THE COPPERBELT UNIVERSITY

School of mines and minerals sciences

CE 560 / MT 580
Process control

Lecture 06
Control strategies

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

January, 2024
 Feedforward & Feedback strategies - summary

Feedforward Feedback
Advantage o Compensates for disturbance before o Effective for all sources of
process output is affected. disturbances.
o Does not affect the stability of a o Does not necessarily need
closed-loop control system. models.
o Can provide zero steady-
state offset.
Disadvantage o Not effective for disturbances of o Does not take control action
unknown sources. until process output variable
o Requires a sensor and a model for each has been affected.
disturbance. o Affects stability of a closed-
o Cannot eliminate steady-state offset loop control system.
(Since it does not use traditional
controllers i.e., PI, PID).
 Single-loop enhancement using feedforward

o Single-loop enhancement takes

advantage of the additional F2

measurement information to improve A2

performance of a control. T2 T3
Feed

o Fig. 6.1 shows a packed-bed reactor. F1 T1

Set point
AC

 Objective is to tightly control the

product concentration by adjusting
preheat flow.
A1

 Product concentration is controlled Product

using a single-loop feedback

strategy.
Fig.6.1 Single-loop feedback control strategy
 Single-loop enhancement using feedforward

o Suppose that feed composition is the single most important disturbance.

o Effectiveness of this strategy in attenuating possible disturbances can be

qualitatively assessed.

o Feed inlet temperature: Change in feed inlet temperature 𝑇1 affects reactor feed
temperature 𝑇3 .

 𝑇3 affects reactor exit concentration, assuming the reaction is endothermic.

 Therefore, an increase in exit concentration will ultimately be detected by the

composition analyser.

 Thus, the controller will respond by closing the preheat valve to reduce the
heating of the feed stream.
 Single-loop enhancement using feedforward

 Single-loop feedback is effective but slow to effect the necessary correction.

 Hence, performance to attenuate feed inlet temperature changes is inadequate.

o Similar analyses can be made for 𝐹1 , 𝐹2 and 𝑇2 with similar findings as for 𝑇1 .

o A change in feed composition also has an effect on the reactor exit concentration.

 Increase in feed concentration may result in increased reaction rate.

 Hence, an increase exit concentration.

 The increase in exit concentration will be detected by the composition analyser.

 Controller responds by closing the heating medium valve and hence reducing
𝑇3 .
 Single-loop enhancement using feedforward

o Since feed composition is the most significant disturbance, the controller must be
able to compensate for this disturbance rapidly.

o Single-loop feedback is therefore effective but not adequate since the response is
slow.

o Hence, performance can be improved by combining with feedforward.

o Fig. 6.2 shows the same packed bed reactor but controlled with feedback-
feedforward combination.

o Feedforward adjustments must be implemented in a way that does not interfere with
feedback adjustments.
 Single-loop enhancement using feedforward

Feedforward controller

AC-2
F1

A2
T2 T3
Feed +

+
F1 T1

AC-1

A1

Product

Fig.6.2 Feedforward-feedback control for packed bed reactor

 Single-loop enhancement using feedforward

o The process is assumed to behave linearly with respect to the two adjustments.

o Hence, the feedforward and feedback adjustments can be calculated independently

and added.

o Two or more controller signals can be combined if they both:

 Adjust the same manipulated variable, i.e., preheat flow-rate in this case.

 Regulate the same controlled variable, i.e., exit concertation in this case.

o Hence, the feedforward controller output AC-2 is added to the feedback controller
output AC-1.

o The combined signal is sent to the preheat control valve.

 Single-loop enhancement suing multiple feedforward loops

o A single feedback loop can attenuate the effect of several disturbances.

o However, for feedforward, each disturbance must be measured for it to be effective.

o This results in multiple measurements of the disturbance variables.

o Assuming linearity in the process response, adjustments can be calculated for each
disturbance.

o The adjustments, as before, can the be combined, provided that:

1) they are using the same manipulated variable

2) the controlled variable being regulated is the same.

 Multiple feedforward loops

o Fig. 6.3 shows a stirred tank heater with two feedforward controllers.
Feed

F1 T1 Product

T
FC TC h T1
Q

Setpoint
TC

+
F

Heating oil

Fig.6.3 Stirred tank with two feedforward controllers and feedback

 Criteria for implementing feedback

o Feedback control is required when

 Feedforward control does not provide satisfactory control performance.
 Measured output (controlled) variable is available and reflect the control objective.
 A manipulated variable is available.
o Criteria for selecting manipulated variable is:
 There must be a causal relationship between manipulated and controlled variables.
 Must have fast dynamics compared to the controlled variable.
 Must be able to compensate for large disturbance changes.
 Able to be adjusted with little upset to the remaining plant or process.
 Criteria for implementing feedforward

o Feedforward control is required when

1) Feedback control does not provide satisfactory control performance.
2) A measured feedforward variable is available.
o Furthermore, a feedforward variable must satisfy the following:
1) Variable must be able to indicate occurrence of the most important disturbance.
2) There must not be a causal relationship between the feedforward and the
manipulated variables.
3) Disturbance dynamics must not be significantly faster than the manipulated
variable dynamics.
 Cascade control

o Feedback has the inherent drawback i.e., fails to control until output variable deviates.

o And with feedforward, disturbance compensation is done before process is affected.

o Thus, feedforward offers better improvement for processes with large time constants
or time delay (dead time).

 Large time constant means process response to a disturbance will be slow.

 Hence, response would be significantly delayed for feedback.

 A large delay in response is undesirable as it may lead to large deviation from

setpoint.
 Cascade control

o But, feedforward requires measurement of every disturbance and a model for each.

o E.g., feedforward control of stirred tank heater, controlled variable is temperature 𝑇.

o Measurements are 𝑇𝑖 or 𝐹𝑖 thus a correlation between 𝑇𝑖 or 𝐹𝑖 and 𝑇 is required.

o I.e. a model for feedforward implementation is

𝑇 controlled variable = 𝑓 𝑇𝑖 or 𝐹𝑖 or both 6.1

o Measuring 𝑇𝑖 and 𝐹𝑖 allows 𝑇 to be determined by the controller.

o If setpoint (𝑇𝑠 ) is provided, the feedforward controller will calculate the adjustments.

o Models depends on the process complexity and may not always be available.
 Cascade control

o May also be difficult to obtain especially for a process in operation.

o Both feedback and feedforward requires no fluctuations in manipulated variable.

o Hence, an alternative control approach such as cascade control may be convenient.

o Cascade control: Control strategy that uses two controllers with the output of the
first serving to adjust the set point for the second controller and the control loop for
the second controller is nested inside the control loop for the first..

 Requires secondary measurement, secondary controller, primary measurement

and primary controller.

 Secondary measurement is selected such that it indicates the key disturbance.

 Cascade control

 Thus, cascade control can be used when the key disturbance occurs in
manipulated variable.
o The two controllers and hence the two loops are referred to as follows:
 Controllers: primary or outer or master controller and secondary or inner or
slave controller.
 Corresponding control loops are primary or outer or master and secondary or
inner or slave.
 Secondary controller dynamics should be faster than primary controller.
o For the stirred tank heater, the cascade control is shown in Fig. 6.4.
o Control objective, i.e., tight control of the exit temperature remains the same.
 Cascade control – stirred tank heater

o Hence, primary measurement is the Fi (ft3/min), Ti (oF)

exit temperature.
CV1
o Suppose that the most significant
disturbance variable is the steam T TT
pressure. h
Q

 Steam flow rate is selected as the TC

SP1 from
secondary measurement. Condensate
operator

SP2=MV1
 Since it responds in a predictable MV2

manner to the disturbance, CV2

Fst (Ib/min)
FT FC
 I.e., changes in steam pressure Steam
which is not measured.
Fig.6.4 Stirred heater tank with cascade control
 Cascade control – stirred tank heater

o In the absence of the secondary loop, a single-loop feedback makes slow

correction for changes in steam pressure.

o Cascade results in faster correction thus provides better performance.

o Initial response to changes in steam pressure can be analysed.

o Initially system is at steady state and exit temperature is constant.

o Also primary and secondary controller outputs are constant.

o But the secondary controller output is such that it gives a valve opening that
results in steam flow rate that is just enough to keep temperature at the desired
value.
 Cascade control – stirred tank heater

o When steam pressure increases, the flow sensor/transmitter measures this increase.

o At this point, exit temperature is still at the desired value and set point for secondary
controller is still constant.

o Thus, the secondary controller sees the increase in flow rate as an increase in error.

o It then adjusts the control valve to reduce steam flow.

o Since secondary sensor and valve must have fast dynamics, flow controller rapidly
achieves desired steam flow.

o This restores steam flow to its pre-existing level at which the system was at steady
state.
 Cascade control – natural draft furnace

o Consider the natural draft furnace in which the exit hot oil temperature is to be
controlled. Stack gas

SP2=MV1

PC
Furnace
PT
Fuel gas TC
CV2 SP1 from
operator

CV1
F (ft3/min), TT
Ti (oF) F (ft3/min),
To (oF)

Cold oil Hot oil

Fig.7.4 Furnace temperature control using cascade control

 Cascade control – natural draft furnace

o A single-loop feedback control would adjust the fuel valve directly.

o Suppose the key disturbance is the fuel gas supply pressure.
 Fuel gas flow-rate will change and hot oil temperature will eventually be
affected.
o Feedback only results in delayed action since hot oil temperature must change.
 Thus, conventional feedback control results in a sluggish response as fuel gas
pressure changes.
o Similar initial response analysis as for the stirred tank heater can be made.
o Hence, cascade control improves control performance.
 Since it results in a faster corrective action as fuel supply pressure changes.
 Criterial for implementing cascade control

o Cascade control is required when

1) Single-loop control does not provide satisfactory control performance.
2) A measured secondary variable is available.
o Furthermore, a feedforward variable must satisfy the following:
1) Secondary variable must be able to indicate the occurrence of an important
disturbance.
2) There must be a causal relationship between the manipulated and the secondary
variables.
3) Secondary variable dynamics must be significantly faster than the primary
variable dynamics.