5

Mode Selection and

Tuning of Common
Feedback Loops

The preceding chapters dealt with the tuning of feedback controllers for gen-
eral processes that can be represented by a single-lag-plus-dead–time
(SLPDT) model. This chapter presents tuning guidelines for the most typical
process control loops, specifically flow, level, pressure, temperature, and com-
position control loops.

Learning Objectives—When you have completed this chapter, you should be

able to:

A. Decide on the appropriate control objective for a loop.

B. Select proportional, integral, and derivative modes for specific control

loops.

C. Design and tune simple feedback controllers for flow, level, pressure,
temperature, and composition.

77
78 Tuning of Industrial Control Systems, Third Edition

5-1. Deciding on the Control Objective

Although the most common objective for feedback control is to maintain the
controlled variable at its set point, there are some control situations, often in
the control of level or pressure, when it is acceptable to just maintain the con-
trolled variable within an acceptable range. The difference between these two
objectives is important because, as Chapter 2 showed, the purpose of the inte-
gral mode is to eliminate the offset or steady-state deviation of the process
variable from the set point. Consequently, integral mode is not required when
it is acceptable to allow the controlled variable to vary over a range. One
advantage of eliminating the integral mode is that it permits higher propor-
tional gain, thus reducing the initial deviation of the controlled variable
caused by disturbances.

There are two situations when integral mode is not required:

• When the process is so controllable—a time lag with insignificant dead

time—that the proportional gain can be set high to maintain the con-
trolled variable in a very narrow range.

• When it is desirable to allow the controlled variable to vary over a wide

range so that the control loop attenuates the oscillations caused by
recurring disturbances.

The first of these situations calls for proportional (P) or proportional-deriva-

tive (PD) controllers with very high gains, or for on-off controllers. These may
be found in the control of level in evaporators and reboilers, and in the control
of temperature in refrigeration systems, ovens, constant-temperature baths,
and air conditioning/heating systems. On-off controllers can be used when
the time constant is long enough that the cycling it necessarily causes is of a
very low frequency; otherwise, P or PD controllers are used to modulate the
operation of the manipulated variable. In either case, the dead band of the on-
off controller or the proportional band of the P or PD controllers can be set
very narrow. Derivative mode can be added to compensate for the lag in the
sensor or final control element and thus improve stability.

The second situation calls for proportional controllers with as wide a propor-
tional band as possible. These are found in the control of level in intermediate
storage tanks and condenser accumulators, and in the control of pressure in
Mode Selection and Tuning of Common Feedback Loops 79

gas surge tanks, because in these cases the purpose of the tank is to attenuate
variations in process flow.

5-2. Flow Control

Flow control is the simplest and most common of the feedback control loops.
The schematic diagram of a flow control loop in Figure 5-1 shows that there
are no lags between the control valve that causes the flow to change and the
flow sensor/transmitter (FT) that measures the flow. Since most types of flow
sensors (orifice, venturi, flow tubes, magnetic flowmeters, turbine meters,
coriolis, etc.) respond very fast, the only significant lag in the flow loop is the
control valve actuator, and most actuators have time constants of the order of
a few seconds.

Figure 5-1. Schematic of a Flow Control Loop

SP

FC

FT

Several controller design theories (Internal Model Control2 controller

synthesis1, etc.) suggest that the controller for a very fast loop should contain
only integral mode. In practice, flow controllers have been traditionally PI
controllers tuned with low proportional gains and very short integral times, of
the order of seconds, which are essentially pure integral controllers. Such an
approach is acceptable when flow is controlled to maintain a constant rate,
with rare changes in the flow set point by the operator.
80 Tuning of Industrial Control Systems, Third Edition

However, when the flow controller is the slave in a cascade control scheme
(see Chapter 7), it is important for the flow to respond quickly to set point
changes. This requires a proportional-integral controller with a gain near
unity, which to maintain stability may require an increase in the integral time
from the few seconds normally used in flow controllers. The IMC2 tuning
rules (see Section 4-1) suggest that the integral time be set equal to the time
constant of the loop, usually that of the control valve actuator. In cascade situ-
ations, tight flow control is indicated.

The proportional gain should also be increased when hysteresis of the control
valve causes variations in the flow around its set point. As mentioned in Sec-
tion 4-2, hysteresis is caused by static friction in the valve packing that creates
a difference between the actual valve position and the corresponding control-
ler output. The error changes direction according to the direction in which the
valve stem must move, and this causes a dead band around the desired valve
position; that is, a band within which the valve does not respond to changes in
the controller output. Increasing the flow controller gain reduces the ampli-
tude of the flow variations caused by hysteresis. A valve positioner also
reduces hysteresis and speeds up the valve response, but positioners are usu-
ally difficult to cost-justify for flow control loops.

Example 5-1. Flow Control with Valve Hysteresis

Figure 5-2 shows responses of a flow control loop with valve hysteresis for
two different tunings of the controller. The top curve is for the traditional
tuning of low gain and fast integral (that is, a short integral time), while the
bottom curve is for a more aggressive tuning of a gain of 1.5 and the same
integral time. As the figure shows, the more aggressive tuning reduces the
variations in flow caused by hysteresis in the valve.

5-3. Level and Pressure Control

There are two reasons for controlling level and pressure:

• To keep them constant because of their effect on process or equipment

operation.
Mode Selection and Tuning of Common Feedback Loops 81

Figure 5-2. Oscillations on a Flow Control Loop with Valve Hysteresis are
Reduced in Amplitude with a Higher Controller Gain

Kc = 0.25 TI = 0.2 min

Kc = 1.5 TI = 0.2 min

• To smooth out variations in flow while satisfying the material balance.

Keeping level and pressure constant calls for “tight” control, while smoothing
out variations in control usually calls for “averaging” control. Pressure is to
gas systems what level is to most liquid systems, although liquid pressure is
sometimes controlled.

Tight Control

One example of tight liquid level control and one example of tight pressure
control are shown in Figure 5-3. The control of level in natural-circulation
evaporators and reboilers is important because too low a level causes deposits
on the bare hot tubes and overheating of the tubes at the top. Conversely, too
high a level causes elevation of the boiling point, reducing the heat transfer
rate and preventing the formation of bubbles, which enhances heat transfer by
promoting turbulence. The example of tight pressure control or pressure regu-
82 Tuning of Industrial Control Systems, Third Edition

lation is the control of the pressure in a liquid or gas supply header. It is

important to keep the pressure in the supply header constant to prevent dis-
turbances to the using processes when there is a sudden change in the
demand of one or more of the using processes.

The design of tight level and pressure control systems requires a fast-acting
control valve, with a positioner if necessary, to prevent secondary time lags
that would cause oscillatory behavior at high controller gains. If the level or
pressure controller is cascaded to a flow controller, the latter must be tuned as
tight as possible, as discussed in the preceding section.

Normally, only proportional mode is needed for tight level or pressure con-
trol. The proportional gain must be set high, from 10 to over 100 (proportional
band of 1 to 10% of range). If the lag of the level or pressure sensor is signifi-
cant, derivative mode could be added to compensate for it and to afford a
higher gain. The derivative time should be set approximately equal to the time
constant of the sensor (see the next section). Integral mode should not be used,
since it would require a reduction of the proportional gain.

Many modern controllers cannot be configured to be proportional only or

proportional-derivative. When the integral mode cannot be removed, a long
integral time should be used for tight level control to permit higher propor-
tional gains. The reason the integral mode can be slow is that the high propor-
tional gain keeps the process variable near the set point all the time.

Averaging Level Control

Two examples of averaging level control are shown in Figure 5-4: the control
of level in a surge tank (a) and in a condenser accumulator drum (b). Both the
surge tank and the accumulator drum are intermediate process storage tanks.
The liquid level in these tanks has absolutely no effect on the operation of the
process. It is important to realize that the purpose of an averaging level con-
troller is to smooth out flow variations while keeping the tank from overflow-
ing or running empty. If the level were to be controlled tightly in such a
situation, the outlet flow would vary just as much as the inlet flow(s), and it
would be as if the tank (or accumulator) were not there.

The averaging level controller should be proportional-only with a set point of

50% of range, a gain of 1.0 (proportional band of 100%), and an output bias of
50%. Note that the “bias” is the controller output when the process variable is
Mode Selection and Tuning of Common Feedback Loops 83

Figure 5-3. Examples of Tight Control: (a) Evaporator Level; (b) Supply Header
Pressure

Vapors

Feed
Steam

LC LT T

Condensate
(a)
Product Supply

PC
PT

(b) Loads
84 Tuning of Industrial Control Systems, Third Edition

Figure 5-4. Examples of Averaging Level Control: (a) Surge Tank; (b) Condenser
Accumulator

Feeds

SP
Surge
tank LT LC

Outlet flow
(a)

Vapors

Condenser

Accumulator LT LC
Column

FC

FT Distillate
(b)
Reflux
Mode Selection and Tuning of Common Feedback Loops 85

at the set point, a term that is only important when the controller has no inte-
gral mode. This will cause the outlet valve to be fully open when the level is at
100% of range and fully closed when the level is at 0% of range, thus using the
full capacity of the valve and of the tank. A proportional gain higher than
unity would reduce the effective capacity of the tank for smoothing variations
in flow, while a gain lower than unity would reduce the effective capacity of
the control valve and create the possibility of the tank overflowing or running
dry. With the proposed design the tank behaves as a low-pass filter to flow
variations; a low-pass filter allows low-frequency input through while it
attenuates high frequency variations. The time constant of such a filter is:

A ( h max – h min )
τ = -----------------------------------------
- (5-1)
K c F max

where:

A = the cross-sectional area of the tank, ft2

hmin and hmax = the low and high points of the range of the level
transmitter, respectively, ft

Fmax = the maximum flow through the control valve when opened fully
(100% controller output), ft3/min

Kc = the controller gain

The controller gain is assumed to be 1.0 in this design. When the level control-
ler is cascaded to a flow controller, Fmax is the upper limit of the range of the
flow transmitter in the flow control loop. Note that a proportional gain greater
than unity results in a reduction of the filter time constant and therefore less
smoothing of the variations in flow. A good way to see it is to note that dou-
bling the gain would be equivalent to reducing either the tank area or the
transmitter range by a factor of two, thus reducing the effective capacity of the
tank. On the other hand, reducing the controller gain to half would be equiva-
lent to reducing the capacity of the valve by half, thus increasing the possibil-
ity of the tank overflowing.

Although averaging level control can be accomplished by a simple propor-

tional controller, most level control applications use PI controllers. This is
because control room operators have an aversion to variables that are not at
86 Tuning of Industrial Control Systems, Third Edition

their set points. The process in a level control loop is unlike most other loops
in that it does not self-regulate; that is, the level tends to continuously rise or
fall when the feedback controller is not in Automatic. This usually means that
for level control loops, a time constant cannot be determined by an open-loop
test. Even when there is some degree of self-regulation, the process time con-
stant is very long, on the order of one hour or longer. Because of this, PI con-
trollers in level control loops have the following characteristics:

• The level, and the flow that is manipulated to control it, oscillate for a
long period. Sometimes the period is so long that the oscillation is
imperceptible, unless it is trended over a very long time.

• The shorter the integral time, the shorter the period of oscillation.

• The level control loop is unstable when the integral time is equal to or
shorter than the time constant of the control valve.

• Unlike most other loops, there is a range of controller gains over which
the oscillations increase as the controller gain is decreased.

This leads to the following general rules for tuning PI controllers for averag-
ing level control:

• Set the integral time to 60 minutes or longer.

• Set the proportional gain to at least 1.0.

Averaging pressure control is not as common as averaging level control

because in the case of gas systems, a simple fixed resistance on the outlet of
the surge tank is usually all that is required to smooth out variations in flow.

Intermediate Level Control

There are intermediate situations that do not require a very tight level control
but where it is important not to allow the level to swing over the full range of
the transmitter, as in averaging level control. A typical example is a blending
tank, where the level controls the tank volume, and therefore the residence
time for blending. If a ±5% variation in residence time is acceptable, a propor-
tional controller with a gain of 5 to 10 or even lower could be used, since the
flow would not be expected to vary over the full range of the control valve
capacity.
Mode Selection and Tuning of Common Feedback Loops 87

Example 5-2. Tight and Averaging Level Control

Figure 5-5 shows responses of the control of the level in a tank such as the
one in Figure 5-4(a) with the level controller tuned for averaging and for
tight level control. The inlet flow into the tank increases by increments of
200 gpm as several batch processes dump their contents into the tank. The
tank has a total capacity of 10,000 gallons, while the valve has a flow capac-
ity of 1,000 gpm when fully open.

Figure 5-5 also illustrates that averaging level control (the continuous lines)
averages out the variation of the inlet flow, resulting in a smooth variation
of the outlet flow. On the other hand, tight level control (the dashed lines)
maintains the level nearly constant, but this requires that the outlet flow
essentially follow the variations in the inlet flow, just as if the tank were
not there. In this example the averaging level controller has a gain of 1.0 and
the tight controller has a gain of 20; both have integral times of 20 minutes.

5-4. Temperature Control

Temperature controllers are usually proportional-integral-derivative (PID),

the derivative mode being required to compensate for the lag time of the tem-
perature sensor, which is usually significant. The sensor time constant can
often be estimated by the following formula:

MC
τ s = ------------p- (5-2)
hA

where:

M = the mass of the sensor, including the thermowell, kg

Cp = the average specific heat of the sensor, kJ/kg-°C

h = the film coefficient of heat transfer, kW/m2-°C

A = the area of contact of the thermowell, m2

88 Tuning of Industrial Control Systems, Third Edition

Figure 5-5. Responses of Averaging Level Control (Continuous Lines) and Tight
Level Control (Dashed Lines) on a Surge Tank

Level
Kc = 1

Kc = 20

TI = 20 min

Outlet flow

Kc = 20

Kc = 1

When these units are used, the time constant is calculated in seconds.

Temperature is the variable most often controlled in chemical reactors, fur-

naces, and heat exchangers. When the temperature controller manipulates the
flow of steam (Figure 3-1) or fuel to a heater or furnace (Figure 5-6), the rate of
heat supplied to the process fluid is proportional to the flow of steam or fuel.
This is because the heat of condensation of the steam and the heating value of
the fuel remain approximately constant with load. However, when the manip-
ulated variable is cooling water or hot oil, the rate of heat removed or sup-
plied to the process fluid is very nonlinear with water or oil flow because the
heat transfer rate requires that the outlet utility temperature moves closer to
its inlet temperature as the heat transfer rate increases. This means that it
requires higher increments in flow for equal increments in heat rate as the
load increases. To reduce the nonlinear nature of the loop, the temperature
controller TC is sometimes cascaded to a heat rate controller (QC), as in Figure
5-7. The process variable for the heat rate controller is the rate of heat transfer
Mode Selection and Tuning of Common Feedback Loops 89

in the exchanger, which is proportional to the flow and to the change in tem-
perature of the hot oil:

Q = FoilCp(Toin - Toout)

where:

Q = rate of heat transfer

Foil = flow rate of the hot oil

Cp = specific heat of the hot oil

Toin = inlet temperature of the hot oil

Toout = outlet temperature of the hot oil

This calculation is carried out in the heat rate controller QC in Figure 5-7 to
determine the process variable of the controller. The process outlet tempera-
ture controller TC sets the set point of the heat rate controller QC.

Example 5-3. Estimate of Temperature Sensor Time Constant

Estimate the time constant of an RTD (resistance temperature device)

weighing 0.23 kg and having a specific heat of 0.15 kJ/kg-°C. The thermow-
ell is cylindrical with an outside diameter of 12.5 mm and a length of 125
mm. The film coefficient of heat transfer between the fluid and the ther-
mowell is 0.5 kW/m2-°C.

The area of the thermowell is:

A = πDL = 3.1416(0.0125)(0.125) = 0.0049 m2.

The time constant, from Equation (5-1), is estimated as:

τs = (0.22)(0.15)/(0.5)(0.0049) = 13.5 s (0.22 min).

90 Tuning of Industrial Control Systems, Third Edition

Figure 5-6. Temperature Control of a Process Furnace

SP
Process TC
flow
TT

Air Fuel

Most industrial temperature controllers can usually be tuned by the methods

outlined in Chapters 2, 3, and 4. There are a few exceptions:

• The control of the outlet temperature from reformer furnaces by

manipulation of the fluid flow involves very fast loops similar to flow
control loops. (Reformer furnaces are used to carry out highly endo-
thermic catalytic reactions. They differ from regular furnaces in that the
tubes are packed with catalyst.) The controllers can be tuned as flow
controllers (see Section 5-2).

• The control of laboratory constant temperature baths by manipulation

of power to electric heaters is usually done with on-off controllers or
high-gain proportional controllers.
Mode Selection and Tuning of Common Feedback Loops 91

Figure 5-7. Temperature Control of Process Heater by Manipulation of the Heat

Rate Temperature Control of Process Heater by Manipulation of the Heat Rate

Hot oil

Toin TT SP
TC
Foil FT
SP
QC
Process TT
in

Process
out
Toout TT

5-5. Analyzer Control

The major problem with the control of composition by analyzer is usually

associated with the sensor/transmitter. Sampling of process streams intro-
duces significant dead time into the loop because sensors are often slow, plus
some measurement noise occurs if the sample is not representative due to
poor mixing. In addition, sensor measurements are sensitive to temperature
and other process variables. Analyses of hydrocarbon mixtures are done by
chromatographic separation, which is discontinuous in time; they also have a
time delay of about the same magnitude as the period of the analysis cycle,
compounding the control problem.

In spite of all the sources for time delays in sampling and analysis, since it is
the ratio of the dead time to the process time constant that determines the
uncontrollability of the loop (see Chapter 4), if the combination of the analysis
sampling period and time delay is less than the process time constant, a pro-
portional-integral-derivative (PID) controller is indicated. The tuning strategy
of Chapters 2 and 4 can be used. On the other hand, if the total dead time is on
92 Tuning of Industrial Control Systems, Third Edition

the order of several process time constants, theories such as IMC2 and control-
ler synthesis1 call for a pure integral controller. This is because the process
responds fast relative to the time frame in which the analysis is done. Chapter
6 discusses the tuning of controllers that make use of sampled, rather than
continuous, measurements.

5-6. Summary

This chapter presented some guidelines for selecting and tuning feedback
controllers for several common process variables. While flow control calls for
fast PI controllers with low gains, level and pressure control can be achieved
with simple proportional controllers with high or low gains, depending on
whether the objective is tight control or smoothing of flow disturbances.
When PI controllers are used for level control, the integral time should be
long, on the order of one hour or longer. PID controllers are commonly used
for temperature and analyzer control.

References

1. Martin, J. Jr., Corripio, A. B. and Smith, C. L. “How to Select Controller

Modes and Tuning Parameters from Simple Process Models,” ISA Transac-
tions, V. 15 (Apr. 1976), pp. 314-319.

2. Rivera, D. E., Morari, M. and Skogestad, S. “Internal Model Control, 4. PID

Controller Design,” Industrial and Engineering Chemistry Process Design and
Development, V. 25 (1986), p. 252.

Review Questions

5-1. Briefly state the difference between tight level control and averaging
level control. In which of the two is it important to maintain the level at
the set point? Give an example of each.

5-2. What type of controller is recommended for flow control loops? Indicate
typical values for the gain and integral times.

5-3. What type of controller is indicated for tight level control? Indicate typi-
cal gains for the controller.
Mode Selection and Tuning of Common Feedback Loops 93

5-4. What type of controller is indicated for averaging level control? Indicate
typical gains for the controller.

5-5. When a PI controller is used for averaging level control, what should the
integral time be? Would an increase in gain increase oscillations or
decrease oscillations?

5-6. Estimate the time constant of a temperature sensor weighing 0.03 kg,
with a specific heat of 23 kJ/kg-°C. The thermowell has a contact area of
0.012 m2 and the heat transfer coefficient is 0.6 kW/m2-°C.

5-7. Why are PID controllers commonly used for controlling temperature?

5-8. What is the major problem with the control of composition?