke −3,6 s
G (s) = 
(5s + 1)(3s + 1)(0,5s + 1) 5s + 1
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Application for (E.1)
- Skogestad approximation method to obtain a 1𝑠𝑡 - order model:
𝜏𝑖
K (−0,1s + 1) Ke − s
G (s) = 
(5s + 1)(3s + 1)(0,5s + 1)  s + 1
𝜏𝑟 𝜏𝑚 𝜏𝑖

1 1
𝜏 = 𝜏𝑟 + 𝜏𝑚 =5+ . 3 = 6,5
2 2
1 1
𝜃 = 𝜏𝑚 + σ 𝜏𝑖 = . 3 + 0,5 + 0,1 = 2,1
2 2

K (−0,1s + 1) Ke −2,1s
G (s) = 
(5s + 1)(3s + 1)(0,5s + 1) 6,5s + 1
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Comparation the actual and approximate models
Actual Skogestad
K (−0,1s + 1) K ( −0,1 s + 1) Ke −2,1s
G (s) = G (s) = 
(5s + 1)(3s + 1)(0,5s + 1) (5s + 1)(3s + 1)(0,5s + 1) 6,5s + 1
Taylor series
K (−0,1s + 1) Ke −3,6 s Conclusion????
G (s) = 
(5s + 1)(3s + 1)(0,5s + 1) 5s + 1

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Example
Temperature Control System
Where,
YIC
55 𝐾𝑝0 𝐾𝑝𝑐
𝐺𝑝0 = ;𝐺 = ;𝐺
𝜏𝑝0 𝑠 + 1 𝑝𝑐 𝜏𝑝𝑐 𝑠 + 1 𝑠
Power
TCC
55 controller TI 𝐾𝑠 𝐾𝑑
= ;𝐺 =
𝜏𝑠 𝑠 + 1 𝑑 𝜏𝑑 𝑠 + 1
55

Heat chamber
Air inlet Air outlet
tin, vin tout, vout
Heat resistor
Heat chamber
𝐶𝑉 𝑠
DV(s) 𝐾𝑝0 𝐾𝑝𝑐 𝐾𝑠
Gd(s) = × × 𝑀𝑉 𝑠
𝜏𝑝0 𝑠 + 1 𝜏𝑝𝑐 𝑠 + 1 𝜏𝑠 𝑠 + 1
SP E(s)
GC(s)
MV(s)
Gpc(s) GP0(s)
CVm(s) 𝐾𝑑 𝐾𝑠
- + × 𝐷𝑉 𝑠
𝜏𝑑 𝑠 + 1 𝜏𝑠 𝑠 + 1
CV GS(s)

𝐾𝑝 𝑒 −𝜃𝑝𝑠 𝐾𝑑1 𝑒 −𝜃𝑑 𝑠

𝐶𝑉 ≈ 𝑀𝑉 𝑠 + 𝐷𝑉 𝑠
𝜏𝑝 𝑠 + 1 𝜏𝑑1 𝑠 + 1

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
F. Interacting process and block diagram representation
Two tanks in series whose liquid levels interact
fi fo h2
(F.1) Rv2
d
A1 A2 A2
d
(F.2) f o1 fo h2
h1 p1 f o1 Rv1 h2 Rv2 f Rv2
p2 o
d
A2
h2 d
fo = (F.1) (F.4) fi f o1 fo
Rv2
h2
Rv 2 &(F.5) d Rv1
A1 g
A2 dh2 d h
= f o1 − f o (F.2) 1

d
p1 =  gh1 A2 s
(F.3)
p1 − p2 = Rv1 f o1 p2 =  gh2 Fi ( s ) Fo1 ( s ) Fo ( s )
Rv2
H 2 (s)
Rv1
g A1dh1 A1s g H 2 (s)
=?
f o1 = ( h1 − h2 ) (F.4) = fi − f o1 (F.5)
d
H1 ( s ) Fi ( s )
Rv1
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
G. Feedback control for process control systems
Illustrative example: Schematic diagram for level control systems
V 200 LC Set point

Feed z
LV 200

LT
h I/P

Area:A

LCV 201

Basic component in feedback control loop are:

- Sensor-transmitter combination (LT)
- Feedback controller (LC)
- Current-to-pressure transducer (I/P)
- Final control element (LCV)
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Simple diagram of a feedback controller V 200 LC Set point

Feed z
LV 200

LT
h I/P

Area:A

Set point 𝑢𝑠𝑝

LCV 201

Input signal 𝑢𝑠 Output signal 𝑢𝑐

(from transmitter) (to control valve)
Controller

The controller compares the measured value to the set point and takes the
appropriate corrective action by calculating the controller output and
transmitting it as an electronic signal to control valve
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control

sp e mv Process cv m
Calculating the function
-
cv Sensor

Error signal: e(τ) = sp 𝜏 − 𝑐𝑣(𝜏)

Laplace MV ( s )
For proportional control, 𝑚𝑣(τ)= 𝑚𝑣𝑐𝑏𝑖𝑎𝑠 𝜏 + 𝐾𝑐 𝑒(𝜏) Gc ( s ) = = Kc
E (s)
Proportional Gain
Gs ( s ) = 1
Steady-state (bias) value − p s
K pe
Controller output G p (s) =
 ps +1
SP( s ) E(s) G c (s) MV ( s ) G P (s) CVm (s)
OR
- Kp
CV ( s ) G (s) = 1 G p (s) =
s
 ps +1
e = lim − e( ) = lim s −0 ( sE ( s) ) = ? How does the 𝐾𝑐 influence the 𝑒∞ ????????
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Simulation of step response 2.67
G p (s) =
180 s + 1

1
e =
1 + Kc K p

𝑲𝒄 = 𝟏

𝑲𝒄 = 𝟎. 𝟓

Conclusion???
𝑲𝒄 = 𝟎. 𝟏
POT=?, 𝝉𝒓 =? ; 𝝉𝒔 =? 𝒆∞ =?

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
For proportional and integral Control,
1 τ Laplace G ( s ) = MV ( s ) = K + K c
𝑚𝑣 (τ) = 𝑚𝑣 𝜏 + 𝐾𝑐 𝑒 τ + න 𝑒 𝜏 𝑑𝜏
𝑏𝑖𝑎𝑠
𝜏𝐼 0
c
E (s)
c
is
Integral time Gs ( s ) = 1
Kp
Proportional Gain G p (s) =
Steady-state (bias) value  ps +1
Controller output OR − p s
K pe
G p (s) =
 ps +1
SP( s ) E(s) G c (s) MV ( s ) G P (s) CVm (s)
-
CV ( s ) G (s) = 1
s

e = lim − e( ) = lim s −0 ( sE ( s) ) = ?

How does the τ𝑖 influence the 𝑐𝑣𝑚 ??
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Integral windup

usp Anti-Windup
us

1
e 1 τ τ
1 usp  us 𝑢𝑐_𝐼 (τ) = 𝐾𝑐 𝜏 ‫׬‬0 𝑒 𝜏 𝑑𝜏= 𝐾𝐼 ‫׬‬0 𝑒 𝜏 𝑑𝜏
𝐼

2 6
3 7
e=0  When a PI controller is used, reset windup
uc (%) 4
5 usp  us (figure) occurs because the integral term
uc _ min = 100%
continues to increase until the error signal
2
3
4
changes sign at τ=τ1
7
1 6
50% 5

uc _ min = 0% 
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
2.67
Simulation of step response G p (s) =
180 s + 1
e = 0

𝑲𝒄 = 𝟏;𝝉𝒊 = 𝟏
𝑲𝒄 = 𝟏;𝝉𝒊 = 𝟏𝟎

Conclusion???
POT=?, 𝝉𝒓 =? ; 𝝉𝒔 =? 𝒆∞ =? 𝑲 = 𝟏;𝝉 = 𝟏𝟎𝟎
𝒄 𝒊

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Summary
▪ Disadvantages: Oscillatory, more parameter to tune
▪ Advantages: fast action, eliminating offset
▪ Reset Windup, integral windup

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
For proportional, integral and derivative control,
1 𝜏 𝑑𝑒(𝜏) Laplace U (s) K
𝑚𝑣 𝜏 = 𝑚𝑣𝑏𝑖𝑎𝑠 + 𝐾𝑐 𝑒 𝜏 + න 𝑒 𝜏 𝑑𝜏 + 𝜏𝐷 Gc ( s ) = c = K c + c + K c d s
𝜏𝐼 0 𝑑𝜏 E (s) s i

Derivative time Gs ( s ) = 1
Integral time Kp
G p (s) =
Proportional Gain  ps +1
Steady-state (bias) value OR
− p s
Controller output K pe
G p (s) =
 ps +1

SP( s ) E(s) G c (s) MV ( s ) G P (s) CVm (s)

-
CV ( s ) G (s) = 1
s

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Case of derivative control,

▪ By providing anticipatory control action, the derivative mode tends to

stabilize the controlled process
▪ Derivative control action also tends to improve the dynamic response
of the controlled variable by decreasing the process settling time
▪ The most disadvantage of derivative control is the sensitivity of the
control calculations to high-frequency noise in the measurement.
Therefore, it is usually used with the derivative filter
The transfer function of PID controller with derivative filter
U c (s) K K s
Gc ( s ) = = Kc + c + c d
E (s)  i s  d s + 1
Where, the constant 𝛼 typically has a value between 0.05 and 0.2, with
0.1 being a common choice
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Simulation of step response 2.67e −19.13 s
G p (s) =
180 s + 1

Conclusion???
POT=?, 𝝉𝒓 =? ; 𝝉𝒔 =? 𝒆∞ =?

𝑲𝒄 = 𝟏;𝝉𝒊 = 𝟓𝟎; 𝝉𝒅 =10

𝑲𝒄 = 𝟏;𝝉𝒊 = 𝟓𝟎;𝝉𝒅 =100

𝑲𝒄 = 𝟏;𝝉𝒊 = 𝟓𝟎

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
F. PID controller design, tuning and stability for closed loop system
DV(s)
1e −4 s
SP( s ) E(s) G c (s) MV ( s ) G P (s) CVm (s) G p (s) =
20 s + 1
- CV ( s )
G s (s) = 1

No!
Yes!? Yes!?

Yes!? Yes!? Yes!?

We influence stability when Yes!? Yes!? Yes!?

we implement control. How
do we achieve the influence
we want? Unit-step responses
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Performance criteria for closed-loop systems
The function of a feedback control systems is to ensure that the close-
loop system has desirable dynamic and steady-state response
characteristics
We would like the closed-control system to satisfy the following
performance criteria:
1. The closed-loop system must be stable
2. The effect of disturbances are minimized, providing good disturbance rejection
3. Rapid, smooth responses to set-point changes are obtained, that is, good set-point tracking
4. Steady-state error (offset) is eliminated
5. Excessive control action is avoided
6. The control system is robust, that is insensitive to changes in process conditions and to
inaccuracies in the process model

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
F.1.PID determination by Integral Error Criteria DV(s)
SP( s ) E(s) G c (s) MV ( s ) G P (s) CVm (s)
- CV ( s )
G s (s) = 1
1.5
sp(τ) cv(τ) Return to set point, “zero offset “
1

0.5
POT

Unit-step responses
= IAE 0
0 5 10 15 20 25 30 35 40 45 50
Time
Rise time
 Settling time
= |sp(τ)-cv(τ)| dτ 2

0 1.5

to find out 1

𝑲𝒄 ; 𝝉𝒊 ; 𝝉𝒅 0.5

0
0 5 10 15 20 25 30 35 40 45 50
Time

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Several means of weighing the response curve, starting at τ = 0 until
equilibrium at τ->∞ are:
Integral Error Criteria

Integral Square Error(ISE) Integral Absolute Error (IAE) Integral of time Absolute Error (ITAE)
  
ISE =  ( sp ( ) − cv( ) ) d IAE =  sp ( ) − cv( ) d IAE =   sp ( ) − cv( ) d
2

0 0 0
− p s
K pe
G p (s) = Fitness evaluation function = min( IAE , ITAE )
 ps +1 For set-point changes (Madhuranthakam et al., 2008)
PI controller For set-point changes
b1
a p  p IAE ITAE
Kc = 1   ;  = a1 = 0.758; b1 = −0.861 a1 = 0.586; b1 = −0.916
K p   p  
i
 a2 + b2  p
  a2 = 1.02; b2 = −0.233 a2 = 1.03; b2 = −0.165
 p 
PID controller For set- point changes IAE ITAE
b1 b3
a p  p p  a1 = 0.965; b1 = −0.855
Kc = 1   ; i = ; d = a3 p   a1 = 1.086; b1 = −0.869

Kp  p  
a2 + b2  p
  p  a2 = 0.74; b2 = −0.13 a2 = 0.796; b2 = −0.147
 
a3 = 0.348; b3 = 0.914 a3 = 0.308; b3 = 0.9292
 p 
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
For disturbance inputs (Madhuranthakam et al., 2008)

P controller For disturbance inputs

b ISE IAE ITAE
a p 
Kc = 1   a = 1.411; b = −0.917 a = 0.902; b = −0.985 a = 0.49; b = −1.084
K p   p 

PI controller For disturbance inputs

ISE IAE ITAE
b1 b2
a1 p   p p  a1 = 1.305; b1 = −0.959 a1 = 0.984; b1 = −0.986 a1 = 0.859; b1 = −0.977
Kc =   ;  =  
 p 2  p
i
Kp  a  a2 = 0.492; b2 = 0.739 a2 = 0.608; b2 = 0.707 a2 = 0.674; b2 = 0.68

PID controller For disturbance input

b1 b3 ITAE
a p  p 
p ISE ISE
Kc = 1   ; i = ; d = a3 p   a1 = 1.495; b1 = −0.945 a1 = 1.435; b1 = −0.921 a1 = 1.357; b1 = −0.947
K p   p    
 p 
a2 + b2  p  a2 = 1.101; b2 = 0.771 a2 = 0.878; b2 = 0.749 a2 = 0.842; b2 = 0.738

 p 
a3 = 0.56; b3 = 1.0006 a3 = 0.482; b3 = 1.137 a3 = 0.381; b3 = 0.995

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
F.2. PID determination by on-line controller tuning
a. Continuous cycling method for on-line controller tuning
(Ziegler-Nichols (Z-N), 1942)-(Z-N2) or Tyreus-Luyben (T-L)
DV(s)
SP( s ) E(s) G c (s) MV ( s ) G P (s) CVm (s)
- CV ( s )
G s (s) = 1

Step 1. Set PID control and wait. After the process has reached steady
state, introduce proportional-only control (set 𝜏𝑖 = 0; 𝜏𝑑 = ∞).

Step 2. Increase 𝐾𝑐 in small increments until continuous cycling occurs

(oscillation with constant amplitude). The value of 𝐾𝑐 is 𝐾𝑢

Step 3. Determine the ultimate gain (𝐾𝑢 ) and the ultimate period (𝑇𝑢 )

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

After 𝐾𝑢 and 𝑇𝑢 have been determined, the controller settings can be

calculated using Z-N or T-L tuning relation:
Ziegler-Nichols Tyreus-Luyben
Controller 𝐾𝑐 𝜏𝑖 𝜏𝑑 Controller 𝐾𝑐 𝜏𝑖 𝜏𝑑
P 𝐾𝑢 /2 --- --- PI 0.35𝐾𝑢 2.2𝑇𝑢 ---

PI 𝐾𝑢 /2.2 𝑇𝑢 /1.2 --- PID 0.45𝐾𝑢 2.2𝑇𝑢 𝑇𝑢 /6.3

PID 𝐾𝑢 /1.7 𝑇𝑢 /2.0 𝑇𝑢 /8

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
b. The process reaction curve method for on-line controller tuning
Process reaction curve - Method II
(Ziegler-Nichols (Z-N), 1942)-(Z-N1), (Cohen-Coon)
CV2
0, 63CV
0, 28CV CV
-θps ΔCV CV1
CVm (s) K pe Kp =
ΔMV τ 28%
G p (s) = = τ p = 1,5( 63% -  28% ) τ 63%
MV(s) τ ps +1 MV2
θ p =  63% -  p ΔMV
MV1
τ
τ 0 τ1 τ 2 time
(Ziegler-Nichols, (Z-N1) (Cohen-Coon)
Controller 𝑲𝒄 𝝉𝒊 𝝉𝒅 Controller 𝑲𝒄 𝝉𝒊 𝝉𝒅
P  p  p  _ _
1 + 
P τp
_ _ K p p  3 p 
K pθ p  3 p 
PI p  p   p  30 +  _
 p 
 0.9 +  
PI 0.9τ p
3.3θ p _ K p p  12 p  20 p
9+
K pθ p p
 3 p   6 p 
PID 1.2τ p PID  16 +   p  32 +  40
2θ p 0.5θ p p  p    p  2
K pθ p K p p  12  8 p 11 + p
  13 + p
  p

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
F.3. PID determination by model-based design methods
a1. Direct Synthesis Method (DS Method) for set-point changes
DV
Gd(s) DV=0 SP G(s) CV
SP E Gc(s) MV Gp(s) CV
-

CV Gc ( s )G p ( s ) 1 G ( s)
G (s) = = Gc ( s ) =
SP 1 + Gc ( s )G p ( s ) G p ( s) 1 − G ( s)
(Chen & Seborg, 2002)
Apply for FOPDT model − s Where 𝜏𝑐 is the
e p
Referent model G (s) = desired closed-
 c s + 1 loop time constant

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
− p s
With PI controller e
 1  1 cs +1
CV(s) K p e
-θ p s
K c 1 +  =  − s
G p (s) = =   i 
s -θ s
K pe p e p
MV(s) τ ps +1 1−
τ ps +1 cs +1
1 G ( s)
Gc ( s ) = τ ps +1 1
G p ( s) 1 − G ( s) =  − s
Kp cs +1− e p
 1 
GPI ( s ) = K c 1 +  τ ps +1 1
  i 
s =  − s
Kp cs +1− e p
− p s
e Taylor
G (s) = τ ps +1 1
cs +1 Series expansion  
Kp  c s + 1 − (1 −  p s)

p
 1  c +p  1 
p
K c 1 + =  1 +  Kc = ; i =  p
  i 
s Kp   s
p  K p ( c +  p )
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
a2. Direct Synthesis-Disturbance Rejection Method (DS-d Method)
DV
Gd(s) SP=0 DV CV
T(s)
SP E Gc(s) MV G (s) CV
p
-

CV Gd ( s ) Gd ( s ) 1  1 
T (s) = = Gc ( s ) = − = K c 1 + 
DV 1 + Gc ( s )G p ( s ) T ( s )G p ( s ) G p ( s )   i 
s
i − p s
se (Chen & Seborg,
Referent model (known) Kc
T (s) = 2002)
( c s + 1)
2

With PI controller
 p +  p p − ( c −  p )
2
i
2
1
Gd ( s ) = G p ( s ) Kc = ; i =
K p c +p  p + p
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Direct Synthesis Method for first-order model
DV
Gd(s)
DV=0 SP CV
G(s)
SP E Gc(s) MV G (s) CV
p
- CV
G (s) =
SP
For process without delay, the first-order model in (*) is a
more reasonable choice CV 1
G (s) = = (*)
SP  c s + 1
Gc ( s )G p ( s ) 1 G ( s)
G (s) = Gc ( s ) =
1 + Gc ( s )G p ( s ) G p ( s) 1 − G ( s)
p
Kc = ; i =  p
1 1 With PI controller  1  K p c
Gc ( s ) = Gc ( s ) = K c 1 + 
G p (s)  c s  is  Selection of 𝝉𝒄 , 𝝉𝒑 ≈ 𝟓𝝉𝒄
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
b. Internal Model Control, IMC Method
DV DV
SP E GIMC(s) MV Gp(s) PV SP PV
GIMC(s) MV Gp(s)
- - +
- PV1 PV1
G p (s) G p (s)

GIMC G p GIMC 1 − G p GIMC

PV = G p ( SP − PV ) + DV PV = SP + DV
1 − G p GIMC 1 + (G p − G p )GIMC 1 + (G p − G p )GIMC

Case Gp = Gp PV = GIMC G p  SP + (1 − G p GIMC ) DV

1

if
GIMC =
Gp
then PV = SP
A
Case of Gp  Gp and DV  0

if GIMC =
1
Gp
then PV = SP
B
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
How to determine the Gc for all case A and case B?
Step 1: Separate the process model G p into two terms
G p = G p +G p −

Step 2:To obtain a practical IMC controller. The simplest form

recommended by Morari and Zafiriou is given by
1
GIMC = F
Gp−

F is a low-pass filter with a steady-state gain of one. It typically has the

form: 1
F=
( c s + 1)
k

τc is the desired closed-loop time constant (DS method) and k is an

integer. The value of k is selected exactly equal to the order of G p −
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Step 3: Determination of parameters of PID controller
Gc DV
SP GIMC MV G (s) CV
p
- 1 − GIMC G p

GIMC
Gc =
1 − GIMC G p

Gc is equivalent to the PID controller (𝐺𝑃𝐼𝐷 ). We have parameter of PID

controller
Example1: Use the IMC method to design two controllers for the
FOPTD model. Consider two approximations for time delay term:
a. Taylor series approximation
b. Pade 1/1 approximation
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Solution:
1.Taylor approximation for FOPTD model
Kp − p s K p (1 −  p s )
G p (s) = e =
 ps +1  ps +1
Step 1:
Kp
G p ( s ) = G p + G p − = (1 −  p s )
 ps +1
G p + = (1 −  p s )
Kp
Gp− =
 ps +1
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Step 2: 1 1  ps +1
GIMC =  =
Kp ( c s + 1) K p ( c s + 1)
 ps +1

Step 3:  ps +1
K p ( c s + 1) p  1 
Gc = = 1 + 
 p s + 1 K p (1 −  p s ) K p ( c +  p )   p s 
1−
K p ( c s + 1)  p s + 1

 1 
Gc ( s ) = GPI ( s ) = K c 1 + 
  i 
s
p
Kc = ; i =  p
K p ( c +  p )
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Selection of τc
The choice of design parameter τc is a key decision in both the DS and IMC
design methods.
- In general, increasing τc procedures a more conservative controller because
Kc decreases while 𝜏𝑖 increases.
- Several IMC guidelines for τc have been published for the FOPTD .
+ τc > 0.8𝜃𝑝 ; τc > 0.1𝜏𝑝 (Rivera et al., 1986)
+ 𝜏𝑝 > τc > 𝜃𝑝 (Chien and Fruehauf, 1990)
+
Recommendations Condition References
0.5 p Default value smooth Grimholt and Skogestad

 c =  p control (2013)

1.5 p
Default value robust Astrom and Hagglund (2006);
 c = 3 p control MCMillan (2015)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
b. Pade 1/1 approximation

With PID controller

 p 
2
  + 1
Kc =
1
 
p  ;  =  +  p ; =  p
Kp  c 
i p
2
D
p
2  + 1 2 +1
 p
 p 

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Summary
IMC controller settings for Parallel-Form PID controller (Chien and
Fruehauf,1990); K= 𝐾𝑝 ;τ = 𝜏𝑝 ; 𝜃 = 𝜃𝑝

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Summary

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
G. Hardware, instrumentation, and digital controller dv(τ)
Digital controller
DAC Process G DV
Digital PID
∆τ mv(τ) cvm ( )
sp(τ) e(τ) Hold G P0
G PID OUT_A1
- ADC Measuring
pv(τ)
IN_A1
cv(τ) element
∆τ pv(τ) Gs

IN_A1 CTRL_PID OUT_A1

. . . CPU modules
. .. .
. .
. . . (hardware)
PV_IN
SP_IN blocks
(software)

I/O modules
Analog (hardware)
Analog Output
Input

process &
4...20 mA field instrumentation
0…100 cm 0…100%
4...20 mA
I L

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
G1. Measuring element (Sensor) for process systems

This topic:
- Different types of sensors
- Principles of sensors
- Selection of sensors
- Noise filter

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
G1. Measuring element (Sensor) for process systems
➢ What is sensor (transducer)?
Sensor converts the physical quantity to signal that can be recognized by
other components such as display, transmitter…

Physical variables Measurement

Sensor
L, P, F, T, x,… V, mA, Data…

➢ Sensor types
▪ Temperature: thermocouple, RTD, thermistor
▪ Pressure: bellows, bourdon tube, diaphragm
▪ Flow rate: orifice, venturi, magnetic, ultrasonic, Coriolis effect
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
▪ Liquid level: float, differential pressure
▪ pH: pH electrode
▪ Viscosity: pressure drop across venturi or vane deflection
▪ Composition: density, conductivity, GC, IR, NIR, UV

Magnetic flow meter

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
The following terms (criteria) are used to define a measurement device:
1. Range and span: The range of a transducer defines the limits between
which the input can vary. The span is the maximum value of the input
minus the minimum value. For example, a loadcell for the measurement
of forces might have a range of 0 to 50 kN and a span of 50 kN
2. Error: Error is the difference between the result of the measurement
and the true value of the quantity being measured:
Error = measured value - true value
3. Accuracy: Accuracy is the extent to which the value indicated by a
measurement system might wrong. A temperature-sensor might, for
example, be specified as having an accuracy of ±2𝑜 𝐶. This would mean
that the reading given by the instrument can be expected to lie within plus
or minus 2𝑜 C of true value. Accuracy is often expressed as a percentage
of the full range output or full-range deflection.
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Example: A temperature-sensor might be specified as having an
accuracy of ±5% of full range output. Thus if the range of the sensor
was, say, 0 to 200𝑜 𝐶, the reading given can be expected to be within
plus or minus 10𝑜 𝐶 of the true reading
4 Sensitivity: The sensitivity is the relationship indicating how much
output there is per unit input. For example, a resistance thermometer
may have a sensitivity of 0.5Ω/𝑜 𝐶.
5 Hysteresis error: Sensor can give

Output
different outputs from the same value

g
sin
of quantity being measured according

rea
Error

dec

g
to whether that value has been

sin
rea
reached by a continuously in

inc
increasing change or continuously
decreasing change. 0
Value of measured quantity

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
6 Non-linearity error: Many sensor a linear relationship between the
input and output is assumed over the working range. Few sensor, have
not a truly linear relationship and thus errors occur as a result of the
assumption of linearity

7 Repeatability: The terms repeatability of a sensor are used to describe

its ability to give the same output for repeated applications of the same
input value.
repeatability = 100% (max values given – min values given)/ full range
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
8 Stability: The terms of sensor is its ability to given the same output
when used to measure a constant input over a period of time.
9 Dead band/time: The terms dead band of a sensor is the range of input
values for which there is no output.
10 Resolution: When the input varies continuously over the range, the
output signals for some sensor may change in small steps.
11 Output impedance: The output impedance since this impedance is
being connected in either series or parallel with that circuit

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Example1: Specification of a strain gauge pressure sensor
The sensor can be used to measure
- Ranges: 70 to 1000 kPa pressure between 70 and 1000kPa
The sensor requires a supply of
- Supply voltage: 10Vdc
10Vdc for its operation
- Full range output: 40mV The sensor will give an output 40mV
when the pressure is 1000 kPa
- Non-linearity and hysteresis:
Hysteresis lead to error ±5 kPa
±0.5% full range output

- Temperature range: − 54 𝑜 𝐶 to +120𝑜 C When temperature

change by 1𝑜 𝐶 the
when operating output of sensor for zero
input will change by
- Thermal zero shift:0.03% full range output/o𝐶 0.03% of 1000 = 0.3 kPa
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Example2:
A standard temperature transmitter with
I
a 4-20 mA output signal might be t
adjusted so that the input rage of a
PT100 (the sensor) is 0−200𝑜 𝐶. In this
case, the following correspondence is 𝐼(𝑚𝐴) Actual
obtained: 20
- This instrument has a lower limit, or
zero = 0(0 𝐶) and a range or span, of Ideal

200 (0 𝐶) - (0 𝐶) = 200 (0 𝐶) 4 Span

𝑡(0 𝐶)
- The steady-state relationship between the 0 Zero 200
measured variable I and transmitter output is a
linear equation, with 𝐾𝑠 is the transmitter
steady-state gain I = I max − I min (t − t ) + I = K (t − t ) + I tmax = 200; tmin = 0
tmax − tmin
min min s min min
I max = 20; I min = 4
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
➢ Dynamic model of measurement device
Most measurement device (sensor + transmitter) can be expressed by a
first-order transfer function

Measurement Device
Input Output
Sensor Transmitter
U(s) Y(s)
Indicator

Y (s) Ks
G (s) = =
U (s)  s s + 1

▪ 𝐾𝑠 : is the transmitter steady-state gain

▪ 𝜏𝑠 : the time constant of the measurement
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
➢ Transducer is a converter
• I/P or P/I transducer
• I/V (I/E) or V/I
• P/E or E/P

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
G1.1 Temperature Measurement
- Thermocouple
Thermocouple basically consist of two junctions of dissimilar metals,
such as copper and constantan that are welded or crimped together.
▪ One junction is kept at a constant temperature called the reference
(Cold) junction,
▪ While the other is the measuring (Hot) junction.
When the two junctions are at different temperatures, a voltage is
developed across the junction which is used to measure the temperature.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Seebeck effect
▪ If two wires of dissimilar metals are joined at both ends and one end is
heated, current will flow.
▪ If the circuit is broken, there will be an open circuit voltage across the
wires.
▪ Voltage is a function of temperature and metal types.
▪ For small ∆𝑇 (s), the relationship with temperature is linear
∆𝑉 = 𝛼∆𝑇
where, 𝛼: Seebeck coefficient
▪ For larger ∆𝑇 (s), non-linearity may occur.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Thermocouple types
▪ K-type: most popularly used
▪ J-type: higher electromotive
force (EMF)
▪ E-type: cryogenic
temperature
▪ R-type: high temperature (>
900 ºC)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
RTD (Resistance Temperature Detector)
Most RTDs consist of a length of fine coiled
wire wrapped around a ceramic or glass core.
The RTD element is made from a pure
material, typically platinum, nickel or
copper.

Platinum often is used since it can be used for a wide temperature

range and has excellent stability

RTDs have positive temperature coefficients (PTC). However, they

have very poor thermal sensitivity.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

Based on the principle that the material resistance is a function of its

temperature
𝑅𝑇 = 𝑅0 (1 + 𝐴𝑇 + 𝐵𝑇 2 )

Where, 𝑅𝑇 : resistance of sensor at T (0C)

𝑅0 : resistance of sensor at 0 (0C)
A, B: temperature coefficients (A>0)

Type A.10-3/0C B.10-6/0C

Pt 3,85 0,59
Ni 6,17
Cu 4,27

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Note:
▪ RTD Cu (Copper) is used in low temperature measurement (below 100
0C) to avoid oxidizing.

▪ RTD Ni (Nickel) that has high non-linearity is usually used to

measure at higher temperature (below 300 0C)

▪ RTD Pt (Platinum) is commonly used with R0=100, 200, 500, 1000

(Ohm) corresponding to Pt 100, Pt 200, Pt 500, Pt 1000.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Wire Configuration
The simplest configuration uses two
wires. (ignoring the resistance of
wires). It is only used when high
accuracy is not required
Adding the lead resistance to minimize
the effects of wire resistances

Four-wire configuration

Eliminating voltages drop in the ADC

measuring leads as a contribution to error
by using an additional current source
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
- Thermistor
▪ Thermistors which made of ceramic semiconductor material measure
the change in resistance temperature.
▪ Thermistors are very sensitive (up to 100 times more than RTDs and
1000 times more than thermocouples) and can detect very small
changes in temperature.
▪ Due to their speed, they are
used for precision temperature
control and very small
temperature differences.
▪ The change in thermistor
resistance with temperature
is very non-linear.
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
G1.2 Pressure Measurement
- Fluid pressure can be defined as the measure of force per-unit-area
exerted by a fluid, acting perpendicularly to any surface it contacts
dF
P=
ds
- The standard SI unit for pressure measurement is the Pascal (Pa) (~
N/m2). In the English system, pressure is usually expressed in pounds per
square inch (psi)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
- Pressure measurements can be divided into three different categories:
absolute pressure, gauge pressure and differential pressure.
▪ Absolute pressure is the difference between the pressure at a given
point in a fluid and the absolute zero of pressure or a perfect vacuum.
▪ Gauge pressure is the measurement of the difference between the
absolute pressure and the local atmospheric pressure (depending on
ambient temperature, altitude and local weather conditions)
▪ Differential pressure is simply the
measurement of one unknown
pressure with reference to another
unknown pressure. The pressure
measured is the difference between
the two unknown pressures.
Pressure terms relationship
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
- There are different types of fluid systems and fluid pressures. There are
two types of fluid systems: static systems and dynamic systems. As the
names imply, a static system is one in which the fluid is at rest and a
dynamic system is on in which the fluid is moving.
Static Pressure: The absolute pressure at a depth H in a liquid is defined:
Pabs = P0 +  gH
where, 𝑃0 : the atmospheric pressure
𝜌: the density of the fluid
g: the acceleration due to gravity
H: the depth at which the pressure is measured

Dynamic Pressure. P = Ps + Pd
where, 𝑃𝑠 : the static pressure
𝑃𝑑 = ρ𝑣 2 /2
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
G1.3 Level Measurement
There are plenty of methods to measure level for both point level
detection and continuous level monitoring

https://en.wikipedia.org/wiki/Level_sensor

Level gauge

Observe directly Tank

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Magnetic level indicator
Inside the stand pipe, a float containing a
permanent magnet moves up and down
following the change of liquid level.
On the outside of the stand pipe, a
transparent rail with indicator flaps are
mounted. The small flaps hold small
permanent magnets. The flaps are bi-
color with colors red and white. When
the float passes the flaps, the flaps are
rotated 180° by the magnetic field in the
float.
The indicator shows the red color up to
the liquid level. Above the color is white.

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Float type detectors (Magnetic level switch)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Based on differential pressure

DP = h( 1 −  2 ) g
where, DP: difference pressure
𝜌1 : the density of the heavier fluid
𝜌2 : the density of the lighter fluid
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Based on time-of-light

D = (transmission velocity x time)/2

Level: H = L - D

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Based on capacitance

2 r  0 h
C=
ln( D / d )

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
G1.4 Flow Measurement
Flow measurement is the quantification of bulk fluid movement (wiki)
▪ Volumetric flow rate, Q, [m3/s], [m3/min],…
▪ Mass flow rate, 𝑚,ሶ [kg/s], [kg/min],…

The most common principles for fluid flow metering are:

▪ Differential Pressure Flow meters
▪ Velocity Flow meters
▪ Positive Displacement Flow meters
▪ Mass Flow meters
▪ Open Channel Flow meters
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Differential Pressure Flowmeters
The flow is calculated by measuring the pressure drop over an
obstructions inserted in the flow.
Its principle is based on the Bernoullis Equation, where the pressure
drop and the further measured signal is a function of the square flow
speed.

1 2 P
mv + mgh + mg = cont
2 g
D1 V1 Dt V D2
2

1 2

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

1 2 P
mv + mgh + mg = cont
D1 V1 Dt V D2 2 g
2

1 2

Applying Bernoullis equation at position (1) and (2)

1 P1 1 P2
mV1 + mgh1 + mg = mV2 + mgh2 + mg
2 2

2 g 2 g
Assuming: h1 = h2

2
 V − V2 =
2 2
( P1 − P2 )
1

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
The mass flow rate at position (1) and (2) are the same:
m 1 = m 2 = V1 A1 = V2 A2
 V1 A1 = V2 A2
2 ( p1 − p 2 )
The velocity at (2): V2 =
 [1 − ( A2 A1 ) ]
2

A2
mtheory = V2 A2 = 2  ( p1 − p2 )
[1 − ( A2 A1 ) ]
2 12

Cd . A t
mactual = 2  ( p1 − p2 ) where, 𝐶𝑑 : coefficient constant
[1 − ( A t A1 ) ]
2 12

Given  = Dt /D1, then (At / A1)2 = (Dt /D1)4 = 4

Cd . A t
mactual = 2  ( p − p ) This equation is applied to many flow
[1 −  4 ]1 2
1 2
measurement device
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control

Orifice Plate

Cd At
mactual = 2  ( p1 − p2 )
[1 −  ]
4 12

F o r 0 .2    0 .7 5 a n d 1 0 4  R e  1 0 7
91 . 71 2 . 5
C d = 0 . 5959 + 0 . 0312  2 . 1 − 0 . 184  8 +
R e 0 . 75

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
The Flow Nozzle

Cd At
mactual = 2  ( p1 − p2 )
[1 −  ]
4 12

6 . 53  0 .5
C d = 0 . 9975 −
R e0 .5
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Venturi Tube

Cd At
mactual = 2  ( p1 − p2 )
[1 −  ]
4 12

0.98 <= Cd <= 0.995

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Pitot tubes

1 P2 2
P1,V V2=0
1

2 ( P2 − P1 )
 V1 =


mactual = Cd A t 2  ( p1 − p2 )

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
G2. Actuator for process systems
sp cvm
Controller Actuator Process
cv

Sensor
What is actuator?
Actuator converts the command signal from controllers into physical
adjustment in adjustable process variable. It is normally valve opening,
power level or displacement.
Actuator types
▪ Control valve: pneumatic, electric, hydraulic
▪ Electric heater output: SCR, thyristor
▪ Pump/Motor speed: inverter
▪ Displacement: pneumatic, electric, hydraulic
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Control valve
Control input

Valve actuator

Stem

Valve body
Inlet stream Outlet stream

Closure member Valve seat

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Valve types
▪ Globe valve: rugged, usually the most
expensive, particularly in the larger
sizes, accurate and repeatable control,
high pressure drop.

▪ Gate valve: sliding disc (gate), ideal for

high pressure and high temperature
applications where operation is
infrequent, multi-turn or long stroke
pneumatic and electro-hydraulic
actuators are needed, poor control.
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Valve types

▪ Ball valve: tight shutoff, high

capacity with just a quarter-turn to
operate.

▪ Butterfly valve: damper valve,

most economical valves, high
torque required

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Valve characteristic
Pv
▪ Basic equation q ( ) = Cv f ( )
gs
where, is the valve stem position, (Độ mở valve)
𝐶𝑣 : valve coefficient, decided by valve size
▪ Valve trim type for different plug
• Linear: 0  1
• Square Root (Quick Opening): f ( ) =
−1
• Equal Percentage: f ( ) = R
where, R: range ability (ratio between minimum flow
and maximum flow)

The bigger R is, the more accurate

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control

▪ Linear: used where fairly

constant pressure drop is
expected, used for liquid level or
flow loop.
▪ Quick Opening: used for
frequently on-off service or where
instantly large flow is needed
▪ Equal Percentage: most
commonly used, used where
large pressure drop is expected

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Dynamic behavior of the control valve
F (s) K v −v s
u Gv ( s ) = = e
U (s)  v s + 1
l
Or
L( s ) K v −v s
f Gv ( s ) = = e
U (s)  v s + 1

τv is the valve time constant, and is typically much smaller than the
largest process time τp . The τv value varied from 0.3 to 7.8 s, while the
θv /τv ratios varied from 0.3 to 0.8 (Emerson Process Management, 2005)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
G3. Digital controller
G3. 1 Sampling and reconstruction signal
∆τ ∆τ
y(τ) y* (τ) y* (τ) Zero-order y H (τ)
hold
(a) The sampler
y(τ) ∆τ: sampling period y* (τ)
y H (τ)
y(3)
y(2)
y(1) ∆τ
y(0) τ
 0 1  2  3  0 1  2  3
(b) Continuous signal
*
y(τ ) and samples DAC: Digital-to-analog converter
y (τ)

Guidelines for selecting the sampling period

of process control
 0 1 2 k s s
( ) (2 ) (k  )    Astrom and wittenmark (2011)
(c) Sampled signal 15 6
ADC: Analog-to-digital converter 𝜏𝑠 is the settling time
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
G3. 2 Digital PID calculation
τ
de τ
mv τ = mvP τ + mvI τ + uD τ = K c e τ + K I න e τ dτ + K D (G3.2.1)
0 dτ

- Discrete transfer function of a proportional part. The proportional of e(τ

) in continuous time can be approximated by:

mvP τ = K c e(τ) = K c e(k) (G3.2.2)

Where e(k) is the error at the kth sampling instant for k=1,2,…
Taking the z-transform,
MVP z = K c E(z) (G3.2.3)

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
- Discrete transfer function of an integral part. The integral of e(τ) in con
tinuous time can be approximated by a summation in to the integral
Kc τ τ Kc
mvI τ = ‫ ׬‬e(τ)dt = K I ‫׬‬0 e(τ)dτ ≈ ∆t σnk=0 e(k) (G3.2.4)
τi 0 τi

Taking the z-transform,

Kc
MVI z = ∆τ σnk=0 z −k E(z) (G3.2.5)
τi

When n is large then,

1 Kc
MVI z = ∆τE(z) (G3.2.6)
(1−z−1 ) τi

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

- Discrete transfer function of a derivative part. This expression is known

as the backward-difference approximation of τ (equivalent to a first-order
Taylor series)
de τ e k −e(k−1)
MVD τ = k c τd ≈ k c τd (G3.2.7)
dτ ∆τ

Taking the z-transform,

1−z−1
MVD z = k c τd E(z) (G3.2.8)
∆τ

MVPID z = MVP z +MVI z +MVD z

1 Kc 1−z−1
= Kc + ∆τ + k c τd E(z) (G3.2.9)
1−z−1 τi ∆τ

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Or,
∆τ τ 2τd τd −2
(1 − z −1 )MVPID z = K c 1 + + d − 1+ z −1 + z E(z) (G3.2.10)
τi ∆τ ∆τ ∆τ

Converting the controller transfer function into difference equation form

gives
MVPID k − MVPID k − 1 =
∆τ τd 2τd τd
= K c (1 + + )e(k) − 1+ e(k − 1) + e(k − 2) (G3.2.11)
τi ∆τ ∆τ ∆τ

∆τ τd
MVPID k = MVPID k − 1 + K c 1+ + e k −
τi ∆τ
2τd τd
− Kc 1 + e(k − 1) + K c e(k − 2) (G3.2.12)
∆τ ∆τ

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Note:
▪ When implementing their equations in a computer program the
equations can be rewrite as shown in equation (G3.2.12). To do this
calculation, previous error and control value must be stored. The
calculation also requires the scan time ∆τ between updates


 0.1 (Isermann,1889)
i

s 
   s (Astrom and wittenmark 2011)
15 6

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Flowchart of digital PID calculation Begin

update_time
Parameter input:
( delta_t)
K_c, tau_i, tau_d, delta_t

+ e(k-2) = e(k-1)
+ e(k-1) = e(k)
+ e(k) = sp(k) - pv(k)
+MV_pid(k-1)=MV_pid(k)

K_c
temp1=delta_t/tau_i
temp2=tau_d/detal_t
tau_i temp3=1+temp1
temp4=temp3+temp2
tau_d temp5=K_c*temp4
temp6=temp5*e(k)
temp7=temp2*2 MV_pid(k)
delta_t temp8=1+temp7
temp9=K_c*temp8
temp10=temp9*e(k-1)
MV_pid(k-1) temp11=K_c*temp2
e(k) temp12=temp11*e(k-2)
temp13=temp6-temp10
e(k-1) temp14=temp13+temp12
e(k-2)
MV_pid(k)= MV_pid(k-1)+temp14

Stop

End

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
H. Cascade control
One of most useful concepts in advanced control is cascade control. A
cascade control structure is composed of two feedback controllers with
the output of the primary (or master) controller can be used to manipulate
the setpoint of the secondary (or slave) controller. Each controller has its
own measurement input, but only the primary controller have an
independent setpoint has only secondary controller has an output to the
process. The manipulated variable, the secondary controller, and its
measurement constitute a closed loop within the primary loop, as shown:

DV2 (s) DV1 (s)

SP1 (s) E1 (s) Primary SP2 (s) Secondary CVm (s)
Process II Process I
- controller - controller
PV1 (s) PV2 (s) Inner loop
Outter loop

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
The principal advance of cascade control are:
1. Disturbances arising within the secondary loop are corrected by the
secondary controller before they can influence the primary variable.
2. Phase lag existing in the secondary part of the process is reduced
measurably by the secondary loop. This improves the speed of
response of the primary loop.
3. Gain variations in the secondary part of the process are overcome
within its own loop.
4. The secondary loop permits an exact manipulation of the flow of
mass or energy by the primary controller.
DV2 (s) DV1 (s)
SP1 (s) E1 (s) Primary SP2 (s) Secondary CVm (s)
Process II Process I
- controller - controller
PV1 (s) PV2 (s) Inner loop
Outter loop

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Product(vapor)
Example Cooling water
t TT

TC
Product (Liquid)

Distillation FC
FT
column

f
z Product (c%)
Feed

Steam f st

Cascade control for distillation column is usually employed to regulate

the temperature (and consequently the concentration) at the top or bottom
of a distillation column. The secondary loop is used to compensate for
flow rate change.
Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương
 Process control
Example of cascade control for level control process
𝑫𝑽𝟐
𝑺𝑽𝟏 H (s) K p01 𝑺𝑽𝟐 G U S _ L (s) KS _ L L( s ) K v e −v s 𝑪𝑽𝒎
K p01
GP01 (s) = =
Fi (s)  p01 s + 1 S_L (s) = =
CVm (s)  S _ L s + 1 G C_Secondary Gv ( s ) = =
U (s)  v s + 1
GP01 ( s ) =
H (s)
= 𝑒 −𝐿𝑠
- - Fi ( s )  p01 s + 1
𝑷𝑽𝟐
𝑷𝑽𝟏

G C_Primary
U S _ L (s) KS _ L
GS _ L ( s ) = =
𝑺𝑽𝟐 CVm ( s )  S _ Ls +1
𝑺𝑽𝟏 LC 𝑷𝑽𝟐
FC
Primary
Secondary

𝑷𝑽𝟏 FT

fi
LT
h Rv
fo

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control
Example of cascade control for level control process

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương

 Process control

Thank you

Biên soạn: Nguyễn Xuân Quang, Võ Lâm Chương