THE COPPERBELT UNIVERSITY

School of mines and minerals sciences

CE 560 / MT 580
Process control
Lecture 05
Control strategies
(feedback & feedforward)

C. Botha (Mr.)
Contacts: Chemical Engineering Department
Email: [email protected]
Alt. Email: [email protected]

January, 2024
 Feedforward control
o Figure 5.1 shows a feedforward control. For the stirred tank heater
Fi (ft3/min), Ti (oF)

Thermocouple
T
T - h
Set point Q
Ts F (ft3/min), T (oF)
+ ϵ

Controller Condensate
+
Fst (Ib/min)
Steam

Fig.5.1 Feedforward control for stirred tank heater.

o The distinguishing characteristic of this strategy is that the disturbance variable

(input) is measured and not the controlled variable.
 Select the control configuration: feedback and feedforward configurations

o I.e., measurement is done before the disturbance affects the process.

o The measurement is then used to adjust the manipulated variable.

o Thus, in feedforward, the controller anticipates the effect that the disturbance will
have on the process.

 Adjustments are then made beforehand to compensate for the anticipated effect
of the disturbance on the process.

 This will keep values of the controlled variable at or near their set points.

 I.e., achieve the control objective.

o In Fig. 5.1, the thermocouple measures the temperature of the feed stream
(disturbance).
 Select the control configuration: Feedback control strategy

o This measurement is then used to adjust the steam flow-rate.

o This control action results in the required correction to the stirred tank heater
temperature.

Positive and negative feedback control

o Feedback control can either be implemented as negative feedback or positive

feedback.

o Negative feedback: Corrective action taken by the controller forces the controlled
variable towards a set point value.

o Usually this is the desirable control action.

 Positive and negative feedback control

o Positive feedback: the controlled variable is forced to move further away from the set
point.

o Controller in this case makes the control worse.

o Typically, positive feedback is not the desirable corrective action but can be in certain
situations.

o To illustrate how positive or negative feedback can arise during control, consider the
blending process in Fig. 5.2.

o Stream 1 is a mixture of two chemical species but its mass flow-rate 𝑤1 is assumed
constant.
 Positive and negative feedback control

o The mass fraction of component A, 𝑥1 , on Control valve

the other hand, varies with time. Mixture of A and B Pure A

x1 , w1 w2,x2 1
o Stream 2 has a mass flow-rate of 𝑤2 and is
made up of pure A.

o Control objective is to regulate the exit h

composition so that it is within a desired
value. x, w
Fig.5.2 blending system
o Stream 2 flow-rate can, in this case, be
manipulated.

o Suppose that the nominal value of the feed stream composition is 𝑥ҧ1 ,
 Positive and negative feedback control

o then the nominal flow rate of 𝑤2 required to produce the desired composition
𝑥𝑠𝑝 in the exit stream can be computed.

o Overall material and component A balances can be used to determine the

required flow rate, 𝑤
ഥ2 .

o At steady state, overall balance is

𝑤ഥ1 + 𝑤ഥ2 − 𝑤
ഥ =0 5.1

o Component A balance is
𝑤
ഥ1 𝑥ҧ1 + 𝑤
ഥ 2 𝑥ҧ2 − 𝑤
ഥ 𝑥ҧ = 0 5.2

o The overbar denotes the nominal or steady state values.

 Positive and negative feedback control

o Since 𝑥ҧ2 = 1 and 𝑥ҧ = 𝑥𝑠𝑝 , combining 5.1 and 5.2 and gives
𝑤
ഥ1 𝑥1ҧ + 𝑤
ഥ2 𝑥ҧ2 − (𝑤
ഥ1 + 𝑤
ഥ2 )𝑥ҧ = 0 5.3
𝑤
ഥ2 − 𝑤
ഥ2 𝑥ҧ = 𝑤
ഥ1 𝑥ҧ − 𝑤
ഥ1 𝑥1ҧ 5.4

o Rearranging
𝑥𝑠𝑝 − 𝑥ҧ1
𝑤
ഥ2 = 𝑤
ഥ1 5.5
1 − 𝑥𝑠𝑝

o Equation 5.5 is the design equation for the blending process.

 I.e., if the assumptions made on the process are correct and that the inlet
composition is equal to the nominal value,
 Positive and negative feedback control

 Then the value of stream 2 flow rate computed using Eq. 5.5 will always produce
the desired concentration, 𝑥𝑠𝑝 .

o However, in a typical operation, the concentration of the feed stream can change
with time.

o E.g., 𝑥1 can increase to a new constant value larger than its nominal value 𝑥1ҧ .

o The exit composition will likewise change with time.

o Therefore, how can the exit concentration remain at or near the desired value?

o As previously discussed, one option to achieve this would be to use a feedback

control strategy.
 Positive and negative feedback control

o Option 1 – Feedback control: In using such a strategy, the outlet composition can be
measured.

o The measurement is then used to adjust the flow-rate of stream 2.

o This control strategy can be implemented by an operator (manual control).

o Although automating (automatic control) this task would be economical and more
convenient.

o If automatic control is used, a simple control algorithm (law) can be used, that is;
𝑤2 𝑡 = 𝑤
ഥ2 + 𝐾𝑐 [𝑥𝑠𝑝 −𝑥(𝑡)] 5.6

o Kc is a constant called the controller gain.

 Positive and negative feedback control

o Eq. 5.6 depicts how flow rate 𝑤2 should be varied from its nominal value.

o I.e., the change in the flow rate 𝑤2 𝑡 − 𝑤

ഥ2 is varied proportionally to the deviation
from set point 𝑥𝑠𝑝 − 𝑥 𝑡 .

o This an example of proportional control – to be discussed later.

o Varying 𝑤2 in this way will keep the exit composition at or near its desired value as
the inlet composition varies.

o For the blending process, positive feedback occurs when Kc< 0.

o Stream 2 flow-rate will increase in this case when exit composition increases.

o This will in turn increase exit composition further away from the desired value.
 Positive and negative feedback control

o Controller in this case is said to be direct acting controller.

o Negative feedback occurs when Kc> 0, i.e., controller gain must be positive.

o 𝑤2 will in this case decrease as 𝑥 increases and vice versa.

o This control action is what is required and the controller is said to be reverse acting.

o Hence, whenever feedback control is implemented, it is important to ensure that

negative feedback is used.

o There are situations where negative values of 𝐾𝑐 i.e., positive feedback is desirable.

o E.g., when the control valve is air-to-close so that flow rate reduces as actuating
signal increases.
 Feedback versus feedforward control configuration

Composition
controller
AC
Control
Mixture of A and B valve Pure A
X1 X2=1
w1 w2

AT
h Composition
analyser/Transmitter

X
w

Fig.5.3 Feedback control for blending system

 Feedforward control

o Option 2: Measure stream 1 composition 𝑥1 and manipulate stream 2 flow-rate 𝑤2 .

o A control law for method 2 can be written based on Eq. 5.5.

o I.e., 𝑥1ҧ and 𝑤

ഥ2 are replaced by 𝑥1 𝑡 and 𝑤2 𝑡 so that
𝑥𝑠𝑝 − 𝑥1 𝑡
𝑤2 𝑡 = 𝑤
ഥ1 5.7
1 − 𝑥𝑠𝑝

o Equation 5.7 describes how stream 2 should be varied when there is a change in
stream 1 composition.

o Note that option 2 is feedforward strategy.

o And the feedforward controller will use Eq. 5.7 as the control law, see Fig. 5.4
 Feedforward control

Composition Composition controller

AC
analyser/Transmitter
AT Control
Mixture of A and B valve Pure A
X1 X2=1
w1 w2

X
w
Fig.5.4 Feedforward control scheme for blending system
 Feedback versus feedforward

o Feedback control has the ability to handle disturbances of unknown origin.

o This is the main reason why it is the dominant control strategy in most industrial
processes.

o Feedback control also reduces the sensitivity of the controlled variable to unmeasured
disturbances and process changes.

 That is, controlled variable is less affected by unmeasured disturbances and,

 process changes such as impeller speed in the case of blending process.

o The disadvantage of feedback control is that no corrective action is taken by the

controller until the disturbance has an effect on the process.
 Feedback versus feedforward

 That is, after controlled variable deviates from set-point.

o An important advantage of feedforward control is that the corrective action is taken

before the controlled variable deviates from set-point.

o Ideally, it is desired that the control action completely cancels the effect of the
disturbance.

o However, complete cancellation is not possible.

o But the presence of feedforward can significantly reduce the effect of measured
disturbances.

o There are, however, three main disadvantages of feedforward control.

 Feedback versus feedforward

1) The disturbance variable must be measured or accurately estimated.

 In the case of unmeasured disturbances, no corrective action is taken by

the controller.

 In Fig. 5.2, no corrective action will be taken by the controller for

unmeasured 𝑤1 .

o This situation could in theory be dealt with by measuring both 𝑤1 and 𝑥1 and
adjusting 𝑤2 accordingly.

 Such a control scheme could also improve the quality of the control.
 Feedback versus feedforward

2) However, in industrial applications, it is not economical to install

instrumentation and measure every possible disturbance.

o A more practical approach would be to use a combination of feedforward and

feedback.

o Feedback in this case will provide corrective action for unmeasured

disturbance while feedforward reacts to eliminate measured disturbances.

o Consequently, industrial applications typically uses a combination of

feedforward and feedback control.
 Feedback versus feedforward

3) Another setback for feedforward control is that a process model is typically

required.

 The inferential control configuration for the distillation column is a special

case of feedforward control.

 A process model is used in this case to link the unmeasured distillate

composition to the accurately measured tray temperatures.

o In addition to feedback and feedforward control configurations, another type of a

control strategy is inferential control configuration.

o This has been discussed already.