Closed Loop Control Systems
To enhance your understanding of the control systems to be discussed after this presenta-
tion, an introduction to the concepts of modern closed loop control systems may be of
help. Application of these control systems is all around us. Examples close to home in-
clude thermal controls for home heating and air-conditioning, speed controls for our cars,
and aircraft auto pilot systems. Each of these control systems can be characterized as a
closed loop control system, having a plant (the controlled device), controlling parameters,
and measured parameters. In electrical terms, the measured parameter is fed back to the
plant, compared to a setpoint, and the plant responds to any error by adjusting the control-
ling parameter.
Figure 1: Diagram of a Typical System
Figure 2: Closed Loop Control System
I. CLOSED LOOP CONTROL SYSTEMS
In application to electric power systems, the most common examples of closed loop con-
trol systems are speed governors and automatic voltage regulators (avr). In each case, the
measured parameter, the plant, and the controlling parameter can be identified. It may be
relatively easy to understand the operation of such control systems, however to precisely
calculate responses of such a control system, simulating the action of the control and
experimenting with its design, requires very sophisticated mathematical techniques. We’ll
not try to teach these techniques, as it is no small task, but we want you to recognize when
such tools may be useful and be able to recognize when someone is using them.
1
System Plant Controller Measured Controlled
Home Heating Gas Furnace Thermostat Temperature Gas Valve
Oven Electric Element Thermostat Temperature Switch
Auto speed Engine Speed Controller Drive shaft speed Fuel injectors
Diesel Genset Engine Governor Shaft speed Fuel injectors
Diesel Genset Alternator AVR Alternator Volts Excitation current
Figure 3: Some Typical Closed Loop Control Systems
Examples
In closed loop systems, there are several general types of controls. The simplest is the on-
off response system typical of temperature controls. When the thermostat detects tempera-
ture is too low, heat is turned on until temperature is restored. If temperature gets too
warm, the cooling system turns on. A dead band of acceptable temperature between the
two setpoints will prevent both from running and reduce the cycling from heating to cool-
ing and back again. The second type is continuous control such as the auto speed control,
speed governor, or voltage regulator. This type of control system is capable of much more
precise control than the on-off type of control, with stable operation and good transient
response.
Figure 4: Balanced Beam, Gain of 1
Figure 5: Balanced Beam, Gain of 2.33
2
Figure 6: Balanced Beam, Gain of 0.54
What Is Gain?
As we look at the concepts of closed loop control, two terms will be used repeatedly. The
first is gain. Gain is defined as the change in output divided by the change in input. For
example, if a balance beam is in balance and one forces the right hand side to be dis-
placed by one inch, the displacement that results on the left side can be divided by the
right hand displacement to obtain the gain of this simple mechanical system. A look at the
illustrations shows the differences in mechanical gain achieved by changing the balance
point from the midpoint to 30% from the left side and to 30% from the right side. These
same concepts are used in electronics, but gains are not uncommonly in the hundreds,
thousands, even hundreds of thousands in some applications.
Figure 7: Operational Amplifier Model
3
Figure 8: Operational Amplifier with Gain of 470
Figure 9: Operational Amplifier with Gain of 10
4
II. WHAT IS TIME CONSTANT?
Time constant refers to the time of response of output compared to input. Time constants
in electrical circuits may be the result of limitations in a device (an amplifier, for example) to
change its output in response to a sudden change in input. Or, the time constant may be
intentional using some combination of linear circuit elements (resistor, capacitor, inductor)
to create a desired time delay. By judicious use of circuit elements, filtering, frequency
selection, integration, differentiation, and time delays are implemented in analog circuit
designs. When we are referring to time constants, the reference may be to a resistor used
with a capacitor (RC network) or a resistor used with an inductor (RL network). Essential to
any time constant, therefore, is some inductor or capacitor, or both. The first diagram
shows only resistor loading that does not produce any kind of time delay. The second
figure shows an RC circuit, and the third shows the kind of voltage that would be mea-
sured on the capacitor if the switch is closed, and voltage is applied to the series RC cir-
cuit. The inductive series RL circuit results in a similar curve for the CURRENT through the
inductor. Using R, L, and C elements in various series and parallel configurations is one of
the tools of the electronics designer, used in creating the closed loop control systems that
we are discussing.
Figure 10: Resistive Circuit
5
Figure 11: Resistive Circuit Network Time Constant
Proportional, Integral and Derivative Control
With a continuous type of control, it is possible to design for several different kinds of
response. The proportional response of a control system responds to a step change in its
measured parameter with a step change in its output to the controlled plant, with a magni-
tude dependent on the error. The relationship between the error magnitude and the re-
sponse magnitude is the gain of the proportional response.
Figure 12: Capacitor Voltage in Resistive Circuit
6
Figure 13: Proportional Controller
The integral response of a continuous control system is one that continuously changes in
the direction to reduce the error until the error is restored to zero.
Figure 14: Integral Controller
7
Figure 15: Derivative Controller
The derivative response measures the speed of the change in the measured parameter
and causes an exponentially decaying output in the direction to reduce the error to zero.
Continuously acting control systems can use any or all of these characteristics in combina-
tion to provide the correct response to any control problem, using the adjustment of inte-
gral and derivative time constants and gains for each. By understanding the controlled
plant and its control characteristics, it is possible to model the plant and create a computer
model of the control system. By adjusting the 5 parameters of the model, a good system
performance can be obtained. Then the model can be implemented in hardware using
analog or digital electronics. Some adjustment will probably be necessary due to some
inaccuracy in the modeling process, but the results should be close to the model’s predic-
tions.
Figure 16: Add All Three - PID Controller
8
Why is PID Control Complex?
Complexity of the simulation math is caused by non-linearity in the characteristics of the
plant. This means that graphic representation of input vs. output is not a straight line. In
math, it means that equations to describe the characteristic are differential equations
requiring an understanding of calculus to work with them. In addition, the plant is often a
time-varying system, affected from moment to moment by temperature, pressure, and
operating conditions. These two characteristics raise the level of complexity and require
powerful computer calculation capability in order to solve the resulting equations to obtain
realistic models.
Figure 17: Generator Control Characteristics
That is what closed loop control systems need if we are to be able to predict performance.
Models that perform as closely as possible to the real world performance of the system
give us tools to study the effects of changes in the closed loop control system and evalu-
ate performance results. And what performance are we looking for? First, the control loop
must be stable under steady state conditions. If the control system hunts continuously,
control action is not acceptable. Second, the control loop must be stable under a transient
change. For a step change, large and small, in the measured parameter, the control action
must restore the measured value to the desired setpoint with a minimum of hunting occur-
ring during the recovery. Finally, the control loop should operate as quickly as possible to
a step change, large and small, to restore the measured value to the desired setpoint.
9
Figure 18: Develop Math Model
The tools put to work on the control loop problem are limits of controlled parameter, gains,
and time constants. The control loops that we will discuss include a reference setpoint,
commanding the level of operation of the measured parameter. This setpoint will be com-
pared with the measured parameter and an error signal will result. The error signal is used
to increase or decrease the controlling parameter in the proper direction to restore the
measured parameter to the setpoint and decrease the error signal to zero. The objective is
to reach a state of balance between the setpoint and the measured parameter with the
error signal at zero. Any disturbance to the system that causes the measured parameter to
deviate from the setpoint or any change in the setpoint introduces an unbalance and
triggers a reaction from the controlled parameter to restore the balance.
Figure 19: Define Math Model for Controller and Plant
A model of the plant must also be prepared describing mathematically the response char-
acteristics of the plant to the controlled input(s). A very simplified model of a generator and
its excitation system follows.
10
Figure 20: Simple Math Model
Perhaps the complexity of the control loop would be simplified if the accuracy of the con-
trol would be 10% of the setpoint. Another simplification could be achieved if the recovery
time of the control loop was allowed to be up to 5 minutes. Our modern expectation for
performance is not so loose as this, however, with expectations of control accuracy within
1/2% or 1/4% and response times measured in parts of a second. Achieving these levels of
performance requires continuously acting controls with high gain and time constants as
fast as possible.
As control systems are presented, the key to understanding them lies in recognizing how
they are described, in terms of accuracy and response time, and how they are calibrated
for setpoint, response time, and stability. Each controlled system has variations in its
response that are often determined by the size and configuration of the equipment. For
example, the control system for a 100 kW generator contrasted with a 100 Mw generator
will have some similarity, but many differences. Control of a 120 rpm hydro turbine speed
will have similarity to, but many differences from, an 1800 rpm diesel engine.
If you have any questions or need Basler Electric Headquarters Basler Electric International
additional information, please contact Route 143, Box 269, P.A.E. Les Pins, 67319
Basler Electric Company. Highland Illinois USA 62249 Wasselonne Cedex FRANCE
Our web site is located at: Phone 618/654-2341 Phone (33-3-88) 87-1010
http://www.basler.com Fax 618-654-2351 Fax (33-3-88) 87-0808
e-mail: info@basler.com
11
