Lynkurs Prosessregulering

(Crash course process control)

Sigurd Skogestad
Institutt for kjemisk prosessteknologi
Rom K4-211
skoge@ntnu.no

1
 Course for 3rd year students:
• Pensum (syllabus): Lectures/exercises
Literature (see www.nt.ntnu.no/users/skoge/prosessregulering_lynkurs):
– 1. Nybraaten og Svendsen, "Kort innføring i prosessregulering" (1986)
(det kan synes gammelt, men det står faktisk mye bra her)
– 2. 12 sider fra F. Haugen, "Anvendt reguleringsteknikk", 1992
– 3. S. Skogestad, "Prosessteknikk", 2./3. utgave, Tapir: Kap. 11.3 (tidsrespons) og 11.8/11.6
(prosessregulering)
– 3. S. Skogestad, ”Chemical and Energy Process Engineering”, CRC Press, 2009: Ch. 11.3 (Dynamic
analysis and time response) + Ch. 11.6 (Process control)

– Slides

• Forelesningsplan
• F1: Oversikt over regulering, forover- og tilbakekobling, dusjeksempel
• F2: Klassifisering av variable, prosedyre for utforming av reguleringssystem
• F3: Eksempler
• F4: Prosessdynamikk, tidskonstant, dødtid, PID-regulering,
• F5/F6: Stabilitet, Tuning PID, Forsøk, Eksempler,

More information (literature, old exams, etc.):

• www.nt.ntnu.no/users/skoge/prosessregulering_lynkurs

English Norsk English Norsk

Control regulering
Operation drift Loop Sløyfe
Measurement måling Valve Ventil
Disturbance DV forstyrrelse Gain Forsterkning
Manipulated var. Pådrag Time delay Tidsforsinkelse
(MV) = input = inngang = dead time (θ) = dødtid (θ)
Controlled variab. Regulert variabel
(CV) = output = utgang
Feedback Tilbakekobling
Feedforward Foroverkobling
Controller Regulator

2
 Why control?
• Until now: Design of process. Assume steady-state
• Now: Operation
Actual value(dynamic)
Steady-state (average)

time

In practice never steady-state:

• Feed changes
• Startup
• Operator changes “Disturbances” (d’s)
• Failures
• …..
- Control is needed to reduce the effect of disturbances – remain at steady state
- 30% of investment costs are typically for instrumentation and control

Countermeasures to disturbances (I)

I. Reduce/Eliminate the disturbance
(a) Design process so it is insensitive to
disturbances
• Example: Use buffertank to dampen disturbances
Tin
Tout

inflow ∞ outflow

(b) Detect and remove source of disturbances

• “Statistical process control” (SPC)
• Example: Detect and eliminate variations in feed
composition

3
Countermeasures to disturbances (II)
II. Counteract the disturbance: Process
control (prosessregulering)
Do something (usually manipulate valve)
to counteract the effect of the disturbances

(a) Manual control: Need operator

(b) Automatic control: Need measurement + automatic valve + computer
Goals automatic control:
• Smaller variations MAX FLOW
• more consistent quality
• More optimal (“squeeze and shift”)
after
• Smaller losses (environment) before
• Lower costs
• More production time
Industry: Still large potential for improvements! By improving control and squeezing
the variations we can shift the
setpoint (average) closer to the
constraint and increase production

Example: Control of shower

temperature
5s delay in pipe (θ = V/q = 100ml/20 ml/s = 5s)

T [K]
q [m3/s] want constant

(valves)

Reaction
time

qc

4
 Classification of variables
d

u Process y
input (MV) (shower) output (CV)

Independent variables (“the cause”):

(a) Inputs (MV, u): Variables we can adjust (valves)
(b) Disturbances (DV, d): Variables outside our control

Dependent (output) variables (“the effect or result”):

(c) Primary outputs (CVs, y): Variables we want to keep at a given
setpoint
(d) Internal variables in dynamic model (“states”) (x)

MV = manipulated vartiable (input u)

CV = controlled variable (output y)
DV = disturbance variable (d)

Example: Control of shower

temperature
5s delay: θ = V/q = 100ml/20 ml/s = 5s

T [K] Control objective. MVs, CVs, DVs

qH[m3/s] qC q [m3/s] 1. Control objective
Keep temperature (y1=T) a given setpoint
Keep flow (y2=q) (”pressure”) at given setpoint
2. Classify variables
MVs (u) = qH, qC (strictly speaking, valve positions zH, zC)
CVs (y) = T, q
DVs (d) = qH, qC (strictly speaking, upstream pressure
which gives “uncontrolled” flow changes)

Reaction
time

qc

5
 Inputs for control (MVs)
• Usually: Inputs (MVs) are valves.
– Physical input is valve position (z), but we
often simplify and say that flow (q) is input

Valve equation
z p2
p1

q [m3/s]

6
 Control
• Use inputs (MVs, u) to counteract the
effect of the disturbances (DVs, d) such
that the outputs (CVs, y) are kept close to
their setpoints (ys)

u y
input (MV) Process output (CV)

Two fundamental control principles

• Feedback: Measure the result (= controlled variable CV; output y)
and keep adjusting the manipulated variable (MV; input u) until the
results is OK
– Example: Measure the temperature T (CV) and adjust the flow of cold
water (MV)

• Feedforward: Measure the cause (= disturbance d; DV) and based

on a prediction (model!) make a ”forward” adjustment of the MV
(input u) to (hopefully) counteract its effect on the result (output y)
– Example: Room mate (disturbance d) says ”I am tapping cold water” -
and you know your friend so well (model) that you can make the correct
increase in your cold water (MV) to counteract d.
– NOT VERY REALISTIC FOR SHOWER EXAMPLE
– BUT a good example of feedforward is coming in time to lecture!

7
 BLOCK DIAGRAMS
FEEDBACK (measure output): d
ys ys-ym Controller u Process y
Desired value error (brain) input (MV) (shower) output (CV)
Setpoint
ym Measurement
measured output device

FEEDFORWARD (measure disturbance):

d
dm Measurement
measured disturbance device

Controller u Process y
(brain) input (MV) (shower) output (CV)

•All lines: Signals (information)

•Blocks: controllers and process
•Do not confuse block diagram (lines are signals) with flowsheet (lines are flows); see below

FEEDBACK
+ Self-correcting with negative feedback (keeps adjusting
until y=ys at steady state)
+ Do not need model (but most know process sign!)
- May give instability if controller overreacts
- Need good and fast measurement of output
MAIN ENEMY OF FEEDBACK: TIME DELAY
(in process or output measurement)

FEEDFORWARD
+ Good when large time delay (in process or output measurement)
+ Reacts before damage is done
- Need good model
- Sensitive to changes and errors
- Works only for known and measured disturbances
USUALLY COMBINED WITH FEEDBACK

8
 Piping and instrumentation diagram
(P&ID) (flowsheet)
• Solid lines: mass flow (streams)
• Dashed lines: signals (control)
Example: Shower
Ts
TC

.
qC T
valves
mixer pipe

q
qH
qs
FC

Notation feedback controllers (P&ID)

Ts
(setpoint CV)
T

TC
(measured CV) MV (could be valve)

2nd letter:
C: controller
I: indicator (measurement)
T: transmitter (measurement)
A: alarm

1st letter: Controlled variable (CV) = What we are trying to control (keep constant)
T: temperature
F: flow
L: level
P: pressure
DP: differential pressure (Δp)
A: Analyzer (composition)
C: composition
X: quality (composition)
H: enthalpy/energy

9
 Example: Level control

Inflow (d)
Hs
H
LC

Outflow (u)

CLASSIFICATION OF VARIABLES FOR CONTROL (MV, CV. DV):

INPUT (u, MV): OUTFLOW (Input for control!)
OUTPUT (y, CV): LEVEL
DISTURBANCE (d, DV): INFLOW

Level control when product rate is given

(less common)
Inflow (u)

Hs
H
LC

Outflow (d)

CLASSIFICATION OF VARIABLES FOR CONTROL (MV, CV. DV):

INPUT (u, MV): INFLOW
OUTPUT (y, CV): LEVEL
DISTURBANCE (d, DV): OUTFLOW

10
 Example: Evaporator with heating
qF [m3/s]
From
TF [K] evaporation
reactor

∞
level measurement
H
temperature measurement
T
q [m3/s]
T [K]
concentrate
qH [m3/s]
TH [K]
Heating fluid

1. Control objective
• Keep level H at desired value
• Keep temperature T at desired value
2. Classify variables (CVs, MVs, important DVs)
3. Process matrix (from MVs to CVs)
4. Suggest pairings and put control loops on the flowsheet

Most important control structures

1. Feedback control
2. Cascade control
3. Ratio control (special case of feedforward)

11
 Cascade control
• Controller (“master”) gives setpoint to another controller (“slave”)
– Without cascade: “Master” controller directly adjusts u (input, MV) to control y
– With cascade: Local “slave” controller uses u to control “extra”/fast measurement (y’).
“Master” controller adjusts setpoint y’s.

• Example: Flow controller on valve (very common!)

– y = level H in tank (or could be temperature etc.)
– u = valve position (z)
– y’ = flowrate q through valve

WITHOUT CASCADE WITH CASCADE

flow in measured
level Hs flow in measured
level Hs
H H
LC LC
master
MV=z MV=qs
valve position
q
FC
z
slave
measured
flow
flow out
flow out

What are the benefits of adding a flow

controller (inner cascade)?
qs

Extra measurement y’ = q
q z

f(z)
1. Counteracts nonlinearity in valve, f(z) 1

• With fast flow control we can assume q = qs

2. Eliminates effect of disturbances in p1 and p2 0

0 1 z
(FC reacts faster than outer level loop) (valve opening)

12
 Example: Evaporator with heating
qF [m3/s]
TF [K]
cF [mol/m3]
Feed: From evaporation
reactor

∞
level measurement
H
temperature measurement
T
T Concentrated product
q [m3/s]
T [K]
c [mol/m3]
c
Heating fluid
concentration measurement
qH [m3/s]
TH [K]

Control objectives
• Keep level H at desired value
• NEW: Keep composition c at desired value
BUT: Composition measurement has large delay + unreliable
Suggest control structure based on cascade control

Ratio control (most common case

of feedforward)
Example: Process with two feeds q1(d) and q2 (u), where ratio should be constant.

Use multiplication block (x):

(q2/q1)s
(desired flow ratio)
q1 q2
(measured x (MV: manipulated variable)
flow
disturbance)

“Measure disturbance (d=q1) and adjust input (u=q2) such that

ratio is at given value (q2/q1)s”

13
 Usually: Combine ratio
(feedforward) with feedback
• Adjust (q1/q2)s based on feedback from
process, for example, composition
controller.
• This is a special case of cascade control

– Example cake baking: Use recipe (ratio control = feedforward),

but adjust ratio if result is not as desired (feedback)
– Example evaporator: Fix ratio qH/qF (and use feedback from T
to fine tune ratio)

EXAMPLE: MIXING PROCESS

RATIO CONTROL with outer cascade (to adjust ratio setpoint)

(q2/q1)s
q1,m q2,s
x
q1 [m3/s] FC
C1 [mol/m3] q2,m
C2=0
Concentrate Water

∞
H
cm LC
CC c
q [m3/s]
c [mol/m3]
cs
Diluted product

14
 Procedure for design of control
system
1. Define control objective (why control?)
2. Classify variables
• MVs (u)
• Disturbances (d)
• CVs (y)
+ measurements Process matrix
3. Process description Input 1 input2
• Flow sheet
• Process matrix
– Qualitative: with 0, +, -, (+)*, (-)* Output 1 + -
Quantitative: transfer matrix (see later
courses)
4. Control structure Output 2 0 +
• Feedforward / feedback
• Pairing of variables (avoid pairing on 0!)
• Cascade loops (MV from one controller
(master) is setpoint for another (slave))
• Put on process & instrumentation
diagram (P&ID)
5. Control algorithm Process engineer (YOU):
• On/off • Responsible for items 1- 4
• PID (proportional-integral-derivative) • The most important is
• Model based (MPC)
process understanding
6. Implementation
• Today: Normally computer + connect
measurements and valves (actuators)

*(has some effect, but too small for control)

Rules for pairing of variables and

choice of control structure
Main rule: “Pair close”
1. The response (from input to output) should be fast, large and in one direction.
Avoid dead time and inverse responses!

2. The input (MV) should preferably affect only one output (to avoid interaction
between the loops; may use process matrix)
3. Try to avoid input saturation (valve fully open or closed) in “basic” control loops for
level and pressure
4. The measurement of the output y should be fast and accurate. It should be located
close to the input (MV) and to important disturbances.
• Use extra measurements y’ and cascade control if this is not satisfied
5. The system should be simple
• Avoid too many feedforward and cascade loops
6. “Obvious” loops (for example, for level and pressure) should be closed first, before
you spend too much time on deriving process matrices etc.

15
 Example: Shower
1. Define control objective (why control?)
• CVs: Control temperature T and flow q T
2. Classify variables q
• MVs (u): qc, qh
• Disturbances (d): Focus on main qc qh
• CVs (y): T, q
3. Process description
• Flow sheet
• Process matrix
4. Control structure
• Pairing of variables (Alt.1, Alt.2)
• Multivariable (Alt 3)
Input 1 Input2
qc qh

Output 1
T - +
Output 2
q + +

In this case the process matrix has no 0’s ) Interactive, so pairing is not obvious!

Multivariable control (“decoupling”) is used in practice:

One handle for total flow (qh+qc), one for ratio (qh/qc)

3x3 pairing example

Pairing: Choose one pairing from each row/column. Avoid pairing on 0’s

Inputs
u1 u2 u3

y1 + + +

y2 0 + -
Outputs
y3 0 + 0

16
 Example: Distillation
• Here: Given feed (i.e., feedrate is disturbance)
1. Objective: “Stabilize” column + keep compositions in top
and bottom constant
– But compositions measurements delayed +
unreliable
2. Classify variables
3. Process description
– Flowsheet
– Process matrix
4. Control structure: Stabilize column “profile” using
sensitive temperature measurement.

Typical distillation control:

• Level control using LV-configuration
(reflux L and boilup V left for CC)
• Two-point composition control (CC)
– with inner T-loop (cascade)
LV

CC xD

Ts
TC
Slave (fast)

CC xB
Master (slow)

17
Typical distillation control:
Two-point composition control (CC)
LV-configurationLV (L and V left for CC)
with inner T-loop (cascade)

CC xD

Ts
TC
Slave (fast)

CC xB
Master (slow)

Control hierarchy based on

“time scale separation”

setpoints

MPC or cascade/feedforward/…
(slower advanced and multivariable control)

setpoints

PID (fast “regulatory” control)

PROCESS
MPC = model predictive control
PID = proportional+integral+derivative control

18
 Inventory control rule (TPM)
• Inventory control: Usually control of level and pressure
• LocateTPM: Find out where throughput is set
• Rule («to keep things flowing»): Inventory control must
be radiating around TPM

QUIZ. Are these structures workable? Yes or No?

TPM

TPM

TPM

TPM

19
Quiz 2. Workable? Yes or No
TPM

TPM

TPM

TPM

20