Chapter Four

Feedforward control &

Ratio control
Feedforward control

• Consider the following feedback loop

Feedforward control

• As different disturbances, 𝒅𝟏 , 𝒅𝟐 &. . . , 𝒅𝒏 , enter the process, the

controlled variable c(t), deviates from set point, and feedback
compensates by manipulating another input to the process, the
manipulated variable, m(t).
• The advantage of feedback control is its simplicity.
• Its disadvantage is that in order to compensate for disturbances, the
controlled variable must first deviate from set point. Feedback acts on
an error between the set point and controlled variable. It waits until
the process has been upset before it begins to take corrective action.
Feedforward control

• The idea of feedforward control is to compensate for disturbances

before they affect the controlled variable.
• Specifically, feedforward calls for measuring the disturbances before
they enter the process and, on the basis of these measurements,
calculating the required manipulated variable to maintain the
controlled variable at set point. If the calculation and action are done
correctly, the controlled variable should remain undisturbed.
Feedforward control

Fig A Fig B
Feedforward control

• To further explain, consider a disturbance d(t), as shown in Figure B,

entering the process.
• As soon as the feedforward controller (FFC) realizes that a change has
occurred, it calculates a new value of 𝒎𝑭𝑭 (𝒕), and sends it to the process
(valve). This is done such that path 𝑮𝑴 negates the effect of path 𝑮𝑫 .
• 𝑮𝑴 is the transfer function that describes how the manipulated variable,
𝒎𝑭𝑭 (𝒕), affects the controlled variable, and 𝑮𝑫 , is the transfer function
that describes how the disturbance, d(t), affects the controlled variable.
Feedforward control

• To attain perfect negation, the feedforward controller must be designed by

taking into account the steady-state characteristics of the process.
• Assume that a change of plus one unit in d(t) affects c(t) by plus ten units
and that a change of plus one unit in 𝒎𝑭𝑭 (𝒕), affects c(t) by plus five units.
Thus if d(t) changes by plus one unit, affecting c(t) by plus ten units, then
the feedforward controller must change 𝒎𝑭𝑭 (𝒕), by minus two units,
affecting c(t) by minus ten units and, therefore, negating the effect of d(t).
Feedforward control

• The performance of feed-forward control is limited by model

uncertainty. Without a perfect process model the controller cannot
completely compensate for disturbances and there will generally
be offset. In practice, feed-forward control is combined with
feedback control.
 Designing Feedforward controller

• Consider the following

feedback-feedforward
control system.
• the objective of feedforward
control is to measure the
input(s) and, if a disturbance
is detected, to adjust the
manipulated variable to
maintain the controlled
variable at set point.
• Where:
HD = transfer function that describes the sensor/transmitter that
measures the disturbance.
FFC = transfer function of feedforward controller.
Consider the following feedback-feedforward control

Fig. C
.

• The output is the sum of signals from disturbance, feedforward controller

and feedback controller.
𝑪 𝒔 = 𝑮𝟏 𝒔 𝑮𝒑 𝒔 𝑪𝒊 𝒔 + 𝑮𝒇 𝒔 𝑮𝒑 𝒔 𝑪𝒊 𝒔 + 𝑮𝒄 𝒔 𝑮𝒑 𝒔 𝑬 𝒔 …. Eq.(1)
• In order to determine the transfer function of 𝑮𝒇 𝒔 that will prevent any
change in the control variable 𝑪 𝒔 from its set point R, which is 0, we
solve Eq.(1) for 𝑮𝒇 𝒔 with C = 0, R = 0. (i.e. if C=0 then R should be zero
even with disturbance).
•𝑬 𝒔 = 𝑪 𝒔 − 𝑹 𝒔 = 𝟎
• 𝟎 = 𝑮𝟏 𝒔 𝑮𝒑 𝒔 𝑪𝒊 𝒔 + 𝑮𝒇 𝒔 𝑮𝒑 𝒔 𝑪𝒊 𝒔 + 𝑮𝒄 𝒔 𝑮𝒑 𝒔 ∗ 𝟎
 𝑮𝒇 𝒔 𝑮𝒑 𝒔 𝑪𝒊 𝒔 = −𝑮𝟏 𝒔 𝑮𝒑 𝒔 𝑪𝒊 𝒔
𝑮𝒇 𝒔 = −𝑮𝟏 𝒔
• For this example:
𝟏
𝑮𝒇 𝒔 = − Eq. (2)
𝟓𝑺+𝟏
• If the load response of the control system in Figure C., with
𝟏
𝑮𝒇 𝒔 = − , were obtained for a step change in 𝑪𝒊 𝒔 , there
𝟓𝑺+𝟏
would be no deviation of C from the set point (i.e., perfect
control). This response is shown as Curve II in Fig. 18.11,
which, of course is a horizontal line at C = 0.
Response to step change in disturbance input 𝑪𝒊 𝒔
Response to step change in disturbance input 𝑪𝒊 𝒔

• Rather than use the 𝑮𝒇 𝒔 of Eq. (2) in the feedforward controller, one
can try using only the constant term of 𝑮𝒇 𝒔 , that is, 𝑮𝒇 𝒔 = −𝟏, The
response for 𝑮𝒇 𝒔 = −𝟏 gives Curve III in Fig. 18.11; this response has
a very large undershoot before the feedback controller returns to the
set point. If we try using 𝑮𝒇 𝒔 = −𝟎. 𝟓, we obtain Curve IV of Fig.
18.11; the undershoot is less in this case, but the response is still
unsatisfactory. As shown by Curves III and IV, omitting the dynamic
part of 𝑮𝒇 𝒔 can give very poor results. The success of using a
feedforward controller depends on accurate knowledge of the process
model, a luxury that may not be available in many applications.
Tuning Rules for Feedforward-Feedback Control

• In the practical application of feedforward control, one does not have

a block diagram with transfer functions as shown in Fig. C. For such a
practical situation, one can still tune the feedforward controller by
introducing a step change in the disturbance that enters the
feedforward controller and then applying some tuning rules.
Feedforward Rules

• In describing these rules, reference will be made to the general block

diagram for a feedforward-feedback system shown in Fig. 18.18. It is
assumed that 𝑮𝒇 𝒔 will be a lead-lag transfer function of the form
𝑻𝟏 𝑺+𝟏
𝑮𝒇 𝒔 = 𝑲𝒇 Eq.(4)
𝑻𝟐 𝑺+𝟏
Where:
𝑲𝒇 −steady state gain of feedforward controller
𝑻𝟏 and 𝑻𝟐 − time constants of dynamic part of the feedforward
controller
Feedforward Rules
Feedforward Rules

• The tuning rules listed below

are explained with the help of
Fig. 18.19. In that figure, a
unit step is selected for the
disturbance 𝑪𝒊 and 𝑲𝒇 has
been taken as - 1. In practice,
𝑲𝒇 will, of course, depend on
the particular process being
controlled.
Feedforward tuning Rules

1. Remove the control action in 𝑮𝒄 𝒔 by setting the controller to manual.

2. Set the feedback controller to the computed steady-state gain (𝑲𝒇 )
necessary to compensate ultimately for a step change in 𝑪𝒊 . This means
that the dynamic portion of 𝑮𝒇 𝒔 will be removed and only the
constant term (𝑲𝒇 ) will remain.
3. Make a step change in 𝑪𝒊 and observe the open-loop transient of C. The
general shapes of the response to be expected are shown in Fig. 18.19.
4. If the response shown in Fig. 18.19a occurs, lead must predominate in
𝑮𝒇 𝒔 of Eq. (18.4) (i.e., 𝑻𝟏 > 𝑻𝟐 ). If the response of Fig. 18.19b occurs,
lag must predominate in 𝑮𝒇 𝒔 (i.e., 𝑻𝟏 < 𝑻𝟐 ). The values of 𝑻𝟏 and 𝑻𝟐
in Eq. (18.4) are found by use of the information in Table 18.1.
Feedforward Rules
Example

• Consider the following non-interacting system,

• The above system has the following closed loop feedback-feedforward
block diagram
solution

• Apply the feedforward tuning rules to the system in Fig. 18.14. Since
this example is concerned with the application of the tuning rules to a
system for which a mathematical model is not generally available, the
reader should assume that the transfer functions for G1(s) and Gp(s)
in Fig. 18.14 are unknown. The determination of Gf(s) is to be
obtained solely by information from open-loop transients.
solution

• We must first determine the steady-state gain (Kf) for the system of
Fig. 18.14. If a step change in Ci is made, C will undergo a transient
and eventually level out at a steady-state value. If the controller
parameters are properly selected, the value of C at the end of the
transient will be the same as it was before the transient occurred. By
computation or experiment, one can determine the value of Kf needed
to obtain no change in C. For the system in Fig. 18.14, one can see that
Kf of Eq. (18.4) must be equal to -1.
• We must now apply the feedforward tuning rules to obtain T1 and T2
in Eq. (4). After removing the feedback controller action [𝑮𝒄 𝒔 ]we
have the equivalent diagram shown in Fig. 18.20. A unit-step change in
Ci produces the transient for C shown as Curve I in Fig, 18.21.
Comparing the shape of the transient with those of Fig. 18.19, we see
that lead must predominate in 𝑮𝒇 𝒔 . The peak value occurs at tp = 2.
Applying the rules in Table 18.1 gives
𝑻𝟏 = 𝟏. 𝟓 𝒕𝒑 = 3
T2 = 𝟎. 𝟕 𝒕𝒑 = 1.4
𝑻𝟏 𝑺+𝟏 𝟑𝑺+𝟏
Therefor, 𝑮𝒇 𝒔 = 𝑲𝒇 = −𝟏
𝑻𝟐 𝑺+𝟏 𝟏.𝟒𝑺+𝟏
FIGURE 18.21. Comparing the
shape of the transient with those of
Fig. 18.19
Ratio control

• A commonly used process control technique is ratio control.

• Ratio control is a strategy wherein one variable is manipulated to keep
it as a ratio or proportion of another.
• There are two different methods of applying ratio control.
I. Using multiplier
II. Using divider
Using multiplier

• Assume that it is required to blend two liquid streams, A and B, in

some proportion, or ratio, R.
• An easy way of accomplishing this task is shown in the figure. Each
stream is controlled by a flow loop in which the set points to the
controllers are set such that the liquids are blended in the correct
ratio.
What if one of the liquid has wild flow

• However, suppose now that stream A cannot be controlled but only

measured. The flow rate of this stream, often referred to as “wild
flow,” is usually manipulated to control something else, such as level
or temperature.
Solution I: Using Multiplier ratio station

• In this scheme, as shown in Fig. I,

consists of measuring the wild flow
and multiplying it by the desired
ratio, in FY102, to obtain the
required flow rate of stream B; that
Fig I.
is, FB = RFA,. The output of the
multiplier, or ratio station, FY 102, is
the required flow rate of stream B,
and it is used as the set point to the
flow controller of stream B.
Solution I: Using Multiplier ratio station

• Thus as the flow rate of stream A varies, the set point to the flow
controller of stream B will vary accordingly to maintain both streams
at the required ratio. If a new ratio between the two streams is
required, the new ratio is set in the multiplier. The set point to the flow
controller of stream B is set from a computation, not locally.
Solution II: using divider ratio station

• The second ratio control scheme,

shown in Fig. II, consists of
measuring both streams and
dividing them, in FY102, to obtain
the actual ratio flowing through Fig. II
the system. The calculated ratio is
then sent to a controller, RC101,
which manipulates the flow of
stream B to maintain set point. The
set point to this controller is the
required ratio and is set locally.