LECTURE NOTES ON EEE 502: RELIABILITY AND MAINTENABILITY OF SYSTEMS

1.1. RELIABILITY
1.2. Explaining of some terms related to reliability
1.2.1. An item
1.2.2. Failure
1.2.2.1. Classification of Failure
Cause
Timing
Degree
Combination
1.2.2.2. Failure Rate
1.2.2.3. Causes of component/Equipment Failure
A. Classification of causes of component/Equipment Failure
1. Environmental
• Temperature
• Heat
• Humidity
• Mechanical Vibrations
• Atmospheric Pressure
• Air impurities
• Electromagnetic Radiation
• Electrostatics
2. Operating Stresses
• Voltage Surge
• Current Surge
• Frequency Change
B. Failure in Electronic Components
• Thermionic Valves
• Fixed Resistors
• Variable Resistors
• Capacitors
• Variable Capacitors
• Inductors and Transformers
• Semi-conductor Devices
• Relays
C. General Approaches in Reducing Equipment/Components Failure
o Part selection
o Use of improved method and New Technology
o Use of redundancy
o Use of improved design methods
o Use of Derating concepts

1
 1.3 MTBF
• Replacement method
• Non-replacement method
1.4 MTTF
1.5 Reliability
1.6 Mathematical inter-relationship of terms
1.6.1 Reliability and Failure Rate
1.6.2 MTBF, MTTF and Failure Rate
1.7 Reliability Equations and Curves when Failure rate is constant.
• Reliability and Unreliability Equations and Curves
• Survivor and failure equations and curves
2. Concept of Reliability Prediction
2.1. Introduction
2.2. Laws of Probability Relevant to Reliability Prediction
2.2.1. Multiplication Rule
2.2.2. addition Rule
2.2.3. Binomial
2.3. The Reliability, MTBF of a Series System
2.3.1. Reliability of Series System
2.3.2. MTBF of a Series System
2.4. The Reliability, MTBF of a parallel System
2.4.1. Reliability of Parallel System
2.4.2. MTBF of a parallel System
2.5. The Reliability, MTBF of a Series-parallel System
2.5.1. Reliability of Series-Parallel System
2.5.2. MTBF of a series-parallel System
3. Redundancy
3.1. Definition, Significance and Limitations of Redundancy
3.2. Classification
3.2.1. Active
3.2.2. Passive or standby
3.3. Mathematical basis of Redundancy
3.3.1. Full(parallel) Active
3.3.2. Partial Active
3.3.3. Conditional (Majority Voting) Active Redundancy
3.3.4. Standby Redundancy
3.3.5. Some Limitations of Redundancy
4. The Concept of Derating
4.1. Derating as Applied to a Transistor
4.2. Worst-case Design Method

2
1.1 RELIABILITY
Reliability is the characteristic of an item expressed as the probability that it will perform a
required function under stated conditions for a stated period of time. Note that other literature
sources may define reliability in slightly different statements. Regardless of the approach, the
operative phrases PERFORM A REQUIRED FUNCTION, UNDER STATED CONDITONS, FOR A
STATED PERIOD OF TIME are always emphasized.
Based on mathematical reasons, reliability lies between 0 and 1
For example
Reliability = 1 means the item will always work as intended
Reliability = 0.9 means 90% likely to work as intended
Reliability = 0 means absolutely certain it will not work as intended.
Thus the reliability of a public address system amplifier might be given as 0.9 over a 10-hour
period, with an ambient temperature of 25 oC. Here the required function of the amplifier is to
amplify electrical signal under the condition of ambient temperature of 25 oC, and for a
specified period of 10 hours.
It may be observed that all possible environmental conditions to which the amplifier may be
exposed have not been stated. Besides, the reliability figure of 0.9 may cease to be valid if the
operating period is increased. Furthermore, the reliability of the amplifier may also affected by
mechanical condition such as vibration and electrical conditions such as variation in the main
voltage. It should therefore be noted that the general definition of reliability given above
means that each reliability figure quoted for an item relates only to the operating period
concerned, and the required function to perform and its working conditions.
DISCUSS ON THE IMPORTANCE OF RELIABILITY
1.2. EXPLANTIONS OF SOME TERMS RELATED TO RELIABILITY
1.2.1 AN ITEM
Is defined as a part, sub-system, or equipment which can be individually considered and
separately tested
1.2.2. FAILURE
Failure is termination of the ability of an item to perform its required function. Every item has
its own operational characteristics which it must possess and exhibit in the course of its use. A
deviation in the characteristics under working condition is considered as a fault. A state of fault
is regarded as a failure.
An item is considered to have failed under any of the following conditions:
1. When it becomes completely inoperative
2. When it is still operative but unable to perform any longer the required function. For
example electric motors rotating at a speed lower than normal.
3. When it becomes unsafe for its continued use for example if electric item produces a
shocking effect on users.

3
1.2.2.1 CLASSIFICATION OF FAILURE
1. Causes
a. Misuse failure
Failures attributable to the application of stresses beyond the stated capabilities of the item
For example application of 230 V ac mains to equipment specified for use with 110 V as mains
only.
b. Inherent weakness failure
Failures attributable to weakness inherent in the item itself when subjected to stress within the
stated capabilities of the item
2. Timing
a. Sudden failure
Failure that cannot be anticipated by prior examination
b. Gradual failure
Failure that could be anticipated by prior examination
3. Degree of failure
a. Partial failure
Failure resulting from deviations in characteristics beyond specified limits but not such as to
cause complete lack of required function
b. Complete failure
4. Combination
a. Catastrophic failure
Failure which is both sudden and complete. Examples (i) blowing of fuse (ii) open circuit in wire
wound resistors or relay coils, or short circuit failure in capacitors.
b. Degradation failure
This is Failure which is both sudden and partial. Example change in the value of resistance of a
resistor due to over-operational stress
1.2.2.2 FAILURE RATE
One indicator of the reliability of an item is the rate at which the item fails and it is called
failure rate. The number of failures occurring per unit time is known as the failure rate. This
appears to be an over simplified definition of failure rate and it is hardly used in any
computation. Strictly speaking, failure rate 𝜆(𝑡) is normally defined by the mathematical
relation:
1 ∆𝑁𝑓 1 𝑑𝑁𝑓
𝜆(𝑡) = lim × =𝑁 × (1.1)
Δ𝑡→0 𝑁𝑠 ∆𝑡 𝑠 𝑑𝑡
𝑁𝑠 = number of surviving items after a life test
∆𝑁𝑓 = number of failed items during the time interval, ∆𝑡
Example 1
20 items have failed out of 2020 put on test during period of 4000 hours.
Calculate the failure rate.

4
Solution:
Using (1.1)
1 20
𝜆 = 2000 × 4000 = 1.25 × 10−6 failures per hour
And the percentage failure rate is 0.0000125 per cent per hour
NOTE
Failure rate is most commonly expressed as a percentage per 1000 hours. The above answer
would then be 0.00125 percent/1000 hours
Component failure rate are normally affected by the mechanical, electrical and thermal
environments in which they are required to operate. Shock and vibration have relatively small
effect on electronic components with small size. For this reason, the effect of environment,
temperature and operational stress on the failure rate of each component in equipment is
accounted for by introducing multiplying coefficients WE, WT and WR known as weighting
factors due to environment, temperature and rating respectively. When an equipment
operates under an unfavorable condition the relevant weighting factor is greater than unity;
whereas for an equipment operates under favorable condition the weighting factor is less than
or equal to unity.
Therefore the overall failure rate of each component, 𝜆0 in equipment can be obtained as:
𝜆0 = 𝜆𝑛𝑊𝑇 𝑊𝐸 𝑊𝑅 (1.2)
𝜆 = 𝑏𝑎𝑠𝑖𝑐 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 𝑟𝑎𝑡𝑒
𝑛 = 𝑞𝑢𝑎𝑛𝑡𝑖𝑡𝑖𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑒𝑞𝑢𝑖𝑝𝑚𝑒𝑛𝑡
𝑊𝐸 = 𝑤𝑒𝑖𝑔ℎ𝑡𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟 𝑑𝑢𝑒 𝑡𝑜 𝑒𝑛𝑣𝑖𝑟𝑜𝑛𝑚𝑒𝑛𝑡(𝑜𝑡ℎ𝑒𝑟 𝑡ℎ𝑎𝑛 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒)
𝑊𝑅 = 𝑤𝑒𝑖𝑔ℎ𝑡𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟 𝑑𝑢𝑒 𝑡𝑜 𝑟𝑎𝑡𝑖𝑛𝑔
𝑊𝑇 = 𝑤𝑒𝑖𝑔ℎ𝑡𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟 𝑑𝑢𝑒 𝑡𝑜 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒
Total failure rate 𝜆 𝑇 for equipment can be obtained by adding up the overall failure rates
contributed by all the individual components in the equipment, i.e.
𝜆 𝑇 = 𝜆01 + 𝜆02 + 𝜆03 + ⋯ + 𝜆0𝑛 (1.3)
Example 2
An electronic system has the following components with failure rates used for its construction
Component Number used Failure rate(% Per 1000 hours)
Resistors 120 0.05
Capacitors 75 0.02
Diodes 17 0.06
Transistors 55 0.08
Connections(soldered) 900 0.001

Weighting factors
Temperature (all components) 1.5
Environment (all components) 2.0
Rating
Capacitors 4.0
Resistors 2.0
Transistors 2.0
Diodes 1.6

5
Calculate the system failure rate.
Weighting factors Overall
Number
Component 𝜆 failure
used 𝑊𝐸 𝑊𝑇 𝑊𝑅
rate 𝜆0
Resistors 120 0.05 2.0 1.5 2.0 36
Capacitors 75 0.02 2.0 1.5 4.0 18
Diodes 17 0.06 2.0 1.5 1.6 4.896
Transistors 55 0.08 2.0 1.5 2.0 26.4
Connections(soldered) 900 0.001 2.0 1.5 - 2.7
87.996
System failure rate = 88%
1.2.2.3. Causes of Component/Equipment Failure
Electronic equipment is made up of wholly electronic components or a combination of
electronic and electromechanical parts; and several equipment constitute a system. Therefore,
a discussion on component failure presumes reference to failure of equipment or systems. It
therefore follows that the failure of a component in equipment may lead to the failure of that
equipment. The failure of equipment within a system may also lead, in turn, to the failure of
the system.
Furthermore, we may wish to recall that in section 1.2.2.2, we indicated how the failure rate
value of equipment may be influenced by different weighing factors, depending on the
environmental and operating conditions, in which equipment is put into use. In this section, we
shall examine in some detail, the types and nature of environmental and operating conditions
which may cause component/equipment failure.
1.2.2.3. (A) Classification of Causes of Component Failure
The stresses acting on components within an equipment or system may be classified into two
major categories, namely: environmental stresses and operating stresses. A further sub-
division of each type is possible as shown in Fig. 1.1.
It should be noted that this sub-classification may not be perfect because combinations of
these factors and overlapping are inevitable. However, this classification is important in
bringing into focus at a glance, the various types of stresses which may affect component
failure.
In the succeeding sections, we shall examine in some detail the nature and effect of both
environmental and operating stresses, and practical ways of minimizing or eliminating their
effect.

6
 Causes of Components Failure

Environmental Stresses Operating Stresses

Voltage Surge Current Surge Frequency

Charnge

Atmospheric Generated Heat Humidity Shock and Atmospheric Wind, Air,Dust

Temperature Vibration Pressure

Radiation Electrostatics

Fig. 1.1.
1. Environmental Stresses
Environmental stresses on equipment may be due to effect of factors which are external to the
equipment such as the weather, (atmospheric temperature, atmospheric pressure, humidity,
wind), sea water, etc. Any equipment which operates outdoors cannot escape the full effects
of the weather in its locality.
(i) The Effects of Temperature Variations
In polar regions the temperature may be as low as -40oC, and as high as +60oC in hot desert
regions. Even within a given region there may be considerable temperature variation day to
day and from one season to another. This type of temperature cycling can accelerate failure
which is mechanical in nature. It has been established a long time ago, (particularly through
the Arrhenius law) that failure rate of a component approximately doubles with every 10 oC rise
in temperature for a fixed applied voltage. Therefore, high temperature represents one form
of stress which increases the failure rate of all electronic components. High temperature
environment may also cause deterioration of insulation materials in equipment. It may also
initiate or accelerate chemical reaction between the material used for a component and some
very small amount of contaminant. The contaminant may have found its way into the
component during manufacture, or may sneak into it as a result of imperfect sealing between
the component and the surrounding atmosphere. The chemical reaction may cause the
component to fail because the reaction may ultimately destroy the composition of the
component. An imperfectly sealed wire-wound or metal film resistor used in a humid
environment is a good example for illustration. In this case, the presence of moisture and
applied voltage may cause electrolysis to take place, thereby reducing the material with which
the resistor is made, and ultimately causing an open-circuit failure. Low-temperature
environments may also cause deterioration of insulation materials; changes of component

7
values; softening, melting or freezing of some component materials and damage due to
mechanical contraction. It is of interest to note that at very low temperature (at about -35oC
some types of electrolytic capacitor have their capacitance reduced to a very little value.
In summary, extremes of temperature invariably increase failure rate because of:
The effect of expansion and contraction of materials with change of temperature;
Possible changes of component value;
The effect of temperature on chemical reaction;
Melting, softening or freezing of some component materials.
Remedy
Against the effect of high temperature environment, it is recommended to dissipate heat by
use of adequate heat sinks and/or ventilation or forced air-cooling. The choice of components
with low temperature coefficient of expansion and temperature characteristics should be
adopted as much as possible in electronic circuits (equipment) which are meant to operate in
high temperature areas.
In order to minimize the effect of low temperature environment, indirect heating of equipment
to control temperature can be applied. The correct choice of materials and components for
equipment design and construction is also a possible solution. The effect of temperature
cycling environment can be checked by the introduction into equipment design and
construction a means of ensuring large thermal delays to prevent rapid changes affecting
internal components. Nowadays, thermal analysis and design techniques are used with the aid
of computer programs to optimize the components and printed circuit board layout.
(ii) The Effects of Generated Heat
In Section (i) above, we are concerned mainly with high temperature variations in the
environment in which equipment was used without specific consideration for the effect of heat
generated internally in semiconductor devices themselves. In actual fact, semiconductor
devices like transistors that carry appreciable current during operation will most likely generate
appreciable amount of heat. This heat generated can cause failure of such devices when it
exceeds the power rating. In addition, it can cause a variation of the normal operational
specification of the device and increase chemical action resulting in rapid ageing.
Remedy
The use of appropriate heat sink is strongly recommended, although other forms of heat
removing methods such as the provision of ventilation or forced air-cooling may be employed
in special cases. Prudent choice of components with low expansion and temperature
characteristics is also a worthy solution. Finally, a reduction in the amount of heat generated
can be achieved by a method called derating. By derating, a device is made to operate at a
level of applied voltage, current or power, less likely to cause excessive heat to be generated
internally. Derating decreases the temperature rise due to the internal heat dissipated, and
thus decreases the failure rate.
(iii) The Effects of Humidity
High humidity can cause increased failure rate, particularly when it is associated with high
temperatures. Under these conditions a thin film of water known as water vapour can be

8
formed on a component and surfaces of printed circuit boards can become consequently
ionized and form an undesirable conducting path. This phenomenon is likely to cause a
component failure due to any of the following factors:
• Reduction in values of insulation resistance leading to possible electrical breakdown
(due to short-circuiting)
• Encouragement of fungi growth, resulting in reduction of insulation
For given moisture content in the air, the humidity rises as the temperature falls, and vice
versa. Therefore, any location where a sharp temperature fall is experienced at night will
certainly have condensation problem i.e. the process leading to the formation of water vapour.
Remedy
In order to minimize humidity effects, insulating materials must be used for equipment casing
that do not absorb moisture or support a water film or encourage growth of fungi. Silicones
and polystyrene are known to be good materials that do not absorb moisture. The most
effective approach is to close up totally any sensitive components or the whole equipment in a
special way.
(iv) The Effects of Mechanical Vibrations and Shocks
All equipment when transported or moved even from one branch to another will be subjected
to some degree of mechanical vibration and shock. Consequently, these may cause the
weakening of equipment supports, loosen wires and connections, bend and possibly damage
components, and promote stresses that may lead to component failure due to fatigue.
However, it should be noted that mechanical vibration and shock have relatively little effect on
electronic components which are small in size, and which can be protected by encasing in an
insulating medium such as epoxy resin to provide mechanical support and protection.
Electromechanical components (such as relays and contactors) and heavy components such as
transformers are more liable to malfunction due to mechanical vibration and shock.
Experiments on environmental testing of electronic equipment show that mechanical vibration
and shock increase failure rate. However, the majority of such failures occur in
electromechanical components, or from the mounting and interconnecting of electronic
components. Consequently, this may cause connector failure.
Remedy
Effect of mechanical vibration and shock can be minimized by careful design techniques using
anti-vibration mountings, shake-proof washers, locking nuts, and by the encapsulation of
sensitive components with some protective material.
(v) The Effects of Atmospheric Pressure
The atmospheric changes very little over the earth’s surface. At very high altitude, atmospheric
pressure becomes extremely low. Therefore, there is a remarkable effect of pressure on
components or equipment in air bone systems, especially in space vehicles, where extremes of
pressure variation may be experienced. Very low atmospheric pressure can cause leaking of
seals in components such as paper-foil and aluminum electrolytic capacitors. This can take
place if equipment is transported in air in an unpressurised casing. Furthermore, low pressure
causes a decrease of breakdown voltage between contacts using air insulation.

9
Remedy
For equipment designed to work in low pressure environment there is need to provide safe
distances between electrical conductors and maintain dust-free and dirt-free paths between
conductors.
(vi) The Effects of Air Impurities
Certainly any equipment that works outdoors on land cannot escape the menace of earthy
materials, such as dust, dirt and sand. Salinity of air over the sea and in coastal regions also has
its effect on electronic equipment. The movement of dust into equipment may cause tracking
across components especially switches, and this may lead to their early failure. Furthermore,
the ingress of dust may result in long-term degradation of insulation and increased contact
resistance. Salt atmosphere may cause corrosion and promote degradation of insulation in
equipment.
Remedy
In order to minimize the effect of these air impurities provisions of suitable sealing is necessary
and periodically removing dust by blowing is desirable. However, mechanical protection of
components is recommended in order to guard against the effect of salt atmosphere.
(vii) The Effect of Electromagnetic Radiation
Electromagnetic radiation due to α-rays and β-rays (especially of a nuclear nature), can cause
malfunctioning of certain components, especially bipolar solid state devices. For this reason,
equipment designed for the remote surveillance or control of nuclear processes in space is
susceptible to the effect of radiation.
Remedy
The major remedy against the effect of radiation is the provision of shielding, and prudent
selection of components for equipment design. Efforts are being made recently to design
certain transistors and integrated circuits which are helpful in reducing shielding problems.
(viii) The Effect of Electrostatics
Static electricity can cause damage to sensitive electronic devices and assemblies, and can
create handling, packaging, sorting and storage problems. It can also be a safety hazard in
explosion and fire risk areas during the manufacture of devices and equipment.
Problems caused by static discharges occur in the most modern of industries such as solid state
electronics. This poses a great problem both in the manufacture and handling of MOS (metal-
oxide-silicon) semiconductor devices. The extremely thin oxide separating the silicon and the
metallization can easily be destroyed by static discharges.
Remedies
Protection from the hazardous effects of static electricity can be achieved by using any or a
combination of the following approaches:
Elimination or reduction of static electricity: - This may be achieved by earthing all possible
electrostatically charged parts by electro conductive materials by increasing the electrical
conductivity of all materials coming into contact by friction. Specifically ICs and MOS-FETs are
protected from case to case by gate safe diodes against destruction or damage by static
discharges. In such cases, the gate safe diodes are either integrated into the electronic
component or they are soldered onto the external connections.

10
Use of electro conductive from system for the protection of electronic components and
finished products:
The use of electro conductive plastic systems can eliminate the risks and damage which may be
caused by static electricity. These materials are capable of eliminating static charge without
dangerous effects and preventing the creation of a static charge from the start.
Protection of working and production areas by electro conductive foam: This can be achieved
by using:
• Leather-like electro-conductive fleece for covering workbenches in the electronics
industries. A typical example is shown in Fig.1.2 with anti-static mats on the workbench
and immediately below it on the floor but linked together.

Figure 1.2: Anti-Static mat at a workplace

Figure 1.3: Antic static rug

11
 • Electro-conductive carpeting material which prevents the generation of electric charges
and help to shut off charges on persons or objects without any spark discharges. Fig.1.3
shows the example of a work place with electro-conductive carpeting material on its
floor. Electrostatic charges are discharged immediately on a person walking on the anti-
static rug even before the person settles at the workbench.
• Special shoes incorporating electro-conductive plastic cylinders which prevent static
charges being produced on the floor. Any charge produced on the wearer’s body is
equally earthed.
• Use of special coatings of products or production areas:
Special coatings are available for the prevention of static charges on plastic products such as
plastic tubes and other plastic containers.
2. Operating Stresses
Operating stresses may be further subdivided into voltage, current and frequency. Each of
these will be treated briefly in the succeeding sections.
(i) Voltage Surges
Voltage surges that occur during switching of electronic equipment may be of short duration
but may have extremely high peak values which will exceed the steady-state value by many
times. Apart from the occurrence of voltage surge, it is essential to maintain the applied
voltage at, or even below, the rated value in order not to increase the failure rate of some
components. Components such as capacitors which are characteristic of static electric field are
known to have a failure rate which is approximately proportional to the fifth power of the
applied voltage at fixed temperature.
(ii) Current Surges
Current surges contribute to component failure in an exactly similar manner as voltage surges
do. Again, sudden fluctuation of current in excess of steady-state values must be anticipated,
but compensation may be effected to some extent by the use of derating procedures,
discussed later in Section 4. This same approach is applicable in case of voltage surges.
(iii) Frequency Changes
It is not uncommon to see items of electronic equipment manufactured for use, say, in the USA
where the mains frequency of electricity supplies is 60Hz being used in another country where
the mains frequency is 50 Hz such as Nigeria. Therefore, any attempt to use such an electronic
item in Nigeria may lead to its malfunctioning immediately and subsequently to its complete
failure. The change in mains frequency may therefore result in component failure. Apart from
the foregoing example, it is possible for electronic equipment (i.e. an amplifier) designed for
audio frequency use to malfunction if subjected to operation at excessively high frequency.
This may lead to the damage of some semiconductor devices which are designed for operation
at relatively low frequency.
1.2.2.3. (B) Failure in Electronic Components
In this section, we shall focus attention on some of the commonest electronic components,
identify the types of failure they usually experience due to environmental stresses, operating
stresses, and manufacturing defects.

12
 (i). Thermionic valves
Of all electronic components, thermionic valves have the highest failure rate. The failures of
thermionic valves may be characterized by open-circuiting of filament, short-circuiting of
electrodes (i.e. grid to cathode) and low cathode emission of electrons. The commonest valve
failures are due to excessive temperature, mechanical shock or vibration. Excessive
temperature occurs mainly due to the generation of heat by surrounding components, and also
due to unsatisfactory cooling arrangement.
(ii). Fixed Resistors
The commonest type of failure of fixed resistors is due either to increase in resistance value or
open-circuitry. Increase in resistance value is a common fault in carbon composition resistor,
and this is due to any of the following reasons:
• Movement of carbon or binder under influence of heat, voltage or moisture
• Absorption of moisture, resulting in swelling or forcing of carbon particles to separate.
Open-circuiting may occur not only in carbon composition resistors, but also in film and wire-
wound resistors. In carbon composition resistors, this may occur as a result of excessive heat
burning out resistor center, and mechanical stress fracturing the resistor. Wire-wound
resistors may become open-circuited due to fracture of the wire resulting from progressive
crystallization of wire because of impurities or resulting from corrosion of wire due to the
electrolytic action set up by absorbed moisture. Film resistors may experience open-circuiting
due to disintegration of film as a result of high temperature, applied voltage or damaging of the
film during manufacture.
(iii). Variable resistor (wire and composition type)
Failure of variable resistors is commonly caused by open-circuiting or intermittent contact
resulting from mechanical wear and/or excessive current flow over a small segment of low
resistance value.
(iv). Capacitors
Failure of capacitors may be characterized by open-circuiting, short-circuiting, intermittent
contact and fluctuations in capacitance value. Intermittent open-circuiting may be caused by
mechanical/thermal shock, as in paper foil capacitors; fracture of connection as in ceramic
capacitors; fracture of internal connections as in aluminum, electrolytic capacitors; and silver
not adhering to mica as in mica capacitors. Short-circuiting may occur due to fracture of
dielectric resulting from shock or vibration caused by high humidity as in mica capacitors,
leaking of seal as a result of mechanical/thermal shock or vibration in pressure as in paper foil
capacitors. Fluctuation in capacitance value may occur due to loss of electrolyte via leaking
seal caused by pressure, thermal or mechanical shock as in aluminum electrolytic capacitors.
(v). Variable capacitors
Variable capacitors are very reliable components and almost failure free. However, occasional
failure in variable capacitors may occur in air-spaced variable types due to loose plates or
mechanical damage of the plates which may change or close the small gap between the plates.
A deposit of dust or moisture between the plates may cause a similar failure.

13
 (vi). Inductors and Transformers
Inductors and transformers are normally characterized by failure due to open-circuiting or
short-circuiting of the coil to the frame (iron-cored types). Failure may occur due to
overloading perhaps through failure of another component. Mechanical damage may also
result in failure.
(vii). Semi-conductor devices
Failures in semiconductor devices are mainly those of an open or short circuit at a junction. For
example, bipolar transistors may fail due to open or short-circuiting between base and emitter,
collector and base or collector and emitter. Apart from failure due to main environmental
stresses such as temperature and humidity, electrical interference is another cause of failure in
semiconductors.
It is particularly important to note that most integrated circuits (ICs) will be damage and fail if
the maximum supply voltage is exceeded or if the IC is removed from or inserted into a test
socket while the power is applied.
(viii). Relay
Failures in relays may be due to corrosion of fine wires and consequent open-circuiting, and
metal fatigue in the armature spring. Dirty contacts caused by action of chemicals produced by
sparking in an enclosed space may also lead to failure.

1.2.2.3 (C) General Approaches in Reducing Equipment/Components Failure

Apart from specific remedies (approaches) in dealing with probable failure arising from
environmental and operating stresses, there are general approaches (methods) which, if
applied, are useful in reducing components failure rate and improving equipment reliability.
There are numerous approaches some of which will be discussed in the succeeding sections.
1. Part Selection
For the design of a reliable system, it is necessary to ensure that the selected components
(parts):
Are insensitive to the environment concerned,
Do not create any harmful effect on other components,
Can be protected individually or in groups of a complete circuit from the harmful effects of the
environment, and
Do not change in normal value outside the allowable tolerance limit.
2. Use of Improved Constructional Methods and New Technology
Many methods and new technology have been developed which are intended to reduce the
quantity of electronics components in any given equipment and protect them against the
effects of the environment. These measures go a long way in improving reliability. Some of the
main developments in construction have been the use of encapsulation and printed circuits.
There have also been tremendous developments in components themselves which make them
less sensitive to the environment. For example, the development of semiconductor diodes and
transistors has reduced the problem of mounting in sockets in order to reduce the effects of
mechanical shock and vibration. This is because they are small in size and dissipate little
‘harmful’ heat to neighboring components. Encapsulation or posing of circuits and

14
components provides a great deal of protection against humidity, mechanical shock and
vibration.
There are many other examples of new technology and developments in component
production which have improved reliability. These include the following:
Encapsulated reed relays have been replaced by solid state switching. This has brought about
increased switch reliability against a more complex control function.
Many discrete electronic components and circuits replaced integrated circuits, with the result
of providing simplified circuit design and reduction in equipment failure rate.
Light emitting diodes have replaced filament devices, thereby providing more convenient and
reliable indicators.
3. Use of Redundancy
A prudent use of redundancy is one of the proven methods of improving reliability.
Redundancy involves the use of additional units in a variety of configurations such as active
redundancy, passive (standby) redundancy, etc. Different configurations of redundancy and
their attendant benefits will be treated in detail in section 3. However, it should be noted that
the decision to use redundancy must be based on an analysis of the possible trade-offs such as
increase in weight, space, cost price, maintenance and spares-holding costs.
4. Use of Statistical Design Method
The statistical design method involves the use of the distribution of component values between
their tolerance limits for the purpose of assessing whether a design meets its engineering
specifications and determining the proportion of the production which will be rejected as
unsatisfactory. This procedure necessitates intensive use of the computer and is based on the
use of random values with the tolerance range for each component. This procedure should not
be confused with the product testing which takes place after manufacture.
5. Use of Derating Concept
Generally, derating is a process by which a component/equipment is operated below the rated
(environmental or operating) stress level of a parameter in order to reduce the
component/equipment failure rate and consequently improve reliability significantly. Derating
is commonly applied where the reduction of operating voltage and temperature is involved.
This leads to the relationship of failure rate (due to stress) against operating voltages and
temperatures. This will be treated in detail in section 4.
1.3 Mean Time Between Failures (MTBF)
MTBF can be defined as the mean value of the length of time which elapses between failures. It
is applicable to repairable items. It can be computed based on two main types of tests known
as the non-replacement and replacement methods.
1.3.1 Replacement Method
The MTBF, m of a system may be measured by testing it for a total period of time 𝑡𝑛 during
which 𝑛 failures or faults occur. Each fault if repaired and the equipment put back on test, the
repair time being excluded from the total time T results in the observed MTBF given by
𝑇 (𝑡1 − 𝑡0 ) + (𝑡2 − 𝑡1 ) + ⋯ + (𝑡𝑛 − 𝑡𝑛 − 1) 𝑡𝑛 − 𝑡0 𝑡𝑛
𝑚= = = = (1.4)
𝑛 𝑛 𝑛 𝑛
(If 𝑡0 = 0 is the reference starting time)

15
Where
𝑛 = number of failures
𝑡0 = time of start of trial
𝑡1 = time to first failure
𝑡2 = time of second failure
𝑡𝑛 = time to nth failure
EXERCISE 3
Continuous test were conducted on electrical item and faults which were repaired immediately
occurred at the following times
Failures 0 1 1 1 1 1 1
Time 𝑡0 = 0 𝑡1 = 2 𝑡2 = 4 𝑡3 = 8 𝑡4 = 10 𝑡5 = 14 𝑡6 = 15
(× 1000 ℎ𝑟𝑠)
Calculate the mean time between failures
Solution
𝑡𝑛 −𝑡0 15−0
Using (1.4) 𝑀𝑇𝐵𝐹, 𝑚 = = × 103 ℎ𝑟𝑠
𝑛 6
1.3.2 The non - replacement method
This requires that a large number of the items be put to under test and observations made for
possible failure at the beginning and the end of the test period. Under this method it is
assumed that the test time should be truncated before the items are subjected to failure
possibly due to wear out. In other words, the test is limited to the useful life. It has been
demonstrated by Epstein that the best estimate of MTBF , m, for a truncated test is given by
𝑡𝑒𝑠𝑡 ℎ𝑜𝑢𝑟𝑠 𝑓𝑜𝑟 𝑓𝑎𝑖𝑙𝑢𝑟𝑒𝑠 + 𝑡𝑒𝑠𝑡 ℎ𝑜𝑢𝑟𝑠 𝑓𝑜𝑟 𝑠𝑢𝑟𝑣𝑖𝑣𝑜𝑟𝑠
𝑚=
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑎𝑖𝑙𝑢𝑟𝑒𝑠
That is
𝑡𝑜𝑡𝑎𝑙 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝑡𝑒𝑠𝑡 ℎ𝑜𝑢𝑟𝑠(𝑠𝑢𝑟𝑣𝑖𝑣𝑎𝑙 ℎ𝑜𝑢𝑟𝑠)
𝑚= (1.5)
𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑎𝑖𝑙𝑢𝑟𝑒𝑠
Example 4
In order to determine the MTBF of a certain component, 50 were tested for a period lasting
200 hours. The times to failure of the components are shown in table 1.
35 components survived without failure. Assuming that wear-out failure can be ignored,
calculate:
(i) Total test hours before failure
(ii) Total test hours without failure
(iii) Total survival hours
MTBF
TABLE 1
Number of components Time to failure(hours)
6 100
5 140
4 175

Solution
(i) Total test hours before failure (for components which failed)

16
 (6 × 100) + (5 × 140) + (4 × 175) = 2000 ℎ𝑜𝑢𝑟𝑠
(ii) Total hours without failure (test hours for survivors)
35 × 200 = 7000 ℎ𝑜𝑢𝑟𝑠
(iii) Total survival hours = 2000 + 7000 = 9000 ℎ𝑜𝑢𝑟𝑠
9000
𝑀𝑇𝐵𝐹 = = 600 ℎ𝑜𝑢𝑟𝑠
15
Another way of expressing MTBF is through failure rate, which is the number of faults per unit
time. If the failure rate, is approximately constant is value as is the case with many electronic
systems, then MTBF is the reciprocal of failure rate
1
I.e. 𝑀𝑇𝐵𝐹, 𝑚 = (1.6)
𝜆
𝑚 is usually expressed In hours and the corresponding units of 𝜆 faults per hour. Failure rates
msy be expressed as a percentage per 1000 hours. Failure per 106 or failure per 109 hours.
1.4 Mean Time to Failure (MTTF)
This is a term which applies to non-repairable items (such as resistors, capacitors, electric
bulbs, etc) and it is the average time an item may be expected to function before failure.
MTBF can be computed after testing a number of items, 𝑁 in a specified way (by
applying certain electrical, mechanical, heat or humidity conditions) until all have failed. If the
times to failure are (t1, t2, t3…tn) then the observed MTTF is given by:
∑𝑁 𝑖=1(𝑡𝑖 − 𝑡0 )
𝑀𝑇𝑇𝐹 = (1.7)
𝑁
(𝑡1 − 𝑡0 ) + (𝑡2 − 𝑡0 ) + ⋯ + (𝑡𝑁 − 𝑡0 )
𝑖𝑒, 𝑀𝑇𝑇𝐹 =
𝑁
Where
𝑡0 = 𝑆𝑡𝑎𝑟𝑡𝑖𝑛𝑔(𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒)𝑡𝑖𝑚𝑒
(𝑡1 − 𝑡0 ) = 𝑝𝑒𝑟𝑖𝑜𝑑 𝑡𝑜 𝑓𝑖𝑟𝑠𝑡 𝑓𝑎𝑖𝑙𝑢𝑟𝑒
(𝑡2 − 𝑡0 ) = 𝑝𝑒𝑟𝑖𝑜𝑑 𝑡𝑜 𝑠𝑒𝑐𝑜𝑛𝑑 𝑓𝑎𝑖𝑙𝑢𝑟𝑒
(𝑡𝑁 − 𝑡0 ) = 𝑝𝑒𝑟𝑖𝑜𝑑 𝑡𝑜 𝑁𝑡ℎ 𝑓𝑎𝑖𝑙𝑢𝑟𝑒
𝑁 = 𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑎𝑖𝑙𝑢𝑟𝑒𝑠

Example 5
Life testing is made on seven (non-repairable) electrical lamps and the following results were
obtained.
Failures 0 1 1 1 2 1 3
Time 𝑡0 = 0 𝑡1 = 4 𝑡2 = 10 𝑡3 = 16 𝑡4 = 20 𝑡5 = 23 𝑡6 = 25
(× 100 ℎ𝑟𝑠)

Calculate the MTTF

(𝑡1 −𝑡0 )+(𝑡2 −𝑡0 )+(𝑡3 −𝑡0 )+(𝑡4 −𝑡0 )+(𝑡5 −𝑡0 )+(𝑡𝑁 −𝑡0 )
Solution 𝑖𝑒, 𝑀𝑇𝑇𝐹 = (1+1+1+2+1+3)
98
= × 102
9
1.5 Reliability
Consider the case in which a fixed number 𝑁0 of identical component are tested.

17
Let 𝑁𝑠 = number surviving up to time 𝑡
𝑁𝑓 = number failed up to time 𝑡
𝑁0 = 𝑁𝑠 + 𝑁𝑓 = total number in operation at 𝑡 = 0
𝑁
𝑅(𝑡) = 𝑁𝑠 = Reliability at any time 𝑡
0
1.6 MATHEMATICAL INTERRELATIONSHIPS OF TERMS
1.6.1 Reliability and failure rate
From (1.1) we know that failure rate
1 𝑑𝑁𝑓 1 𝑑𝑁𝑓
𝜆=𝑁 × =𝑁 ×
𝑠 𝑑𝑡 0 −𝑁𝑓 𝑑𝑡
1 𝑁 𝑑𝑁
Or ∫0 𝜆 𝑑𝑡 = ∫0 𝑓 𝑁 −𝑁𝑓 𝑑𝑡
0 𝑓
𝑁𝑓
−𝜆𝑡 = [ln (𝑁0 − 𝑁𝑓 )]0
−𝜆𝑡 = [ln(𝑁0 − 𝑁𝑓 ) − ln (𝑁0 − 0)]
𝑁0 − 𝑁𝑓
−𝜆𝑡 = ln ( )
𝑁0
Thus
𝑁𝑓
𝑒 −𝜆𝑡 = (1 − ) (∗)
𝑁0
𝑁 𝑁0 −𝑁𝑓 𝑁
But 𝑅(𝑡) = 𝑁𝑠 = = 1 − 𝑁𝑓 (∗∗)
0 𝑁0 0
Comparing (*) and (**) we conclude that
𝑅(𝑡) = 𝑒 −𝜆𝑡 (1.8)
Example 6
Consider the data of exercise 1. If the system works for 13 hrs for 15 days and 10 hrs for 4 days,
compute the reliability of the system
Solution
The system failure rate was obtained as 88%/1000 hrs
Or 𝜆 𝑇 = 0.88/1000ℎ𝑟
Total operating time =13 × 15 + 10 × 4 = 235 ℎ𝑟
System reliability 𝑅(𝑡) = 𝑒 −𝜆𝑡 = 𝑒 −0.88×0.001×235 = 0.81
1.6.2 MTBF, MTTF and Failure Rate
∞
The general expression for MTBF, 𝑚 is: 𝑚 = ∫0 𝑅(𝑡)𝑑𝑡………………… (***)
For the case when 𝜆 is constant 𝑅(𝑡) = 𝑒 −𝜆𝑡 and (***) becomes
∞
1 −𝜆𝑡 ∞ 1 1
𝑚 = ∫ 𝑅(𝑡)𝑑𝑡 = − [ 𝑒 ] = − [𝑒 −∞ − 𝑒 −0 ] =
𝜆 0 𝜆 𝜆
0
If failures are due to chance and if the FAILURE RATE IS CONSTANT, then it follows that
1
𝜆 = 𝑀𝑇𝑇𝐹 for non-repairable items
1
𝜆 = 𝑀𝑇𝐵𝐹 for repairable items
The following important points should be noted:
The assumption that 𝜆 is constant is nearly always true of electronic components.

18
If failure rate is constant and 𝑅(𝑡) = 𝑒 −𝜆𝑡 and 𝜆 = 1/𝑚, therefore 𝑅(𝑡) = 𝑒 −𝑡/𝑚 .
Consequently, after one MTBF(𝑖. 𝑒. 𝑡 = 𝑚) , the probability of survival 𝑅(𝑡) = 𝑒 −1 = 0.37
1.7 RELIABILITY EQUATIONS AND CURVES WHEN FAILURE RATE IS CONSTANT
(i). Reliability and Unreliability Equations and Curves
If failure rate is constant, the probability of no failures occurring in a given time is:
𝑅(𝑡) = 𝑒 −𝜆𝑡
𝑤ℎ𝑒𝑟𝑒 𝑅 𝑖𝑠 𝑡ℎ𝑒 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑛𝑜 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 𝑖𝑛 𝑡𝑖𝑚𝑒 𝑡 𝑖.e. the reliability.
The unreliability, Q is defined as the probability of total failure. It follows from statistical theory
that:
𝑅 + 𝑄 = 1 (1.9)
And that
𝑄 = 1 − 𝑅 = 1 − 𝑒 −𝜆𝑡 (1.10)
A graph of R and Q against time yields the familiar exponential curves as shown below.

Reliability and Unreliability Curve

𝑎𝑠 𝑡𝑖𝑚𝑒 , 𝑡 → 0, 𝑅 → 1, 𝑄 → 0
At time 𝑡 = 1/𝜆, 𝑅 → 0.37, 𝑄 → 0.63
(ii). Survivor and failure equations and curves
From equation (**) and (1.5)
𝑁
𝑅(𝑡) = 1 − 𝑁𝑓 = 𝑒 −𝜆𝑡 (+)
0

𝑚𝑢𝑙𝑡𝑖𝑝𝑙𝑦𝑖𝑛𝑔 (+) 𝑏𝑦 𝑁0 , 𝑡ℎ𝑒𝑟𝑒𝑓𝑜𝑟𝑒, 𝑁0 − 𝑁𝑓 = 𝑁𝑠 = 𝑁0 𝑒 −𝜆𝑡 (1.11)

Equation (1.11) is the equation of the graph of survivors against time (survivor curves). Similarly
from equation (+) it can be easily shown that
𝑁𝑓 = 𝑁0 (1 − 𝑒 −𝜆𝑡 ) (1.12)
Equation (1.12) is the equation of the graph of failure against time.
The survivor and failure curves are shown below.

19
Survivor and Failure Curves

It can be inferred from equation (1.11) that

𝑁𝑠
𝑅(𝑡) = = 𝑒 −𝜆𝑡
𝑁0
Where R is the reliability
Similarly, from equation (1.12)
𝑁𝑓
𝑄= = (1 − 𝑒 −𝜆𝑡 )
𝑁0
Where Q is the unreliability
Example 7
One thousand similar components each of constant failure rate 5%/1000hr are put into test
together calculate the time lag before failure of the following number of components (i) 100 (ii)
500
Solution
𝑁
We know that 𝑅(𝑡) = 𝑒 −𝜆𝑡 = 1 − 𝑁𝑓
0
5 1 −5
But 𝜆 = 100 × 1000 = 5 × 10
𝑡 = 𝑡1 = time required before failure of 100 components
𝑁0 = 1000
𝑁𝑓 = 𝑁𝑓1 = 100
−5 𝑡1) 100
Then 𝑅(𝑡) = 𝑒 (−6×10 = 1 − 1000
t1 = 2107
Let 𝑡2 be the time required for 500 components to fail, then
−5 500
𝑅(𝑡) = 𝑒 (−5×10 𝑡2) = 1 −
1000
−5
−5 × 10 × 𝑡2 = 𝑙𝑛2
𝑡2 = 13,863 ℎ𝑜𝑢𝑟𝑠

20
2. Concept of Reliability Prediction
2.1 Introduction
Reliability prediction is the process of calculating the anticipated system reliability from
assumed component failure rates. However, in calculating the reliability of a system made up
of a number of sub-units each having their own reliability, the type of system must first be
defined. There are two major types, namely; series and parallel systems, although there may
be a combination of the two systems to give a series-parallel system. It is necessary to
represent the system as a number of functional blocks. These are interconnected according to
the effect of each block failure on the total system.
A series system is one in which the failure of one of the sub-units, (represented by functional
blocks), or component, means failure of the system as a whole. Fig. 2.1 (a) represents the block
diagram of a series system. Here, the system fails if any of the blocks A, B, or C fails.
A parallel system is one which does not fail until all the sub-units or components have failed.
Fig. 2.1 (b) shows a parallel system. Failure of the system occurs if all the blocks A, B, and C fail.
A series – parallel system is one which fails if the block in series connection (block A) fails or if
both blocks (blocks B and C) in parallel connection fail. The failure of B or C alone is insufficient
to cause system failure. Fig. 2.1 (c) represents a series-parallel system.

A B C
Fig. 2.1(a)

C
Fig. 2.1(b)

C
Fig. 2.1(c)

21
2.2 Law of Probability Relevant to Reliability Prediction
It will be recalled that reliability, simply put, is the probability of successful operation. In
calculating system reliability, various well known laws in probability theory are used. The
relevant are discussed below.
2.2.1 Multiplication rule
We refer, for example to Fig. 2.1 (a) and imagine that there are two sub-units A and B
connected in series. If both sub-units can operate simultaneously, and their individual
probabilities of operating successfully are known, then the probability of successful operation
of the series-system is the product of the individual probabilities. Hence, the probability of
both sub-units A and B operating successfully is:
Pab = PaPb …………………… (2.1) where;
Pa = probability that A will operate successfully
Pb = probability that B will operate successfully
Pab = probability that both A and B will operate successfully
If we extend our imagination to n sub-unit (i.e., denoted by A, B, C, …, n) connected in series,
then we can state the probability of successful operation of the new series system as:
Pab = Pa ×Pb× …………× Pab……..Pn ………… (2.2)

Equation (2.2) is generally referred to as the multiplication rule of probabilities. For example, if
Pa =0.85 and Pb = 0.9, therefore
Pab = 0.85 × 0.9 = 0.765
Note that the probability of successful operation of the series system P ab is less than either Pa
or Pb. Note also that the concept of multiplication rule of probabilities is useful in determining
the reliability of a series system.
2.2.2 Addition Rule
Here, we refer to Fig. 2.1(b) and imagine that there are only two sub-units A and B connected
in parallel. If each of the sub-units can operate separately and their individual probabilities of
operating successfully are known as Pa and Pb for sub-unit A and B respectively; then the
probability of either sub-unit A OR sub-unit B or BOTH operating successfully is:
P(A or B) = Pa + Pb -PaPb …………………… (2.3a)
Alternatively,
P (A or B) = 1-[(1-Pa)(1- Pb)] …………………… (2.3b)
It should be noted that equations (2.3a) and (2.3b) are essentially the same. The expansion
and simplification of the right hand side (RHS) of equation (2.3b) will lead to the same
expression on the RHS of equation (2.3a). However, for computational purposes, the formula
given by Equation (2.3b) is more convenient to use.
Generally, the probability of successful operation of one or more of the n sub-units is:
P (a or b or c or ….or n) = 1-[(1-Pa) (1- Pb)….. (1-Pn)] …………………… (2.4)
The expression given by Equation (2.4) is known generally as the addition rule of probabilities.
Note that the addition rule of probabilities is particularly useful in finding the reliability of
parallel systems.

22
2.2.3 Binomial Distribution
Some experiments consist of repetitions of independent trials each of which has two possible
outcomes (success or failure) with fixed probabilities of occurring. In n trials of such an
experiment, the distribution of the two classes of possible outcomes is discrete and of the
binomial type.
Let us take an example to illustrate this idea. Suppose we have an electric generating set which
has a probability of successful operation of 0.95 for a single day, and it is expected to operate
for 5 days (not continuously). Suppose we are required to do the following:
Discuss the probability distribution of the two possible outcomes.
Find the possibility that the generating set will operate successfully for (i) 3days; (ii) at least 4
days.
In answering this question, we should note the following points. First, there are two possible
outcomes, namely; ‘success’ or ‘failure’. Second, the possibility of success may be represented
by p and of failure by q, such that p + q = 1. In other words, the probability of failure may be
represented by (1 – p). Third, the number of trials is equal to the number of days the
generating set is put on operation i.e. n = 5. Four, the probability distribution of the two
possible outcomes, can be determined through binomial expansion (p + q)5. Five, using the
idea of Pascal’s triangle, the coefficients of expansion of (p + q) 5 can be obtained as:
1, 5, 10, 10, 5, 1.
Now, the expansion of (p +q) 5 gives:
P5 +5p4q + 10p3q2 + 10p2q3 + 5pq4+ q5
In discussing the probability distribution of the two possible outcomes, we can say that:
P5 = probability that the generating set will operate successfully for all the 5 days.
4
5p q = probability that the generating set will operate successfully for 4 days and fail for a
day.
10p3q2 = Probability that the generating set will operate successfully for 3 days and fail 2 days.
10p2q3 = probability that the generating set will operate successfully for 2 days and fail 3 days.
5pq4 = probability that the generating set will operate successfully for 1 day and fail 4 days.
q5 = probability that the generating set will operate without success for the 5 days.
The sum of terms in the RHS of the expansion (p+q) 5 is equal to 1, since on the LHS, (p+q) = 1
and (p+q) 5 = 1.
(i). Probability that the set will operate successfully for 3 days = 10p 3q2. Since p = 0.95
and q = 1 – 0.95 = 0.05, therefore
10p3q2 = 10×0.953×0.052 = 0.02
(ii). Probability that the set will operate successfully for at least 4 days is the sum of the
probabilities that it will operate for 4 days and for the 5 days period, i.e.
P(at least 4 days) = p5 + 5p4q = 0.955+(5×0.954×0.05) = 0.978
In a more general case, if the probability of success is p and that of failure is (1 – p), and if there
are n independent trials, then the probability of r successes in n trials is given by
Pr = nCr× pT × (1+P) n-r
Where

23
 𝑛! 𝑛(𝑛 − 1)(𝑛 − 2) … (𝑛 − 𝑟 + 1)
n𝐶𝑟 = =
𝑟! (𝑛 − 𝑟)! 𝑟(𝑟 − 1)(𝑟 − 2) … × 3 × 2 × 1
n
In words, Cr determines the number of ways that exactly r successes and (n – r) failures can
occur in n trials. It should be noted that, in essence, Pr determines each term of the expansion
(p + q)5 mentioned in our earlier example (i.e. p5, 5p4q, …,q5). Now, let us test the strength of
the above formula through an example. Once again, let us determine the probability that the
generating set will operate successfully for 3 days (out of the five days). Using the above
formula; i.e.
i.e. P3 = 5C3 × p3 × (1-P) 5-3.
P3 = 10p3q2.
(This gives the same result as before)
In general, the expansion of (p + q) n is given as:
𝑛(𝑛 − 1) 𝑛−2 2
(𝑝 + 𝑞)𝑛 = 𝑝𝑛 + 𝑛𝑝𝑛−1 𝑞 + 𝑝 𝑞 + ⋯ + 𝑛𝐶𝑞 𝑝𝑛−𝑟 𝑞 𝑟 + ⋯ + 𝑞 𝑛
2!
NOTE
(i). Since p + q = 1, therefore (p +q)n = 1. Consequently, the sum of all the terms in RHS of
the expansion (p +q)n is always equal to 1.
(ii). This idea of binomial distribution will be useful in Section 2.7.2 in finding the reliability
of systems with partial redundancy and conditional (majority voting) redundancy.
2.3 Reliability, MTBF of a Series System
2.3.1 The Reliability of a Series System
If R, is the reliability of a series system while R1, R2… Rn are the respective reliabilities of the
individual n sub-units of the system for the same time period, then using equation (2.2)
𝑅𝑠 = 𝑅1 𝑅2 𝑅3 … … … 𝑅𝑛
However, if the series system is made up of exponentially failing units, such that the sub-unit
reliabilities are
𝑅1 = 𝑒 −𝜆1 𝑡 , 𝑅3 = 𝑒 −𝜆3𝑡 , … … … . 𝑅𝑛 = 𝑒 −𝜆𝑛𝑡
Where 𝜆1 , 𝜆2 , 𝜆3 are the respective failure rates of the sub-units. Thus
𝑅𝑠 = 𝑅1 𝑅2 𝑅3 … … … 𝑅𝑛 = 𝑒 −𝜆1𝑡 × 𝑒 −𝜆2 𝑡 × … … … 𝑒 −𝜆𝑛𝑡
= 𝑒 −(𝜆1 +𝜆2+⋯……𝜆𝑛)𝑡 = 𝑒 −𝜆𝑠 𝑡
Where 𝜆𝑠 is the system failure rate given by:
𝜆𝑠 = 𝜆1 + 𝜆2 + ⋯ … … 𝜆𝑛 ………………… (2.5)

2.3.2 MTBF of a Series System

The MTBF of a series system, 𝑀𝑠 , is equal to the reciprocal of the system failure rate.
1 1
𝑀𝑠 = 𝜆 = 𝜆 ………………… (2.6)
𝑠 1 +𝜆2 +⋯……𝜆𝑛
Note that for a series system containing n similar sub-units of equal reliability the system
reliability,
𝑅𝑠 = 𝑒 −𝑛𝜆𝑡
Consequently, system failure rate 𝜆𝑠𝑛 = 𝑛𝜆
1
System MTBF, 𝑀𝑠𝑛 = 𝑛𝜆

24
2.4 Reliability, MTBF of a Parallel System
2.4.1 The Reliability of a Parallel System
In line with equation (2.4), the reliability of a parallel system with n different sub-units is:
𝑅𝑝 = 1 − [(1 − 𝑅1 )(1 − 𝑅2 ) … … (1 − 𝑅𝑛 )] ………………………(2.7)
𝑅1 , 𝑅2 , 𝑅3 , … … … 𝑅𝑛 are the reliabilities of the respective sub-units.
NOTE
If
𝑅1 = 𝑅2 = 𝑅3 = ⋯ = 𝑅𝑛 = 𝑅,
∴ 𝑅𝑝 = 1 − (1 − 𝑅)𝑛
As before, if we are given that the system is made of exponentially failing sub-units such that
𝑅1 = 𝑒 −𝜆1 𝑡 , 𝑅3 = 𝑒 −𝜆3𝑡 , … … … . 𝑅𝑛 = 𝑒 −𝜆𝑛𝑡
Then the system reliability for a system with n sub-units is given by
𝑅𝑝 = 1 − [(1 − 𝑒 −𝜆1 𝑡 )(1 − 𝑒 −𝜆2 𝑡 ) … … (1 − 𝑒 −𝜆𝑛𝑡 )]
For a system with two sub-units in parallel connection:
𝑅𝑝2 = 1 − [(1 − 𝑅1 )(1 − 𝑅2 )] = 𝑅1 + 𝑅2 − 𝑅1 𝑅2
= 𝑒 −𝜆1𝑡 + 𝑒 −𝜆2𝑡 − 𝑒 −𝜆1𝑡 𝑒 −𝜆2 𝑡 = 𝑒 −𝜆1 𝑡 + 𝑒 −𝜆2𝑡 − 𝑒 −(𝜆1+𝜆2 )𝑡
For a system with three sub-units in parallel:
𝑅𝑝3 = 1 − [(1 − 𝑅1 )(1 − 𝑅2 )(1 − 𝑅3 )]
Expanding and simplifying the expression on the RHS, we obtain
𝑅𝑝3 = 𝑅1 + 𝑅2 + 𝑅3 − 𝑅1 𝑅2 − 𝑅2 𝑅3 − 𝑅1 𝑅3 + 𝑅1 𝑅2 𝑅3
𝑅𝑝3 = 𝑒 −𝜆1𝑡 + 𝑒 −𝜆2𝑡 + 𝑒 −𝜆3 𝑡 − 𝑒 −(𝜆1+𝜆2)𝑡 − 𝑒 −(𝜆1+𝜆3)𝑡 − 𝑒 −(𝜆1+𝜆3)𝑡 + 𝑒 −(𝜆1 +𝜆2+𝜆3 )𝑡
Definitely, these equations are not of a simple exponential form, and therefore the overall
system reliability cannot be expressed in the form e…, as we did in the case of a series system.
2.4.2 MTBF of a Parallel System
We recall from section 1.5 (equation 1.7) that MTBF can be expressed as the integral of
reliability, with the limits of integration from 0 to ∞.
∞
i.e 𝑚 = ∫0 𝑅(𝑡)𝑑𝑡
Therefore, the MTBF can be obtained by the integration of the reliability expression. This
method of finding the MTBF does not rely on the expression being of exponential form.
Consequently, the MTBF of a parallel system with two sub-units is:
∞ ∞ ∞

𝑚𝑝2 = ∫ 𝑅𝑃2𝑑𝑡 = ∫ (𝑅1 + 𝑅2 − 𝑅1 𝑅2 )𝑑𝑡 = ∫ (𝑒 −𝜆1𝑡 + 𝑒 −𝜆2𝑡 − 𝑒 −(𝜆1+𝜆2 )𝑡 )𝑑𝑡

0 0 0
1 1 1
+ −
𝜆1 𝜆2 𝜆1 + 𝜆2
Similarly, for a three unit system in parallel connection:
1 1 1 1 1 1 1
𝑚𝑝3 = + + − − − +
𝜆1 𝜆2 𝜆3 𝜆1 + 𝜆2 𝜆2 + 𝜆3 𝜆1 + 𝜆3 𝜆1 + 𝜆2 + 𝜆3
Where, ʎ1, ʎ2, ʎ3 are the unit failure rates respectively. Furthermore, for an n-unit system each
unit having the same failure rate ʎ, the MTBF is:
1 1 1 1
𝑚𝑝𝑛 = + + +⋯+ (2.9)
𝜆 2𝜆 3𝜆 𝑛𝜆

25
N.B: The proof of this result is left as an exercise to the reader.
2.5 Reliability, MTBF of a Series-Parallel System
2.5.1 The Reliability of a Series-Parallel System
Suppose we have a series-parallel system represented by the diagram shown in Fig. 2.2, and we
need to find its reliability.
Ra

Rb

R1 R2 R00
Rz
A general guideline for the determination of the system reliability involves the following steps:
Identify units in series within the system, and calculate the reliability of a single equivalent unit,
using the relationship.
𝑅𝑠 = 𝑅1 × 𝑅2 × 𝑅3 × … … … × 𝑅𝑛
Reduce each parallel arrangement to a single equivalent unit using equation (2.7) to give:
𝑅𝑝 = 1 − [(1 − 𝑅1 )(1 − 𝑅2 ) … … (1 − 𝑅𝑛 )]
Find the product of the results in (i) and (ii) respectively to give the final system reliability, Rsp.
In this case, series-parallel system reliability is.
𝑅𝑠𝑝 = 𝑅𝑠 × 𝑅𝑝 = (𝑅1 × 𝑅2 × 𝑅3 × … … … × 𝑅𝑛 )[1 − (1 − 𝑅1 )(1 − 𝑅2 ) … … (1 − 𝑅𝑛 )]
2.5.2 MTBF of a Series-Parallel System
In order to find the MTBF of a series-parallel system in Fig. 2.2 we have to use the method
discussed in the preceding paragraph i.e. integrating the reliability expression for the series-
parallel system. Simply put, the MTBF will be
∞

𝑚𝑠𝑝 = ∫ 𝑅𝑠𝑝(𝑡)𝑑𝑡
0
∞

= ∫ ((𝑅1 × 𝑅2 × 𝑅3 × … … … × 𝑅𝑛 )[1 − (1 − 𝑅1 )(1 − 𝑅2 ) … … (1 − 𝑅𝑛 )])𝑑𝑡

0
(2.10)
We shall further explain this point and others previously discussed through worked examples.
2.6 Worked Examples
Example 2.1
Compare the reliability of a series system with a parallel system, if each system contains 3 sub-
units having reliabilities 0.95, 0.85 and 0.75 respectively.
Solution:
For a series combination of the 3 sub-units, the reliability
𝑅𝑠 = 0.95 × 0.85 × 0,75 = 0.605
For a parallel combination of the 3 sub-units, the reliability,
𝑅𝑝 = 1 − [(1 − 0.95)(1 − 0.65)(1 − 0.75)] = 0.998

26
Example 2.2
A complex communication satellite has an in-built microwave repeater unit, having a mean
time to failure of 40,000 hours. The link is operative if one channel is working and the
reliability of the switching unit is 0.95. Calculate the reliability for one year operating period
using.
A single channel
Two parallel channels
Three parallel channels
Solution
The diagrammatic illustration of the connection is shown in Fig. 2.3 below:
SINGLE CHANNEL R

R
TWO PARALLEL
CHANNEL

R
R

THREE PARALLEL R
CHANNEL

R
From the data given
1
Failure rate 𝜆 = = 0.25 × 10−4 /ℎ𝑟
4×104
Operating period 𝑡 = 1𝑦𝑟 = 8760 ℎ𝑟𝑠
Using single channel
−4
𝑅1 = 𝑅 = 𝑒 −𝜆1𝑡 = 𝑒 −0.25×8760×10 = 0.803
Using two parallel channels
𝑅2 = 1 − (1 − 𝑅)(1 − 𝑅) = 0.961
Taking the presence of the switching unit into consideration, then we have a parallel-series
combination, and so
𝑅2𝑠 = 𝑅2 × 𝑅𝑠 = 0.961 × 0.95 = 0.913
Using three parallel channels:
𝑅3 = 1 − (1 − 𝑅)3 = 0.992
Similarly,
𝑅3𝑠 = 𝑅3 × 𝑅𝑠 = 0.992 × 0.95 = 0.942

27
Examples 2.3
An electrical power system consists of three sections connected in series. The sections have
mean times between failure of 10,000 hours, 5,000 hours and 4,000 hours respectively.
Calculate the MTBF of the system.
Solution:
If ʎ1, ʎ2, ʎ3 represent the failure rates of each section respectively and the sections are
connected in series, therefore
𝜆𝑠 = 𝜆1 + 𝜆2 + 𝜆3 (from Equation 2.5)
1 1
𝑀𝑠 = 𝜆 = 𝜆 (from Equation 2.6)
𝑠 1 +𝜆2 +𝜆3
1
𝜆1 = = 0.1 × 10−3 /ℎ𝑟
10 × 103
1
𝜆2 = = 0.2 × 10−3 /ℎ𝑟
5 × 103
1
𝜆3 = = 0.25 × 10−3 /ℎ𝑟
4 × 103
1 1
𝑀𝑠 = = = 1818 ℎ𝑟𝑠
𝜆𝑠 (0.1 + 0.2 + 0.25) × 10−3
Example 2.4
Calculate the MTBF of the system shown in Fig. 2.4, if the failure rate of each unit is ʎ.
λ

λ
Solution
Reliability of the parallel blocks,
𝑅𝑝 = 1 − (1 − 𝑅)(1 − 𝑅) = 2𝑅 − 𝑅 2
Reliability of the series blocks,
𝑅𝑠 = 𝑒 −𝜆𝑡
Reliability of the whole system,
𝑅𝑇 = 𝑅(2𝑅 − 𝑅 2 ) = 2𝑅 2 − 𝑅 3
MTBF of the system,
∞ ∞
1 1 2
𝑀𝑇 = ∫ (2𝑅2 − 𝑅3 )𝑑𝑡 = ∫ (2𝑒−2𝜆𝑡 − 𝑒−3𝜆𝑡 )𝑑𝑡 = − =
𝜆 3𝜆 3𝜆
0 0
3 Redundancy
3.1 Definition, Significance and Limitations of Redundancy
The application of redundancy is a method of improving equipment or system reliability.
Redundancy can be defined as the provision of more than one means of getting an
equipment/system to perform a given function. In the context of reliability, application of
redundancy implies that a system will continue to function satisfactorily in spite of failure of

28
some items from which it is built up. However, redundancy represents only a means of
enhancing system reliability. There are other methods.
The following example will further illustrate this concept. The pilot on an aircraft needs an
accurate altimeter to know how high he is flying and since he regards this information very
vital, provision must be made for the situation when there is a fault. A way out of this odd
situation is to provide three identical altimeters in the aircraft so that if one goes faulty, any
two which give the same reading can be identified as correct. In this example, it is clear that
one altimeter might be enough for the height measurement but provision is made for
additional two redundant altimeters in order to achieve system reliability.
The decision to use redundant design techniques must be based on a careful analysis of the
trade-offs involved. Redundancy may prove the only available methods when other techniques
of improving reliability have been exhausted or when methods of part improvement prove to
be more costly than duplications. Its use may offer an advantage when preventive
maintenance is planned. The existence of a redundant element can allow for repair, in some
cases, with no system downtime. Occasionally, situations exist in which equipment cannot be
maintained e.g. space-craft. In such cases, redundant elements may prolong operating time
significantly. However, the application of redundancy is not without limitations. It increases
weight, space, complexity, cost and time to design and maintenance cost.
3.2 Classification of Redundancy
There are two main types of redundancy in common use viz: active redundancy and standby
(passive) redundancy. However, redundancy may further be sub-divided under different
headings as summarized in Fig. 3.1.

REDUDANCY

ACTIVE STANDBY

FULL PARTIAL CONDITIONAL NON-OPERATING OPERATING

SIMPLE BINOMIAL
Fig. 3.1: Classification of Redundancy Techniques
Redundancy does not lend itself to categorization exclusively by element complexity. Certain
types of redundancy are more applicable at the part or circuit level than at the equipment
level. In other words, it may be sufficient to build redundancy into a system not merely by
duplicating the system with another of its type but simply by duplicating certain parts or
circuits in the system. This is so not because of inherent limitations of a particular redundancy
configuration but because of such factors as cost, weight, and complexity.
Active Redundancy

29
In this case, all the units are energized simultaneously to perform a given function. In the
context of active redundancy, the expected function can be performed even if only one out of
the several units is working. This concept is represented diagrammatically in Fig. 3.2.

Figure 2.2: BLOCK DIAGRAM ARRANGEMENT OF ACTIVE REDUNDUNCY

It is worthy of note that the units are drawn as though they are in parallel in the electrical
sense. The electrical analogy is that current will flow X to Y so long as there is an unbroken
circuit through at least one unit/sub-system.
An example of the application of active redundancy concerns the technique of diversity
reception in long-distance radio transmission. Generally three well-spaced aerials are used as
shown in Fig. 3.3.

Fig. 3.3: Illustrating method of Diversity Reception in long distance Radio Transmission

After signal amplification a combining unit automatically selects the largest of the three signals
received by the aerials for subsequent processing. In this case, all the three aerials are actively
working at the same time.
Passive or Standby Redundancy

30
In this case one of the alternative units or systems is energized at a time, and there is provision
to switch in another unit if one fails. For example, in many establishments where continuous
supply of electricity is necessary (e.g. operating theatre of a hospital) provision is made to
switch over to a standby generator if the mains electricity supply fails.
3.3 Mathematical Basis of Redundancy
Here, mathematical expressions will be obtained relating different redundancy configurations
(techniques).
3.3.1 Full (Parallel) Active Redundancy
In line with the concept of active redundancy, in this case a number of units are energized
simultaneously but system operation is maintained so long as at least one unit remains
operating. Here, the addition rule treated in section 2.2.2 applies because the reliability of
such a system is the probability of non-failure of one or more of the units. Hence for n items in
parallel, active redundancy from equation (2.7), is:
𝑅𝑝 = 1 − [(1 − 𝑅1 )(1 − 𝑅2 ) … … (1 − 𝑅𝑛 )]
From equation (2.9), the MTBF for n-units system is:
1 1 1 1
𝑀𝑝𝑛 = + + +⋯+
𝜆 2𝜆 3𝜆 𝑛𝜆
3.3.2 Partial Active Redundancy
In the case of partial active redundancy, the number of units permitted to fail is less than in full
active redundancy. Unlike the full redundancy, more than one path must continue to function
if the system as a whole is to work. An example is an aircraft with four engines. It could almost
certainly land safely if three were still working, but safe landing might be difficult, if not
impossible, with only one engine. Calculation of reliability involving partial redundancy is
based on the use of binomial expression discussed in section 2.2.3. In general, if r items may
fail out of n then the reliability is given as the sum of the first (r + 1) terms of the binomial
expansion
(𝑅 + 𝑄)𝑛 , 𝑊ℎ𝑒𝑟𝑒 𝑅 = 𝑟𝑒𝑙𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑦, 𝑄 = 𝑈𝑛𝑟𝑒𝑙𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑎𝑛𝑑 (𝑅 + 𝑄) = 1. 𝑇ℎ𝑖𝑠 𝑖𝑚𝑝𝑙𝑖𝑒𝑠 𝑡ℎ𝑎𝑡 𝑄
= 1 − 𝑅. 𝑇ℎ𝑒𝑟𝑒𝑓𝑜𝑟𝑒 𝑡ℎ𝑒 𝑟𝑒𝑖𝑙𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑖𝑠:
𝑅𝑠𝑝 = [𝑅 + (1 − 𝑅)]𝑛
𝑛(𝑛 − 1) 𝑛−2
= 𝑅 𝑛 + 𝑛𝑅 𝑛−1 (1 − 𝑅) + 𝑅 (1 − 𝑅)2 + ⋯ + 𝑛𝐶𝑞 𝑅 𝑛−𝑟 (1 − 𝑅)𝑟 + ⋯
2!
+ (1 − 𝑅)𝑛

This concept is further illustrated through a worked example shown immediately below,
Example 2.5
The unreliability of an aircraft engine during flight is 0.01. What is the reliability of successful
flight if the aircraft can complete the flight on at least three out of its four engines?

Solution:
For a successful flight, there are two possibilities:
All the four engines are working i.e. no failure
One of the engines fails.

31
From Binomial expansion
(𝑅 + 𝑄)4 = 𝑅 4 + 4𝑅 3 𝑄 + 6𝑅 2 𝑄 2 + 4𝑅𝑄 3 + 𝑄 4 = 1 (∗)

In this case, the sum of the first 2 terms of the binomial expansion of
(R + Q)4 will give the required reliability.
𝑅𝑠𝑝 = 𝑅 4 + 4𝑅 3 𝑄
𝐵𝑢𝑡 𝑄 = 0.01, 𝑅 = 0.99
𝑅𝑠𝑝 = 0.994 + 4 × 0.993 × 0.01 = 0.999
Example 2.6
A power transmission system shown in Fig. 3.4 consists of a step-up transformer T1, two circuits
of power transmission lines L, and two step-down transformers T2. Full power can be
transmitted to a consumer through any of the two transmission line circuits. The step-down
transformer can handle 50% power only. The unreliability of transformer T1, each transmission
line circuit and each step-down transformer are QT1 = 0.005, QL = 0.03 and QT2 = 0.004
respectively.

T2
LX
T1

Figure 3.4: POWR TRANSMISSION SYSTEM

Failures in all these components are random and independent. Assuming that constant power
is transmitted, determine the reliability of the system under the following conditions:
100% power transmission
50% power transmission

Solution
(i) We note that 100% power transmission is possible on condition that the step-up
transformer T1, at least one of the transmission lines L and both step-down transformers T2
are operational. Reliability of the system for 100% power transmission.
𝑅(100%) = 𝑅𝑇1 × (1 − 𝑄𝐿 2 ) × 𝑅𝑇2 2
𝑅𝑇1 = 1 − 𝑄𝑇1 ; 𝑄𝐿 = 1 − 𝑅𝐿 𝑎𝑛𝑑 𝑅𝑇2 = 1 − 𝑄𝑇2
Using the data given;
𝑅(100%) = 0.995 × (1 − 0.032 ) × 0.9962 = 0.986
(ii) Reliability of the system for 50% power transmission is obtainable on condition that
transformer T1 and at least one of the transmission lines L are functioning while one of the
T2 transformers is non-operational. First, we should obtain the reliability of T2
transformers under its present working condition. To do this, we recall that by binomial
distribution for two events

32
 (𝑅𝑇2 + 𝑄𝑇2 )2 = 𝑅𝑇2 2 + 2𝑅𝑇2 𝑄𝑇2 + 𝑄𝑇2 2
We note that only the middle term 2𝑅𝑇2 𝑄𝑇2 gives the probability that one of the T2
transformers is operational. Therefore, reliability of the system for 50% power transmission is
𝑅(50%) = 𝑅𝑇1 × (1 − 𝑄𝐿 2 ) × 2𝑅𝑇2 𝑄𝑇2
Using the data given as before
𝑅(50%) = 0.995 × (1 − 0.032 ) × 2 × 0.996 × 0.004 = 0.008

N.B.

(1 − 𝑄𝐿 2 ) is the reliability of a unit consisting of two parallel sub-units. This is the alternative
form of the formula given by Equation (2.7) for two items in parallel.

3.3.3 Conditional (Majority Voting) Active Redundancy

In the case of conditional active redundancy, (sometimes called majority voting redundancy) m
out of the n units may be required to be working for the system to function. This is called m-
out-of-n (or m/n) redundancy. This concept is illustrated in Fig. 3.5.
A1

A2
m/n

A3 MVG Output

A4

An
Figure 3.5: Majority Voting (Conditional m/n Redundancy)
This redundancy configuration should not be confused with the idea of partial redundancy just
discussed in the preceding section. Here, we have for example, five processing sections
represented by A1, A2, A3, A5 and A5 and their readings are sent to, say 4/5 majority voting
agate (MVG) which is more or less a decision maker. The output reading of the MVG will be the
same reading as that recorded by the majority (in this case, only 4) of the 5 units. A practical
example can also be found in the aircraft. Aircraft electrical flying controls typically utilize triple
voting active redundancy. A sensing system automatically switches off one unit if it transmits
signals which do not match those transmitted by the other two.

The reliability of an m/n system which components in which all the unit reliabilities are equal, is
the binomial reliability function:
𝑚−1

𝑅𝑚−𝑛 = 1 − ∑ 𝑛𝐶𝑖 𝑅 𝑖 (1 − 𝑅)𝑛−𝑖 (2.12)

𝑖=0
(See later Example 2.7 below for further illustration).

33
In using the formula given by equation (2.12), the following points should be noted:

(i) Reliability of each of the units A1, A2…An is the same and each is equal to R.
(ii) The majority voting gate (MVG) is assumed perfect and therefore has a reliability 𝑅𝑉 = 1.
(iii) If the majority voting gate is not assumed perfect then its reliability 𝑅𝑉 ≠ 1. Consequently
the combined reliability of the system is given by Rm – n x Rv.
Perhaps, it should be mentioned that there is an alternative approach in finding the reliability
of an m/n system without the direct application of the formula given by equation (2.12). An
example will be helpful as an illustration. Suppose, we need to find the reliability of 3-out-of-4
(or ¾) system and the reliability of each of the 4 units is R while the unreliability of each unit is
Q. In this case, for each unit R + Q = 1. Since there are 4 units in the system working at the
same time, then
(𝑅 + 𝑄)(𝑅 + 𝑄)(𝑅 + 𝑄)(𝑅 + 𝑄) = (𝑅 + 𝑄)4 = 1
Expansion of (𝑅 + 𝑄)4 gives:
(𝑅 + 𝑄)4 = 𝑅 4 + 4𝑅 3 𝑄 + 6𝑅 2 𝑄 2 + 4𝑅𝑄 3 + 𝑄 4 = 1 (∗)
This expansion is called binomial distribution and each term has its meaning a explained earlier
in section 2.2.3. Now, we are interested in finding the probability that at least 3-out-of-4 units
will function successfully, and which is equal to the sum of the probability that 3 units will
function successfully and the probability that all the 4 units will function successfully. This is
equal to the sum of the first two terms in the LHS of the binomial expansion.
𝑖. 𝑒. 𝑅3−4 = 𝑅 4 + 4𝑅 3 𝑄
For each unit, 𝑄 = 1 − 𝑅 therefore
𝑅3−4 = 𝑅 4 + 4𝑅 3 (1 − 𝑅) (∗∗)
Equation (**) gives the expected answer according to the alternative approach.

NOTE:
It is not difficult to observe also from equation (*) that
𝑅3−4 = 𝑅 4 + 4𝑅 3 𝑄 = 1 − (6𝑅 2 𝑄 2 + 4𝑅𝑄 3 + 𝑄 4 )
Since for each unit 𝑄 = 1 − 𝑅 therefore
𝑅3−4 = 1 − [6𝑅 2 (1 − 𝑅)2 + 4𝑅(1 − 𝑅)3 + (1 − 𝑅)4 ] (∗∗∗)
The expression given by equation (***) is what should be obtained by directly using the
formula in equation (2.12). In conclusion, it can be stated that we can obtain the value for R3-4
either by adding up the first two terms or by subtracting from 1 the sum of the last three terms
of the expansion (R + Q)4.
So far in our discussion on how to find the reliability of an m/n (m-out-of-n) system with regard
to Fig. 2.8, we have worked under the assumption that each of the units A1, A2, A3, …, An of
the system has the same reliability R and unreliability Q. Suppose this assumption ceases to
hold. Instead, let units A1, A2, …, An have reliabilities of R1 and unreliability Q1; R2 and Q2, …,
Rn and Qn respectively. Now let us consider a simple case when there are 3 units and we are
to find the reliability of 2-out-of-3 (i.e. 2/3) system. As explained earlier, for units A1, A2 and
A3, we have …

34
 𝑅1 + 𝑄1 = 1 𝑅2 + 𝑄2 = 1 𝑎𝑛𝑑 𝑅3 + 𝑄3 = 1
Respectively. Since there are three units in the system, then
(𝑅1 + 𝑄1 )(𝑅2 + 𝑄2 )(𝑅3 + 𝑄3 ) = 1
Expanding the expression on the LHS, we have
𝑅1 𝑅2 𝑅3 + 𝑅1 𝑅2 𝑄3 + 𝑅2 𝑅3 𝑄1 + 𝑅1 𝑄2 𝑄3 + 𝑅2 𝑄1 𝑄3 + 𝑅3 𝑄1 𝑄2 + 𝑄1 𝑄2 𝑄3 = 1
As expected each term in the expression (immediately above) has a meaning
For example:
𝑅1 𝑅2 𝑅3 = 𝑟𝑒𝑙𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑢𝑛𝑖𝑡𝑠 𝐴1 , 𝐴2 𝑎𝑛𝑑 𝐴3
(𝑖𝑒 𝑡ℎ𝑒 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑠𝑢𝑐𝑐𝑒𝑠𝑠𝑓𝑢𝑙 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑢𝑛𝑖𝑡𝑠 𝐴1 , 𝐴2 𝑎𝑛𝑑 𝐴3 )
𝑅1 𝑅2 𝑄3 = 𝑟𝑒𝑙𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑢𝑛𝑖𝑡𝑠 𝐴1 , 𝐴2 𝑎𝑛𝑑 𝑢𝑛𝑟𝑒𝑙𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑢𝑛𝑖𝑡 𝐴3
(𝑖𝑒 𝑡ℎ𝑒 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑠𝑢𝑐𝑐𝑒𝑠𝑠𝑓𝑢𝑙 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑢𝑛𝑖𝑡𝑠 𝐴1 , 𝐴2 𝑎𝑛𝑑 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 𝑜𝑓 𝐴3 )
Now for one case, the reliability of 2-out-of-3 is
𝑅2−3 = 𝑅1 𝑅2 𝑅3 + 𝑅1 𝑅2 𝑄3 + 𝑅1 𝑅3 𝑄2 + 𝑅2 𝑅3 𝑄1
NOTE
Other terms in the LHS of the expression above, containing more than one unreliability (ie
𝑅1 𝑄2 𝑄3 + 𝑅2 𝑄1 𝑄3 + 𝑅3 𝑄1 𝑄2 + 𝑄1 𝑄2 𝑄3) are not useful in finding 𝑅2−3

Example 2.7
Determine the reliability of the system shown in fig 3.6

Fig. 3.6:
Assuming the Majority Voting Gate (MVG) is perfect, then system reliability
𝑅1 × 𝑅𝐵 × 𝑅𝐶
𝑅𝐵 = 1 − (1 − 𝑅1 )(1 − 𝑅3 ) (𝑓𝑟𝑜𝑚 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2.7)

𝑅𝐶 = 1 − [𝑅4 0 (1 − 𝑅4 )3 + 3𝑅4 (1 − 𝑅4 )2 ] (𝑓𝑟𝑜𝑚 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2.12)

= 1 − [(1 − 𝑅4 )3 + 3𝑅4 (1 − 𝑅4 )2 ]
NOTE
From equation 2.12, m = 2, n = 3. Therefore we shall have two terms. (First, when i = 0 and
second, when i = 1).
Alternatively, RC which is the reliability of the 2-out-of-3 system can be obtained by adding the
first two terms of the binomial expansion.
(𝑅 + 𝑄)3 = 𝑅 3 + 3𝑅 2 𝑄 + 3𝑅𝑄 2 + 𝑄 3

35
To get
𝑅𝐶 = 𝑅 3 + 3𝑅 2 𝑄
𝑜𝑟 𝑅𝐶 = 𝑅 3 + 3𝑅 2 (1 − 𝑅) 𝑠𝑖𝑛𝑐𝑒 𝑄 = 1 − 𝑅
The expansion and simplification of the expression for Rc obtained by these two methods will
lead to the same result.
3.3.4 Standby Redundancy
Standby redundancy involves the use of additional units which are only energized when the
operating unit fails. Fig. 3.7 shows n identical units with unit 1 operating. When a failure is
detected, then, unit 2 will be switched in its place.
1

Fig. 3.7: Standby Redundancy

The reliability for standby redundancy is given by the first n terms of the Poisson expression.
(assuming perfect switching)
−𝜆𝑡
𝜆2 𝑡 2 𝜆(𝑛−1) 𝑡 (𝑛−1)
𝑅𝑎𝑏 = 𝑅(𝑡) = 𝑒 (1 + 𝜆𝑡 + +⋯+ ) (2.13)
2! (𝑛 − 1)!
For two units, (one out of which is on standby) the expression reduces to
𝑅𝑎𝑏 = 𝑒 −𝜆𝑡 (1 + 𝜆𝑡)
1. Equation (2.13) makes two important assumptions, amongst others. The assumptions
are:
The sensing device that detects a failure when it occurs and for switching from the defective
unit to the standby unit is assumed to be perfect i.e. failure-free.
2. The standby units are identical and have constant failure rates as the main unit.
Suppose we have two units in standby redundancy with different but constant failure
rates of ʎ1 and ʎ2 respectively, then the reliability of the system.
𝜆2 𝑅1 + 𝜆1 𝑅2 𝜆2 𝑒 −𝜆1𝑡 + 𝜆1 𝑒 −𝜆2𝑡
𝑅𝑎𝑏 = 𝑜𝑟 𝑅𝑎𝑏 = (2.14)
𝜆2 − 𝜆1 𝜆2 − 𝜆1
NOTE:
A simple formula cannot be readily obtained for reliability of a system with more than two
standby redundant units with different failure rates. The following worked exampled will assist
in shedding more light on the concept of standby redundancy.
Example 2.8
A computer and two other similar computers in standby redundancy are available for the
control of a chemical process, such that if a computer fails, another is instantaneously switched

36
on for operation. The failure rate of each computer is 0.01 failure per hour. Calculate the
system reliability if the operating period is 50 hours and the switch is perfect, using.
1. A single computer
2. One standby computer
3. Two standby computers

Solution:
For a single computer, (from equation 2.13 for n = 1)
𝑅(𝑡) = 𝑒 −𝜆𝑡 = 𝑒 −(0.01)(50) = 0.606
For one standby computer, (n = 2)
𝑅(𝑡) = 𝑒 −𝜆𝑡 (1 + 𝜆𝑡) = 𝑒 −0.5𝑡 (1 + 0.5) = 0.909
For two standby computers (n = 3):
𝜆2 𝑡 2
𝑅(𝑡) = 𝑒 −𝜆𝑡 (1 + 𝜆𝑡 + ) = 𝑒 −0.5𝑡 (1 + 0.5 + 0.125) = 0.985
2!
NOTE
The system reliability is improved individually by a factor of approximately 1.5 and 1.625
respectively with the introduction of standby redundant computer(s).

Some Limitations of Standby Redundancy

It is, however, dangerous to rush to conclusion about the merits of standby redundancy as
clearly demonstrated in the preceding three sections without due consideration of certain
limiting factors against standby redundancy. Some of these are as follows:
For many systems, particularly in the field of electronics, stray surges (transient process) may
appear and wander round the circuits at the moment of switch-on. This may be responsible for
the damage of any part with some inherent weakness. With active redundancy, however, the
second unit is switched on with the first and there is time to check that it works properly
before it is finally put into operation.
Some parts of a system may show signs of deterioration if left unloaded for long periods.
It is usually necessary to install further devices (such as a fault detector and a switching unit) in
order to detect the moment when a failure occurs and switch over to a standby unit. This
requirement invariably means more cost to the standby redundant system.
The assumption that both the failure detecting and the switching-over devices are failure-free
is a bit of an over-statement. In reality they may fail, and that may consequently erode the
advantage which standby redundancy claims over active redundancy.

37
4. THE CONCEPTS OF DERATING
As stated in section 1.2.2.3C, Derating is commonly applied where the reduction of operating
voltage and temperature is involved. This leads to the relationship of failure rate (due to
stress) against operating voltages and temperatures.
The mathematical relationship is sometimes called the Arrhenius law, and is expressed as 𝜆2 =
𝑉 𝑛
𝜆1 (𝑉2 ) 𝐾 (𝑡2−𝑡1 ) (4.1)
1
Where, V1, V2; t1 and t2 are voltage and temperature levels respectively, and 𝜆1 , 𝜆2 are the
corresponding failure rates at those levels; K and n are constants. 𝐾 is a constant showing the
effect of temperature for a particular component. If the value of n is assumed to be 5 then
equation (4.1) may be referred to as the fifth power law for a particular component.
However, if the values of constants in K are unknown, then they may be determined by tests
involving, firstly, keeping the voltage constant, and thus determining K, and secondly, keeping
temperature constant, and thus determining n. Mechanical and other stresses are assumed
constant throughout. This may be clarified by considering Equation (4.1), firstly for assumed
𝑉 𝑛
constant voltage, (i.e. V2 = V1 = V), giving 𝜆2 = 𝜆1 (𝑉) 𝐾 (𝑡2 −𝑡1)
Since 𝑛 > 0
𝜆2 = 𝜆1 𝐾 (𝑡2−𝑡1) (4.2)
Hence,
𝜆
𝑙𝑜𝑔𝑒 ( 2 )
𝜆1
𝑙𝑜𝑔𝑒 𝐾 = (4.3)
(𝑡2 − 𝑡1 )
Furthermore, for constant temperature
𝑉 𝑛
𝜆2 = 𝜆1 (𝑉1 )
2
𝜆
𝑙𝑜𝑔𝑒 ( 2 )
𝜆1
𝑛= 𝑉 (4.4)
𝑙𝑜𝑔𝑒 ( 1 )
𝑉2

A worked example is provided immediately below to further illustrate the concept of derating.
Example 1
In a test, to determine failure rate, three sets of observations were recorded under the
following conditions:
(a) Failure rate for rated voltage and rated temperature,
(b) Failure rate for rated voltage and twice rated temperature, and
(c) Failure rate for half rated voltage and rated temperature.
If the failure rates for these conditions are 0.025, 0.100, and 0.00125 respectively, and the
rated temperature is 30oC, calculate the probable failure rate at one-third rated voltage and
two-third rated temperature.
Solution
If the rated temperature is 30oC (given); then twice rated temperature = 60oC and two-third
rated temperature = 2/3 x 30oC = 20oC.
𝐼𝑓 𝑉1 𝑖𝑠 𝑒𝑞𝑢𝑎𝑙 𝑡𝑜 𝑡ℎ𝑒 𝑟𝑎𝑡𝑒𝑑 𝑣𝑜𝑙𝑡𝑎𝑔𝑒 𝑉, 𝑎𝑛𝑑 𝑉2 𝑖𝑠 𝑒𝑞𝑢𝑎𝑙 𝑡𝑜 𝑜𝑛𝑒 𝑡ℎ𝑖𝑟𝑑 𝑟𝑎𝑡𝑒𝑑 𝑣𝑜𝑙𝑡𝑎𝑔𝑒, 𝑡ℎ𝑒𝑛
𝑉 𝑉1 𝑉
𝑉2 = . 𝐶𝑜𝑛𝑠𝑒𝑞𝑢𝑒𝑛𝑡𝑙𝑦, 𝑡ℎ𝑒 𝑟𝑎𝑡𝑖𝑜 𝑖𝑠 𝑒𝑞𝑢𝑎𝑙 𝑡𝑜 /𝑉 𝑜𝑟 1/3 .
3 𝑉2 3

38
From 4.1 for test (b),
(i.e. 𝑉1 = 𝑉2 = 𝑉, 𝑡1 = 30°𝐶 , 𝑡2 = 60°𝐶)
0.100 = 0.025(𝐾)30
𝐾 30 = 4; 𝐾 = 1.047
For test (c),
0.00125 = 0.025(1/2)𝑛 × 𝐾 0 ; 𝑛 = 4.32
𝑉2 4.32
𝑇ℎ𝑒 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 𝑚𝑎𝑦 𝑡ℎ𝑢𝑠 𝑏𝑒 𝑎𝑠𝑠𝑢𝑚𝑒𝑑 𝑡𝑜 𝑏𝑒 𝜆2 = 0.025 ( ) × 1.047(𝑡2−𝑡1)
𝑉1
And for one third rated voltage and two third rated temperature,
𝜆2 = 0.025(1/3)4.32 × 1.047(20−30) = 0.0001371
Comment: There is a remarkable decrease in failure rate from 0.025 to
0.0001371, which proves convincingly the benefit derived from derating.
Derating as Applied to a Transistor
An operating component, particularly a high-power semiconductor device, generates heat
which in turn causes an internal temperature rise. As discussed in Section 3.3.2, the use of a
heat sink is a method of efficiently removing excess heat so that the temperature rise is kept
within normal operational limits.
In this section we shall concern ourselves with further knowledge of heat sink as it relates to
reliability studies in determining whether transistors should be derated that is, used below
their specified power dissipation. For avoidance of doubt, it should be mentioned that all
semiconductor devices have a collector dissipation rating that is stated as a function of
temperature. These dissipation ratings are established by the semiconductor manufacturer. At
a certain temperature, the crystalline structure of the semiconductor may be destroyed due to
excess heat. The critical point in the transistor is the junction between the collector and the
base. The maximum allowable junction temperature is usually given as T j, maximum in degrees
centigrade. The lower limit of the operating temperature of a semiconductor is taken as -65oC.
Therefore the operating temperature range of a semiconductor is …
−65°𝐶 ≤ 𝑇𝑗 ≤ 𝑇𝑗,𝑚𝑎𝑥
This specification may read, for example,
−65°𝐶 ≤ 𝑇𝑗 ≤ 100°𝐶
Furthermore, it should be mentioned that a heat sink is a mechanical device connected to the
case of the semiconductor that provides a path to lead away the generated heat, as shown in
Figure 4.1. The heat flows through the heat sink and is carried off to the surrounding air. If a
heat sink is not used, all the heat must transfer from the case to the surrounding air. This
situation, from the point of view of failure rate, is dangerous for the semiconductor. However,
the use of a heat sink causes the temperature of the case (external metal enclosure) of the
semi-conductor to be considerably lowered.
All the heat generated at a collector junction could not be transferred out of the transistor
instantaneously because there is a finite thermal lag. Therefore, heat can only flow in a path
where there is a temperature difference. The existence of a temperature difference between
two points implies that there is a thermal resistance between the points.

39
Figure 4.1 (a) Details of the heat-sink mounting of a power transistor.
(b) Typical high-power heat-sink.
N.B. The mica insulator or washer between the transistor and the metal heat sink ensures
that the transistor collector (which is internally connected to the case) does not short to the
heat sink.
This heat problem is very similar to a simple electric circuit and suggests the application of the
Ohm’s law. The quantity of heat flow corresponding to electric current, I, is the heat flow Pc,
(Watts) and the heat unit which corresponds to voltage (potential difference between two
points), V, is the difference in temperature (T2 – T1) across the element. The analogy between
thermal and electrical circuits is further illustrated in Figure 4.2.
T1 V1

Pc I

θ R

T2 V2

Figure 4.2

Then by Ohm’s law concept, we obtain thermal resistance as:

𝑇2 − 𝑇1
𝜃= (4.5𝑎)
𝑃𝑐
Other useful forms in which the last expression may be put are:

40
𝑇2 = 𝑇1 + 𝜃𝑃𝑐 (4.5𝑏)
𝑇2 − 𝑇1 = 𝜃𝑃𝑐 (4.5𝑐)
𝑇2 −𝑇1
𝑃𝑐 = (4.5𝑑)
𝜃
In relation to a transistor, it can be said that a transistor junction at temperature 𝑇𝑗 , having a
thermal resistance 𝜃 will be able to dissipate heat to its surroundings at ambient temperature
𝑇𝐴 , where
𝑇𝑗 = 𝑇𝐴 + 𝜃𝑃𝑐
In working with heat-sink problems, it is usual to make use of a series thermal circuit shown in
Figure 4.3. The junction temperature Tj at the base-collector junction (J) of a transistor is the
specified maximum value. The heat dissipation at the collector Pc (in Watts) flows through the
transistor to the case (c) and results in a case temperature, T c. There is a separation between
the transistor case and the heat sink that creates a thermal resistance. Special silicon grease is
often used to establish a good heat-conducting path between the case and the heat sink. As a
result, the sink temperature, Ts differs from the case temperature, T c. Example of thermal
circuit for a BD131 transistor operating at 10 W in an ambient of 50 oC is shown in Figure 4.4.

J(Junction) C(Case) S(Sink) A(Ambient)

θjc θcs θsA

Tj(Junction Ts
Tc TA
temperature)
Figure 4.3

J(Junction) C(Case) S(Sink) A(Ambient)

θJc = 8°C/W θcs = 1.8°C/W θsA = 2.5°C/W

180°C 90°C 80°C

Figure 4.4

41
The heat sink (S) has fins designed to transfer the heat to the surrounding air (A) which is at the
ambient temperature (TA).
As in electric circuit theory, the total thermal resistance is the sum of the individual resistance
in series,
𝜃𝑗𝐴 = 𝜃𝑗𝑐 + 𝜃𝑐𝑠 + 𝜃𝑠𝐴 (4.6)
And the overall equation can be expressed as:
𝑇𝑗 − 𝑇𝐴 = 𝜃𝑗𝐴 𝑃𝑐 (4.7)
Combining equations (3.6) and (3.7) we obtain:
𝑇𝑗 − 𝑇𝐴 = 𝑃𝑐 (𝜃𝑗𝑐 + 𝜃𝑐𝑠 + 𝜃𝑠𝐴 ) (4.8)
Equation (3.8) is often used for reliability studies to determine whether a transistor should be
derated. A worked example is provided immediately below to further illustrate some of the
ideas already discussed.
Example 2
A batch of transistors tested at a junction temperature of 150oC gave a failure rate of
0.10%/103hr. If the internal thermal resistance from junction to case is 50oC per Watt, and the
heat resistance between the case and the surrounding (ambient) is 180oC per Watt, calculate
the maximum transistor junction dissipation which can be permitted if the failure rate is not to
exceed 0.01 %/103hr. Assume that the failure rate increases by a factor of 2.1 for every 10 oC
increase in junction temperature. The ambient temperature is 45 oC.
Solution
The relevant thermal circuit is shown in Figure 4.5
TJ = 150°C/W

θJc = 50°C/W

Pc

S θcA = 180°C/W

A TA = 40°C/W

Figure 4.5

42
In this problem, we have no information concerning change of applied voltage. Therefore we
can apply equation (3.1) by putting V1 = V2 = V, and so
𝜆2 = 𝜆1 𝐾 (𝑡2−𝑡1) (∗)
Next, we find the constant K from the last expression using the fact that failure rate increases
by a factor of 2.1 for every 10oC temperature rise.
𝜆2
= 𝐾 (10)
𝜆1
𝑇ℎ𝑒𝑟𝑒𝑓𝑜𝑟𝑒 𝐾 = 1.077.
𝑈𝑠𝑖𝑛𝑔 𝜆1 = 0.10%/103 ℎ𝑟;
𝜆2 = 0.01%/103 ℎ𝑟
𝐾 = 1.077 𝑎𝑛𝑑 𝑇𝑗1 = 150°𝐶 𝑖𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 (∗), 𝑤𝑒 𝑔𝑒𝑡:
0.01 = 0.10 × 1.077(𝑇2𝑗−150)
0.1 = 1.077(𝑇2𝑗 −150) (∗∗)
𝐼𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 (∗∗), 𝑎𝑠𝑠𝑢𝑚𝑒 𝑇2𝑗 − 150 = 𝑥
0.1 = 1.077(𝑥) (∗∗)
𝑥 = −31°𝐶
𝑇2𝑗 = 119°𝐶
𝑈𝑠𝑖𝑛𝑔 𝑡ℎ𝑒 𝑔𝑖𝑣𝑒𝑛 𝑖𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 (3.8), 𝑇2𝑗 − 45 = 𝑃𝑐 (𝜃𝑗𝑐 + 𝜃𝑐𝐴 )
119 − 45 = 𝑃𝑐 (50 + 180)
𝑃𝑐 = 0.32 𝑊
Worst-case Design Method
The concept of worst-case design is based on the assumption that all component values in an
electronic circuit are at the extreme acceptable value and are so chosen that the circuit is most
likely to fail. Sometimes a designer resorts to the use of worst-case design method in order to
predict how an equipment would perform if all the tolerances were unfavorable. In other
words, he wants to be able to predict a design situation that may cause failure of an electronic
circuit. In most cases where there are large numbers of components and parts, it is almost
impossible to have a situation when all the parts will be at the worst end of their tolerance
band in the same unit. In practical situations, some tolerance may be unfavorable while others
will be favorably disposed.
The worst-case design requires the use of components with high ratings, taking into account
the unlikely situation that all the tolerances are at their extreme values. This method has two
disadvantages. Firstly, the requirement of components with high ratings is synonymous with
high cost. Secondly, the probability of a degradation failure is increased by a reduction in the
allowable tolerance. For these reasons the use of statistical design method, (earlier discussed)
is preferred. In that situation, the design is carried out on the assumption that extreme values
will not occur in the same circuit, and the design is based on a calculated probability that
certain ratings will not be exceeded. However, it should be noted that statistical design
method requires generally a greater amount of computation than the worst-case design and so
its use is justified for circuits which are produced in large quantities. Conversely, the worst-
case design is more suitable for production of items needed in small quantity.
A worked example provided immediately below is helpful in throwing some light on this topic.

43
Example 3
A series electrical circuit consists of a d.c. power supply voltage E and two resistors R1 and R2,
as shown in Figure 4.6. Given that R1 = 50 Ohms ± 20 percent, R2 = 100 Ohms ± 10 percent,
and E = 12V ± 10 percent. Determine the worst-case dissipation of power in R2.
R1

E R2

Figure 4.6
Solution
If the power dissipation in 𝑅2 is 𝑃2 then
𝐸 2 𝑅2
𝑃2 =
(𝑅1 + 𝑅2 )2
𝑅1 = 𝑅10 (1 ∓ 𝐶1 ) = 50(1 ∓ 0.2) 𝑂ℎ𝑚𝑠
𝑅2 = 𝑅20 (1 ∓ 𝐶2 ) = 100(1 ∓ 0.1) 𝑂ℎ𝑚𝑠
𝐸 = 𝐸0 (1 ∓ 𝐶3 ) = 12(1 ∓ 0.1) 𝑂ℎ𝑚𝑠
𝐸02 (1 ∓ 𝐶3 )2 𝑅20 (1 ∓ 𝐶2 )
𝐻𝑒𝑛𝑐𝑒 𝑃2 =
(𝑅10 + 𝑅20 ∓ ∆𝑅1 ∓ ∆𝑅2 )2
𝑊ℎ𝑒𝑟𝑒 ∆𝑅1 = 𝑅10 𝐶1 𝑎𝑛𝑑 ∆𝑅2 = 𝑅20 𝐶2
𝑅𝑒𝑎𝑟𝑟𝑎𝑛𝑔𝑖𝑛𝑔 𝑡ℎ𝑒 𝑙𝑎𝑠𝑡 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 𝑔𝑖𝑣𝑒𝑠:
𝐸02 𝑅20 (1 ∓ 𝐶3 )2 (1 ∓ 𝐶2 )
𝑃2 = 2
(𝑅10 + 𝑅20 )2 𝑅 𝑅
[1 ∓ (𝑅 +10𝑅 ) 𝐶1 + (𝑅 +20𝑅 ) 𝐶2 ]
10 20 10 20

𝑃20 (1 ∓ 𝐶3 )2 (1 ∓ 𝐶2 )
𝑖. 𝑒. 𝑃2 = 2 (∗)
𝑅10 𝑅20
[1 ∓ (𝑅 + 𝑅 ) 𝐶1 + (𝑅 + 𝑅 ) 𝐶2 ]
10 20 10 20

𝐸02 𝑅20
𝑊ℎ𝑒𝑟𝑒 𝑃20 =
(𝑅10 + 𝑅20 )2
From Equation (*), power 𝑃2 in resistor 𝑅2 will be maximum when E is maximum, and 𝑅1 is
minimum. By the maximum power transfer theorem, it is expected that 𝑅2 should be equal to
𝑅1 . (i.e. 𝑅1 = 𝑅2 ) for 𝑃2 max., but in this case, the values of 𝑅1 and 𝑅2 are already fixed.
Therefore, it stands to reason to deduce that 𝑃2 is maximum in this case when the value of 𝑅2
is closest to the value of 𝑅1 ; we shall therefore take the lower limit of 𝑅2 . Hence
𝐻𝑒𝑛𝑐𝑒 𝑃2 = (𝑃𝑜𝑤𝑒𝑟 𝑑𝑖𝑠𝑠𝑖𝑝𝑎𝑡𝑒𝑑 𝑏𝑦 𝑅2 𝑤ℎ𝑒𝑛 𝑎𝑙𝑙 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡𝑠 ℎ𝑎𝑣𝑒 𝑛𝑜𝑟𝑚𝑖𝑛𝑎𝑙 𝑣𝑎𝑙𝑢𝑒𝑠)
× 𝑇𝑜𝑙𝑒𝑟𝑎𝑛𝑐𝑒 𝑓𝑎𝑐𝑡𝑜𝑟

44
 122 × 100
𝐼𝑛 𝑡ℎ𝑖𝑠 𝑐𝑎𝑠𝑒 𝑃2 = = 0.64 𝑊
1502
For maximum value of tolerance, we take only the positive signs in the numerator and the
negative signs in the denominator of Equation (*). Hence,
(1.2)2 × 1.1
𝑇𝑜𝑙𝑒𝑟𝑎𝑛𝑐𝑒 𝑓𝑎𝑐𝑡𝑜𝑟 = 2 = 2.1
50 100
[1 − (100 × 0.2) − ( × 0.1)]
150
Hence, Power P2 in the worst-case = 0.64 x 2.1 = 1.34 W
From this result, we can deduce that the power P2 in R2 is equal to 0.64 W if all the nominal
values of R1 and R2 are used. However, P2 in the worst-case is equal to 1.34 W.

45