MODULE I- PROCESS CONTROL STRATEGIES

BASIC PROCESS CONTROL STRATEGIES

The purpose of a process control system is to maintain the controlled variable at the setpoint.
The complexity of the system will depend upon the process under control. There are different types
of variables involved in process control and are discussed here.

DIFFERENT VARIABLES IN PROCESS CONTROL

Single variable process control

It is the basic process control strategy. There is only a single variable that is under control.
The variable is measured directly and is kept at the set point, regardless of other process variables. Is
the simplest and the cheapest process control strategy.
In the example, there is a
process control tank. Water
is fed to the tank through
the inlet tube. The flow
control loop maintains the
flow at a desired set point.
This is one single variable
loop. A second single-
variable control loop, as in
the figure, regulates the
temperature of liquid in the
tank by adjustment of heat
input. This also is a single-
variable loop that maintains the liquid temperature at the desired setpoint value.

Independent variable process control

In certain process control applications, it is desirable to keep the process variable at the
setpoint, regardless (independent) of the change is other process variables. The control of this type is
known as independent variable process control.
In the previous example, when there is a load change in flow, the controller adjusts the control
valve to bring the flow to the desired setpoint. This is not dependent on the temperature control loop
in the process tank. Thus, in the example, control of flow is independent variable process control.
Interactive variable process control
In certain processes, there can be more than one control loop. The action of one control loop
may affect the action of other. The loops can interact. In such processes, the control of one of the
variables will depend upon the change in other variable.
In the previous example, when flow is held constant (under nominal conditions), the
temperature is driven towards setpoint using the temperature control loop. Both flow and temperature
are held constant. But when the flow changes, it will appear as a load change for the temperature
control system, because, the change in inlet flow will vary the temperature of the fluid in the tank.
That is, the variable temperature is interacting with the variable flow and hence the temperature
control is interactive variable process control.

Compound variable process control

In some process control applications, a single loop is used to provide control of the
relationship between two or more variables. This is termed as compound variable process control.
This is done by applying measurements from two sensors as input to the controller. Sometimes a
signal conditioning system is used to scale the measurements from two sensors. The analysis of such
systems are generally complicated.

A common example is the control of chemical reaction. In a chemical reaction, it is desirable

to keep the ratio of two reactants as a constant. In this example, flow rates of both the reactants and
measured, scaled and added and is given as input to the controller. The controller compares it with
the set point (set point will be scaled sum of two desired flow rates) and adjusts the control valve
accordingly. It is to be noted that, the flow rate of only one of the reactants (reactant B) is controlled,
and not both the flow rates.

Multivariable process control

In several practical control applications there are several variables to be controlled and this is
done by adjusting several manipulated variables. Such systems involving several independent or
interacting controlled/manipulated variables are known as multivariable systems. An important
characteristic of multivariable systems is process interaction. That is, each manipulated variable can
affect all the controlled variables. In-line blending system and distillation column are two examples
of multi variable systems.

a) In-line blending system

Two streams A and B are blended (mixed) to

produce a product stream. WA and WB denote
the mass flow rates of A and B respectively. At
the product side, w and x represents mass flow
rate and composition respectively of the product
stream. Varying either WA or WB affects both w
and x. Hence this is a multivariable system with
process interaction.

b) Distillation column
 A distillation column is used to separate the components in a mixture, based on the difference
in their volatilities. The feed indicates the mixture which is supplied to the column. Steam (S) is
passed through the bottom of the column to heat the mixture. Components having high volatility rises
up the column and is collected at the output as the distillate product. 𝑥𝐷 denotes the composition of
distillate product. Those components having less volatility will remain at the bottom of the distillation
column and is collected as the bottom product. 𝑥𝐵 denotes the composition of bottom product. A
portion of the distillate product known as Reflux R is condensed and supplied back to the top of the
column, which again undergoes the process and gets separated to either distillate product or bottom
product. Here the two main input variables are reflux R and Steam S and output variables of the
process are distillate composition 𝑥𝐷 and bottom composition 𝑥𝐵 . Varying either S or R will vary
both 𝑥𝐷 and 𝑥𝐵 . Hence this is another multivariable process with significant process interaction.

IMPORTANT CONTROL STRATEGIES FOR A PROCESS

Feedback control

In feedback control system, the controlled

variable is measured using sensors or
transducers. The measured value of controlled
variable is compared with the setpoint. Based on
the deviation of controlled variable from the
setpoint, the controller changes the action of
final control element, on the process. It has the
following advantages:

1. Corrective action takes place as soon as the controlled variable deviates from the setpoint,
regardless of the type of disturbance.
2. Development of a process model is not required for feedback control.
3. The most common form of feedback controller is a PID controller. It is simple, less complex
mathematically and robust. It gives satisfactory control for almost all systems.

A feedback controller has the following disadvantages:

1. Corrective action is taken only when the controlled variable deviates from the setpoint. Even
if the deviation is known in advance, no action can be performed until it actually occurs.
2. There is no predictive control in the presence of disturbances.
 3. It is not suitable for slow processes, that is for process with dead time or cycling.
4. Feedback control requires measurement of the controlled variable on-line. It may not be
possible if the process environments are dangerous.

Feedback control of liquid level in a boiler drum

Figure represents feedback control of liquid level in a boiler drum. Level is measured using a
Level Transmitter (LT) and is compared with the setpoint at the Level Control (LC). Based on the
deviation of Level from the setpoint, the control valve is adjusted accordingly. The use of feedback
control for controlling level in boiler drum has two disadvantages. One is that, steam is continuously
withdrawn from the boiler drum dur to requirements downstream the process. Hence, the level
fluctuates very often. Second disadvantage is that, it is difficult to measure the level of boiling liquid.
The steam output act as a disturbance in this system. Hence, feedforward control is preferred for
controlling the level in a boiler drum.

Feedforward control

In feedforward control, instead of measuring the

controlled variable, disturbances affecting the
process is measured. Using a proper process model,
the effect of disturbance in the variation of
controlled variable is calculated. The measured
value of disturbance is applied to the feedforward
controller. The feedforward controller evaluates the
effect of disturbance on controlled variable and based on the setpoint, generates a control signal U,
which is used to control the process. It has the following advantages:

1. In feedforward control, disturbance is measured and the corrective action is taken before the
controlled variable deviates from the setpoint.
2. It is suitable for slow processes.
3. The effect of disturbances is taken into consideration
4. It does not require the measurement of controlled variable online.

Feedforward control has the following disadvantages:

1. The disturbance variables are to be measured online. Sometimes it is not feasible.

2. Feedforward control needs a mathematical model for the process called the process model.
Developing a mathematical model for an industrial process is a difficult task.
3. If there are unmeasurable disturbances, feedforward control is not effective.

Feedback control of liquid level in a boiler drum

The feedforward control scheme is better suited for the control of liquid level in a boiler drum. In this
method, the controlled variable is the liquid level, but it is not measured. Steam flow output is a
disturbance variable. It is measured and given to the feedforward controller. Based on the amount of
steam, the flow of feedwater to the boiler drum is controlled. The liquid level is maintained at the
setpoint, before any deviation or error occurs and without actually measuring it.
 Ratio Control

Ratio control is special type of feedforward control. It has a lot of applications in process
industries. Its aim is to maintain the ratio of two process variables at a constant specified value. The
two variables are usually flow rates, a manipulated variable 𝑢 and a disturbance variable 𝑑. Thus, the
ratio 𝑅 is maintained at the setpoint ratio, instead of controlling the individual variables.

𝑢
𝑅=
𝑑

Applications of ratio control include the following:

• Specifying the relative amount of components in blending operations.

• Maintaining the stoichiometric ratio of reactants in a reactor.
• Keeping specified reflux ratio in a distillation column.
• Maintaining air-fuel ratio for combustion in a furnace at an optimum value.

Ratio control scheme can be implemented using two schemes.

Ratio Control – Method I

In this method, the flow rates of both the disturbance stream and manipulated stream are measured.
The ratio of measured flow rates is then calculated.

𝑢𝑚
𝑅𝑚 =
𝑑𝑚

The output of the divider

element is sent to the ration
controller (RC) and is
compared with the desired
ratio 𝑅𝑑 of flow rates. If the
desired ratio is not equal to the
ratio of measured flow rates,
the flow of manipulated
stream is controlled
accordingly. A PI controller is
usually used as the ratio
controller (RC). The
advantage of this method is that the measured ratio 𝑅𝑚 s calculated. The disadvantage is that it needs
to include a divider element in the loop. In order to avoid this, the second ratio control method is
usually preferred over the first one.

Ratio Control – Method II

In this method, the flow rate of both

disturbance stream and manipulated
stream is measured. The measured value
of disturbance flow rate is transmitted to
the ratio station (RS). It is multiplied
with the desired ratio of flow rates. The
output signal from the Ratio Station is
then used as the set point 𝑢𝑠𝑝 for the flow
controller. The flow controller then
adjusts the flow rate of manipulated
stream accordingly. Both ratio control
schemes are a type of feed forward
control.

Split Range control

In split range control, there is only one measurement (controlled output) and more than one
manipulated variable. The control signal is split into two parts. Each will affect one of the two
manipulated variables. That is, a single process output is controlled by coordinating the action of
several manipulated variables. Such systems are not common in industry. But, they increase safety of
the system.

Split Range Control of chemical reactor

In the figure, split range control of a chemical reaction is demonstrated. The reactants are in gaseous
phase. There is only one measurement, which is the pressure inside the reactor. Based on the variation
of pressure, the flow rates of both the feed and the product streams are manipulated. That is, there are
two manipulated variables. When the pressure in the reactor increases, the controller’s output signal
increases. It is split into two parts and affects the two valves simultaneously.
The following actions take place.

1. When the controller’s output signal is less that 3 psi, Valve V1 is completely open and Valve
V2 is completely closed.
2. As pressure increases, when it range is between 3 psi-9 psi, valve V1 remains fully open. The
Valve V2 will gradually begin to open.
3. At 9 psi, both the valves are fully open.
4. As pressure further increases (between 9 psi and 15 psi), valve V2 will remain fully open. On
the other hand, valve V1 will begin to close.
5. When the pressure is large (>15 psi), Valve V1 is fully closed and valve V2 is fully open.
 Cascade Control

In cascade control, there is only one manipulated variable and more than one measurement. That is,
only one output can be controlled. Cascade control can be explained with the example of CSTR.

Consider the Continuous

Stirred Tank Reactor shown in
figure. The reaction is
exothermic (Generates heat).
The heat generated is removed
by the coolant, which flows in
a jacket around the tank. The
control objective is to keep the
temperature of the reacting
mixture, 𝑇, at a constant
desired temperature. The disturbances in this system are the feed temperature 𝑇𝑖 and the coolant
temperature 𝑇𝑐 . The manipulates flow rate is the coolant flow rate 𝐹𝑐 . That is, the temperature of the
reactant mixture is kept at the desired value by controlling the flow of coolant through the jacket.

For this system, if feedback control (Refer figure in your notebook) is effective only to control
the changes in 𝑇 due to 𝑇𝑖 . In order to control the changes in 𝑇 due to 𝑇𝑐 , cascade control must be
used.

In cascade control of CSTR, the coolant temperature 𝑇𝑐 is measured. The control action is
provided before its change affects the reacting mixture. We can see that there are two measurements,
𝑇 and 𝑇𝑐 , but there is only a single manipulated variable 𝐹𝑐 . The cascade control loop works in the
following two steps:

1. The first loop measures the temperature of the reacting mixture 𝑇, compares it with the desired
value of temperature at the first Temperature controller and gives it s output as the setpoint to
the second Temperature Controller. This is the primary loop or Master loop.
2. The second loop measures the coolant temperature 𝑇𝑐 and controls the flow of cooant 𝐹𝑐
accordingly. This is the secondary or slave loop.

This control configuration consisting of two loops is the cascade control. It is very common in
chemical processes.
Generalized Block Diagram of Cascade Control

Consider a process consisting of two parts, Process I and Process II. The output of process I
is the main controlled variable which we want to control. Its control is done using the primary
controller. The output of Process II affects the main controlled variable, hence it is also controlled,
using a secondary controller. The measurement of both the processes are done using separate
measuring devices. The loop consisting of primary controller, Process I and the corresponding
measuring device is the primary loop. The loop consisting of Secondary controller, Process II and the
corresponding measuring device is the secondary loop. In the CSTR system, reaction in the tank is
the Process I and the controlled output is temperature T. Process II is the jacket and its output 𝑇𝑐
affects process I. One important characteristic of cascade control is that the disturbances in the
secondary control loop are corrected by the secondary controller before it affects the value of primary
controlled output.

Adaptive Control

An adaptive control system is one whose parameters are automatically adjusted to meet the
corresponding variations in the process under control, in order to optimize the response of control
loop. There are different types of adaptive control schemes, depending upon the way the parameters
are updated. The two main reasons for using adaptive control method are,

1. Most processes are nonlinear in practical. Controller parameters are designed for linear
processes. Hence they need to be updated.
2. The characteristics of the process changes with time and environment. The parameters of the
controller should be updated according to the changes in process.
 The above changes require a revise of control parameters, that is a change in P, PI, PD or PID
controller gains. The tuning of an adaptive mechanism means the adjustment of the parameters of the
controller whenever there are changes in the process characteristics. The implementation of adaptive
controller requires complex mathematical computations.

Method I – Programmed or Scheduled Adaptive Control

The programmed or scheduled adaptive control is used when a measurable process

variable produces a predictable effect on the process control loop. Compensation for this effect by a
programmable adjustment mechanism. This is done in the following steps.

 Let there be an auxiliary process variable which reflects the changes in process.

 Program the adjustment mechanism based on the measurement of auxiliary process

variable.

 Adjust the controller parameters based on the adjustment mechanism.

There is an inner loop and an outer loop. The inner loop is a normal feedback loop consisting of
process and controller. The outer loop consists of measurement of auxiliary variables and the
adjustment mechanism. Examples include gain scheduling adaptive controller and adaptive
control of air-fuel ratio in combustion system.
Example (a) - Gain scheduling Adaptive Control

The above block diagram represents a typical feedback control system. The gains of controller, valve,
Process and Measurement sensor are 𝐾𝑐 , 𝐾𝑓 , 𝐾𝑝 and 𝐾𝑚 respectively. Let the overall gain of the
process control loop is to be kept as a constant. Overall gain is given by,

𝐾𝑜𝑣𝑒𝑟𝑎𝑙𝑙 = 𝐾𝑐 𝐾𝑝 𝐾𝑚 𝐾𝑓 = 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡

Let the process gain 𝐾𝑝 and measurement sensor gain 𝐾𝑚 are constant. The valve gain 𝐾𝑓 changes
according to the position of valve stem. If the overall gain is to be kept at a constant, the controller
gain 𝐾𝑐 will have to be updated according to the change in valve gain 𝐾𝑓 . The update of the controller
gain is done using the equation as follows:

𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝐾𝐶 =
𝐾𝑝 𝐾𝑚 𝐾𝑓

In order to adjust the controller parameters according to change in valve position, gain scheduling
adaptive control can be used. In this, the valve position is measured and is given as input to the
adaptation mechanism. The controller parameter 𝐾𝑐 is thus updated according to the above equation.
 Example (b) - Combustion System.

In furnaces where air-fuel ratio is

controlled using a Ratio Controller, there is
significance for programmed adaptive
controller. The desired value of air-fuel ratio
that is suitable for combustion is dependent
on temperature of the air. As temperature
changes, the desired air-fuel ratio changes.
Hence, used a sensor, temperature of the sir
(auxiliary variable) is measured and is given
to a Ratio Adjustment mechanism. Thus,
according to the change in temperature, a
new value of desired ratio is updated. The
flow of air is thus controlled, according to
the new value of desired air-fuel ratio.

Method II – Self Adaptive Control

This is used during situations when the controlled output is measured online. The adjustment
mechanism is designed according to the measured value of controlled output. Two examples are
Model Reference Adaptive Control (MRAC) and Self Tuning Regulator.

Example (a) - Model Reference Adaptive Control (MRAC)

In this method, a reference process model (a mathematical model of the process-like a transfer
function), is developed first. If the setpoint is given as input to the reference model, it will give an
ideal model output. This is compared with the actual process output. The difference between model
output and actual process output is the error denoted by 𝑒𝑚 . The objective is to drive the process
output towards actual output, that is the error should be minimized. In MRAC, the Integral Square
Error (ISE) is reduced with the help of an adaptation mechanism. ISE is given by,
The controller parameters are adjusted by the adaptation mechanism in such a way to minimize ISE.
The MRAC consists on two loops. The inner loop is an ordinary feedback loop consisting of process
and controller. The outer loop has the adaptation mechanism, which is used to update the controller
parameters by minimizing ISE.

Example (b) – Self Tuning Regulator (STR)

 Self-tuning Regulator method is an example of self-adaptive controller. There is an inner loop,
which is normal feedback control consisting of the process and the controller. The controlled variable
is measured and given to a parameter estimator. Another input to the parameter estimator is the control
signal from the controller. The parameter estimator calculates the values for controller parameters
using y and m. The calculated values of the controller parameters are given to a parameter adjustment
mechanism. It updates the new values of controller parameters. Both the parameter estimator and
adjustment mechanism involve mathematical computations. Hence Self Tuning Regulator method
can be implemented only using digital computers.