Control technology

Answers for infrastructure.

Contents
1

Introduction to control engineering .......................................................... 6

1.1

Introduction ................................................................................................... 6

1.2

Open-loop control ......................................................................................... 6

1.3

Closed-loop control ....................................................................................... 8

1.3.1

Principle of closed-loop control ..................................................................... 8

1.4

The control loop and its elements ............................................................... 11

1.4.1

Control loop ................................................................................................ 11

1.4.2

Control loop elements ................................................................................. 12

1.5

Controller properties ................................................................................... 16

1.5.1

Direction of control action ........................................................................... 16

1.5.2

Types of reference variables ...................................................................... 16

1.5.2.1

Fixed setpoint control .................................................................................. 16

1.5.2.2

Schedule or program control ...................................................................... 17

1.5.2.3

Slave control ............................................................................................... 17

1.6

Application examples .................................................................................. 18

1.6.1

Heating control ............................................................................................ 18

1.6.1.1

Room temperature control .......................................................................... 18

1.6.1.2

Weather-compensated supply temperature control .................................... 19

1.6.2

Partial air conditioning plant with temperature control ................................ 21

Controlled system .................................................................................... 22

2.1

Introduction ................................................................................................. 22

2.2

Non-self-regulating controlled systems (integral systems) ......................... 22

2.3

Self-regulating controlled systems (proportional systems) ......................... 24

2.3.1

Steady-state response of self-regulating controlled systems ..................... 25

2.3.2

Dynamic response of self-regulating controlled systems ............................ 28

2.4

Evaluating the controlled system ................................................................ 29

2.5

Investigation of various plant types via the step response ......................... 31

2.5.1

System without storage elements (zeroth order system) ............................. 31

2.5.2

System with one storage element (1st order system) .................................. 32

2.5.3

System with dead time (delay element) ...................................................... 34

2.5.4

System with one storage element and dead time (1st order system with
delay element) ............................................................................................ 35

2.5.5

System with multiple storage elements (nth order system) ......................... 36

2.6

Evaluation of the step response.................................................................. 37

2.6.1

Degree of difficulty of a controlled system .................................................. 38

2.6.2

Characteristics of various controlled systems in the HVAC field ................ 39

2.7

Description of the controlled system via the oscillation test ........................ 40

2.8

Description of the controlled system via the frequency response ............... 40

Controllers .................................................................................................42

3.1

Introduction .................................................................................................42

3.2

Classification of controllers .........................................................................43

3.2.1

Classification according to application ........................................................43

3.2.2

Classification according to controlled variable type ....................................43

3.2.3

Classification according to the auxiliary energy ..........................................44

3.2.3.1

Self-actuated controllers .............................................................................44

3.2.3.2

Powered controllers ....................................................................................44

3.2.4

Control action of the effective direction .......................................................45

3.2.5

Classification according to control action of the manipulated variable ........46

3.2.5.1

Continuous-action (modulating) controllers .................................................46

3.2.5.2

Discontinuous-action (step) controllers .......................................................46

3.2.5.3

Quasi-continuous controllers .......................................................................46

3.2.6

Response characteristic ..............................................................................47

3.2.6.1

Differentiation of response characteristics (control characteristics) ............47

3.2.6.2
3.2.6.2.1

Steady-state and dynamic response of the controller .................................47

Steady-state response of the controller ......................................................47

3.2.6.2.2

Dynamic response of the controller .............................................................48

3.3

The P, PI, PID response characteristic .......................................................49

3.3.1

The proportional-action controller (P-controller) ..........................................49

3.3.1.1

Operating principle ......................................................................................49

3.3.1.2

Characteristic quantity and steady-state characteristic of the P-controller 51

3.3.1.3

Dynamic response of the P-controller .........................................................53

3.3.1.4

Shifting the offset ........................................................................................54

3.3.2

The integral-action controller (I-controller) ..................................................55

3.3.2.1

General Example 1 .....................................................................................55

3.3.2.2

Mode of Operation Example 2 ....................................................................56

3.3.2.3

Characteristic quantity of the I-controller .....................................................57

3.3.2.4

Steady-state characteristic of the I-controller ..............................................57

3.3.2.5

Dynamic response of the I-controller ..........................................................58

3.3.3

The proportional-plus-integral controller (PI-controller) ...............................59

3.3.3.1

Operating principle and step response .......................................................59

3.3.3.2

Characteristic quantities of the PI-controller ...............................................60

3.3.4

The derivative-action element (D-element) .................................................62

3.3.4.1

General Example 1 .....................................................................................62

3.3.4.2

Mode of Operation Example 2 ....................................................................63

3.3.4.3

Step response of the D-element .................................................................63

3.3.4.4

Ramp response (ramp-forced response) of the D-element .........................65

3.3.4.5

Application of the D-element .......................................................................65

3.3.5

The proportional-plus-derivative controller (PD-controller) .........................66

3.3.5.1

Operating principle and step response .......................................................66

3.3.5.2

Ramp response of the PD-controller ...........................................................67

3.3.5.3

Characteristic quantities of the PD-controller ..............................................67

3.3.6

The proportional-plus-integral-plus-derivative controller (PID-controller) ....68

3.3.6.1

Operating principle and step response .......................................................68

3.3.6.2

Characteristic quantities of the PID-controller ............................................ 69

3.4

The two-position controller .......................................................................... 70

3.4.1

Operating principle ...................................................................................... 70

3.4.2

Characteristic quantity and steady-state characteristic of the two-position

controller ..................................................................................................... 71

3.4.3

Two-position controller with feedback ......................................................... 73

3.4.3.1

Design and operating principle ................................................................... 73

3.5

The three-position controller ....................................................................... 75

3.5.1

A simple three-position controller................................................................ 75

3.5.2

Three-position controller with quasi-continuous control-action result ......... 76

3.5.2.1

Design ......................................................................................................... 76

3.5.2.2

Operating principle ...................................................................................... 76

Control Loop ............................................................................................. 79

4.1

Loop gain VO ............................................................................................... 79

4.2

Stability investigations ................................................................................ 81

4.2.1

Response to interference ............................................................................ 81

4.2.2

Response to setpoint changes ................................................................... 83

4.3

Setting rules according to Ziegler and Nichols............................................ 84

4.3.1

Known controlled systems .......................................................................... 85

4.3.2

Unknown controlled systems ...................................................................... 85

4.4

Characteristics of different controllers on a controlled system.................... 85

4.4.1

P-controller and controlled system ............................................................. 85

4.4.2

I-controller and controlled system ............................................................... 88

4.4.3

PI-controller and controlled system ............................................................ 89

4.4.4

PID-controller and controlled system .......................................................... 90

4.4.5

Two-position controller and controlled system ............................................ 91

Digital Control ........................................................................................... 92

5.1

Introduction ................................................................................................. 92

5.2

Principle of digital control ............................................................................ 92

5.3

Universal digital controller ........................................................................... 93

5.3.1

Example of a universal digital controller ..................................................... 95

5.4

Digital controller with three-position actuator .............................................. 96

Terms and Symbols .................................................................................. 97

6.1

List of terms ................................................................................................ 97

6.2

List of symbols .......................................................................................... 100

Introduction to control engineering

1.1

Introduction

Closed-loop control and open-loop control are (technical) processes with

considerably different chains of action.

1.2

Open-loop control

Open-loop control is defined as follows according to DIN 19226:

Open-loop control is a process in a system whereby one or more variables, the input
variables, influence other variables, the output variables, according to the specific
regularities of the system. Open-loop control is characterised by an open chain of
action.
The principle of open-loop control is exemplified by the outside/recirculated air mixing
system shown in Fig. 1-1. The outside air temperature is detected by sensor 1 and
transmitted to controller 2. There the signal is converted into a control signal for damper
actuators 3 according to the regularity defined in the controller so that each outside
temperature gives rise to a specific damper position. The temperature of the mixed air 5
is adjusted accordingly; there is no feedback of this temperature to the controller.
Open-loop control always has an open chain of action.

B31-9

2
4

Fig. 1-1
1
2
3
4
5

Recirculated air mixing as an example of open-loop control

Outside temperature sensor

Open-loop control device
Damper actuators
Recirculated air
Mixed air

The control of this system can be presented using a block diagram (Fig. 1-2). It consists
of the open-loop controlling equipment 1 and the controlled system 2.
z
xe

1
y

Fig. 1-2
1
2
xe
xa
y
z

xa
B31-11

Block diagram of an open-loop control system (general)

Controlling equipment (open-loop control function)

Controlled system
Input variable (control variable)
Output variable
Manipulated variable
Interference variable

Difference to closed-loop
control:

The output variable xa of the controlled system has no backward effect on the
controlling equipment (the control result is not monitored).
Further refinement of the block diagram (Fig. 1-2) gives rise to the signal flow diagram
shown in Fig. 1-3.

9
6

2
4

Fig. 1-3

Signal flow diagram of the open-loop control system

10

1
2
3
4
5
6
7
8
9
10
11

11

B31-10

Outside temperature
Controlling equipment
Outside temperature sensor
Open-loop control device
Damper actuators
Controlled system
Air mixing box
Dampers
Outside air
Recirculated air
Mixed air

Possible open-loop control processes in HVAC:

Generation of the reference variable (setpoint): use of the manipulated variable y (or
output variable xa) to control a subsequent closed-loop control system.

Monitoring of ...

Frost protection

Air filters

Fan belts

Window contacts

Room occupancy

etc.

1.3

Closed-loop control

The definition as per DIN 19226 is as follows:

Closed-loop control is a process in which the variable to be controlled (controlled
variable), e.g. air temperature, air humidity etc., is continuously measured, compared
with another variable (reference variable) and influenced according to the result of that
comparison in the sense of adaptation to the reference variable. The resultant chain of
action takes place in a closed circuit, the control loop.
Therefore, the task of a closed-loop control system is to set a physical variable to a
desired value and to keep it at that value in spite of any interference that may occur.

1.3.1

Principle of closed-loop control

The principle and major terms of closed-loop control will first be explained using the
example of the simple manual room temperature control system as shown in Fig. 1-4.
The room temperature of room 1 with radiator 2 is measured using thermometer 3. Let
the desired room temperature be 20 C. If Mr. Control Guy is assigned the task of
keeping the room temperature constant, he will first compare the measured room
temperature with the desired room temperature. He will than make the following
decision according to that comparison:
a)

Measured room temperature = desired room temperature

do not adjust valve 4

b)

Measured room temperature < desired room temperature

open valve 4 further

c)

Measured room temperature > desired room temperature

close valve 4 further

Mr. Control Guy will check the room temperature again at appropriate intervals. One
reason for doing so will be to verify the effect of any valve adjustment he has
performed.

za

C
24
22
20
18
16

zb

zc

w = 20 C

y
4
2

Fig. 1-4
1
2
3
4
w
x
y
za, b, c

Manual room temperature control

Room
Radiator
Thermometer
Manual valves
Desired room temperature (reference variable)
Measured room temperature (controlled variable)
Action on the control element (manipulated variable)
Interference variables

If a radiator valve is properly adjusted, why do room temperature changes that

necessitate further valve adjustments still occur? Three primary reasons for this can be
named in our example:
a) Fluctuating heat dissipation through walls, windows and doors
(mainly dependent on the outside temperature)
b) Fluctuating heating water temperature (no single heat output level at the radiator is
associated with each valve position)
c)

Interference variable(s) z

Changes of the heat energy arising in the room itself due to occupants, machines,
lighting and so on.

Influences that give rise to a deviation from a required value; these are referred to as
interference variables z. Constant interference variables can be compensated from the
start by supplying a greater or lesser quantity of energy (e.g. adjustment of the radiator
to the respective climatic zone or building structure).
However, closed-loop control is always necessary if variations of the interference
variables can occur within the system.
Further closed-loop control terms can derive from Fig. 1-4:

Controlled variable x
(actual value)

The variable to be controlled, in this case the room temperature, is referred to as the
controlled variable x, and its current value is referred to as the actual value xi.

Reference variable w
(setpoint)

The desired value of the controlled variable, which must be maintained in spite of the
variations of the interference variables, i.e. for which the controlled variable should be
compensated, is referred to as the reference variable w. A constant value of reference
variable w is referred to as the setpoint xs.

Control difference e

In order to compare the controlled variable x with the reference variable w, Mr. Control
Guy has to perform a simple subtraction, i.e. determine the difference w x. This
difference is called the control difference.
e=wx
If w x = 0, the measured and desired room temperatures, i.e. the controlled variable
and the reference variable, are the same, so the heating valve does not need to be
adjusted. If w x = +1 K, i.e. the measured room temperature is 1 K too low, the
heating valve must be opened further. If w x = -2 K, i.e. the measured room
temperature is 2 K above the desired value, the heating valve must be closed further.

Manipulated variable y

The signal transmitted to the control element is referred to as the manipulated variable
y.
In Fig. 1-5, Mr. Control Guy has been replaced by controller 1. The conditions already
explained must also be met in the case of automatic control.
The controlled variable x is measured with sensor 2, and the reference variable w must
be input at the controller.

The sequence of the

control process is as
follows:

We will take the steady-state condition as the starting point, where the measured and
desired room temperatures are the same (e.g. 20 C). Heating valve 3 assumes a
position in the middle of its operating range. Due to the change of an interference
variable, the measured room temperature falls. Temperature sensor 2 passes this
changed value to the controller's (1) comparison equipment. The control difference is
now not equal to zero. The controller receives the control difference as an input signal,
amplifies it, adding a certain time variance, then passes it as an actuating signal to the
controller output. This output variable, which is referred to as the manipulated variable
y, adjusts control element 3 in the heating water pipe via an actuator, thus adjusting the
heat output. This results in a new room temperature, which is measured by the sensor
and communicated to the controller. It is a closed circuit; the described process is
continually repeated.

za

zb

zc

T 2

T
B31-2

1
w
Fig. 1-5
1
2
3
w
4
y
za, b, c

A closed-loop control
process must fulfil the
following tasks:

Automatic room temperature control

Closed-loop temperature controller
Temperature sensor
Actuator with valve
Reference variable
Controlled variable
Manipulated variable
Interference variables

1. Measuring

2.
3.
4.
5.

Comparison
Amplification and creation of the dynamic response
Positioning
Measuring (check)

In closed-loop control, the amount by which the controller adjusts the valve in case of a
control difference is highly significant. It is also important how the controller should
intervene in case of a control difference, e.g. quickly or slowly. The way in which a
controller reacts to a control difference is referred to as its dynamic response. This
dynamic response must be generated in the controller and is in many cases adjustable.
The correct setting depends on the system to be controlled and is normally determined
experimentally.
In the control process, it must be ensured that a negative control difference (measured
room temperature less than the desired room temperature) gives rise to an increase in
the heat supply, i.e. the heating valve must be opened further with a decrease in room
temperature. This is referred to by control engineers as reverse action. Assuming that
the direction of control action has been correctly selected in our example, the increased
water flow volume produced by the valve adjustment will change the room temperature
such that, after a certain time, the controlled variable and the setpoint will be equal,
making the control difference zero. At this point in time, the controller has fulfilled its
task, provided no additional interference variable changes occur.
In order to take a closer look at this specific closed-loop control process, we will now
consider the control loop and its elements.

10

1.4

The control loop and its elements

1.4.1

Control loop

The control process acts within a closed circuit, the so-called control loop. Signals pass
through the loop in one direction. A control loop as per Fig. 1-6 is already identifiable in
the simple example of manual room temperature control shown in Fig. 1-4 (bottom
right). The control loop is divided into two parts: the controlling equipment and the
controlled system.
za

C
24
22
20
18
16

zb

zc

w = 20 C

y
4

Fig. 1-6
1
2
w
x
y
z

Control loop (general)

B31-5

Fig. 1-4

Controlling equipment
Controlled system
Reference variable (setpoint)
Controlled variable (actual value)
Manipulated variable
Interference variable(s)

Object

Input

Controlling equipment

Controlled variable x

(example)

(temperature)

Controlled system

Manipulated variable y
(stroke 0..100%)

Output

(example)

Manipulated variable y
(stroke 0..100%)

Controlled variable x
(temperature)

The controlling equipment and controlled system belonging to the control loop can be
further refined with control loop elements. This provides a so-called signal flow diagram
or block diagram of the control loop (Fig. 1-7).
1
2
3

6
9

Fig. 1-7
1
2
3
4
5
6
7
8
9

Control loop elements within the control loop (general)

Control loop
Controlling equipment
Sensor
Controller
Actuator
Controlled system
Control element
Transfer device (e.g. heat exchanger, radiator)
Process (e.g. room)

11

Action of the control loop:

The actuation repeatedly affects the measurement. Therefore, the term closed-loop
control circuit is used.
The control process is continually repeated within the control loop. This is the task of
the controller.
Each control loop element contributes to the control quality, because a control loop is
like a chain: it is only as strong as its weakest link.
The control quality depends not only on the controlling equipment but also to a
considerable degree on the controlled system. The controlled system and the
controlling equipment must be suited to each other.
In actual practice, the finished controlled system is usually a given for the control
engineer, who then has to equip it with appropriate controlling equipment to make a
control loop and then adjust it correctly.

1.4.2
Control loop elements are
function blocks

Control loop elements

As already illustrated in Fig. 1-5, each control loop element can be represented by a
simple box on the signal flow diagram. If each of these boxes is appropriately labelled,
e.g. temperature sensor, controller, control element etc., the signal flow diagram
becomes readily comprehensible. However, the control engineer does not identify the
control loop elements with labels, because the internal relationship, i.e. the interaction
of the input and output variable of each control loop element the so-called response
characteristic, is of greater interest to the engineer. In control engineering, therefore,
each of the boxes is identified with the function (Fig. 1-8) that indicates the respective
response characteristic. As we will see, this can be done in various ways. For more
simple presentation, numerous control loop elements are combined to a single element.
However, complete devices, e.g. controllers, can also be divided up into individual
elements.

Fig. 1-8
e
a

Control loop element shown as a block

Input
Output

Since the signals always flow through the control loop in one direction, a distinction can
always be made between an input side and an output side on any control loop element
that is taken from the control loop at random. In order to illustrate the specification of
the input and output of a control loop element, a selection of control loop elements are
listed from this point of view in Fig. 1-9.
It is also possible for one and the same control loop element to have different output
variables. This depends simply on the purpose for which it is used. If, for example, an
electric motor is used to drive a fan whose speed must be kept constant, the speed of
rotation is the output variable. If the motor is used as a damper actuator, however, the
output variable is the angle of rotation or the stroke. Although an electric motor is used
in both cases, its control behaviour is different if it is used as a damper actuator than if
it is used to drive a fan.
In the case of each control loop element, therefore, careful consideration is required as
to what is the input variable and what is the output variable.

12

Symbol

Output

Control loop
element

Input

Sensor

Temperature

Resistance

Controller

Resistance

Voltage

Actuator

Voltage

Stroke

Control element

Stroke

Transfer device

Flow volume

Heating energy

Room

Heating energy

Temperature

Fig. 1-9

(example)

(010 V)

(0100%)

(example)

(010 V)
(0100%)

Flow rate

Input and output variables of various control loop elements

With the help of control loop elements, therefore, all control processes, regardless of
equipment and plant design, can be represented in a generally applicable manner, i.e.
in the form of a mathematical formula, using only 6 basic elements. These six basic
functions are as follows:

P-element

proportional action element

PT1-element

proportional action element with first order delay

I-element

integral action element

D-element

derivative action element

DT1-element

derivative action element with first order delay

T1-element

lag element

The description of a control loop element includes both its steady-state response and
its dynamic response (see also chapters 2 and 3):
The steady-state response

shows the relationship between the input variable and the output variable in the steadystate condition. The steady-state condition of a control loop element occurs when the
input variable is kept constant until the resultant constant value of the output variable is
achieved.

The dynamic response

shows the relationship between the input variable and the output variable over time.
There are various ways of demonstrating this dynamic response. The most interesting
for the practitioner is the so-called step response. This is obtained by making an abrupt
change to the input variable (step function) and recording the progression of the
resultant output variable change over time.

13

Example
Fig. 1-10 shows a schematic representation of the heating control system from Fig. 1-5.
w
3

e=w-x

-e y

5
1

Fig. 1-10
1
2
3
4
5
6
w
x
e
y
z

Schematic of a heating control system

Temperature sensor
Temperature controller
Actuator
Control element
Radiator
Room
Reference variable
Controlled variable
Control difference
Manipulated variable
Interference variable

The control view of a control loop like this does not consider its external structure but
takes a higher level view. The individual control loop elements can be represented by
simple boxes, e.g. sensor (1), controller (2), actuator (3), control element (4), radiator
(5), and room (6).
This provides a so-called signal flow diagram of the control loop (Fig. 1-11). It clearly
shows the unidirectional signal flow and the closed control loop.

Fig. 1-11
1
2
3
4
5
6

14

Signal flow diagram of a control loop from the control point of view

Temperature sensor
Controller
Actuator
Control element
Radiator
Room

A comparison of Fig. 1-11 with Fig. 1-7 shows that there are differences in the
allocation of the individual control loop elements to the controlling equipment and
controlled system. A distinction must be made between the control view and the
equipment view:

In the control view, the controlled system consists of the following control loop
elements: control element 4, radiator 5, room 6 and sensor 1. The sensor belongs to
the controlled system, because its inertia can have a considerable effect on the
controlled system's dynamic response; the control element belongs to the controlled
system, because it can affect its linearity.
The controlling equipment includes the remaining elements, i.e. controller 1 and
actuator 4. The actuator is included in the controlling equipment, because it has a
contributory effect on behaviour with certain controller designs.

In the equipment view, the controlled system consists of the part of the control loop
that is to be controlled, i.e. the part that is built by the systems contractor. The
controlling equipment is the part of the control loop that is necessary for control, i.e.
the part supplied by the control equipment vendor. These are the sensor, controller,
actuator and control element.

15

1.5

Controller properties

1.5.1

Direction of control action

A controller can be direct or reverse acting. The function of the respective controller in
the control loop is to
keep the controlled variable high

(Fig. 1-12 left:

keep the controlled variable low

(Fig. 1-12 right: direct action)

100 %

100 %

0%

0%

x
w

Fig. 1-12

reverse action)

x
w

Direction of control action in a control loop

In some controllers, the direction of control action is selectable.

Application examples

Reverse action is used for heating or humidify controllers, for example.

Direct action is used for cooling or dehumidify controllers, for example.

1.5.2

Types of reference variables

A distinction is made between the following three possibilities with regard to the
reference variable w:
1.5.2.1

Fixed setpoint control

The reference variable w is set as a fixed setpoint xs at the controller's setpoint selector
(Fig. 1-13), e.g.:

xs = 70 C in the case of supply temperature control

xs = 80 C in the case of boiler temperature control

xs = 55 C in the case of boiler return temperature high limit control

xs = 55 C in the case of hot water tank temperature control

w
y
x
Fig. 1-13

16

Fixed setpoint control

B31-6

1.5.2.2

Schedule or program control

In this type of control, different values of the reference variable w are set by a scheduler
(Fig. 1-14) or a program controller. This can be done continuously or in steps, e.g.:
Room temperature control with a

Daytime setpoint of
and a night-time setpoint of

Closed-loop temperature control of a drying process with various temperature

values.

w1 = 20 C from 06.00 to 22.00 hours

w2 = 16 C from 22.00 to 06.00 hours

w1
y

w2

B31-7

x
Fig. 1-14

1.5.2.3

Schedule control

Slave control

The reference variable w is set by another measured variable which itself cannot be
influenced by the controlling equipment and which varies the value of the reference
variable w continuously according to an adjustable regularity (Fig. 1-15) e.g.:

Weather-compensated supply temperature control where the heating supply

temperature is the controlled variable x and the outside temperature is the reference
variable which varies (compensates) the value of the reference variable w according
to a set heating curve.

w
y
B31-8

x
Fig. 1-15

Slave control

The schedule control and slave control types can also be combined to provide schedule
slave control.

17

1.6

Application examples

1.6.1

Heating control

In the following examples from the domain of heating, closed-loop and open-loop
control will again be compared and the differences indicated.
1.6.1.1

Room temperature control

Fig. 1-16 shows a simple room temperature control system in which, unlike the system
shown in Fig. 1-5, controller 1 acts on a mixing valve (3).
The room temperature is measured by temperature sensor 2 and passed to the
controller as the controlled variable x. The controller compares the reference variable
with the controlled variable and, in case of a control difference, the valve is further
opened or closed depending on polarity until the controlled variable and the reference
variable are equal again. The closed control loop is clearly identifiable, as the
corresponding block diagram also shows.

x
T

2
5

w
x

y
7

z
Fig. 1-16 Room temperature control
1
2
3
4
5
6
7
8

18

Temperature controller
Temperature sensor
Mixing valve
Circulation pump
Radiator
Boiler
Controlled system
Controlling equipment

B31-12

1.6.1.2

Weather-compensated supply temperature control

The most frequently used type of heating control, that is weather or outside
temperature dependent control, includes two open control loops and one closed control
loop (Fig. 1-17).
Open-loop control functions:

Open control loop 1:

The outside temperature sensor signalises its measured value to open-loop control
device 2. This control device converts the signal into the desired supply temperature
setpoint according to an adjustable heating curve, and the setpoint is passed to
closed-loop controller 4 as reference variable w.

Open control loop 2:

The supply temperature is adjusted accordingly by closed-loop controller 4 and
control element 5, and this supply temperature controls the room temperature x2.

Therefore, open control loop 1 is an open-loop, weather-compensated supply

temperature control system, and open control loop 2 is an open-loop, weathercompensated room temperature control system. The open chain of action can be seen
in the block diagram.
Closed-loop control function:
The closed-loop control applies to the supply temperature, which is measured by
sensor 3 and passed to closed-loop controller 4. If closed loop controller 4 detects a
deviation from the setpoint calculated by open-loop control unit 2, it effects an
appropriate change of the manipulated variable y via actuator 5 in order to adjust the
supply temperature. Supply temperature sensor 3 signalises the changed supply
temperature value to closed-loop controller 4, thus closing the control loop (see block
diagram). This is weather-compensated supply temperature control, also referred to as
slave control, because the supply temperature is based on weather conditions.
The external heat gain from occupants, equipment and lighting in the room as well as
solar radiation through the window acts as interference variable z1.
The fluctuating boiler water temperature, for example, acts as interference variable z2.

19

z1
1
x2

w
2

x1

z2

B31-13

1
x2
y
d

x2

z1
A

Fig. 1-17
1
2
3
4
5
a
b
c
b+c
d
A
B
C
w
x1
x2
y
z1
z2

20

z2
B

Weather-compensated supply temperature control

Outside temperature sensor

Heating curve setting
Supply temperature sensor
Supply temperature controller
Control valve (actuator + control element)
Open-loop control equipment 1
Closed-loop control equipment (supply)
Closed-loop controlled system (supply)
Open-loop control equipment 2 (room)
Open-loop controlled system 2 (room)
Open control loop 1
Closed-loop control
Open control loop 2
Supply temperature setpoint (open-loop control variable 1)
Supply temperature actual value (closed-loop controlled variable)
Room temperature actual value (open-loop control variable 2)
Manipulated variable
Interference variable 1 (external heat gain)
Interference variable 2 (fluctuating boiler water temperature)

1.6.2

Partial air conditioning plant with temperature control

The system shown in Fig. 1-18 is equipped with an air heater and an air cooler. This
allows the room temperature to be maintained within the desired range (wH..wK) not
only in winter but also in summer and in case of internal heat gain.
5
2

4
1

y1
y2
Fig. 1-18
1
2
3
4
5

wH
wK

B31-18

Ventilation system with heating coil and cooling coil

Temperature sequence controller (1 heating sequence + 1 cooling sequence)

Room temperature sensor
Heating valve
Cooling valve
Frost protection thermostat

The purpose of this closed-loop controlling equipment is to keep the room temperature
within the dead band xdz between setpoints wH and wK.
Temperature controller 1 compares the room temperature measured by sensor 2
against the selected setpoints wH and wK. In case of a deviation (Fig. 1-19), the
controller adjusts heating valve 3 or cooling valve 4 until the room temperature is within
the dead band xdz.
The purpose of separating cooling and heating operation with the so-called dead band
xdz is to save cooling energy. This means that the controller operates to the lower
heating setpoint wH during heating operation and to the higher cooling setpoint wK
during cooling operation.

y
Xdz
100 %
3

0%
wH
Fig. 1-19
3
4
x
wH
wK
xdz

wK

B31-19

Heating valve cooling valve sequential control

Heating valve
Cooling valve
Room temperature
Heating setpoint
Cooling setpoint
Dead band (e.g. 4 K)

The purpose of this controlling equipment is that no energy must be supplied between
the setpoints wH and wK. The room occupants are comfortable between the setpoints.

21

Controlled system

2.1

Introduction

The controlled system is the part of the control loop that is controlled. From the control
point of view, the controlled system begins at the manipulation point, i.e. where the
manipulated variable acts on the mass or energy flow, and ends at the measuring point,
i.e. where the sensor for measuring the controlled variable is located. The controlled
system is usually a given for the control engineer. However, the control engineer has
an interest in statements about the controlled system, because the adjustment of the
controlling equipment and the effort involved to achieve a satisfactory control result
largely depend on these.
The controlled system is a chain of 1..n control loop elements whose input variable is
the manipulated variable y and whose output variable is the controlled variable x (Fig.
2-1).

1
y

Fig. 2-1
1
2
x
y

B32-1

Block diagram of a controlled system

Control element
Controlled system
Controlled variable (output)
Manipulated variable (input)

A distinction is made between 2 types of controlled system:

non-self-regulating systems, also referred to as astatic or integral systems

(I-systems)

self-regulating systems, also referred to as static or proportional systems

(P-systems)

2.2

Non-self-regulating controlled systems

(integral systems)

The water level controlled system shown in Fig. 2-2 provides an illustrative example of
a non-self-regulating system. The input variable in this case is the water inflow v& in l/h,
and the output variable is the water level h in the container in m (overflow at 4 m).
The output variable remains constant at 1 m if the inflow and outflow are equal.
If at time T0 the inflowing water volume is increased, for example, from 400 I/h to 500
I/h (more water is flowing in than out), the water level will rise in a linear fashion until
the container overflows after a certain time. No self-regulation, i.e. no new steady-state
condition, occurs (apart from the natural limitation when the container overflows)
Assuming that an excess of inflowing water of 50 l/h raises the water level by 0.1 m per
hour, the hourly rise in water level at an excess of 100 l/h will be:
0.1m/h 100 l/h
50 l/h

0.2 m/h

and the time until the container overflows will be:

3m
0.2 m/h

22

15 h

If the inflowing water volume is increased from 400 l/h to 600 l/h, i.e. an increase of 200
l/h, the container will fill twice as fast (dashed lines).
In the case of an abrupt input variable change v& in a non-self-regulating controlled
system, the output variable h moves away from the previous value without taking on a
new steady-state condition. The rate of change of the output variable is proportional to
the change of the input variable. This behaviour is referred to as an integral system.
h

400 l/h

4m
3
2
1
0

400 l/h

l/h
600
500
400

t
T0

10

15

20 h

10

15

20 h

h
m
4
3
2
1

Fig. 2-2
v
h
t
T0

B32-2

Water level controlled system as an example of a non-self-regulating system

Flow volume (input)

Water level (output)
Time
Start time

Since non-self-regulating controlled systems practically never occur in heating,

ventilation and air-conditioning, this topic will not be covered in any greater detail.

23

2.3

Self-regulating controlled systems

(proportional systems)

In the case of a self-regulating controlled system, the output variable x always tends
towards a new steady-state condition following a change in the input variable y.
So a balanced state occurs again (self regulation). All controlled systems for the
controlled variables temperature, pressure, differential pressure, flow volume and
humidity occurring in heating, ventilation and air-conditioning are self-regulating
systems (P-systems).
An example of a self-regulating controlled system is a room that is heated by an
adjustable electric radiator (Fig. 2-3). If the heat output is increased from level 1 to level
3 by turning the switch, the room temperature does not keep rising ad infinitum; instead
it takes on a new, higher value (x3) after a certain time (the progression of the
temperature rise over time is not currently of interest to us, so it is only indicated by a
dotted line).
If the heat output is only increased by one level, the room temperature rises by a
correspondingly lesser degree (x2).
y
3

2
1
0

22

T0

20
18

x3
x2

x1

16

Fig. 2-3
x
x1
x2
x3
x
y
y
T0

Electrically heated room as an example of a self-regulating controlled system

Controlled variable (output)

initial steady-state condition
new steady-state condition for x = 1 level
new steady-state condition for x = 2 levels
Change of controlled variable
Manipulated variable (input)
Change of manipulated variable
Start time

The graphical presentation of the input variable (manipulated variable y) and the output
variable (controlled variable x) shows that the controlled variable changes in a given
proportion to the manipulated variable, i.e. the change of output variable x is
proportional to the change of input variable y in this case. Therefore, self-regulating
control systems (with regard to the steady-state condition) are also referred to as
proportional controlled systems.
The steady-state response and dynamic response are highly significant for the choice
of controller type with regard to adaptation to the controlled system and stability in the
control loop (topics which will be covered in later chapters).

24

2.3.1

Steady-state response of self-regulating controlled

systems

The steady-state response of a self-regulating controlled system refers to the

relationship between the output variable (controlled variable x) and the input variable
(manipulated variable y) in steady-state condition.
This relationship was illustrated for an electric room heating system in Fig. 2-3. A
change in the manipulated variable y (step switch from position 1 to position 3) gave
rise to a given change in the controlled variable x (room temperature rise from 18 C to
22 C, x = 4 K).
The transfer coefficient KS

It is important for the control engineer to know how large the change of controlled
variable x is in proportion to the change made to the manipulated variable y in the
steady-state condition. This proportion, which is referred to as the transfer coefficient
KS of the controlled system, can be calculated as follows:
Transfer coefficient K s =

x
y

output var iable

input var iable

Therefore, the transfer coefficient KS indicates by how many units the controlled
variable x changes if the manipulated variable y is changed by one unit (after the
steady-state condition is reached).
The transfer coefficient of the electrical room heating system shown in Fig. 2-3,
therefore, is:
Ks =

x
4K
=
= 2 K per level
y 2 levels

Moving the step switch of the heater from position 0 (off) directly to the highest level,
position 3, gives rise to the maximum possible room temperature change under the
current ambient conditions (outside temperature, external heat gain). In control terms,
this is the:
Adjustment range

The change of the manipulated variable y from 0 to 100%, i.e. through the adjustment
range Yh of the manipulated variable, gives rise to a change of the controlled variable x
through the controlled variable's adjustment range Xh, also referred to as the control
range Xh (Fig. 2-4).
y
1
Yh

0
T0

x
C
22
20

Xh

18
16

Fig. 2-4

T0

t
B32-4

Room temperature change in the system as per Fig. 2-3 with a unit step from 0...1

Yh Adjustment range of the manipulated variable

Xh Adjustment range of the controlled variable, control range
T0 Start time

25

If the value of y as the unit step from 0...1 (= Yh) is entered in the formula KS = x/y,
the result is a value of x that corresponds to the control range Xh. Therefore:
Ks =
Control range

x Xh 6 K
=
=
= 6 K = Xh
Yh
1
y

The control range Xh vividly illustrates the effect that the controller can have within the
controlled system if the control element travels through the entire adjustment range Yh.
The control range Xh of the controlled system is an important criterion for the
controllability of a plant. It must never be too large but also not too small.

The plant should not be

oversized

If the control range is too large, the control loop tends to oscillate. If it is too small, a
good control quality is possible in the partial load range, but the setpoint can no longer
be achieved if the maximum interference variable (load) occurs.
In actual practice, plant parts (heat exchangers, radiators, air heaters, control elements,
burners, pumps, fans etc.) tend to be oversized (allowances due to uncertainty).
Therefore, the control range tends to be on the large side, which gives rise to a
deterioration of the controllability and efficiency of the plant.

Control characteristic

As can be seen from Fig. 2-3, in self-regulating controlled systems there is a specific
value of the controlled variable x for every value of the manipulated variable y. This
relationship can also be presented graphically by joining all of the respective points on
a chart. This gives rise to the steady-state characteristic of the controlled system,
which is also referred to as the control characteristic (Fig. 2-5).
As an example, let us imagine the step switch from Fig. 2-3 not with 3 but with an
infinite number of positions (infinitely variable). Each position of the control knob gives
rise to a corresponding room temperature. Therefore, the progression of the control
characteristic is linear (control characteristic b) in this case. The slope of the control
characteristic x/y = tg corresponds to the respective transfer coefficient.
x
b)

C
22

a)

20
x

tg
18

16

y
0
0

Fig. 2-5
a)
b)

26

2
50

3
100

[Stufe]
[%]

B32-5

Control characteristic of the electric room heater as per Fig. 2-3

3-position control
Infinitely variable control

Unfortunately, a linear control characteristic is rare in practice. As an example, Fig. 2-6

shows the steady-state characteristic of a room heated by a radiator. In this case, the
input variable is the valve stroke y, which varies the flow of hot water at constant inlet
temperature in an assumed linear proportion; the output variable is the room
temperature x. The control characteristic of this controlled system is obtained by
plotting the relationship between y and x on a chart.

x
y

x
C
24

x2

22
20
18

x1

16

14
0

20 40 60 80 100 %

y1

y2
B32-6

Fig. 2-6

Control characteristic of a radiator room heating system

As can be seen, this control characteristic is not linear, i.e. a valve stroke change y1 in
the range 0...20% gives rise to a change in room temperature x1 from 14 C to 19 C,
i.e. of 5 K, whereas a valve stroke change y2 in the range 40...60% only produces a
room temperature change x2 of 1.4 K. The transfer coefficient KS of the controlled
system is no longer constant in this case; instead it is relatively high in the lower stroke
range and decreases to an increasingly great degree in the upper range.

27

The more non-linear the control characteristic is the more difficult it becomes for the
controller to quickly and stably compensate for interference under all load conditions. It
is attempted, therefore, to linearise the control characteristic, e.g. with a corresponding
opposite non-linearity of the control valve, so that a valve stroke of 50% will actually
produce an output as close as possible to 50%. Fig. 2-7 shows the appropriate
functional principle. This topic is covered in greater detail in the Hydraulics training.
V

100 %

100 %

100 %

50 %

50 %

25 %

Y
50 %

Fig. 2-7

1
2
3

V
25 %

Y
50 %

B32-7

Linearisation of a deformed heat exchanger characteristic via opposite non-linearity

of the valve characteristic

Valve characteristic (relationship between valve stroke and fluid flow)

Heat exchanger characteristic (relationship between fluid flow and output)
Resultant control characteristic (relationship between valve stroke and output)

However, the steady-state view of the controlled system is insufficient to describe it

completely. A control process consists of time-dependent functions, so the temporal
relationship between the change of the output and input variables, i.e. the so-called
dynamic response of the controlled system, is equally important.

2.3.2
Storage properties

Dynamic response of self-regulating controlled systems

As already seen in Fig. 2-3, the output variable of a controlled system that corresponds
to a change in the input variable is generally not achieved until a certain time has
passed. This delay can be caused by the flow or travel time of the energy medium on
the one hand and by the storage properties of the control loop elements, to which the
controlled system belongs, on the other. In this case, the term storage properties refers
to energy storage in the form of the thermal response of the plant parts themselves
(e.g. mass of the pipes, valves, containers etc.) and not that of the energy medium they
contain (e.g. volume of a hot water tank).
Examples of energy storage:

Heating of a hotplate on a stove by electricity

Heating of the water in a boiler

Heating of a room by a radiator

Filling of a pressure vessel with air

The most common controlled systems in practice exhibit the behaviour of 2 or more
storage elements (e.g. water heats an air heater, the air heater heats the air flowing
through it, the heated air heats a room).
Therefore, controlled systems can be classified as follows according to the number of
storage elements they contain:

28

Controlled systems without storage elements

Controlled systems with one storage element: single storage element systems

Controlled systems with two or more storage elements: multiple storage element
systems

Ordinal number

Degree of difficulty

Another classification is based on the ordinal number, where the order is the same as
the number of storage elements present:

System without storage elements: zeroth order system = self-regulating system (Psystem) without storage elements (T0) = PT0-system

System with one storage element: first order system = self-regulating system (Psystem) with one storage element (T1) = PT1-system

System with n storage elements: nth order system (higher order) = self-regulating
system (P-system) with multiple storage elements (Tn) = PTn-system

All of these systems can naturally also have dead times Tt , in which case a
controlled system may be referred to, for example, as a PT1Tt-system.

Alongside the steady-state response, the order of a controlled system (number of

storage elements) characterises to a large degree the difficulty of a control task.
It can be generally said that higher order controlled systems are more difficult to control
than lower order systems. In other words: The controllability of a system decreases with
an increasing number of storage elements.
We will return to this relationship in chapter 2.6.1.

Temporal behaviour

The dynamic response of a controlled system refers to the relationship between the
change in the input variable x and the change in the output variable y over time, i.e.
to its temporal behaviour.

2.4

Evaluating the controlled system

There are several ways of describing the dynamic response of a controlled system,
e.g.:

Description via the step response

Description by the Ziegler-Nichols method

Description via the frequency response (harmonic response)

The description of a controlled system via the step response is the simplest method for
the practitioner and the most common. It provides a simple tool for evaluating a
controlled system.
In recording the step response, a distinction is made between the manipulated variable
step response and the interference variable step response.

The manipulated variable step response provides information on the response to

setpoint changes of the controlled system.

The interference variable step response provides information on the response to

interference of the controlled system, i.e. how the controlled system reacts to a
sudden ingress of external heat energy, for example.

The type of step response used in practice depends on the application case and on the
possibilities offered by the plant.
Recording of the manipulated variable step response is considerably more common in
practice, which is why only this step response will be described here.

29

The step response (Fig. 2-8) of a controlled system is obtained by stepping up input
variable 1 (manipulated variable y) by a random value and plotting the progression of
output variable 2 (controlled variable x) over time. Any interference variables z must be
kept as constant as possible during the experiment. The curve that is obtained in this
way shows the transient behaviour of the controlled system and is referred to as the
step response. The step of the input variable is referred to as a step function.
y

y2

y1
x
x

T0
2

x2
x

x1
T0
Fig. 2-8

t
B32-16

Example of the step response of a room air heating system

1 Input variable
2 Output variable
T0 Start time

If the step response refers to an input variable change from 0...100% (0...1), the
transient function of the controlled system is obtained. This shows not only the
system's dynamic response but also the magnitude of the control range Xh. In the case
of a linear control characteristic, the transfer coefficient Ks can be calculated from the
control range Xh.
In the following examples, the step responses and transient functions of various
controlled systems will be considered, and an example for the evaluation of the step
response is explained in chapter 2.6.

30

2.5

Investigation of various plant types via the

step response

2.5.1

System without storage elements (zeroth order system)

As an example for a controlled system of this kind, we will consider a flow control
system for flow volume (Fig. 2-9). In this case, the control element is a manual valve 1
with a linear flow characteristic. The flow volume is indicated on indicating instrument 3
using flow measuring device 2.
If at time T0 the opening of the manual valve at the system input is abruptly increased
by y = 4 mm = 40%, for example, (step function of the manipulated variable y), an
abrupt increase in fluid flow occurs simultaneously at the system output (step response
of the controlled variable x), which in this case will be assumed to be x = v = 2 m3/h.
Therefore, the controlled variable change x follows the manipulated variable change
y without delay.
The controlled system operates with the following transfer coefficient:
x
2 m3 / h
m3 / h
=
= 0 .5
y
4 mm
mm

Ks =

2 m3 / h
m3 / h
= 0 . 05
40 %
%

or

y [m/h]

10
8

4
2

0
0

T0

x [m/h]
5
4
3

2
1

0
0

Fig. 2-9

T0

B32-8

Example of the transfer function of the flow volume in a system with no storage elements

1
2
3

Manual valve
Flow measuring device
Flow indication
y Change of manipulated variable
x Change of controlled variable
T0 Start time

31

If you are not only interested in the transfer coefficient KS of the controlled system, i.e.
the controlled variable change per millimetre or percent of the manipulated variable
change (KS), but also in the resultant controlled variable change if the control element is
adjusted over the entire adjustment range Yh (y = 100%) the control range Xh of the
system is obtained; in this case it is:
Xh = K s Yh = 0.05 m 3 / h % 100 % = 5 m3 / h

In the case of a valve with a linear flow characteristic, KS can be assumed to be

constant over the entire adjustment range.
Controlled systems with no storage elements are quite rare in heating, ventilation and
air-conditioning.

2.5.2

System with one storage element (1st order system)

A controlled system with one storage element is illustrated in Fig. 2-10, which shows an
electrically heated hot water tank. The agitator ensures that the water temperature is
evenly distributed throughout the container. The only significant storage element is the
water. The storage effects of the immersion heater and of the thermometer are so small
that they will be neglected here. The input variable of the controlled system is the
position y of electrical switch 1, and the output variable is the water temperature x,
which is measured with thermometer 4. If the switch is open (manipulated variable y1),
let the water temperature x1 be assumed to be 20 C.
4

1
y = Yh
0

T0

C
50

X2

40

x = Xh

30

X1

20
T0

Ts

t
B32-9

5 Ts

Fig. 2-10

Example and transient function of a system with one storage element

1
2
3
4

Electrical switch
Immersion heater
Agitator
Temperature indication (thermometer)
y Step function
x Step response

If the switch is closed at time T0, the temperature of the water in the container begins to
rise with practically no delay. The temperature change is initially relatively fast (initial
rate of change) and then becomes increasingly slower until the final temperature of e.g.
50 C is achieved. Apart from insulation losses, the heat-up time (storage tank charging
time) in this example depends entirely on the quantity of water (container size):

32

Small volume:

short heat-up time

Large volume:

long heat-up time

Regardless of whether heat, pressure of electricity (capacitor) is being stored, an

abrupt change in the input variable always gives rise both physically and
mathematically to the same charging curve for a storage element a so-called
exponential function (e-function).
This step response (Fig. 2-11) is characterised by the time constant T which is
determined by the storage capacity and the transfer coefficient KS.
The time constant of a controlled system is referred to as TS.

y
%

2 100
63,2

0
Ts1

Ts2

Ts3
5 * Ts

Fig. 2-11

Ts4 Ts5
B32-10

Relationship between a system with single storage element characteristic and the time constant

v Initial rate of change

TS Time constant

The time constant TS is the time that the output variable would require to change by the
amount x corresponding to the input variable change y if the initial rate of change
were maintained. The initial rate of change is obtained by plotting a tangent to the start
of the step response.
In practice:

time constant TS is indicated as the time required for roughly 2/3 of the output
variable change x to be achieved (exactly: 63.2%)

after 5 time constants, roughly 100% of the output variable change is achieved
(exactly: 99.3%)

33

2.5.3

System with dead time (delay element)

Fig. 2-12 shows an example of a controlled system with dead time. It is a mixed water
system with a manual mixing valve 1, a pipe 2 and a temperature sensor 3 installed
some distance away (the storage capacity of the pipe is neglected). The input variable
(manipulated variable y) is the position of the manual mixing valve, and the output
variable (controlled variable x) is the measured mixed water temperature.
y
%
1
60

60 C

v
y

AB

A
B

10 C
Tt =

l
v

t0

x
[ C]
40

10

t0
Tt

Fig. 2-12
1
2
3

t
B32-11

Controlled system with dead time (mixed water temperature control system)

Manual mixing valve

Pipe
Temperature sensor

If at time T0 control port A of the mixing valve at the system input is abruptly opened
from 0% to 60% (at the same time closing bypass port B from 100% open to 40%
closed), the result is an abrupt rise in the water temperature at the mixing valve outlet
(constant port AB) from 10 C to 40 C (as per mixing equation).
Since, however, the measuring point with temperature sensor 3 at the controlled
system output is some distance away from the manipulation point, the water with the
changed temperature must first flow as far as the temperature sensor before the sensor
can detect a temperature change. This travel time is referred to as the dead time Tt.
For the sake of simplification, it can be considered as the dead time Tt of the controlled
system. The dead time Tt is the time that elapses before the manipulated variable
change y executed at the manipulation point starts to act as a controlled variable
change x at the measuring point, i.e. until the output variable can react to a change in
the input variable. Therefore, the dead time is the travel, running or flow time of the
water, steam, air etc. and as such is dependent on the distance between the
manipulation point and the measuring point.
Dead time has a disadvantageous effect on a controlled system, because it prevents
quick correction of control differences. Therefore, it should be kept as small as possible
(e.g. install valve close to heat exchanger).

34

2.5.4

System with one storage element and dead time

(1st order system with delay element)

The step response of the controlled variable as shown in Fig. 2-12 rarely occurs in
practice. The mass of the plant parts (valve, pipe) and sensor stores heat energy,
which gives rise to a slower transient behaviour of the controlled system.
If at time T0 the mixing valve is abruptly opened from 0% to 60%, initially nothing
happens at the measuring point. When the dead time has elapsed (which includes the
dead time of the temperature sensor) the controlled variable x changes in the same
manner as in a single storage element system, i.e. according to an e-function (Fig. 213). The time constant TS is determined by the storage capacity of the controlled
system.
The step response is characterised by the dead time Tt, the time constant TS and
the transfer coefficient KS.
y
%
1
60

60 C

v
y

AB

A
B

10 C
Tt =

l
v

t0

x
[ C]
40

10

t0
Tt TS

Fig. 2-13

1
2
3

B32-17

Controlled system with one storage element and dead time (mixed water temperature control
system)

Manual mixing valve

Pipe
Slow temperature sensor

Genuine single storage element systems and single storage element systems with
dead time are quite rare in heating, ventilation and air-conditioning. Normally multiple
storage element systems are involved.

35

2.5.5
Multiple storage element
system

System with multiple storage elements

(nth order system)

Controlled systems with two or more storage elements are normally encountered in
practice. Examples of such controlled systems are:

Heating system with valve, hot water supply, radiator, room

Ventilation system with valve, heating register, supply air duct, room

In order to consider the dynamic response of a multiple storage element system we will
take an example from the heating field (Fig. 2-14). Four significant storage elements
are identifiable here: the boiler, the pipe, the radiator and the room.
The oil or gas burner of the boiler acts as the input variable (manipulated variable y) of
the controlled system. The output variable (controlled variable x) of the system is the
room temperature.
If the burner (manipulated variable y in the diagram) is started at time T0, the room
temperature no longer changes according to an e-function but according to a higher
order function. It is evident from this that the step responses of multiple storage
element systems have the following common features:

They initially rise gently from the time axis.

They then climb with an increasing slope to an inflexion point P where the rate of
change is greatest.

After the inflexion point, the rate of rise decreases again, i.e. the curve tends
towards the steady-state condition asymptotically.

In principle, the step response is gentler the greater the number of storage elements in
the system.
y
[%]
100

3
2

y = Yh

x
0

[C]

40

T0

x
X2

y
P

10

X1
T0

Fig. 2-14
1
2
3
x1, x2
P

x = Xh

[min]

Model and transient function of a multiple storage element system

Boiler
Radiator
Room temperature sensor
Steady-state conditions
Inflexion point

In multiple storage element systems, the greatest rate of change of the controlled
variable is found at the inflexion point P (change of direction) of the S-shaped step
response.

36

2.6

Evaluation of the step response

We have already learned that the transient behaviour (step response) of controlled
systems with one storage element, which occur relatively rarely in practice, is
determined by the dead time Tt and the time constant TS.
Inflectional tangent

In multiple storage element systems, the transient behaviour is characterised by the

delay time Tu and the balancing time Tg (Fig. 2-15).
In order to determine Tu and Tg, a tangent is plotted at the inflexion point P (change of
direction) of the step response. This inflectional tangent intersects the two steady-state
conditions x1 and x2 (initial and final value of the controlled variable). If we mark the
lower point of intersection T1 and the upper point of intersection T2 on the time axis,
then:

time difference T0 T1:

is the delay time Tu of the controlled system

time difference T1 T2:

is the balancing time Tg of the controlled system

The delay time Tu also includes any existing dead time Tt.
y
[%]
100
y =Y h
0

T0

x
[C]
40

10
T0

T1
Tu

Fig. 2-15

T2
Tg

B32-13

Determining the delay time TU and balancing time Tg of a multiple storage element system

P Inflexion point
Tu Delay time
Tg Balancing time

Fig. 2-16 shows the step responses of controlled systems containing up to n = 6

identical storage elements (same time constant TS). It can be seen that the more
storage elements that are connected in series the less steep the step response
becomes.

37

Yh

0
Tg 6

Tg 5
Tg 4
Tg 3
Tg 2
Ts

1'

2'

3'

4'

5'

6'

Xh

n=0

n=1

Tu
2

n=2

n=3

n=4

n=5

n=6

B32-14

Tu
3
Tu
4
Tu
5
Tu
6

Fig. 2-16
n

Number of storage elements

2.6.1
Degree of difficulty S

Transient functions of controlled systems with 0...6 identical storage elements

Degree of difficulty of a controlled system

The controllability of a controlled system is of particular interest to the control engineer.

This controllability is expressed as the degree of difficulty S and is calculated from the
relationship between dead time Tt and time constant TS or delay time Tu and balancing
time Tg:
Degree of difficulty S =

Tt
Ts

or

Tu
Tg

The greater the degree of difficulty the more difficult a system is to control. It depends
on the number of storage elements connected in series, their time constants and the
magnitude of any dead times. The greater the proportion of the delay time to the
balancing time the later the controlled variable will react to a change in the manipulated
variable; on the other hand, the reaction may already be too great. A system is easiest
to control if Tu = 0. The ratio of Tu/Tg is then also 0, which means that the controlled
system consists of a single storage element. In this case, a change in the manipulated
variable gives rise to an immediate reaction of the controlled variable.
The degree of difficulty S also gives an indication of the type of controller to be used
and assists in the selection of controller setting parameters.
With regard to the controllability of a system, the following values of S mean:

38

S < 0.1:

easy controlled system

S = 0.1...0.3:

medium difficulty controlled system

S > 0.3:

difficult controlled system

In order to provide a rough definition of controlled systems, a distinction is made with

regard to:

Time constant or balancing time:

fast vs. slow systems

Degree of difficulty:
easy - medium difficulty - difficult systems

2.6.2

Characteristics of various controlled systems in the

HVAC field
Controlled variable

Characteristics
Tu

S (=Tu/Tg)

Xh

Boiler water temperature

15 min

0.050.15

20 K/E

Flue draught

13 s

0.31

310 mbar

Mixed water temperature

520 s

0.20.5

2070 K

1. Heating

(supply and return side mixing,

including sensor)
Room temperature:
Living area

35 min

0.10.3

610 K

Greenhouse/tent

35 min

0.10.3

2030 K

Mixing

0.52 s

0.10.4

3060 K

Heat exchanger

530 s

0.10.8

3060 K

Supply

1060 s

0.10.3

28 K

Return

68 h
1060 s

0.10.6

10100 K

Mixing temperature (air)

1s

0.30.8

040 K

Supply air temperature

15 s4 min

0.20.6

2050 K

Room temperature

0.53 min

0.10.3

610 K

Exhaust air temp.

0.53 min

0.20.4

610 K

Washer outlet temperature

10 s2 min

0.10.3

2025 K

Room humidity incl. sensor

0.53 min

0.20.5

2 0 . . . 70% r.h.

Flow volume (air)

<1s

0.10.5

Hot water temperature

Swimming pool water temp.

Heat exchanger
2. Air-conditioning

(depending on plant)

(depending on plant)

Overview of the characteristics of various controlled systems

In Fig. 2-17 the above table has been converted to a chart showing the dependency of
S and Tg of the most common controlled systems in the HVAC field. It can be seen that:

controlled systems in the heating field tend to have lower degrees of difficulty and
greater balancing times

controlled systems in the ventilation field tend to have higher degrees of difficulty
and lesser balancing times

39

1
2

Tg
Fig. 2-17
S
Tg
1
2

Most common controlled systems in heating, ventilation and air-conditioning

Degree of difficulty
Balancing time
Controlled systems mainly in the ventilation and air-conditioning field
Controlled systems mainly in the heating field

2.7

Description of the controlled system via the

oscillation test

The description of a controlled system via the oscillation test (Ziegler-Nichols method)
is performed in a closed control loop. By increasing the controller gain Kp and varying
the setpoint, the control loop is set to steady-state periodic oscillation T0 critical, i.e. it is
taken to its stability limit.
The time factor T0 critical of this oscillation is typical for the respective plant (controlled
system) and provides a measure for the setting parameters of the controller according
to Ziegler and Nichols.
Typical setting values are explained in the control loop chapter.

2.8

Description of the controlled system via the

frequency response

In the description of a controlled system via the frequency response, sinusoidal

oscillation of constant amplitude and variable frequency is used instead of the abrupt
input variable change. If the input oscillation is not too fast for the respective controlled
system, the output variable, i.e. controlled variable x, will oscillate in the same rhythm.
This oscillation of the controlled variable represents the harmonic response.
If the frequency of the input oscillation is modified, i.e. made faster or slower, the same
frequency will occur at the controlled system's output after a certain delay. At the same
time, however, the amplitude of the output oscillation and its phase angle with respect
to the input oscillation will also change.
These two characteristics, amplitude and phase angle, are now plotted over the
frequency ...

40

either on a common locus diagram

or separately on a special diagram, the Bode diagram. This gives rise to two
characteristic curves. One is the amplitude over the frequency, the so-called
amplitude response, and the other is the phase angle over the frequency, i.e. the
phase response.

The following general statements can be made:

The harmonic response is the progression of the output variable over time in the
steady-state condition with a sinusoidal input variable.

The frequency response is the sum of all harmonic responses of a controlled

system for various frequencies.

The frequency response has considerably greater resolution than the step response but
it is also more complicated. It is used to calculate control loops over the frequency
range. Therefore, it will not be covered in greater detail here. The step response
provides the practitioner with a more readily comprehensible and illustrative description
of the dynamic response of a control loop element and is quite sufficient for evaluation
in practically every case.

41

Controller

Controllers

3.1

Introduction

As already mentioned in chapter 1, the task of a controller is to automatically effect a

change in the manipulated variable y in order to correct a control difference e caused
by a reference variable change or by interference.
In order to fulfil this task, the controller must continually measure (via a sensor) the
value of the controlled variable and compare it with the setpoint w. Depending on the
result of this comparison, an appropriate control element is then adjusted such that the
control difference that occurred is corrected, i.e. the controlled variable equals the
setpoint again.

Controlling equipment

Controlling equipment consists mainly of the following components, which are shown
as blocks in Fig. 3-1:

3
w

3.1

+
2
x

3.2

B33-1a

Fig. 3-1

Components of controlling equipment

1
2
3
3.1
3.2
4
e
w
x
y

Setpoint adjuster
Sensor
Controller
Comparator
Control algorithm (amplifier and control action)
Actuator
Control difference (w - x)
Reference variable
Controlled variable
Manipulated variable

The desired value of the setpoint w for the controlled variable is set at the setpoint
adjuster (e.g. digital setting value, potentiometer or change of length on an adjusting
screw).

Sensor 2 to measure the controlled variable x (e.g. change in the resistance of a

nickel wire coil or change in the length of a bimetallic element).

Controller 3:

The comparator 3.1 compares the setpoint with the actual value of the controlled
variable and determines the control difference e = w x (e.g. from 2
temperatures or 2 lengths).

The amplifier 3.2 amplifies the control difference (e.g. electronic amplifier or
mechanical leverage) and influences the control response (dynamic response) of
the controller (not always present).

The actuator 4 varies the manipulated variable y.

In terms of practical implementation, two or more of the blocks shown are often
designed as a single component, e.g. controller, control valve (the control element of
the control valve belongs to the controlled system and not to the controlling equipment).

42

3.2

Classification of controllers

There are many different designs of controllers. However, they can be assigned to a
relatively small number of groups. This classification can be performed according to
various criteria. Controllers can be distinguished, for example, according to:

according to the application

according to the type of controlled variable

according to the necessary auxiliary energy

according to the effective direction

according to the manipulated variable

according to the response characteristic

3.2.1

Classification according to application

For the control of HVAC systems there are single-purpose controllers and universal
controllers. They are distinguished, for example, by the selectability of function,
physical type of controlled variable, sensor type, setpoint and sensor range, direction of
action, control parameters, manipulated variable output (possible actuator design) etc.
These can be fixed, or they can be selectable within limits or universally.
Single-purpose controllers,
standard controllers

These are designed for specific applications (e.g. heating control). Their adjustability is
restricted to the application and also limited. Parts of the above settings are fixed. The
setting values in the supply state correspond to typical operating values.
The purpose of these limitations is to optimise the costs of the controller (product) and
commissioning.

Universal controllers

These are designed for varied of HVAC applications (e.g. complex air treatment
system, large-scale heating system with multiple boilers and heating groups). They are
universally adjustable.
The function of the controller is often configurable (tailored, extensive and complex
applications). The costs of universal controllers are higher than those of single-purpose
controllers (for simpler applications). The engineering and commissioning costs can be
optimised using pre-configured and preset applications (e.g. downloadable from an
application library).

3.2.2

Classification according to controlled variable type

Depending on whether the variable to be controlled is a temperature, pressure,

humidity etc. the following terms are used:

Temperature controller

Pressure controller

Humidity controller

Air quality controller

Air speed controller

etc.

43

3.2.3

Classification according to the auxiliary energy

The following distinction is made according to the origin of the energy used to actuate
the control element:

Self-actuated controllers

Powered controllers

3.2.3.1

Self-actuated controllers

Self-actuated controllers take the energy for actuating the control element from the
measuring mechanism (sensor) itself. The comparison of the actual value and setpoint
is achieved mechanically by the force-balance or motion-balance method, e.g. level
controller with float (Fig. 3-2). In this system, a change in the height of float 1 acts via
lever 3 to produce a change in the stroke of gate valve 2.
3
2
1
w
Q

2
1
B33-2

Fig. 3-2
1
2
3

Flow of force in a self-actuated controller

Sensor (float)
Control element (gate valve)
Lever

Self-actuated controllers are mainly used for temperature, pressure or level control (e.g.
thermostatic radiator valve). Their advantage is that they operate independently of any
energy supply, they are robust and unsusceptible to faults.
3.2.3.2

Powered controllers

In most measuring mechanisms (sensor for acquiring the controlled variable) the
available energy is insufficient to actuate the control element directly. Therefore,
auxiliary energy is required to adjust the control element. For this purpose, an amplifier
controls the amount of actuating energy supplied to the actuator and with it the valve
stroke according to the specifications of the sensor or the comparator.
Appropriate types of auxiliary energy are especially electricity, compressed air or
pressurised oil. Depending on the type of auxiliary energy used, a distinction is made
between:

44

Electrical (electronic) controllers

Pneumatic controllers

Hydraulic controllers

etc.

Auxiliary energy electricity

Actuators

The use of electricity as the auxiliary energy provides the following main advantages:

The transmission of measured values and control commands, as well as remote

setpoint adjustment in many cases, is possible over great distances.

Electrical cables can be installed almost everywhere.

The installation can be performed by any electrical installation contractor.

Since the controllers contain practically no moving parts, they require little
maintenance and have an almost unlimited service life.

Electric motor, electrohydraulic, solenoid and electrothermal actuators are used with
powered controllers whose auxiliary energy is electricit

3.2.4

Control action of the effective direction

The direction of control action determines the direction in which the controller changes
the output variable y based on the direction of change of the input variable x. It can be
direct acting or reverse acting. The function of the respective controller in the control
loop is to
keep the controlled variable high

(Fig. 3-3 left:

keep the controlled variable low

(Fig. 3-3 right: direct action)

reverse action)

100 %

100 %

0%

x
w

0%

x
w
B31-17

Fig. 3-3

Direction of control action in a control loop

In some controllers, the direction of control action is selectable.

Direction of control action,
definition:
Reverse action

Direct action

Direction of control action,

applications

The manipulated variable rises if the controlled variable (or the actual value x) falls.

The controller only activates its output if e = w x > 0

The manipulated variable rises if the controlled variable (or the actual value x) rises.

The controller only activates its output if e = w x < 0

Reverse action is used for heating controllers, for example.

Direct action is used for cooling controllers, for example.
There are also controllers whose direction of control action can be changed during
plant operation (the term changeover is used). They are used in plants where either
hot water/air for heating or cold water/air for cooling circulates in the same distributor
and consumer circuit depending on the season.
Depending on the direction of control action, the control algorithm changes the sign of
the received control difference e.

45

3.2.5

Classification according to control action of the

manipulated variable

Each controlled system has a specific dynamic response, which can be ascertained by
metrological or theoretical means, but which is beyond the influence of the control
engineer. Therefore, it is necessary to be able to adapt the controller to the controlled
system as closely as possible. For this purpose, different controller types with regard to
the output signal (control response and direction of control action) are available which
can be roughly divided into three groups:

Continuous-action controllers

Discontinuous-action controllers

Quasi-continuous controllers

3.2.5.1

Continuous-action (modulating) controllers

Continuous-action controllers (almost) steplessly detect the control difference e with

sign across the entire range of values. Their positioning speed can be continuously
varied in both directions between zero and a maximum value, and the manipulated
variable can assume any intermediate position within the entire positioning range Yh,
i.e. it can be brought into the necessary position to maintain the setpoint.
3.2.5.2

Discontinuous-action (step) controllers

Step controllers can only position their output variable (manipulated variable y) at two
or more predefined values (e.g. 0, 33, 66, 100%) within the entire positioning range Yh.
The transition from one value to the next is abrupt (discontinuous).
Step controller types

Depending on the number of actuating positions that the step controller can set, a
distinction is made between:

Two-position controllers (0-100%, ON/OFF, open/closed)

Multi-position controllers or two-position controllers with two or multi-step design

(e.g. 0-50-100%)

3.2.5.3

Quasi-continuous controllers

In HVAC systems, control elements with electric motor actuators that are controlled by
quasi-continuous controllers are frequently used (issue of price). In the case of these
controllers, the manipulated variable can assume any value, but their positioning speed
can only have one predefined value apart from zero.
Quasi-continuous
controller types

46

Three-position controllers (open-stop-close)

Pulse width controlled two-position controllers (TOn / TCycle = 0...100%)

3.2.6

Response characteristic

3.2.6.1

Differentiation of response characteristics (control characteristics)

The temporal behaviour of a controller's positioning output based on a change at the

actual value input can be classified as follows:

Proportional-action controller

P-controller

Integral-action controller

I-controller

Proportional-plus-integral controller

PI-controller

Proportional-plus-derivative controller

PD-controller

Proportional-plus-integral-plus-derivative controller

PID-controller

3.2.6.2

Steady-state and dynamic response of the controller

A controller 1 (Fig. 3-4) can be considered as a control loop element where the control
difference e is the input variable and the manipulated variable y is the output variable.
Since, however, the comparator 2 is an internal circuit of the controller, making e an
internal value of the controller, the controlled variable x is also referred to as the input
variable in case of a constant reference variable w.
2

y
1
x

Fig. 3-4
1
2
e
w
x
y

B33-3

Input and output variables of a controller

Controller
Comparator
Control difference
Reference variable
Controlled variable
Manipulated variable

Decisive for the selection of a controller and for the control quality to be achieved in
combination with the controlled system is how quickly, by how much and for how long
the controller varies the control element in case of a control difference e. This can, as in
the case of the controlled systems, be described in terms of the steady-state and
dynamic response of the controller.
3.2.6.2.1

Steady-state response of the controller

The steady-state response of a controller refers to the relationship between the output
variable a (manipulated variable y) and the input variable e (controlled variable x or
control difference e) in steady-state condition.
What is important is the magnitude of the manipulated variable change y in proportion
to a controlled variable change x. This proportion, which is referred to as the transfer
coefficient KR of the controller, can be calculated as follows:
Transfer coefficient KR

KR =

Manipulated variable change y

Controlled variable change x

e.g. [mm stroke K ]

The relationship between the manipulated variable change y and the controlled
variable change x in the steady-state condition can also be presented graphically by
the steady-state characteristic (See 3.4.2, P-controller, for example).

47

3.2.6.2.2

Dynamic response of the controller

The dynamic response of a controller, which is also referred to as the response

characteristic, describes the progression of the controller's output variable (manipulated
variable y) over time in case of a change in its input variable (controlled variable x).
Since the characteristics of the plant to be controlled (Tu, Tg, KS) are usually given and
cannot be influenced, the dynamic response of the controller must be adapted to the
plant in order to achieve optimal control quality. This dynamic response must be
generated in the controller and is usually adjustable.
Step response

48

A controller's step response provides information on its dynamic response. In order to

obtain the step response, the controller's input variable (control difference e) is abruptly
changed by a random value (e.g. via a setpoint change) with the control loop open, and
the progression of the output variable (manipulated variable y) over time is plotted on a
recording device. The curve that is obtained in this way shows the dynamic response
(transient behaviour) of the controller.

3.3

The P, PI, PID response characteristic

3.3.1

The proportional-action controller (P-controller)

3.3.1.1

Operating principle

The operating principle of a P-controller can be explained most simply using the
example of a mechanical water level control system as per Fig. 3.5. The illustration
shows an open water tank with adjustable inflow (gate valve 1) and variable outflow
(gate valve 2). The control system consists of float 3 (sensor), lever a + b (comparator
and amplifier) and gate valve 1 (control element). The controller's input variable,
controlled variable x, is the water level measured by the float. The output variable,
manipulated variable y, is the position of gate valve 1 in the inlet pipe. The variable
water outlet (depending on the position of gate valve 2) represents the load Q on the
system. The control task is to keep the water level in the tank as constant as possible in
spite of varying load (drawing of water). The desired water level (setpoint w) is set by
the attachment height h of the lever.
h
4
1

Yh

V1

x
3

Xp

230 cm

200
170

B33-4

V2
Fig. 3-5
1
2
2
4,5
V1
V2
a+b
h
w
x
Xp
Yh

Model of a P-controller
Gate valve in inlet
Gate valve in outlet
Float (sensor)
Float attachment points
Inlet volume
Outlet volume
Lever
Attachment height of the lever
Reference variable
Controlled variable (water level)
Proportional band
Positioning range

49

At constant pressure in the supply pipe, the inflowing water volume V1 is given by the
position of gate valve 1. Therefore, a specific fluid flow is associated with each valve
position. Let us assume that the characteristics of gate valves 1 and 2 are linear so that
a stroke change of 10%, for example, always gives rise to a flow volume change of
10%. Therefore, the fluid flow is proportional to the position of the valve.
At the beginning of our observation (Fig. 3-5, Fig. 3-6) gate valve 2 is in the middle
position, and the outflowing water volume per unit time V2 is half of the maximum
possible volume. Therefore, the system is working at half load at the specified
operating point P.
Float 3 measures a water level of 200 cm, for example, which corresponds to the
setpoint w. Via the lever, the stroke of gate valve 1 (manipulated variable y) is also set
to 50%, i.e. in the middle position. Let the inflowing water volume V1 be equal to the
outflowing water volume V2 in this state, so the water level in the tank remains constant
(steady-state condition, rest).
If the water outflow, i.e. the load, is now increased, for example to 75%, by opening
gate valve 2 further, gate valve 1 (manipulated variable y) must also be opened further
in order to prevent a drop in the water level by increasing the inflow. In case of an
increased outflow volume, the water level can only be kept constant by increasing the
inflowing water volume to this new value.
V1, Y
%
100
75
50

0
0

50

75

100 %

V2, Q
B33-5

P
Fig. 3-6
Q
P

Load-dependent controller

Proportional relationship between water outlet V2 (load) and water inlet V1; i.e. manipulated variable y

Load
Specified operating point

However, in order for gate valve 1, which is actuated by float 3, to open further, the
float, and therefore the water level (controlled variable x), has to go down by a certain
amount (Fig. 3-6, dashed line). What this means is that a value of the controlled
variable x that differs from setpoint w is maintained.
This means that P-controllers have the marked disadvantage that they are unable to
keep the controlled variable constant at the desired level in every load state, i.e. they
are load-dependent.
As can also be seen from Fig. 3-6, each load state Q has an associated stroke of gate
valve 1 (manipulated variable y) and, therefore, an associated float level (controlled
variable x).
Therefore, the output variable of the P-controller, manipulated variable y, is
proportional to the input variable, controlled variable x or control difference e, i.e.
each value of the input variable has an associated fixed value of the output
variable. A load change gives rise to a deviation of the controlled variable x from the
setpoint w that rises proportionally with the load and, therefore, with the opening of gate
valve 1 (manipulated variable y). This residual deviation from the setpoint is referred to
as the steady-state deviation.

50

3.3.1.2

Characteristic quantity and steady-state characteristic of the

P-controller

Fig. 3-5 shows that the water level (controlled variable x) must change over a specific
range in order for gate valve 1 (manipulated variable y) to travel through its full stroke
Yh. This range, the characteristic quantity of the P-controller, is referred to as the
proportional band Xp, or P-band Xp for short.
P-band Xp

The P-band Xp is the range through which the controlled variable x has to change in
order to adjust the manipulated variable y over the entire positioning range Yh.
This relationship is represented by the steady-state characteristic of the P-controller
(Fig. 3-7).
y
Xp

KP =

%
100

y
x

50

Yh

0
0

185

Q
Fig. 3-7

200

100 %

215 cm

0%

B33-6

Steady-state characteristic and P-band of the P-controller as per Fig. 3-5

x Controlled variable change

y Manipulated variable change

Xp
Yh
KP
Q
P

Proportional band
Positioning range
Transfer coefficient
Load
Offset

The P-band Xp is indicated in units of the controlled variable x or as a percentage of the

range of reference variables Wh, i.e. of the setpoint potentiometer's adjustment range.
If, for example, the water level, i.e. the controlled variable at setpoint w, has to change
by +/- 15 cm for gate valve 1 to be adjusted over its full stroke Yh of e.g. 20 cm, the
controller has a P-band of 30 cm. With an assumed range of reference variables Wh of
60 cm (i.e. from 170...230 cm), therefore:

Xp (%) =

Xp [cm] 100 %
Wh [cm]

30 100%
=
60

50 %

The proportional band Xp is normally adjustable to allow the controller to be adapted to

the respective controlled system. This is done in Fig. 3-5 by moving float 3 from point 5
to point 4 This increases the transmission ratio a:b from 1:1 to 2:1 so that a control
difference of only +/- 7.5 cm is now required for the gate valve to travel through the full
positioning range Yh (Fig. 3-7, dashed characteristic).
The smaller the proportional band Xp that is selected the smaller the offset will be
but the greater the tendency of the control loop to oscillate.
Transfer coefficient KP

Fig. 3-7 also shows that, with a change of the P-band, the steady-state characteristic of
the controller simply rotates about the controller's offset P, i.e. around the point on the
steady-state characteristic at which the actual value corresponds to the setpoint.
Therefore, the P-band Xp becomes smaller with greater transmission ratios and vice
versa. But the transmission ratio is nothing other than the transfer coefficient KP of the
controller. This is defined by the magnitude of the manipulated variable change y with
respect to the controlled variable change x.

51

Manipulated variable change y

Controlled variable change x

Transfer coefficient KP =

This equation also applies to the values Yh (positioning range of the manipulated
variable) and Xh (adjustment range of the controlled variable) assigned to each other
and to Xp:

KP =

y Yh xh
=
=
x Xp Yp

For the manipulated variable change, this means:

y = KP x

Y
KP = h
Xp

and since

Y
y = h x
Xp
The following general statements can be made about the transfer coefficient: A large Pband Xp corresponds to a small transfer coefficient KP and vice versa.
Therefore, the transfer coefficient KP is equal to the reciprocal (opposite) of the P-band
Xp:
KP =

1
Xp

KP =

100%
Xp

Xp in % inserted

Examples:
KP =

52

100%
=3
33 13 %

or

KP =

100%
=2
50%

3.3.1.3

Dynamic response of the P-controller

The dynamic response of the P-controller can be demonstrated by the already familiar
step response, i.e. the change over time of the output variable (manipulated variable y)
in case of an abrupt change of the input variable (controlled variable x) in an open
control circuit.
The ideal P-controller reacts to an abrupt change of the controlled variable x with an
equally abrupt change in the manipulated variable y, where the magnitude of the
change depends on the magnitude of the controlled variable change x and on the
selected P-band Xp.
As shown in Fig. 3-8, the manipulated variable change y for a given controlled variable
change x with a P-band Xp = 100% is only half as large as with Xp = 50%. Therefore,
the manipulated variable y (output variable) is proportional to the controlled variable x
(input variable).

a)

x
0

b)

Xp = 100 %

KP =

y
x

t
y

d)

c)

KP =
Xp = 50 %

t
Fig. 3-8
a)
b)
c)
d)

Properties of the
P-controller

4 mm
4K

= 1 mm/K

B33-7

Step response of a P-controller with various P-bands Xp (dynamic characteristic)

Input
Output for Xp = 100%
Output for Xp = 50%
Example

The dynamic characteristic of a P-controller shows the following important points:

The step response corresponds to the proportional relationship between the input
and output signals

This proportional relationship is the reason why the P-controller is load-dependent

The control correction immediately follows the corresponding control difference

This makes the P-controller a fast controller.

The magnitude of the correction is limited (proportional to the deviation), i.e. the
controller is intrinsically stable, so it can also stably control non-self-regulating
controlled systems.

The magnitude of the output signal in proportion to the input signal can be adjusted in
the controller (P-band Xp, or Kp), which adapts the controller to the controlled system.

53

3.3.1.4

Shifting the offset

As the water level control system as per Fig. 3-5 shows, the selected setpoint w of a Pcontroller can only be achieved at a specific load at the so-called offset P (load Q =
50% in this case). At every other load, control differences occur within the selected Pband, i.e. the offset eb.
In some P-controllers, both the magnitude of the P-band can be adjusted, and the
offset can be shifted (Fig. 3-9).
Therefore, the P-controller can be calibrated such that the setpoint is achieved at load
Q = 0% or Q = 80%, for example. However, the consequence of this is that with
calibration for load Q = 0%, for example, the P-offset is as large as the P-band and,
therefore, twice as large as it would be with calibration for load Q = 50%. For a given Pband, therefore, the maximum possible P-offset depends on the position of the offset.
If the offset can be adjusted, it will be defined such that both positive and negative
interference can be compensated (e.g. at load Q = 50%). If the most frequently
occurring load state is known, the offset will be set to that value.
y
Xp
100 %

a)

Yh

0%

w
Q
100 %

y
Xp
100 %

b)

Yh

0%

x
w

Q
100 %

y
Xp
100 %

P
Yh
x

0%

c)

w
Q
100 %

Fig. 3-9
a)
b)
c)

54

Examples of shifting of the offset in a P-controller.

Offset at load Q = 0%
Offset at load Q = 50%
Offset at load Q = 100%

3.3.2

The integral-action controller (I-controller)

3.3.2.1

General Example 1

Integrate, in this context has the sense of combining or adding. In control loop elements
with integral action, the output variable is formed from the sum of consecutive input
variables over a period of time.
This can be demonstrated with a simple example
xe
xa

200 l/h

S0169

xe

xe2
xe 1

300 l/h

200 l/h

xa

berlauf

xa2
200 l/h

Fig. 3-10

xa1
t

Container as level controller, Example of a control loop with integral action response

The controlled system in the example is used to control level, represented by a tank
with an inlet and an outlet. The input variable is the flow rate at the inlet, and the output
variable is the level (not the flow rate) at the output. If we assume that the flow rate at
the inlet is equal to theat at the outlet, then the level will remain constant. To record the
step rsponse, we need mentally to increase the flow by a sudden large amount. For
example, the inlet valve could be opended manually, so that now, the flow rate at the
inlet is 300l/h instead of 200l/h. The discharge rate will remain unchanged. The output
variable, i.e. the level, will increase continuously until the tank overflows.
We can now no longer maintain that there is a given output variable for every input
variable, since in our example, for any input variable which is greater than the
discharge rate, the level will rise until it overflows. This type of response is defined as
integral action. The contents of the tank are given by the sum of the liquid flowing into
the tank minus the quantity discharged. Only the length of the time taken to fill the tank
will be affected by the input variable. In other words, it is not the output variable itself,
but the rate of change of the outut variable, which depends on the input variable. This is
expressed by the formula:
va = KI . xe
KI is the so called integral control action constand and in this example u.a. is
dependent upon the container.
N.B.

For an integral action controller the speed of the output variable is dependent upon the
input variable. For a step change in the input variable then the output variable changes
consecutively.

55

3.3.2.2

Mode of Operation Example 2

The operating principle of the I-controller will be illustrated using the pressure control
system as per Fig. 3-11. The illustration shows a pressure chamber 1 with a supply
pipe 2, valve 3, several air outlets (load Q), a manometer 4 and a pressure sensor 5.
The pressure sensor consists of a counterweighted piston which changes position with
variations in the chamber pressure. The purpose of the control system is to keep the
chamber pressure constant in spite of varying air draw-off. The pressure is specified by
the setpoint selection represented in this case by the counterweight for the chamber
pressure measured by the pressure sensor.
The controlled variable x, and therefore the controller's input variable, is the chamber
pressure. The output variable, manipulated variable y, is the stroke of the control
element.
n [N/m2] = w

KI

8
10

6
x

y
4

Q
Q

Q
1
Fig. 3-11
1
2
3
4
5
6
7
8

Model of an I-control system

Pressure chamber
Supply pressure
Control valve
Manometer
Pressure sensor
Hydraulic controller
Auxiliary energy, oil pressure
Close supply pipe

9
10
w
x
y
KI
Q

Open supply pipe

Hydraulic actuator
Setpoint
Actual value
Manipulated variable
Metering valve
Load

In the steady-state condition, i.e. when the chamber pressure corresponds exactly to
the selected setpoint w, both pistons of the hydraulic controller are positioned exactly in
front of the supply ports to the hydraulic actuator. No oil can flow to the drive, so it
remains in its current position. Let the stroke of the valve be 50%, for example.
If the chamber pressure drops due to an increased air draw-off Q, the piston of the
pressure sensor move downwards. This forces the two control pistons downwards as
well. The result is that oil now flows through the lower pipe to the drive, so that the
drive's piston is forced upwards, opening the valve further to compensate for the
pressure drop in the chamber.
The rate of change vy of the valve stroke (positioning speed) is, notwithstanding the
position of metering valve KI (basic setting of the controller for the system to be
controlled), given by the magnitude of the control difference e.

56

This means: the greater the deviation between the setpoint and the actual value the
greater the rate of change of the manipulated variable will be (Fig. 3-12).
+

vy

xw

KI
b)
a)

B33-10

Fig. 3-12
KI
a)
b)

Load-independent
controller

Proportional relationship between the control difference e and the positioning speed vy

Metering valve
Low rate of change
Higher rate of change

Due to the opening of valve 3, the pressure in the chamber gradually rises again. The
control difference e becomes increasingly smaller, so the positioning speed vy becomes
increasingly lower. However the hydraulic actuator continues to operate, if only very
slowly, until the sliding contact of the pressure sensor returns to its original position, i.e.
until there is no longer a deviation and, therefore, no more compensatory flow through
the valve actuator.
Therefore, the I-controller corrects control differences completely; it is loadindependent.
3.3.2.3

Integral-action coefficient
KI

Characteristic quantity of the I-controller

In Fig. 3-12, metering valve Kl specifies the magnitude of the compensatory flow and,
therefore, the running time of the drive for a specific control difference e. This basic
setting, which is the characteristic quantity of the I-controller, is referred to as the
integral-action coefficient Kl and is used to adapt the controller to the respective
controlled system.
The integral-action coefficient Kl expresses the ratio of the positioning speed vy to the
control difference e, thus indicating the magnitude of the positioning speed
(e.g. 1 mm stroke/min) at a specified control difference e (e.g. 1 mbar); in this case,
Kl = 1 mm/min 1 mbar.
3.3.2.4

Steady-state characteristic of the I-controller

Since, in the case of the I-controller, the manipulated variable continues to change as
long as the controlled variable deviates from the setpoint, the controller is loadindependent. With any control difference e and an appropriate duration, therefore, any
value of the manipulated variable y within the positioning range Yh can be reached, i.e.
the control element can travel to its end stop. At rest, therefore, there is no direct
relationship between the control difference e and the manipulated variable y as in the controller. This relationship is also shown by the steady-state characteristic of the Icontroller (Fig. 3-13). On the other hand, there is of course a proportional relationship
between the load Q and the manipulated variable y.
y
100%
Yh

0
w

Fig. 3-13

B33-11

Steady-state characteristic of the I-controller

57

3.3.2.5

Dynamic response of the I-controller

The dynamic response of an I-controller can also be represented by the step response
(Fig. 3-14).
e
0

t
y

P
y

t
t

Fig. 3-14
e
I
P

Properties of the
I-controller

B33-12

Dynamic characteristic of the I-controller

Input (control difference)

Output of the integral-action controller
Output of a proportional-action controller

The following major properties of the I-controller can be determined from the step
response:
Dynamically, the I-controller behaves in the same manner as a non-self-regulating
system: In case of an abrupt change in the input signal (controlled variable x or control
difference e) the output signal (manipulated variable y) begins to change in a timelinear manner until it is limited by the range limits (e.g. valve stroke = 0...20 mm).
The rate of change of the output variable is proportional to the change of the input
variable, or in other words: the positioning speed vy is proportional to the control
difference e. On the other hand, there is no relationship between the control difference
e and the valve position y.
Since even a small control difference e gives rise, after a certain time, to the
manipulated variable change required to achieve the setpoint, the I-controller is loadindependent.

Application of the
I-controller

In contrast to the P-controller, the control action builds up only slowly, because a
certain time t must elapse before the manipulated variable change reaches a certain
value y. Therefore, the I-controller is a time-dependent controller.
The positioning speed vy can be calculated for any control difference e using the
integral-action coefficient Kl:
v y = KI e =

y
t

In order to determine the manipulated variable change y that occurs after a certain
time, it must be taken into consideration that it depends not only on the magnitude of
the control difference e but also on its duration t:

y = KI e t
Pure I-controllers are not used in the HVAC field.

58

3.3.3

The proportional-plus-integral controller

(PI-controller)

3.3.3.1

Operating principle and step response

A PI-controller can be imagined as a P-controller and an I-controller connected in

parallel (Fig. 3-15). This combines the advantages of the P-controller (fast) with those
of the I-controller (load-independent).

+
e

PI
Fig. 3-15

Combination of P-controller and I-controller to a PI-controller

Therefore, the step response of a PI-controller (Fig. 3-16) is the sum of the step
responses of a P-controller and an I-controller, i.e. it consists of a proportional
component (P-component) with a superimposed integral-action component (Icomponent).
As in the P-controller, the P-component initially gives rise to an adjustment yP of the
manipulated variable y that is proportional to the control difference e:
yP = K e
P

However, due to the I-component, the valve does not remain in that position but
continues to travel at a positioning speed vy that is dependent on the integral-action
coefficient Kl of the I-controller and proportional to the control difference e until it is
limited by the end stops, e.g. valve stroke = 0...20 mm.
vy = K e
I
Therefore, after a certain time (t0t1), the manipulated variable change yPI is made up
of the sum of the two variables:
y PI = y

)(

+ y = K P e K I x I t
I

e
0

t
y
PI

yI
yPI

yP

t
t0

Fig. 3-16
yp
yI
yPI

t1

B33-14

Step response of the PI-controller

P-component
I-component
P-component and I-component

59

Load-independent
controller

Since, in the case of the PI-controller, the I-component continues to change the
manipulated variable y as long as a control difference e exists, the controller is loadindependent. As with the I-controller, there is no relationship between the control
difference e and the valve position y.
The step response shows that the reaction of the PI-controller to an input signal is
faster/greater than that of the I-controller. The I-controller requires a certain time to
achieve a manipulated variable change of yP, for example, which, due to the Pcomponent, is not the case with the PI-controller.
3.3.3.2

Characteristic quantities of the PI-controller

The dynamic response of the PI-controller is determined by two adjustable

characteristic quantities (Fig. 3-17):
P-band Xp

1. The first characteristic quantity is the proportional band Xp (or the transfer coefficient
KP). This determines the magnitude of the P-component.

Integral-action time Tn

2. The second characteristic quantity is the integral-action time Tn, which is derived
from the I-controller's integral-action coefficient Kl. The integral-action time is
indicated in seconds or minutes and can be defined in two different ways:

Tn is the time required by the I-component, with a given input signal, to effect the
same manipulated variable change yI as immediately produced by the Pcomponent yP (Fig. 3-17a).

With the same input signal, a pure I-controller would have had to have intervened
time Tn earlier in order to achieve the manipulated variable change yP immediately
produced by the P-component. Therefore, Tn represents the time saved by the Pcomponent (Fig. 3-17b).

a)

y I = y P
1
= KP
XP

y P

t
Tn

b)

1
= KP
XP

y P

t
Tn

Fig. 3-17

B33-15

Definition of the integral-action time Tn

The integral-action time Tn can be determined by the following formula:

Tn =

60

KP
KI

The two characteristic quantities Xp and Tn are the setting parameters of the PIcontroller and are usually adjustable. Fig. 3-18 shows a number of step responses for
the same input signal but with different controller settings. The setting possibilities allow
the controller's dynamic response to be adjusted to the characteristics of the controlled
system. The result is an almost universally applicable, high-quality controller which
requires control engineering knowledge in order to achieve optimal adjustment for the
controlled system.
e
0

a)

b)
y

c)

d)
y

Fig. 3-18
e
y
a)
b)
c)
d)

Effect of Xp

Step response with various controller settings

Input signal
Output signal
Xp small, Tn short
Xp small, Tn long
Xp large, Tn short
Xp large, Tn long

The step responses shown in Fig. 3-18 show that, in the case of a PI-controller,
changing the P-band Xp has a much greater effect than changing the integral-action
time Tn, because the change in Xp not only gives rise to a different P-adjustment but
also to a different I-adjustment (compare a with c and b with d). Changing the integralaction time Tn gives rise to a different I-adjustment while the P-adjustment remains the
same (compare a with b and c with d).
This relationship is explained and illustrated in the model shown in Fig. 3-26.

61

3.3.4

The derivative-action element (D-element)

3.3.4.1

General Example 1

An example from the electrical engineering branch is used as an example for a control
loop with differential response.
S0172A

Ue
Ue

i R

Ua

t
Ua

B33-19

t
TD
Fig. 3-19

RC circuit and associated step response as an example of a derivate action control element
Tv = Derivative action time

The input variable is the voltage to the capacitor and resistor. The output variable is the
drop in voltage at the resistor.
As we know, a capacitor blocks direct current, but allows changes in current to pass. If
a direct voltage is supplied to the input, no current can build up. As a result of which the
output variable will be zero. If we now apply a step change to the input variable, there
will be a very large change when this step change occurs, i.e., the capacitor will allow a
hight current to pass, resulting in a large drop in voltage at the resistor, and hence a
very hight outut signal. The change in input variable is over within a very brief period.
As soon as the capacitor is fully charged, its charge current, and hence, the output
variable, will return to zero.
A derivative action control element, i.e., one without a time lag, has an infinitely long
and infinitely fine needle as its step response. In practice, of course, infinite values
cannot be achieved, but for practical purposes it is possible to achieve a step response
which rises quickly, and falls slowly, depending on the sizing of the circuit. This is
referred to as a D-T element (D-element with time delay). It can now be seen, that in
control elements with a derivative action, an output variable will be obtained only when
the input variable changes. The more quickly the input variable changes, the greater
the output variable will be.
The output variable is thus dependent on the rate of change of the input variable. This
can be expressed with the simple formula:
xa = KD ve

KD where is the derivative control action constant

In a derivative action control element, the output variable depends upon the rate of
change of the input variable. A single step change at the input produces a needle point
at the output.
Control loop elements are rare. Controllers with a purely derivative or D-T action are
not practicable, because they only respond to changes in the control deviation, while
any constant control deviations would not be registered at all. D-action control elements
do have their uses, however, particularly in conjunction with P- and I-controllers.

62

3.3.4.2

Mode of Operation Example 2

In a closed-loop control circuit, a rapid and correspondingly large adjustment of the

manipulated variable on occurrence of a control difference has a beneficial effect on the
transient recovery time. However, the previously discussed P, I and PI-controllers only
partially meet this requirement. This is particularly noticeable in controlled systems with
relatively long dead times; it is impossible to get a control difference back to zero within
a short time in such controlled systems.
In order to eliminate such undesirable and long-lasting control differences, the P or PIcontroller is given a head start to make its initial reaction faster and greater than that
of a pure P or PI-controller.
It is given a derivative-action element (D-element) which enables a greater control
impulse to be initiated on occurrence of a control difference. Therefore, the D-element
makes it possible to recover very quickly from rapid controlled variable changes.
A derivative-action element does not measure the magnitude of the control
difference e but its rate of change. Therefore, it produces a change in the
manipulated variable corresponding to the rate of change of the control difference. The
more rapid the change in the controlled variable the greater the correction via the
control element will be.
3.3.4.3

Step response of the D-element

An abrupt change in the input signal e as shown in Fig. 3-20a gives rise in an ideal Delement to an infinitely large output signal a which immediately drops back to zero (Fig.
3-20b) because no further change occurs after the step function, so no rate of change
is present any more. However, a spike function of this kind is not achievable in practice,
because no device can supply signals of infinite magnitude.

a)

b)

t
Fig. 3-20 Step response of the ideal D-element
a)
b)

Input signal e: step function

Output signal y: spike function

In actual practice, an electronic circuit is used as the D-element which supplies an

output signal like that of the model shown in Fig. 3-21. On the right-hand side of the
lever, this model consists of a cylinder that is suspended on a spring, is filled with oil
and has a perforated piston. On the left-hand side of the lever there is basically the
equivalent of a stroke indicator.
If at time t0 the piston is pulled abruptly downwards by the ring, the entire cylinder is
pulled down with it due to the inertia of the oil, which causes the indicating arrow on the
left-hand side to rise by a corresponding amount. However, the tensioned spring gives
rise to a positive pressure on the lower side of the piston, so the oil flows through the
perforations into the upper chamber of the cylinder. This causes the positive pressure
to gradually decrease, and the cylinder can be pulled by the spring tension back into its
original position. If plotted over time, the stroke change of the indicating arrow
describes an exponential curve that tails off with derivative-action time constant TD (Fig.
3-21b). The speed at which the oil flows back, and therefore the derivative-action time

63

constant TD, can be varied in the model by changing the size of the piston's
perforations.

a)

B33-18

b)
TD
Fig. 3-21
a)
b)
1
2
3
4
TD

64

Model of a derivative-action element

Model
Magnified step response
Lever
Oil filled cylinder with perforated piston
Return spring
Stroke indicator
Derivative-action time constant

3.3.4.4
Ramp response

Ramp response (ramp-forced response) of the D-element

The ramp response indicates the progression of the output variable in case of an input
signal that suddenly rises steadily (ramps up). This type of input signal is referred to as
a ramp function. The corresponding behaviour of the controller's output y is referred to
as the ramp response or ramp-forced response.
The ramp response (Fig. 3-22) is more suitable for characterising a D-element than the
step response. It shows that a manipulated variable change yD always occurs when
the rate of change vx of the controlled variable changes (at the beginning of the
change). However, if this rate of change is constant (during the ramp function), the
manipulated variable y remains at the value it assumed at the last change in the rate of
change of the controlled variable.
In the case of the D-element, therefore, the magnitude of change of the manipulated
variable yD is proportional to the rate of change vx of the controlled variable.
e
0

vx

t
y

y D

t
Fig. 3-22

Response of a D-element to a ramp function

vx Rate of change of the controlled variable x

yD Manipulated variable change in proportion to the ramp function

Derivative-action
coefficient KD

The characteristic quantity of the D-element is the derivative-action coefficient KD. It

expresses the relationship between the magnitude of change of the output signal
(manipulated variable y) and the rate of change of the input signal (controlled variable
x), thus indicating the amount yD by which the manipulated variable will change if the
rate of change vx of the controlled variable changes by one basic unit (e.g. 1 K/min).
KD =

yD
vx

Since the output variable change yD is a function of the rate of change of the input
variable, the former can be calculated as follows:
yD = v x K D
3.3.4.5

Application of the D-element

Since the D-element is unable to detect the absolute value of the controlled variable, it
cannot be used on its own for control purposes. However, as an additional element of a
P or PI controller (for which the term PD or PID-controller is then used) it has a certain
significance in the control of systems with relatively long dead times. The D-element
gives the controller a certain lead, i.e. it reacts as if it had detected the change in the
controlled variable earlier. This diminishes the damaging effects of the dead time.

65

3.3.5

The proportional-plus-derivative controller

(PD-controller)

3.3.5.1

Operating principle and step response

A PD-controller can be imagined as a proportional-action controller (P-controller) and a

derivative-action element (D-element) connected in parallel (Fig. 2-21a). It detects not
only the magnitude of the control difference e but, due to the D-element, also its rate of
change vx. This makes the already fast P-controller even more effective.

a)

PD

b)

t
y
D

TD
Fig. 3-23
a)
b)
P
D
TD

t
B33-20

P-controller and D-element as a PD-controller

Parallel connection of a P-controller and a D-element

Resultant step response
P-component
D-component
Derivative-action time constant

Therefore, the step response of a PD-controller (Fig. 3-23b) consists of a P-component

with a superimposed D-component. It shows that, due the derivative action, the
controller produces a large change in the manipulated variable in case of an abrupt
change of the controlled variable x. After the initial step, the D-component falls back to
zero along an exponential curve with derivative-action time constant TD, because the
rate of change of the controlled variable is now zero. However, it is now the Pcomponent that determines the controller's behaviour, i.e. after the decay of the Dcomponent, the manipulated variable assumes a value that is proportional to the
controlled variable.
PD-controllers can correct a control difference more quickly than P-controllers, but they
also give rise to a residual deviation (offset). However, the load-dependence of the PD
controller is less pronounced, because the stabilising effect of the D-component allows
a smaller Xp to be selected.
Application of the
PD-controller

66

PD controllers are hardly used in HVAC systems.

3.3.5.2

Ramp response of the PD-controller

As with the D-element, the ramp function (see section 3.7.2) is also more suitable for
describing the PD-controller than the step response.
The dashed line in Fig. 3-24 shows the reaction of a pure P-controller to the ramp
function. In the case of the PD-controller, however, a proportional adjustment yD of the
control element corresponding to the rate of change vx occurs at the start of the ramp
function. Afterwards the manipulated variable progresses according to the proportional
component, i.e. parallel to the dashed line of the P-controller, because the rate of
change vx of the controlled variable is now constant.
e
0

t
y

PD
P

yP

yP D

yD

t
Tv

Fig. 3-24
yP
yD
yPD

Tv

B33-21

Ramp response of a P-controller and a PD-controller

P-adjustment
D-adjustment
Ramp response of the PD-controller
Derivative-action time

3.3.5.3

Characteristic quantities of the PD-controller

As already mentioned and as visible in Fig. 3-24, the manipulated variable change after
time Tv is the sum of the P-adjustment and the adjustment by the D-element:

yPD = yp + yI = KP e + (K D v x )
The characteristic quantities of the PD-controller, and therefore the possible settings for
adjusting the controller to the controlled system, are as follows:

Derivative-action time Tv

The proportional band Xp (or the transfer coefficient KP) of the P-component

The derivative-action coefficient KD or the derivative-action time Tv derived from it

Since the terms transfer coefficient KP and derivative-action coefficient KD are already
known, only the derivative-action time Tv will be described here: The derivative-action
time Tv indicates how long before the beginning of the ramp function a pure P-controller
would have to begin changing the manipulated variable in order to achieve the value
yD reached immediately by the D-element (Fig. 3-24). Therefore, Tv represents the
time saved by the D-element; it is calculated as follows:
Tv =

KD
KP

Replacing the derivative-action coefficient KD with the derivative-action time Tv has the
following result for the manipulated variable change of the PD-controller:

) (

yPD = KP e + KP TV v x

The derivative-action time Tv governs the effect of the D-component. It is indicated in

seconds or minutes. The greater the value of Tv the greater the initial reaction of the
controller will be.

67

3.3.6

The proportional-plus-integral-plus-derivative controller

(PID-controller)

3.3.6.1

Operating principle and step response

A PID-controller can be imagined as a proportional-action controller (P-controller), an

integral-action controller (I-controller) and a derivative-action element (D-element)
connected in parallel (Fig. 3-25a). The controller's output signal (manipulated variable
change) is generated not only from the magnitude of the control difference e but, due to
the D-element, also its rate of change.
As already known, with a D-component, the magnitude of the manipulated variable
change is proportional to the rate of change of the controlled variable or control
difference.

a)
P

PID

b)

1
= KP
XP

t
Tn
Fig. 3-25
a)
b)
P
I
D

B33-22

PID-controller

Model
Step response
P-component
I-component
D-component

The step response of the PID-controller (Fig. 3-25b) consists of the additive
superimposition of the P-component, I-component and D-component. The illustration
shows that an initial large change of the manipulated variable is produced by the Delement. This means that interference will not give rise to too great a control difference.
The manipulated variable y is then reduced almost to the level of the P-controller, from
where it rises again in a linear fashion under the influence of the I-component. Due to
the I-component, the PID controller compensates for setpoint changes and
disturbances with no residual deviation and is faster than a PI or P-controller. It is
especially suitable for difficult controlled systems.

68

In specialist literature, the relationship is usually presented as a parallel connection of

the P, I and D-components (Fig. 3-25). In the actual signal processing within the
controller, however, it should be noted that the P-component (KP) also affects the I and
D-components (Fig. 3-26).

PID

e (t)

K P *e(t )

Fig. 3-26

3.3.6.2

Representation of a PID controller in accordance with actual practice

Characteristic quantities of the PID-controller

The dynamic response of the PID-controller is determined by three adjustable

characteristic quantities:

Proportional band Xp (or transfer coefficient KP): this determines the magnitude of
the P-component. The smaller the selected value of Xp (or the greater the value of
KP) the greater the P-component will be.

Integral-action time Tn: this determines the rate of rise of the I-component. The
smaller the value of Tn the steeper the slope.

Derivative-action time Tv: this determines the effect of the D-component. The
greater the selected value of Tv the greater (higher and broader) the surface area,
and therefore the effect, of the D-component will be.

Effect of Xp

As with PI-controllers, changing the P-band Xp has a much greater effect than changing
the integral-action time Tn or the derivative-action time Tv, because the change in Xp not
only gives rise to a different P-effect but also influences the I and D-effect (as visible in
the model shown in Fig. 3-26).
Changing the integral-action time Tn or derivative-action time Tv, however, only gives
rise to a change in the I- or D-adjustment while the P-effect remains the same.

Application of the
PID-controller

The three setting variables provide for optimal adjustment of the controller for the given
controlled system. In order to set them correctly, detailed control engineering
knowledge is required. Therefore, PID controllers are only used in cases where the
high demands of the controlled system justify the effort required for commissioning.
A typical application for a PID controller in the HVAC field is domestic hot water
heating. The demands on this control task are high because, on the one hand, the
controlled system is fast and difficult (it consists only of the section of pipe between the
valve and the sensor, i.e. a major part of this system consists of dead time the time
the water takes to flow through the pipe), and on the other hand relatively fast
interference can occur.
This application requires not only a PID controller but also a fast-acting sensor and
actuator in order to be able to quickly compensate for the fast interference variables.
People are highly sensitive to sudden water temperature variations.

69

3.4

The two-position controller

3.4.1

Operating principle

In the case of two-position controllers, the manipulated variable can only assume one
of two states: ON or OFF. They are usually electrical contact controllers (e.g.
thermostats). The operating principle is illustrated by the boiler temperature controller
shown in Fig. 3-27.
NO
C
NC

6
5
4
3

B33-23

Fig. 3-27
1
2
3
4
5
6
7
8
C
NC
NO

Design of a mechanical two-position controller

Sensor
CapiIlary tube
Aneroid diaphragm
Ball bearing
Transmission lever
Fulcrum of the transmission lever
Setpoint adjusting screw
Microswitch:
common
normally closed
normally open

As the temperature in sensor 1 rises, the fluid it contains expands through capillary
tube 2 causing a stroke action of aneroid diaphragm 3. This stroke acts on via ball
bearing 4 on transmission lever 5. Setpoint adjusting screw 7 determines the position of
fulcrum 6. When the sensor temperature reaches the selected setpoint, the ball bearing
rotates the transmission lever about the fulcrum. This then actuates microswitch 7. The
C - NC contact opens, which switches off the burner connected to it, for example.
The setpoint can be selected via the setpoint adjusting screw such that the switch-on or
switch-off point of the microswitch is reached at a specific diaphragm stroke, i.e. at a
specific sensor temperature.
This simple two-position controller design may give rise to very slow contact making,
causing excessive contact erosion, which in turn can produce negative effects in the
electrical network. In order to prevent this from occurring, an additional throw-over
mechanism (e.g. an -spring, not shown) is required as an energy store in the
microswitch to ensure abrupt contact switching.

70

3.4.2
Switching differential SD

Characteristic quantity and steady-state characteristic

of the two-position controller

The throw-over mechanism of the microswitch for rapid switching also has the effect
that the contact does not open and close at the same sensor temperature. A distinction
is made between the upper switching point xo and the lower switching point xu, and the
resultant temperature difference is referred to as the switching differential SD (Fig. 328).
The switching differential SD is the characteristic quantity of the two-position controller
and is permanently set by the design of the throw-over mechanism in the case of
mechanical setpoint/actual comparison. In the case of electronic two-position
controllers (no mechanical energy storage required) it is usually selectable as a setting
value.
The steady-state characteristic of the two-position controller (Fig. 3-28) shows the
relationship between the controlled variable x (e.g. boiler water temperature) and the
manipulated variable y (e.g. burner operation). It can be seen that the manipulated
variable y can only assume one of two values with different outputs:
y = 100% (ON) and y = 0% (OFF):

If the controlled variable x (boiler water temperature) exceeds the selected setpoint
w when the burner is running, the value of the manipulated variable drops from y =
100% to y = 0% (burner off).

After the boiler water temperature has fallen by the amount of the switching
differential SD, the controller starts the burner again (y = 100%).

xu

xo

SD
100 %

w
Fig. 3-28
x
xu
xo
y
w
SD

B33-24

Steady-state characteristic of the two-position controller

Controlled variable
Lower switching point
Upper switching point
Manipulated variable
Setpoint
Switching differential

Two-position controllers are normally calibrated so that the switch-off point, i.e. the
upper switching point, corresponds to the selected setpoint.

71

Due to the operating principle of the two-position controller, full power is applied
periodically to the controlled plant (e.g. boiler water). The ON-time depends on load, i.e.
the more hot boiler water is required, the longer the burner remains in operation. The
quotient of the ON-time te and the cycle time tz is referred to as the control factor
(epsilon) (Fig. 3-29).
Control factor

Control factor =

te
tz

y
100 %

Yh

y = Yh

te

ta

B33-25

tZ
Fig. 3-29
te
ta
tz

Switching cycle of the two-position controller

ON-time, pulse duration

OFF-time
Cycle time
Control factor

The theoretical manipulated variable (shown dashed in Fig. 3-29) is as follows in units
of the manipulated variable:
y = Yh
and the mean power supplier to the plant is:
Q = Qmax
Since, on occurrence of interference or in case of setpoint adjustment, the two-position
controller immediately switches to full power and remains on (or off) until the setpoint is
reached again, it can be considered a very fast type of control. However, the switching
differential SD has the effect in the control loop that the controlled variable is
continuously subjected to oscillation. The magnitude of these oscillations corresponds
to at least the magnitude of the switching differential, and it increases with increasing
delay time Tu of the controlled system. If the magnitude of the switching differential is
adjustable (electronic two-position controller), the amplitude of the control fluctuations
and their cycle time can be influenced.
The two-position controller is generally used as a low-cost and reliable controller for
simple control tasks and as a monitoring device. It is, for example, used as a
thermostat to switch the burner of a boiler water temperature control system or the
immersion heater of a hot water tank, as a hygrostat to switch the humidifier of a simple
room humidity control system, as a pressostat to switch the compressor of a
compressed air station, as a supply temperature low-limit controller in a floor heating
system or as a frost protection thermostat in a simple ventilation system.

72

3.4.3

Two-position controller with feedback

The considerable fluctuations of the controlled variable x that arise with simple twoposition controllers are undesirable or not permissible in many control loops. They are
mainly caused by the fact that the controller leaves the manipulated variable (e.g. the
burner) ON and OFF for too long.
If a pure two-position controller is equipped with a delaying thermal or electronic
feedback, the variation range of the controlled variable can be considerably reduced,
i.e. made considerably less than the controller's switching differential.
3.4.3.1

Design and operating principle

In the case of a simple bimetallic room thermostat (Fig. 3-30), the feedback consists of
a small heating resistor 3 which is installed close to the bimetallic element 1. This
resistor is switched on at the same time as the room heating 5, i.e. a voltage is applied
to it, and it heats up the bimetallic element together with the room temperature. This
additional, more rapid heating of the bimetallic element (the time constant of the
feedback is very short in relation to that of the room) simulates the reaching of the room
temperature setpoint for the bimetallic element. Therefore, the controller switches the
room heating and the feedback off again long before the room is overheated due to
dead time, thermal inertia and the switching differential.

4
N

Ph

3
1
N

2
21 C

Fig. 3-30
1
2
3
4
5

20 C

19 C

Design of a two-position controller with thermal feedback

Bimetallic element
Setpoint adjustment
Feedback resistor
Snap-action magnet, throw-over mechanism
Electric room heater

Conversely, the feedback resistor cools much faster (bimetallic element is passing
through its switching differential) than the room temperature can react to the switching
off of the heating. Therefore, the heating is switched back on long before the room
temperature begins to fall again.
Installing the thermal feedback has increased the switching frequency of the twoposition controller to such an extent that now it is no longer the individual impulses that
act on the controlled variable but only their mean value or the control factor.

73

The switching frequency, and therefore the fluctuations of the controlled variable, now
depends on the switching differential SD of the two-position controller, the maximum
feedback variable Yrh and the feedback time constant Tr. The maximum feedback
variable is the maximum controlled variable change that the feedback can simulate for
the bimetallic element if it is permanently on. It is indicated in units of the controlled
variable or as a percentage of the adjustment range, and it can only be fully effective if
the control difference is greater than Yrh.
Fig. 3-31 should explain the interaction of the controlled variable x and the feedback
variable yr in greater detail. It shows observations at the room temperatures (controlled
variable x) 17, 18, 19 and 20 C while the maximum feedback variable Yrh = 4.5 K and
the switching differential SD = 2 K. The diagram shows that, apart from the long initial
impulse, the control factor is proportional to the control difference e. From this, the
mean manipulated variable ym and the proportional behaviour of the two-position
controller can be determined. As with the P-controller, the two-position controller also
has a proportional band and a residual deviation. Within the proportional band, the
controller switches from permanently ON (full load) through long to short switching
impulses to permanently OFF (load 0).
The cycle time depends on the switching differential, the feedback variable yr and the
control difference e. With respect to the control difference, it is shortest with a control
factor = 0.5 (ON half the time). It becomes very long towards = 0 (permanently OFF)
and = 1 (permanently ON).
The initial longer control impulse, which occurs when yr is considerably greater than the
switching differential SD, has a similar effect to the derivative-action impulse of the PDcontroller. Therefore, the two-position controller with feedback is said to have a quasiPD control action (greatly extended mean value line PD in the pulse sequence). In this
regard, the feedback variable yrh corresponds to the proportional band Xp, and the
feedback time constant Tr corresponds to the derivative-action time Tv.
x

x
Tr

C
22
21
20
w 19
18
17
16

SD

y
1

yr

w
e

x
ta

te

1
ym

y rh

ym

PD

tz

Fig. 3-31
Yrh
yr
Tr
SD
te

t
B33-27

Operating principle of the two-position controller with thermal feedback

Max. feedback variable

Feedback variable
Feedback time constant
Switching differential
ON-time

ta
tz

ym
PD

OFF-time
Cycle time
Control factor
Mean manipulated variable
Resultant PD action

Modern two-position controllers with feedback operate with electronic amplifiers and
feedback. This makes it possible to even achieve a quasi-PID control action and,
therefore, load-independent two-position control.
Two-position controllers with feedback are used for room temperature control. In this
application they control solenoid valves of atmospheric gas burners, electrothermal
actuators, circulation pumps, direct electric heaters, zone valves etc.

74

3.5
Three-position actuators

The three-position controller

Electric motor actuators are usually designed as three-position actuators. They have
the following electrical connections:

Supply

Open connection

y1

(e.g. switched neutral conductor 230 VAC)

Close connection

y2

(e.g. switched neutral conductor 230 VAC)

(e.g. phase 230 VAC)

Three-position actuators are often equipped with synchronous motors. Since their
rotational speed is load-independent, they can only assume three positioning speeds:

Open

Stop

Close

(at constant speed)

(at constant speed)

Electric motor actuators are universally applicable in the HVAC field and offer very
good value for money. However, special controllers three-position controllers are
required to control them.
In addition to electric motor three-position actuators, electrohydraulic three-position
actuators are also available.
Three-position controllers

Three-position controllers are designed to control three-position actuators in a closedloop control circuit. Two positioning outputs are required on the controller: y1 (for
opening) and y2 (for closing). y1 and y2 are power outputs that can close the supply
circuits of the actuator's two running directions. However, they must not both be closed
at the same time. The positioning function for the actuator (the so-called positioning
algorithm) is contained in the controller.
Three-position actuators are generally slow-running actuators. Their response
characteristic is integral, which is taken into account in the functional design of a threeposition controller.

3.5.1

A simple three-position controller

Let us imagine this three-position controller as a combination of two thermostats: one

for heating (contact y1) and one for cooling (contact y2). y1 switches an electric heater
on at room temperatures below 19 C and off again at 19.5 C. y2 switches a
refrigeration unit on at room temperatures above 21 C and off again at 20.5 C.
Therefore, contact outputs y1 and y2 give rise to a cycle: heating - off - cooling. Fig. 3-32
shows the steady-state characteristic of this three-position controller.
The major characteristic quantities of this three-position controller are the dead zone
xDZ and the switching differential SD:

The dead zone xDZ is the range within which the controlled variable can change
without one of the contacts closing and emitting a correction signal. It is indicated as
a percentage of the adjustment range Wh or directly in units of the controlled
variable. Instead of the dead zone, the responsiveness emin is often indicated.
This corresponds to xDZ /2.

The switching differential SD is the controlled variable difference of a switching

contact between its switch-on and switch-off points. The arrangement of the switchon and switch-off points at different control differences (hysteresis) means that the
switching state even remains stable if the control difference is close to the switching
thresholds.

The dead zone and switching differential contribute to the accuracy of the control-action
result.

75

a)

SD

SD
xDZ

b)
y1

y2
x

B33-28

Fig. 3-32 Simple three-position controller (three-position switch)

a)
b)

Steady-state characteristic
Block diagram

This type of three-position controller is (now) hardly ever used in control loops.
However, in the form of a three-position switch it is a functional element of newer threeposition controllers.

3.5.2

Three-position controller with quasi-continuous controlaction result

The electric motor and electrohydraulic actuators frequently used to drive the valves
and dampers of HVAC systems have a constant runtime. However, the I-component of
a PI-controller requires an actuator positioning speed that is proportional to the input
signal.
Quasi-continuous three-position controllers solve this problem by dividing up the
motor's runtime into steps that correspond to the change in the I-component. This
reduces the mean positioning speed of such actuators.
3.5.2.1

Design

The quasi-continuous PI-controller consists of a three-position controller that actuates a

modulating control element (Fig. 3-33a) via an actuator with fixed runtime.
However, the output signal of the three-position controller does not give rise to heating
ON or OFF as in a two-position controller but to valve open stop close. The
output signal of the three-position controller is, at the same time, also looped back to its
input via a time-delay element, where it counteracts the input signal. In this case, the
motor acts as an integral part of the controller function, because the PI-characteristic of
the manipulated variable would not be achieved without it.
3.5.2.2

Operating principle

As long as the control difference remains within the dead zone xDZ of the three-position
controller, the two outputs y1 and y2 are inactive, and the motor is at a standstill.
If the control difference goes beyond the dead zone of the three-position controller, its
output signal y1 or y2 (depending on the sign of the control difference) acts on the
actuator which now starts to move in the appropriate direction at constant speed (Fig.

76

3-33b). The same output signal is looped back to the controller's input via a delay
element (elastic feedback) and switched opposite to the input signal. When the
difference between the input signal and the gradually rising feedback signal yr is less
than half the dead zone of the three-position controller, the controller resets the output
signal, and the motor stops. The length of the first correction step is determined by the
feedback variable, the feedback time constant and the magnitude of the input signal.
Therefore, it is proportional to the magnitude of the input step, and the resultant
manipulated variable change produces the P-component of the PI-controller.
When the three-position controller stops the actuator, the feedback signal also begins
to decay. As soon as the difference between the input signal and the feedback variable
is greater than half the dead zone of the three-position controller again, the switch-on
process is repeated. However, the actuating signal is only issued for brief moments,
because the feedback signal no longer has to build up from zero. The actuator now
only changes its position in steps, whereby the manipulated variable is adjusted at a
reduced mean positioning speed (I-component). If the control difference decreases, the
motor receives short impulses at increasing intervals. This reduces the mean
positioning speed. If the control difference increases, the mean positioning speed also
increases because of the shorter intervals of the positioning impulses.
Therefore, the step response of this quasi-continuous PI-controller consists of a longer
initial control step, whose duration is proportional to the input signal, and therefore
represents the P-component, and a series of additional steps, whose control factor and
the resultant mean positioning speed are also proportional to the input signal, thus
representing the I-component.
Major characteristic
quantities of the quasicontinuous three-position
controller

The characteristic quantities of the quasi-continuous PI-controller are the dead zone xDZ
and the switching differential SD of the controller as well as the feedback variable yr
(achievable effect of the feedback, expressed in units of the control variable or as a
percentage of the adjustment range Wh), the feedback time constant Tr and the runtime
Ty of the motor. In most cases, the controller is adjusted for the controlled system via
the feedback variable yr and the feedback time constant Tr. This gives rise to the
following relationships:

The controller's P-band depends on the feedback variable yr, the feedback time
constant Tr and the runtime Ty of the actuator:
Xp =

Ty yr
Tr

This means:

The greater yr the greater Xp

The greater Ty the greater Xp

The greater Tr the smaller Xp

The integral-action time Tn is determined by the feedback time constant Tr. Where the
following applies:
T n = Tr
The quasi-continuous PI-controller corrects relatively quickly and accurately, apart from
the small dead zone. Because of its simple design it is very widespread. It is mainly
encountered in the heating, ventilation and air-conditioning industry.

77

y1

y2

x
1

yr

a)
X p ,T n
2

t
y1

b)
t
y
I
p
Tn
Fig. 3-33
a)
b)
1
2
3
x
y1
y2
y

78

Quasi-continuous PI-controller

Block diagram
Operating principle
Three-position controller
Feedback
Three-position actuator (electric motor or hydraulic)
Input signal (controlled variable)
Relay output signal open valve
Relay output signal close valve
Manipulated variable, valve stroke

Control Loop

In the previous chapters we have considered the controlled systems and controllers
individually. This was meaningful in the sense that it enabled the respective control
engineering terms to be introduced and the basic design principles of controllers to be
presented. Additionally, the characteristic control engineering values of both parts had
to be defined in order to make the processes in the control loop that will now be
covered accessible.
A control loop (Fig. 4-1) is produced by connecting a controlled system with a closedloop controller. The control loop is represented by a block diagram that contains no
equipment details of either the plant or the controlling equipment.
2

2.2

2.4

2.3

2.5

(1.1)

2.1

1.2
1.3

1.1

1.2
w

x
e

w
B34-1a

Fig. 4-1
1
1.1, 2.5
1.2
1.3
2
e
w
x
y

Ventilation system with block diagram of the control loop

Controlling equipment
Sensor, measuring equipment
Controller
Control element
Controlled system
Control difference
Reference variable
Controlled variable
Manipulated variable

The connecting lines between the blocks indicate that the controller and controlled
system influence each other mutually via the controlled variable and manipulated
variable, whereby the influence can only take place in the direction shown. The clearly
specified direction of transmission gives rise to the abstract concept of a closed cycle.
This is where the term control loop comes from.
Oscillatory system

Since the controlled variable and the manipulated variable exert a mutual influence on
each other, a control loop is an oscillatory system. This manifests itself in the fact that,
under certain circumstances, the controlled variable does not assume the required
constant value but constantly varies between a higher and a lower value.
The controlled variable escalates into oscillation, and the process gets out of control
the system becomes unstable.
This is an operating state of the control loop that must be avoided at all costs in
practice, because it is absolutely useless. Only stable control loops are of interest in
control engineering. This means that a closed-loop control circuit must be prevented
from performing self-excited oscillations, i.e. oscillations whose amplitude remains
constant or even increases, and that interference and setpoint changes must be
compensated quickly and without a long period of transient recovery.

4.1

Loop gain VO

For the reasons given above, an investigation of the behaviour of a control loop
becomes an investigation into its oscillatory characteristics. It is found that the stability
of the closed-loop control circuit with a P-controller depends on the degree of difficulty
S of the controlled system (= Tu/Tg) (Transient characteristic ) and the Transient
characteristic from the controller (Control equipment) on the loop gain VO.

79

The loop gain VO of a control loop with a P-controller is the product of the transfer
coefficient of the controller KR and the transfer coefficient of the controlled system KS:
VO = KR KS
The loop gain VO of a control loop with a P-controller is the product of the transfer
coefficient of the controller KP and the transfer coefficient of the controlled system KS:
VO = KR KP KS
The term loop gain will be explained in greater detail using a control loop that has been
opened (open control loop) between the P-controller and the controlled system (Fig.
4-2):
The loop gain VO is based on the concept that a change of the manipulated variable
that is made as e at the input of the controlled system will emerge as a at the output
of the controller after one pass through the control loop. In order for the control loop to
have the desired effect, a must have the opposite sign to e, because a correction of
a control difference can only take place if the controller initiates a change of the
controlled variable in the opposite direction to the deviation via the controlled system. It
should also be noted that the correction has a greater effect the greater a is for a
given e, or in other words: the greater the loop gain VO = a/e is made.

x
w
Fig. 4-2
1
2
e
a
x
y

KR
KS
VO

1
R
KR =

y
x

2
S
a

KS =

x
y

x
B34-2

Definition of loop gain VO

Controller (controlling equipment)

Controlled system
Change of input variable, manipulated variable
Change of output variable, manipulated variable
Controlled variable change
Manipulated variable change
Transfer coefficient of the P-controller
Transfer coefficient of the controlled system
Loop gain

If the sign reversal is ignored, these processes can be described using the following
equations:

An abrupt manipulated variable change e is produced manually at the input of the

controlled system. This manipulated variable change gives rise in the controlled
system to a change in the controlled variable of
x = KS e

The change in the controlled variable x gives rise in the P-controller to an output
signal a (manipulated variable change) of
a = KR x

(1)

(2)

Replacing x in the formula (2) by KS e (1) gives rise to

a = KR KS e
which results in
a/e = KR KS = VO

The dimensions of KR and KS must always be used in the formulae in such a way that
VO becomes dimensionless. If the manipulated variable is indicated in mm, for example,
the resultant units of KR are mm/K and the units of KS are K/mm.

80

For a given controlled system, therefore, the loop gain VO is decisive for the control
quality. Since the transfer coefficient KS of the system can be considered as given, the
necessary loop gain VO is set simply by adjusting the transfer coefficient KR of the Pcontroller.

4.2

Stability investigations

As already known, the degree of difficulty S of the controlled system and the loop gain
VO are decisive for the oscillatory characteristics of a control loop.
In stability investigations, a distinction is made between the response to interference
and the response to setpoint changes:

The response to interference of a control loop is obtained by investigating the

reaction of the controlled variable x to a sudden interference or interference variable
change.

The response to setpoint changes of the control loop describes the reaction of the
controlled variable x to a setpoint step w.

4.2.1

Response to interference

If, without external intervention, the controlled variable in the control loop fails to
assume a constant value even after an infinite period of time since the occurrence of
the last interference (sustained oscillation), the term self-excitation is used, and the
control loop is said to be unstable. This state must be avoided at all costs by correct
adjustment of the controller.
In unstable control loops two different types of oscillation can be distinguished
periodically undamped oscillation and periodically excited oscillation:

In the case of periodically undamped oscillation, the amplitude and frequency of

the sustained oscillation occurring after occurrence of the interference z are
constant (Fig. 4-3).
The control element must not travel to either the upper or lower limit position.
The time required for one full oscillation is referred to as the period of oscillation TP.
The loop gain that gives rise to undamped oscillation in the control loop is referred
to as the critical loop gain Vokrit. The proportional band Xp and the transfer coefficient
KR KP of the P-controller that produce this undamped oscillation are referred to as Xp
krit. and KR KP krit. respectively.
TP

t
Fig. 4-3

Undamped oscillation

Tp Period of oscillation
z Interference

81

Periodically excited oscillation is characterised by a continuously increasing amplitude

of the sustained oscillation that occurs (Fig. 4-4). It occurs if the loop gain VO of the
control loop is set higher than its critical value.
x

t
Fig. 4-4
z

Excited oscillation

Interference

In practice, the amplitude of oscillation continues to increase as described until it is

limited by the end stops, e.g. valve fully open or valve closed.
In a stable control loop, oscillations of the controlled variable also occur after
interference, but their amplitude decreases over time until they disappear completely.
This is referred to as damped oscillation. A steady state, i.e. rest, is achieved
(Fig. 4-5).
x

t
Fig. 4-5
z

Damped oscillation

Interference

A steady state, and therefore rest, occurs if the loop gain VO of the control loop is set
lower than its critical value:

In order to achieve sufficient damping of a control process, a simple, generally

applied rule of thumb recommends that the loop gain VO should be set to half of the
critical loop gain.

In higher order controlled systems and in systems with dead time the loop gain VO
should not be too high, otherwise the control process will exhibit poor damping or
can even become unstable.

On the other hand, too low a loop gain makes the control loop sluggish and
imprecise.

In many cases, fast correction is desirable. Therefore, a certain amount of overshoot is

accepted. So there is always a trade-off between speed and overshoot.

82

4.2.2

Response to setpoint changes

It is the task of the controlling equipment to compensate for setpoint changes and
interference as quickly as possible. Undesirable oscillations must be avoided in the
process. However, this requirement depends greatly on the type of controlled plant.
For example, several oscillations are perfectly acceptable in a supply air temperature
control system. In a boiler plant on the other hand, an overshoot can cause the safety
thermostat to trip.
The response to an abrupt change of the setpoint w can have numerous, very different
forms. A selection are shown in Fig. 4-6.
w

a)
t
x

b)

1)
t
x

c)
t
x
2)

d)
t
x

e)
t
x

f)

3)
t
x

g)

t
B34-6

Fig. 4-6
a)
b)
c)
d)
e)
f)
g)
1)
2)
3)

Different transient responses

Setpoint step
super-critically damped, gradual approach without overshoot
critically damped, approach as fast as possible without overshoot
sub-critically damped, single overshoot
periodically damped, multiple, decreasing overshoots
periodically undamped, sustained oscillation
periodically excited, escalating oscillation
Too slow
Usable to good depending on plant, i.e. on criterion for the transient response
Stability limit

83

It has been found in practice that the most suitable setting values for controllers (shortest recovery time)
according to Ziegler and Nichols are based on the following process: after an abrupt setpoint change, the
controlled variable overshoots by about 20% during the transient response, and two changes of direction of the
manipulated variable and controlled variable occur before the steady state is achieved (Fig. 4-7).

This represents a sub-critically damped, periodic transient response.

It is suitable for all applications where a fast transient recovery without large or
prolonged overshooting is required.
x, w
120 %
100 %

t
Fig. 4-7

Optimal transient response based on Ziegler and Nichols

w Change of reference variable

x Controlled variable

For an aperiodic transient recovery process without overshooting of the controlled

variable (e.g. suitable for high-temperature boiler plants, where the controlled variable
must not overshoot because of the safety thermostat) setting rules have been
established by Chien, Hrones and Reswick. These will not be discussed in detail
here.

4.3

Setting rules according to Ziegler and

Nichols

A controller must be adapted to the controlled system such that, on the one hand,
deviations of the controlled variable x from the setpoint w are corrected as quickly and
precisely as possible, and that, on the other hand, the control loop cannot become
unstable in any operating state. This adaptation is performed by adjusting the following
controller setting values:
P-controller:

Proportional band Xp (or transfer coefficient KP):

PI-controller:

Proportional band Xp (KP), integral-action time Tn

PD-controller:

Proportional band Xp (KP), derivative-action time Tv

PID-controller:

Proportional band Xp (KP), integral-action time Tn and

derivative-action time Tv

The formulae for controller adjustment in the following provide standard values. Fine
tuning must be performed by observing the controlled variable or the control element
with an abrupt change in the reference variable.

84

4.3.1

Known controlled systems

Based on the controlled system data Tt, TS and KS or Xh, the following setting rules
according to Ziegler and Nichols apply for the optimal adaptation of the controller to a
single storage element system with dead time:
Controller

PI

PD

PID

Xp

Xh S

1,25 Xh S

0,83 Xh S

0,83 Xh S

Tn

3 Tt

Tv
KR
Fig. 4-8

T / (KS Tt)

0.9 T / (KS Tt)

2 Tt
0.25 Tt

0.42 Tt

1.2 T / (KS Tt)

1.2 T / (KS Tt)

Setting rules for known controlled system data

For higher order systems Tt and TS are replaced in the formulae by Tu and Tg in
practice.

4.3.2

Unknown controlled systems

In case of a controlled system with unknown dynamic response, the optimal adjustment
of the controller can be determined experimentally. In order to do so, the controller is
initially set to proportional action, or Tn is set as large as possible and Tv as small as
possible. After a reference variable change, the proportional band Xp is then reduced
until undamped oscillation only just occurs (see 4.2.1). The resultant proportional band
is Xp crit., and the time required for a full oscillation is referred to as the period of
oscillation Tp. According to experiments by Ziegler and Nichols, the resultant setting
guidelines for the various controller types are as follows:
Controller

PI

PD

PID

Xp

2 Xp crit.

2.2 Xp crit.

1.5 Xp crit.

1.7 Xp crit.

Tn

0.85 TP

Tv
Fig. 4-9

0.5 TP
0.1 TP

0.125 TP

Setting rules for unknown controlled system data

A disadvantage is the fact that, while the critical value Xp crit. is being determined,
unstable plant behavior must be tolerated for a certain length of time. Therefore, this
method is especially suitable if Tp is not longer than about 5 minutes. In case of a
longer period of oscillation, it is recommended that the control system data should be
determined by plotting the transient function, and then the procedure as per 4.3.1
should be applied.

4.4

Characteristics of different controllers on a

controlled system

4.4.1

P-controller and controlled system

As already known from the description of the P-controller (see 3.4), this controller is not
able to fully compensate for setpoint changes and interference variables. There always
remains a residual deviation (P-offset) which depends on the magnitude of the
respective setpoint change or interference variable, on the size of the selected P-band
Xp and on the position of the operating point. The controller is adapted to the controlled
system by adjusting the P-band Xp.

85

As shown in Fig. 4-10, while a P-controller can quickly compensate for a setpoint
change or for interference with a rapid and large adjustment of the control element, it
cannot eliminate it completely (setpoint not achieved). As a result of the proportional
relationship between the controlled variable and the manipulated variable y,
manipulated variable y1 will assume a different value, e.g. y2, but controlled variable x1
must also change proportionally and become x2. The difference between x1 and x2 is
the residual deviation or proportional offset eb.
In order to keep the residual deviation as small as possible, the P-band Xp would have
to be set as small as possible (large transfer coefficient KR of the controller).
This, however, means severe intervention of the controller in the control process, i.e.
the P-controller becomes very sensitive.

Therefore, a small P-band gives rise to a reduced residual deviation, but the control
loop tends to oscillate (unstable and, therefore, unusable control response).

On the other hand, a large P-band gives rise to greater stability, but also to
increased sluggishness and load-dependence, i.e. large residual deviation.

x2 =
w,x

w2

K P K S
1 + (K P K S )

w 2

x wb

x2

a)
w1

x1

xe
b =
wb
x

KS
1 + (K P K S )
w1
x2

b)
z

x wb

c)

y
y2
y1

t
B34-10

Fig. 4-10
a)
b)
c)
w1, w2
x1, x2
eb
y1, y2

86

Transient recovery processes with a P-controller

In case of a setpoint change (response to setpoint changes)

In case of interference z (response to interference)
Manipulated variable change for interference z
Old and new setpoint
Old and new value of the controlled variable
Residual P-offset
Old and new value of the manipulated variable

Therefore, the selection of the P-band Xp is always a trade-off between stability and
control precision. The most suitable value depends on the plant data: The greater the
ratio of the delay time Tu to the balancing time Tg (= degree of difficulty S) of a
controlled system the greater the transfer coefficient KS of the system will be, so the
larger the P-band required for stable control will be (see setting rules under 4.3).
The effects of interference or of a setpoint change in a control loop with a P-controller
are explained in more detail in the following (operating point of the P-controller under
load = 0%):

If the effect of interference z on the controlled variable x without control is

considered, the resultant controlled variable change xz2 is
xz = KS z

If this interference z is corrected by a P-controller, it is attenuated by the factor

1/1 +( KR KS). This factor is referred to as control factor R.
1
1 + (K P K S )

R=

The control factor R indicates the portion to which an interference z is attenuated by

the P-controller (x is the value of interference z that remains). Control factors of
approximately 0.1...0.01 are normally achieved.
x =

KS
z
1 + (K P K S )

For a selected setpoint w the resultant controlled variable x is

x=

KR K S
w
1 + (K P K S )

and the residual deviation is, therefore,

eb =

1 + (K P K S )

With a setpoint change w, the controlled variable x changes by the value

x =

KR K S
w
1 + (K P K S )

and the residual deviation of the setpoint change w is again

eb =

1
w
1 + (K P K S )

The P-controller is probably the most widespread continuous-action controller. It is very

undemanding in operation, and it can be adapted to a plant by simply adjusting the Pband Xp. Its load-dependence is often insignificant, e.g. in boiler control or converter
control, or it does not manifest itself because no major load variations occur in
operation.
It is especially the P-controller's fast action that is appreciated, because in many control
tasks a quick reaction to load changes is more important than precise control to
setpoint. However, the P-controller should not be used for fast, difficult controlled
systems, because the necessarily large setting of Xp would give rise to excessively
great load-dependent P-offsets.

87

4.4.2

I-controller and controlled system

From the description of the I-controller (see 3.5) it is known that a certain rate of
change of the manipulated variable is associated with each value of the control
difference in this controller. Therefore, a control loop with an I-controller has no residual
deviation, i.e. the manipulated variable y continues to be adjusted until the setpoint is
achieved again. The manipulated variable can assume any value of its positioning
range in the process. The I-controller's dynamic response can be optimised by
adjusting the positioning speed vy or the integral-action coefficient Kl to the controlled
system.
If, in a closed-loop control circuit, the controlled variable x drops under the effect of an
interference variable z (Fig. 4-11), the momentary rate of change of the manipulated
variable y is always proportional to the control difference e. In comparison with the Pcontroller, the valve positioning is noticeably slower. The consequence of this is that the
recovery time and maximum transient deviation are greater for the same controlled
system, i.e. the dynamic control quality is poorer. However, on completion of the
transient recovery process (the controlled variable and the control element can execute
several oscillations depending on controller setting) the setpoint is achieved again.
x

a)
z

b)

I
y2
y
y1

t
Fig. 4-11
a)
b)
b)
I
z

Transient recovery process of an I-controller in case of interference

Progression of the controlled variable

Progression of the manipulated variable
P-controller
l-controller
Interference

In spite of the advantage of load-independence, the I-controller is rarely used in

practice. Provided the integral-action time is selected correctly, the controller will control
any self-regulating system stably and precisely, but its recovery time is long. In the
case of non-self-regulating controlled systems, the I-controller oscillates no matter how
long the selected runtime of the control element is. Therefore, its use is restricted to
fast controlled systems with relatively slow load variations. In slow systems, the control
element runtime necessary for stabilisation is so long that the control action can no
longer be considered effective. However, a short runtime would give rise to ON/OFF
operation, because it takes too long for the effect of the control element adjustment to
work its way through the controlled system.

88

4.4.3

PI-controller and controlled system

The PI-controller combines the advantages of the P-controller (fast) with those of the Icontroller (load-independent). If a control difference occurs, it is initially corrected
quickly and roughly by the P-component. The integral-action component does not take
effect in the control process until afterwards. This component brings the controlled
variable back to the setpoint (Fig. 4-12). So the manipulated variable y continues to be
adjusted until it assumes a value that is sufficient to correct the control difference, i.e.
until the setpoint is achieved. During this process, the rate of change of the
manipulated variable is proportional to the remaining deviation. It decreases markedly
as the controlled variable approaches the setpoint.
x

z
w
z
PI
P
z
TZ

Fig. 4-12
Pl
P
z

t
B34-12

Setpoint step and interference z corrected by a PI and a P-controller

Pl controller
P-controller (operating point corrected)
Interference

The PI-controller is adapted to the controlled system by adjusting the proportional band
Xp and the integral-action time Tn. The transfer coefficient KS and the degree of difficulty
S of the controlled system are decisive for determining the most suitable P-band,
whereas the delay time Tu is the only decisive variable for the integral-action time Tn
(setting instructions, see 4.3). The following points should be noted:

Reduction of the P-band Xp and/or the integral-action time Tn beyond their optimal
values gives rise to unstable, i.e. unusable, control action.

The larger the P-band Xp that is selected the more the PI-controller exhibits the
behaviour of an I-controller, i.e. the control process becomes much to slow because
it takes much too long for the I-component to correct the excessively large P-offset.

The longer the integral-action time Tn that is selected the more the PI-controller
exhibits the behaviour of a P-controller, i.e. the P-offset remains present far too long
(it takes too long for the I-component to correct the optimal P-offset).

In case of unknown controlled system data, it is recommended that the PI-controller

should be set to the correct value of the P-band Xp while the integral-action time Tn
is at its longest setting, and that the integral-action time Tn should not be adjusted
until afterwards.

As already mentioned under 3.9.2, changing the P-band Xp has a much greater
effect than changing the integral-action time Tn, because the change in Xp not only
gives rise to a different P-adjustment but also to different I-adjustment. Changing Tn,
however, only gives rise to a change in the I-adjustment while the P-adjustment
remains the same.

As shown in Fig. 4-12, the PI-controller is slightly slower than a P-controller for the
same task. It is used if the use of a P-controller would give rise to too great a loaddependence because of the necessary Xp setting, i.e. when precise, load-independent
maintenance of the setpoint is required.

89

4.4.4

PID-controller and controlled system

As already mentioned in the description of the D-element, the rapid and large
adjustment of the control element in case of a control difference has a beneficial effect
in terms of a short recovery time. Therefore, PID controllers are mainly used if the
controlled system has a high degree of difficulty S (relatively long delay time) and the
controller must correct fast-acting interference variables.
Due to the D-element, not only the magnitude of the control difference is detected but
also its rate of change. For this reason, a PID-controller compensates for setpoint
changes and especially for interference more rapidly then a PI or P-controller (Fig. 413). Due to the I-component, it also has no residual deviation.
x
z
w
z
PID
PI
z
TZ
Fig. 4-13

t
B34-13

Setpoint step and interference z corrected by a PID and a PI-controller

Correct adaptation of the PID-controller to a controlled system is relatively difficult,

since the transfer coefficient KR KP of the P-component (or the P-band Xp), the integralaction time Tn of the I-component and the derivative-action time Tv of the D-element
must be set correctly. Appropriate instructions are given in the setting rules under 4.3.
The following additional points should also be noted:

90

An unstable, i.e. unusable, control action is caused by reducing the P-band Xp

and/or the integral-action time Tn and/or by increasing the derivative-action time Tv
beyond their optimal values.

If excessively large values of Xp and/or Tn are selected, the control action will be
stable, but it will take too long to compensate for a setpoint change or interference.

If too small a value is set for the derivative-action time Tv, the D-element can only
take insufficient effect, i.e. the controller is initially too slow (insufficiently large
counteraction of the control difference that has occurred).

As already known, changing the P-band Xp has a much greater effect than changing
the integral-action time Tn or the derivative-action time Tv, because the change in Xp
not only gives rise to a different P-adjustment but also to different I and Dadjustments. Changing Tn or Tv, however, only gives rise to a change in the Iadjustment or D-adjustment while the P-adjustment remains the same.

4.4.5

Two-position controller and controlled system

In the case of a two-position controller, the manipulated variable can only assume one
of two values, which are normally y = 0% and y = 100% (OFF/ON). This controller is
characterised by a defined switching differential SD between the ON and OFF
positions, which gives rise to a periodic variation of the controlled variable x (Fig. 4-14).
x
0
w = 80 %
1

SD

0
0
w = 50 %
1

Xh

0
w = 20 %
1

t
y
1
0

t
B34-14

Fig. 4-14

Fluctuation of the controlled variable, and control factor at several different

Setpoints (load)

The term control precision cannot be used here in the same sense as with the
continuous-action controllers. The fluctuations of the controlled variable x become
greater:

the greater the controller's switching differential SD is,

the more difficult the controlled system (i.e. the greater the delay time Tu) is, and

the greater the positioning range Xh (in units of the controlled variable) is.

Due to the fluctuations of the controlled variable, its mean value is generally not located
in the middle of the switching differential. This is only the case with equally long switchon and switch-off impulses ( = 0.5) or with a delay-free controlled system and SD = 0.
However, oscillation of the controlled variable also occurs with a switching differential
SD = 0 if the controlled system has a delay time Tu.
In order to reduce the control fluctuations in systems with large load changes, twostage or multistage two-position controllers can be used.
The prerequisite for this is, of course, that the manipulated variable (energy supply) can
be divided into the appropriate number of power levels.
The two-position controller without feedback can be encountered in numerous
applications in actual practice, because it can fulfil certain control tasks satisfactorily at
relatively modest expense. In order to decide whether the operating principle can fulfil
the requirements of the control task, two properties of the control fluctuation must be
taken into consideration: the maximum fluctuation of the controlled variable and the
oscillation frequency. It is conceivable that the use of a two-position controller may be
out of the question because the maximum fluctuation is too large, so the control loop
operates with insufficient precision. It could also be the case that, although the
precision is sufficient, the oscillation frequency is too high to guarantee an acceptable
device life for practical use.

91

Digital Control

5.1

Introduction

Controllers are today increasingly developed with digital electronics: the control function
is executed by a microcomputer according to a computer program (controller software).
A comparison of the quality of the classic control functions (P, PI or PID-controller) in
digital and analog controllers shows that a digital controller is not actually better. So
why use digital controllers?
The use of digital technology offers very different advantages:

It provides for the combination of open-loop, closed-loop and optimisation functions

in fairly simple hardware.

It provides for the integration of positioning algorithms (two-position pulse width

controlled, three-position, step switch etc.) into the controller product.

Sharing of sensors for various functions

New control functions (e.g. adaptive controllers with self-learning setting values)

Unlike analog electronics, a process computer can control multiple control loops
simultaneously (or more accurately: quasi-simultaneously).

A higher integration density of functions in a product saves money (compared to a

number of simpler products and their interconnecting wiring).

Improvement of the user dialog by integration of a digital user interface (display and
operating panel) on the device

Communication via network, modem etc. provides for further energy and cost
savings (functional optimisation in interconnected systems, remote operation,
remote diagnostics).

Although current digital products are still referred to as controllers, they often actually
include a whole range of control and monitoring functions in a single device: they are
process units (which applies both to standard controllers and to universal controllers).

5.2

Principle of digital control

Let us assume that the operating environment of the controlling equipment is the same
in the case of a digital and an analog controller:

Same sensors (analog operation)

Same actuators

The output signals of the sensors used in building services installations are usually
analog resistance, voltage or current signals. These can be processed directly by
analog controllers. In the case of digital controllers, however, they must first be
converted into a form that can be understood by microprocessors, i.e. into digital
signals. This is done using an analog-to-digital converter (A/D converter).
In the same way, the digital signals of the microprocessor must be converted to analog
signals so that they can be processed by the plant (e.g. by actuators). This is done
using a digital-to-analog converter (D/A converter).
Therefore, in addition to a microcomputer (including the entire controller function in the
form of a computer program), a digital controller needs an analog-to-digital converter to
acquire the measured value of the controlled variable and, in the case of a continuousaction controller, also a digital-to-analog converter to output the manipulated variable to
the actuator. With this peripheral equipment, the microcomputer becomes a process
computer. It executes the following actions periodically (Fig. 5-1):

92

w
xAnalog

xDigital

yDigital D

yAnalog
A

1
2
3
4

B35-1

Fig. 5-1
A/D
C
D/A
1
2
3
TS
4

Sampling period TS

TS

TS

Block diagram and periodic actions of a digital controller

Analog-to-digital converter
Microcomputer (process computer)
Digital-to-analog converter
Measuring (A/D conversion)
Calculation (control algorithm)
Control output (D/A conversion or digital output (in the case of 3-position control
e.g. open / stop / close)
Sampling period of the digital controller
The C is available for other tasks, or it waits for the beginning of the next sampling period.

The execution of the measuring 1, calculation 2 and positioning 3 actions is performed

sequentially and is repeated at intervals of TS. This is referred to as time-discrete
control. If a sufficiently small value of the sampling period TS is selected, the
characteristics of a digital controller become almost the same as those of a continuousaction controller.
Measured values are stored for processing after each A/D conversion. They remain
unchanged until the next conversion.
Each control output is maintained until the next value is issued.

5.3

Universal digital controller

Universal controllers for HVAC tasks are configurable or programmable digital

controllers. They are designed to be able to execute multiple control tasks in a single
process unit.
Unlike analog controllers, where each control loop is assigned to one controller, in
digital control a process computer can support more than one control loop. It processes
them sequentially and periodically.
Whereas in the analog controller the response characteristic (P, PI, PID) is realised via
an electronic circuit, in digital control it is realised via an appropriate computer program
the so-called control algorithm.
In conventional control systems, the open-loop and closed-loop control functions
required for plant operation are implemented by wiring to other components, such as
timing and auxiliary relays, enabling, interlocking and monitoring contacts etc. (wiredprogram controllers).
In the case of configurable, universal digital controllers, the control functions are
entered in a processing list. This list contains the necessary function block instances,
data links (database and function block interfaces) and setting values (parameters).
The actual executable functions are stored as procedures in the program memory. The
process computer processes the processing list periodically. This approach allows even
complex control tasks to be accomplished with relatively simple means. However,
certain safety functions (e.g. overheating protection) must still be configured by
conventional wiring technology so that they cannot be influenced by the universal
controller.

93

A
6

y1...yn

xn..x1
10

10

8
B35-2

Fig. 5-2

Signalflow in a universal digital controller

P Microprocessor
1 Program memory
2 Database (data memory for configuration, measured values, setting values, intermediate results,
Manipulated variables etc.)
3 Input/output controller
4 Input multiplexer
5 Measuring transducer
6 Output multiplexer
7 Holding element
8 Actuator 1 n
9 Controlled system 1...n
10 Sensor 1 n

Fig. 5-2 shows the signal flow and operating principle of a universal digital controller.
The example shows a controller that operates a number of heating groups, for
example. The analog measured variable x1to be acquired for the first controlled system
is routed to the A/D converter by the input collector (input multiplexer). The digital
output signal of the A/D converter now passes via the process computer to the
database. From there it is compared with reference variable w1. The control algorithm
assigned to this controlled system now calculates the value of the manipulated variable
y1. This digital value is converted into a corresponding analog signal by the D/A
converter and is routed via the output distributor (output multiplexer) to a holding
element. The holding element passes the analog value of the manipulated variable y1 to
the control element of the first controlled system. The holding element stores the value
until, after the other controlled systems have been processed, it is again the turn of the
first controlled system (1 cycle time).
With this approach, of course, the input and output multiplexers, measured values and
setting values, processing, manipulated variable and holding element must operate
synchronously according to a plan so that the correct controlling equipment with the
correct data is really assigned to the respective controlled system.

94

5.3.1

Example of a universal digital controller

Depending on the control system, the signal converters of the inputs and outputs are
either incorporated in the base unit or in separate devices so-called I/O modules
(input/output modules). The I/O modules are connected to the process unit via a bus
(Fig. 5-3). They convert plant data into digital signals for the process unit or digital
information from the process unit into analog or logical signals for output to the plant.
Some I/O modules also contain external control functions. Example: continuous/threeposition output (positioning algorithm for three-position actuators).

T1
AC 24 V

N1

U1

Island

bus

U4

M1

M2

M3

10562z149

U3

U2

Fig. 5-3
T1
N1
U1
U2
U3
U4
M1
M2, M3
___

Example of a universal digital controller: Desigo control and monitoring system

Transformer AC 230 V / AC 24 V
Automation station
Power supply module
I/O row with outgoing connections to devices located inside the control panel
Bus connection module
I/O row with connections to the external field devices
Heating coil pump
Supply air fan, extract air fan
Controls cabinet

95

5.4
Minimum actuating step

Digital controller with three-position actuator

Each control output of a digital controller is maintained unchanged throughout the entire
sampling period TS. In case of three-position output (open / stop / close), therefore, the
sampling period determines the smallest increment that a three-position actuator can
travel.
The smallest possible actuating step of an actuator should be so small that the variation
of the controlled variable that it gives rise to is not much greater than half the dead
zone.

Dead zone xDZ

The dead zone is the bandwidth of a minimum control difference (emin) to which a
controller must not yet react. With regard to the symbol xDZ : DZ = dead zone.
Example for a temperature controller: xDZ = 0.2 K (emin = 0.1 K)

Step oscillation

If the smallest possible actuating step causes the control difference to vary by more
than the dead zone, this can give rise to so-called step oscillation in the steady-state
condition of the controlled variable. The control difference changes its sign after each
actuating step, and its value becomes so large that the controller continually has to
compensate (Fig. 5-4).

t
TS

TS

xDZ

emin
0

-emin
Fig. 5-4

Step oscillation of a controller with three-position actuator

Step oscillation gives rise to an unnecessarily large number of start-up cycles of the
actuator. This should be avoided because it can reduce the actuator's service life.
The factory settings of standard controllers are such that step oscillation will rarely
occur. If step oscillation occurs (e.g. in the case of customised solutions with universal
controllers), the following action should be considered:

96

Reduce the sampling period TS (if adjustable)

Increase the controller's dead zone xDZ or its responsiveness emin (if adjustable)

Select a greater runtime of the actuator

Select continuous manipulated variable output (e.g. 0...100%) and use a special I/O
module with the continuous/three-position converter function to control the threeposition actuator.

Terms and Symbols

6.1

List of terms

Actual value xi

Value of the controlled variable x that is currently measured by the sensor

Actuator

This device adjusts the control element in the assigned direction in accordance with the
controller's output signal (e.g. electric motor, solenoid)

Actuating device

This device adjusts the control element in the assigned direction in accordance with the
controller's output signal (e.g. electric motor, solenoid)

Adjustment range of the

controlled variable Xh

Change in the controlled variable when the manipulated variable y changes through
0...100% (in units of the controlled variable)

Adjustment range of the

manipulated variable Yh

Range of possible manipulated variable change:

General 0...1 or 0...100%

Specific e.g. 0...20 mm

Balancing time Tg

Characteristic quantity of a multiple storage element system

Block diagram

General view of control loop elements or of the control loop. This form of presentation
enables the functions and data flow to be explained simply (without physical variables).

Control difference e

Difference between the reference variable w and the controlled variable x in units of the
controlled variable:
e=wx

Controlling equipment,
closed-loop

Equipment whose task it is to influence the (closed-loop) controlled system. It begins at

the measuring point and ends at the manipulation point. It consists of the sensor,
controller and actuator.

Controlled variable x

The physical variable to be acquired (measured) in the controlled system and to be

kept at a desired quantity or value, i.e. to be controlled (temperature, humidity etc.) It is
the output variable of the controlled system and the input variable of the controller.

Control characteristic

Graphical presentation of the steady-state relationship between the controlled variable

x and the manipulated variable y

Control loop

Combination of the controlled system and the controller with a closed chain of action

Closed-loop control

Process in which the variable to be controlled (controlled variable) is continuously

measured, compared with the reference variable and influenced according to the result
of that comparison in the sense of adjustment to the reference variable. The resultant
chain of action takes place in a closed circuit the control loop.

Controlled system,
closed-loop

The plant to be controlled, i.e. the part of the control loop in which the controlled
variable x is to be kept constant by the controller in spite of unintentional interference. It
begins at the manipulation point (where the control element acts) and ends at the
measuring point (where the controlled variable is measured).

Input variable: manipulated variable y

Output variable: controlled variable x

97

Closed-loop control
system

Its purpose is to bring a physical variable (controlled variable x) to a specified value

(reference variable w) and to keep it at that value in spite of any disturbing influences.
In order to do so, the controlled variable x must be continually acquired, compared with
the reference variable w and influenced in the sense of adjustment to the reference
variable.

Controlled system,
open-loop
Controlling equipment,
open-loop
Control variable

The part of the plant that it is the task of the (open-loop) control to influence

Controller, closed-loop

Complete equipment that brings about the control process in the controlled system, i.e.
that measures the difference between the setpoint and actual value of the controlled
variable and then actuates the control element appropriately in order to eliminate the
deviation.

Equipment whose task it is to influence the (open-loop) controlled system

The physical variable in the (open-loop) controlled system that is influenced by the
open-loop control (temperature, humidity etc.)

Input variable: controlled variable x

Output variable: manipulated variable y

Control valve

Combination of actuator and control element

Control element

Mechanism installed in the control loop to control an energy or mass flow (e.g. a valve)

Dead time Tt

Characteristic quantity of a single storage element controlled system (travel time)

Degree of difficulty S

Ratio of the dead time Tt to the time constant TS, or of the delay time Tu to the
balancing time Tg:
S = Tt/TS or Tu/Tg

Dead zone xDZ

Range within which a control difference (emin) has no effect on the response
characteristic. The boundaries of the dead zone (emin) represent the responsiveness.

Delay time Tu

Characteristic quantity of a multiple storage element system

Interference variable z

Variable acting on the control loop from the outside which unintentionally influences the
controlled variable, e.g. external heat sources, solar radiation etc.

Manipulated variable y

Variable that can be adjusted by the controller, which in turn intentionally influences the
value of the controlled variable (e.g. valve stroke). It is the output variable of the
controller and the input variable of the controlled system at the same time.

Manipulation point

Place where the energy flow is influenced

Measuring point

Place where the sensor is located, i.e. where the controlled variable is measured

Open-loop control

In an open chain of action, one or more input variables influence (without feedback)
other variables the output variables based on the specific regularity of the system

Open-loop control
system

In an open chain of action, one or more input variables influence (without feedback)
other variables the output variables based on the specific regularity of the system

Response, dynamic

Relationship between the output variable change and the input variable change over
time

98

Response, steady-state

Relationship between the output variable change and the input variable change in the
steady-state condition

Range of reference
variables Wh
Responsiveness emin

Generally corresponds to the scale range of the setpoint adjuster

As long as the control difference remains within the boundaries of a controller's
responsiveness -emin...+emin, it does not affect the controller's response characteristic.
emin corresponds to half the dead zone xDZ .

Ramp response

Progression of the output signal over time as a result of a ramp function at the input

Reference variable w

Variable applied to the control loop from the outside. It determines the current setpoint.
A distinction is made between:

Fixed setpoint control reference variable is constant and identical to the setpoint

Schedule control

time-dependent setpoint

Slave control

the control setpoint is determined by another measured

variable

Switching differential SD

Distance between the switch-on point and the switch-off point of the controlled variable
of a two-position controller or switch

Sensor

Device that measures the value of the controlled variable

Setpoint w

Currently required value of the controlled variable x which should be kept constant in
spite of interference variables (e.g. setting value of the setpoint adjuster)

Step response

Progression of the output signal over time as a result of a step function at the input

Step function

Abrupt change of the input signal by any amount

Transient function

Progression of the output signal over time as a result of a unit step at the input

Transient behaviour

This describes the progression over time of the output signal if characteristic temporal
progressions of the input signal are applied

Transfer coefficient K

Ratio of the output variable change to the input variable change:

For the controller:

KR = y/x

For the controlled system:

KS = x/y

Time constant TS

Characteristic quantity of a single storage element controlled system

Temporal behaviour

Shows how an output signal follows a variable input signal over time

99

6.2

List of symbols

Semantically undefined symbols

a
e

output signal
input signal

Semantically defined symbols

e
emin
KD
KI
KP
KR
KS
Q
R
S
SD
T
t
T0
TD
Tg
Tl
Tn
TP
TS
TS
Tt
Tu
Tv
Ty
V&
VO
vX
vy
w
Wh
x
xDX
Xh
Xp
xwb
y
Yh
z

100

control difference (= w - x)
responsiveness of the controller
derivative-action coefficient
integral-action coefficient
transfer coefficient of the P-controller
transfer coefficient of the complete controller (controlling equipment)
transfer coefficient of the controlled system
Load
control factor
degree of difficulty
switching differential
time constant (general)
time (current)
start time
derivative-action time constant
balancing time
integral-action time constant
integral-action time
period of oscillation
system time constant
sampling period
dead time
delay time
derivative-action time
positioning time of the actuator
volume, mass flow
loop gain
rate of change of the controlled variable
positioning speed
reference variable
range of reference variables, adjustment range
controlled variable
dead zone
adjustment range of the controlled variable (control range)
proportional band
residual P-offset
manipulated variable
adjustment range of the manipulated variable
interference variable
difference
control factor
time constant

More technical brochures

Order no.
Titel

0-91899-de
0-91899-en

Das h,x-Diagramm
The psychrometric chart

0-91910-de
0-91910-en

Messtechnik
Measuring technology

0-91911-de
0-91911-en

Regeln und Steuern von Heizungsanlagen

Control of heating plants

0-91912-de
0-91912-en

Regeln und Steuern von Lftungs-/Klimaanlagen

Control of ventilation and air conditioning plants

0-91913-de
0-91913-en

Regeltechnik
Control technology

0-91914-de
0-91914-en

Kltetechnik
Refrigeration technology

0-91915-de
0-91915-en

Wrmerckgewinnung im Kltekreislauf
Heat recovery in the refrigeration

0-91916-de
0-91916-en

Einfhrung in die HLK- und Gebudetechnik

Introduction to building technology

0-91917-de
0-91917-en

Hydraulik in der Gebudetechnik

Hydraulics in building systems

0-91918-de
0-91918-en

Stetige Leistungsregelung im Kltekreislauf

Modulating capacity control in the refrigeration cycle

0-92166-de
0-92166-en

Bedarfsgeregelte Lftung
Demand-controlled ventilation

0-92189-de
0-92189-en

Gebudeautomation Einfluss auf die Energieeffizienz

Building automation impact on energy efficiency

101

Siemens Switzerland Ltd

Infrastructure & Cities Sector
Building Technologies Division
International Headquarters
Gubelstrasse 22
6301 Zug
Switzerland
Tel +41 41 724 24 24
Siemens Building Technologies
Infrastructure & Cities Sector
Brunel House
Sir William Siemens Square, Frimley
Camberley
Surrey, GU16 8QD
United Kingdom
Tel +44 1276 696000
Siemens Ltd
Infrastructure & Cities Sector
Building Technologies Division
22/F, Two Landmark East
100 How Ming Street, Kwun Tong
Kowloon, Hong Kong
Tel +852 2870 7888

The information in this document contains general descriptions of technical options available,
which do not always have to be present in individual cases. The required features should therefore
be specified in each individual case at the time of closing the contract.
Siemens Switzerland Ltd, 2013 Order no. 0-91913-en 11301

Answers for infrastructure.

Our world is undergoing changes that force us to think
in new ways: demographic change, urbanization, global
warming and resource shortages. Maximum efficiency
has top priority and not only where energy is concerned.
In addition, we need to increase comfort for the well-being
of users. Also, our need for safety and security is constantly
growing. For our customers, success is defined by how
well they manage these challenges. Siemens has the
answers.

www.siemens.com/buildingtechnologies

We are the trusted technology partner for

energy-efficient, safe and secure buildings and
infrastructure.