Reliability Engineering
System Reliability Modelling
Dr. Monalisa Sarma
Subir Chowdhury School of Quality and Reliability
IIT Kharagpur
This presentation includes…
Introduction to System Reliability Modelling
Series Model
Parallel Model
Series Parallel Model
Parallel Series Model
k out of m Model
Non Series-Parallel Model (Complex System Modelling)
What is System Reliability Modelling?
System Reliability Modelling
How to evaluate the Reliability of a System
Requirement for reliability modelling
Input Output conditions are specified
Conditions of use are known in advance
Interactions among the constituent components
System Reliability Modelling
Reliability Logic Diagram
Reliability Block Diagram
System Reliability Modelling
RLD/RBD may not be synonymous with the physical interconnections of
various constituent elements within a system in relation to the output
performance.
System Reliability Modelling
Assumptions of Modelling
The components fail independently. This means that the failure
of one element does not affect the probability of failure of
other elements.
Each component may be represented as a two state device.
Series Model
Series Model
The system reliability, Rs, for the series model is expressed as :
where, E1, E2,…En refer to the events of units (or components) 1,2,
…n being functionally good, respectively.
where pi is the reliability of the ith component
in the system
Series Model
Reliability of a series model can also be derived from the time to failure
consideration. Let t1 be the random variable which represents the time to
failure of the ith element, the system reliability of an n element system can be
written as
where t is the mission time.
Assuming the independence of events, it can be expressed by
since pi = P(t1>t)
Tutorial 1
Consider a system having three components in series. The
reliabilities of these components are 0.7, 0.6, 0.9
respectively. Find the system reliability .
Series Model
Following observations can be made for a series model :
A series model provides a lower limit of system
reliability.
The reliability is worse than the worst element.
Since on any one unit failure, the system fails, the
dependency or independency of failures would make no
difference in the series reliability model.
Also for the same reason, maintenance does not have any
effect on the system reliability.
Tutorial 3
The failure rate of three components are 0.065 X 10-3, 0.18
X 10-3 and 0.96 X 10-3 per hour. Evaluate the failure rate,
MTTF of the system and the reliability at 500 hours if these
components are connected in series.
Parallel Model
Parallel Model
The reliability of a parallel system is given by
where E1,E2,..., Em are the events of 1,2, …,mth
unit being good.
Tutorial 3
The reliability of each unit in a parallel system is 0.75. Find
out the minimum number of parallel units required to meet
the mission reliability of 0.9950.
Parallel Model
Suppose we want a mission reliability of 0.99
This can be achieved by two unit of reliability 0.9 each
The same can be achieved by three units of 0.8 each
The reliability of the parallel system is greater than
the reliability of the most reliable unit
Parallel Model
=
=
=
• Adding the first component to an existing unit is the most beneficial
• The benefit of adding subsequent parallel unit progressively reduces as
the number of parallel component increases
• Further, weight, space and cost increase with each additional unit
Tutorial 5
The failure rate of three components are 0.065 X 10-3, 0.18
X 10-3 and 0.96 X 10-3 per hour. Evaluate the failure rate,
MTTF of the system and the reliability at 500 hours if these
components are connected in parallel.
Series Parallel Model
This arrangement is also called high level redundancy
Parallel Series Model
This arrangement is also called low level redundancy
Tutorial 6
Find out which configuration is better, series parallel or
parallel series configuration, assuming the same set of
components are used in both the configuration.
Hierarchical Series-Parallel System
This type of system is also called partially redundant system
Derive a general expression for the reliability of the model and hence evaluate the
unreliability of the system if all components have a reliability of 0.8.
k-out of-m system
In many situations, a system functions properly if any k out of m units function
properly. If each of the units is identical, the probability of exactly k successes out
of m is given by,
where p is probability of success of any unit.
The system remains functional as long as k, k+1….m-1, or m units function.
Therefore, the probability of system success is given by,
Such models are sometimes called as k-out-of-m:Good or simply k-out-of-m:G
models since k units are supposed to be good, out of the m units, for the system to
be good.
It may be sometimes useful to compute Rs from the complementary relationship:
If k units must fail for the system to fail, system unreliability is:
k-out of-m system
In k-out of-m system if the components or units are not identical, then
enumeration of all successful combinations are desired
The reliability logic diagram will have paths
Each path will have k elements in series.
k-out of-m system
Standby Model
Redundancy modes.
(a) Parallel redundancy. (b) Standby redundancy
Standby Model cont…
In parallel model, all units of the subsystems are operating
in parallel at any given time
However, in a standby subsystem, one or more units are
in standby mode waiting to take over the operation from
the main operating unit or basic unit as soon as the failure
of the basic unit takes place.
The standby units are not operative until a decision
switch (DS) senses the failure of the basic unit and
switches on the operation to the first standby unit and at
the same time takes the basic unit off the line.
Standby Model cont…
The function of the decision switch is critical and for this
reason, the DS should be highly reliable.
The decision switch can be a human being, a mechanical
gadget, an electromechanical or an electronic device depending
upon the sophistication and automation required.
If there are more than one unit, they would be switched into
operation one-by-one in sequence upon the failure of the
operational unit.
It is also assumed that the operation of failure sensing and
switching on to the standby unit is without loss of time or is
instantaneous such that the proper system operation continues
uninterrupted.
Reliability of Standby Model
Equation (2) appears to be identical
to the Equation for parallel system
Hence it gives the impression that the
probability of failure of a standby
redundant system is identical to that
of a parallel redundant system.
This is not true however since B is
The probability of system failure is used only for short periods it is not
……….(1) likely that its failure probability will
(considering perfect switching) be the same as if it is used
continuously.
which, if it is assumed that A and This leads to the necessity of
B are independent, reduces to: considering time dependent
. ……………(2) probabilities.
Reliability of Standby Model cont…
Assuming two identical units, perfect switching, and both normally operating and
standby components have a constant hazard rate λ.
(a) 2-component system
• The arrangement can be regarded as an equivalent single unit which is allowed to fail
once.
• After the first failure of the equivalent unit (failure of A), the standby component (B)
takes over for the remainder of the mission and therefore the system does not fail.
• If there is a second failure of the equivalent unit (failure of B), the system also fails .
• The logic of this system operation implies that the Poisson distribution can be used to
evaluate the probability of system.
• In this case it is necessary to find the probability of atmost one failure.
Reliability of Standby Model cont…
Using Poisson distribution:
P[no components fail] = (t) =
P[ exactly one component fails] = (t) =
Therefore, the reliability of the system is:
Reliability of Standby Model cont…
2 standby components
n standby components
Mean Time to Failure
For a single standby component
For n standby components
Example
Compare the reliability of a 2-component system each
having a failure rate of 0.02 f/hr after a time of 10 hr if they
are (a) parallel redundant and, (b) standby redundant with a
100% reliable sensing and changeover device. Also,
compare the MTT'Fs of the two systems.
Reliability of Standby Model
Imperfect switching (single standby component)
Let be the probability of successful operation of the
sensing and changeover device.
P[no components fail] = (t) =
P[ exactly one component fails and the sensing and
switchover device operates] = (t). =
Therefore, the reliability of the system is:
Complex System Modelling
Decomposition Method
Also known as Conditional probability approach.
A D
C E
(a) A complex system
Decomposition Method
A D
C E
(a) A complex system
A D
B (b) > What system is this?
> It this same as figure (a)?
C E
Decomposition Method
A D
C E
(a) A complex system
Based on the theorem of total probability (Bayes’ theorem)
Selecting a keystone component say B
R = P(System good|)P(B) + P(System good|)P()
Example 1
Input 1 2 3
4 Output
6
Input 5 7
Calculate the reliability of the system if all the individual
components have a reliability of 0.98
Example
Input 1 3
Input 1
Output
Output
6
Input 5 7
Input 5
8
Input 1 2 3
6
Output
Input 5 7 Output 6
8 Input 5 7
8
Cut-Set Method
A cut set is a set of system components which, when
failed, causes failure of the system.
A set of components which must fail in order to disrupt all
paths between the input and the output of the reliability
network.
A minimum cut set is the one that contains no other cut-
sets within it.
The unreliability of the system is given by the probability
that at least one minimal cut-set fails.
Cut-Set Method
It can be easily programmed on a digital computer for the
fast and efficient solution of any general network.
The cut sets are directly related to the modes of system
failure and therefore identify the distinct and discrete
ways in which a system may fail.
Approximations can be used to simplify evaluation.
Cut-Set Method
Cut-Set Method
Cut-Set Method
Cut-Set Method
Example 2
Find the cut sets of the following figure:
2 3
Input 1 Output
4
Path-Set Method / Tie-Set Method
Path-Set Method
A path set is the complete path through the reliability
block diagram
Minimum path set is defined as the path set that contains
no other path set within it
Reliability of a system is given by the union of all
minimal path sets
Path-Set Method
Path-Set Method
Event Space Method
This method is also called Event tree.
This method is based on listing all possible logical
occurrences of the system.
The reliability of the system is then determined by the
union of all successful occurrences
The number of occurrences depends on the number of
components in the system.
For example a system with 5 components will have
occurrences.
Boolean Truth Table Method
Based on the construction of Boolean truth table for the
system
Very easily can be done using a computer
Similar to the event space method where every possible
state of the system is listed
A state refers to the condition of a component as
functioning or not.
Reliability of the system is obtained by adding all
functioning state probabilities
Boolean Truth Table Method
Reduction Method
It is based on the standard truth table method and then
applying the resulting mutually exclusive sum-of-products
terms.
The procedure starts by constructing a truth table (n is the
number of components in the system)
Each row in the table is then examined, and rows
resulting in system success are indicated.
A reduction table is then constructed by listing all success
rows in Column 1.
Reduction Method
Now we look for two terms from column 1 which differ in
only one literal and combine them into a single term at
column 2.
We can construct the rest of column 2 entries in this manner.
This process is once again repeated for combining the
entries of column 2 to obtain the entries of column 3 and
subsequently in the same manner for column 4.
Once a term is used in a combination, it is eliminated from
all further considerations, to ensure that all the retained
terms are mutually exclusive.
Reduction Method
This process is repeated until no further comparisons are
possible.
The reliability of the system is the union of all terms that
cannot be further compared.
The order of the terms selected for the comparison process has
no effect on the estimation of system reliability.
Functional
States
ABCDE
ABCD
ABCDE
Reduction Table ABC
ABCDE
ABCD
ABCDE
ABCDE
ABCE
E ABCDE
ABCDE
A B C ABCDE ABCD ACE
ABCDE
D ABCDE
ABCE
ABCDE
ABCDE
ABCD
ABCDE
ACD
ABCDE
ABCD
ABCDE
