Process Control

Dead Time,
Lead-Lag, and
Inverse
Seborg: Chapter 6.2
Response
Marlin: Chapter 5
 Learning Objectives

Understand the following process responses:

• Dead time: symbol is q (or td )
• Lead - Lag
• Inverse response
• Higher orders
• Be able to generate an empirical model
• Understand statistical modelling to handle real data

2
 Process Characteristics
“Dead Time” or “(Transport) Delay”
Flow in
Initially at ss TA = TB = 25oC
Q m3/s Tss = 25oC
Flow out Q m3/s
Tss = 25oC
Mixed Tank TB Tss = 25oC
Volume Flow velocity U
= V m3
Pipe, area A m2 = Q/A m/s
35
TA Dm 33
K = 1,
31 t = V/Q
Step change
in inlet T to
TAB 29
35oC at t =0
27 t=t
Dead time or 25
delay q = D/U s 0 20 40 60 80 100
Time t
3
 Dead Time or (Time) Delay

• Occurs when material or energy is transported through a process without

perfect and instantaneous mixing
• Symbol = q
• Units = time
• Time domain equation: y(t) = u(t – q)
• Output is delayed by q time units

4
 First Order Plus Dead time
• Laplace:
L  f (t  q)   e F(s)
qs

• For a 1st order system with dead time q:

Y ( s  Ke qs

U ( s  (ts  1)
• Called “FOPDT”: First Order Plus Dead Time
• Three parameters: Gain K, time constant t and dead
time q

5
6

FOPDT = FO + DT
(First Order plus Dead Time)

1st order response

Shifted along time axis
by t=q

q
DeadTime
 Modeling real, high order processes

PC
• To control real processes
LC we need dynamic models
CC
• Transfer functions would
be very high order
• We could not develop
FC
accurate values for the
LC order or the time
constants
CC • We need a simplified
dynamic model

7
 Modeling real,
high order processes

PC • Aside … the CC devices

LC will have VERY long dead
CC
times.

FC

LC

CC

8
 Higher orders &
Model Approximations
• Have n 1st order processes in series, you have a “n” order system
• Distillation column with 10 components and 50 trays
• 500 equations needed !!!
• Can empirically model these systems
• Only interested in outputs
• No detail of each step
• “Black box” models but still useful

9
 Higher order response example
e.g. 10 tanks in a
line.
Step change in
tank#1 inlet
Can model as:
FOPDT or SOPDT
(SOPDT = Second
order plus dead
time)

Response of tank #10 to step change

10
Empirical Models: FOPDT vs SOPDT

Experimental
FOPDT
response

SOPDT FODT
FOPDT
SODT
SOPDT

Reasonable representation
• SOPDT better, but harder to
estimate parameters

11
 Empirical modelling processes
• Most chemical processes can be modelled by FOPDT
• Second order plus dead time (SOPDT) is used less often
• Underdamped can be useful
• For FOPDT
• 3 parameters - K, t and q).
• Can also use statistics to estimate parameters

12
 Using Dead Time to Model
Real Processes

• High order processes usually 1

have “s-shaped” step (1 s)(1 2s)(1 3s)(1 4s)(1 5s)
response curves 1

• We can approximate them

0.9

e 6.5s
0.8

with a dead time plus a low 0.7

10.5s+1 5th order
order transfer function

Response
0.6
process
• Two options:
0.5

0.4
In practice: Curve fitting
• First order + dead time 0.3 by non-linear least
(FOPDT) 0.2 squares (minimises
• 2nd order + dead time 0.1 effects of noise)
(SOPDT) 0
0 5 10 15 20 25 30 35 40

Time

13
 Manually Fitting an FOPDT Model
to Step Response Data
Find, for known input step
d: 1

• Total ss change in output D Ke  qs

0.9

ts  1
0.8
• Time for 63% of this change, t63
0.7
t28
• Time for 28% of this change t28 0.6

Response
K = D/d 0.5
D
t = 1.5*(t63 - t28) t63
0.4

0.3

q = t63 - t 0.2

0.1

0
0 5 10 15 20 25 30 35 40

Time

14
 Statistical Methods

• Compare response to the model calculations

• Example - least squares
• Estimate parameters K, t, q and calculate response as function of time
• Take squared difference between real data and estimated value for each point – then sum
total error
• Use solver to adjust K, t, q until the “sum of the squares of the errors” is minimised
• This is the best fit of model parameters to the model

15
 Statistical Methods

See Marlin textbook 16

 Parallel Structures – Lead-Lag

If both elements are first

U
order, the overall model is

6 U
5

4 Lead-Lag occurs when one function is

3
faster (the Lead) than the other function
2
(the lag)

t ,K t and t .
1

0 3 p are functions of K1, K2, 1 2

Can you derive Kp and t3??

0 5 10 15 20 25 30 35 40 45
17
 Parallel Structures – Lead-Lag

U
U

Y K
More generally:  1st order LAG unit
U ts  1
Y
 K(ts  1) 1st order LEAD unit
U
Numerator (ts+1) = lead unit;
denominator (ts+1)=lag unit
18
Lead lag response to step change:
depends on time constants ratio
Y (tlead s  1)
 K lag K lead
U (tlag s  1)

NOTE: useful later for feedforward

control

19
 Lead lag response to ramp:
depends on time constants ratio

Lead-Lag is important
Lead Input ramp
for feedforward control

Lag
U

20
 Inverse response
A special kind of 2nd order lead-lag dynamics where
response initially moves in the opposite direction to the
steady state response.

•One TF Gain is positive,

•The other is negative.
•Time constants differ (one
fast, one slow)
•Summed to produce
overall response

21
 Inverse Response –
Thermometer example
Liquid initially dips down then increases as expected – why?
• Put the thermometer in a hot fluid
• First, heat causes glass to expand
• Liquid height initially dips a little.
• Then heat is conducted from the liquid through the glass to the
fluid.
• Liquid expands and rises, and you can read the new temperature

• 2nd order system, inverse response

22
 Inverse Response example:
Boiler Drum
• Boiler drum (water tube boiler) of a
steam power plant.
• High-pressure feedwater is pumped to
the drum.
• Water from the drum circulates
through the boiler tubes, heats to form
steam.
• Steam reenters the drum before exiting
via the steam flow line (goes to the
turbine).

Drum is filled with high pressure boiling water and steam

23
 Inverse Response Boiler Drum –
sudden decrease in steam flow
Need to control drum level, by
controlling the feed water flow,
whilst steam demand varies.
Q: What happens to the level if
the steam flow suddenly
decreases?
A: Drum level will initially drop
before it rises and reaches
steady state.

24
 Inverse Response Boiler Drum –
sudden decrease in steam flow
• Drum level will initially
drop before it rises and
reaches steady state.

25
 Explanation
(for you to read outside of lecture)
• Let us consider a simple example of a system with inverse response.
Consider the dynamic characteristics of a boiler drum in a water tube
boiler of a steam power plant. High-pressure feedwater is pumped to
the drum. Water from the drum circulates through the boiler tubes,
gets heated and is converted to steam. This steam again comes back to
the drum and subsequently is taken out through the steam flow line to
the turbine. So the drum is filled up partially with water and partially
with steam, both at high pressure. It is very important to control the
water level of the drum at a desired level, by controlling the feedwater
flow, with the varying demand of steam.
26
 Explanation
(for you to read outside of lecture)
• The instantaneous level of water at the boiler drum is decided by the steam flow
rate and the feedwater flow rate and it would reach a steady state when both
are equal. Now suppose, the steam flow rate suddenly decreases, the feedwater
flow rate remaining constant. At a first glance, it would appear that the drum
level should rise. But actually the drum level will initially drop for some time
before it rises and reaches a steady value. This is because of the fact that drop in
steam flow rate will initially cause the rise in steam pressure in the drum. Due to
the rise in pressure, the bubbles present inside the water in the drum will
momentarily shrink. This will cause the temporary fall in the drum level. Similarly,
for a sudden increase in steam flow rate, the drum level will momentarily swell
before it drops down to a steady state vale. A typical response curve of the drum
level due to the sudden fall in steam level
27
 Inverse response - Boilers

• Basic boiler control videos:

• https://www.youtube.com/watch?v=_NOftnDDOyE&list=PLc0HvraDpVExZIIgih
XKvdqLKvpZ9Dc2m
• ~ 10 mins
• Runs through different boiler control modes and process implications

28
 Inverse Response Block Diagram

Lead-lag units BUT one has a negative K

U
X
G1

U
X + Y
G2
+

29
 Another way to determine
inverse response

Inverse response occurs when τ3 <

0.

30
 Summary

• FOPDT
• Three parameters: Gain, time constant and dead time
• Leant how to model it
• SOPDT also possible, but no easy method to model.
• Lead-lag systems, where one process is much faster and “leads”
the other “lagging” process
• Inverse systems are lead lag with a negative gain on one unit –
response moves initially in opposite direction
• Statistical modelling needed to handle real data.
31