Frequency and duration concepts
We consider a single repairable component, the state space diagram of
which is shown in the Fig-1. below.
Fig.1
Using continuous Markov process, the probability of residing in the
operable state (availability) and the probability of residing in the
failed state (unavailability) are given by:
P0 = ------- (1)
P1 = ------- (2)
The method of evaluating the frequency and duration indices is based
on the following concepts.
From the above equations, the probability of residing in any state
of the system is equal to the mean residence time of that state divided
by the mean cycle time for that state to occur. This concept, although
self-evident for the two state system shown in Figure -1, it is
applicable to all repairable systems no matter how many system states
exist.
Therefore, if P(S) is the probability of residing in state S, m(S) is the
mean time spent in state S and T(S) is the mean time between
encounters of state S, then,
( )
( )
( )
Also, P0 = --------- (3)
And P1 = --------- (4)
1
From equation (3) & (4), we can write,
i.e. frequency of encountering the ‘up’ state = = (probability of
being in the state) x (rate of departure from the state)
or, = (probability of NOT being in the state) x (rate of entry
into the state)
Therefore if f(S) is the frequency of encountering a state, P(S) is the
probability of being in the state, ̅ ̅̅̅̅
( ) is the probability of NOT being
in the state, λd(S) is the rate of departure from the state and λe(S) is the
rate of entry into the state.
( ) ( ) ( ) ̅ ̅̅̅̅
( ) ( ) -------- (5)
( )
Since T(S) = 1/f(S), ( )
( ) ( )
Frequency of encountering individual states
We consider a two-component repairable system, the state space of
which is shown below.
The state probabilities are:
(λ )(λ )
λ
(λ )(λ )
2
λ
(λ )(λ )
λ λ
(λ )(λ )
The rates of departure and entry to each state can be identified from
the state space diagram. The rate of departure from one state is the
sum of individual rates of departure to other states.
The rates of departure and entry of each state is tabulated under.
Thus for state -1, the rate of departure is (λ1 + λ2) and the rate of entry
is (µ1+µ2).
Frequency of encountering state-1:
Let f1 = frequency of encountering state-1.
= P1 x (rate of departure from state 1) [From equation 5]
Since state probability of state -1 is
(λ )(λ )
therefore,
( )
(λ )(λ )
Similarly,
( )
(λ )(λ )
( )
(λ )(λ )
( )
(λ )(λ )
3
In the case when both components are identical, that is, λ1 = λ2 = λ
and µ1= µ2= µ,
Mean duration of individual states
The mean duration of each of the system states, i.e., the mean time of
residing in each of the states, can be evaluated directly and simply
from the above Table showing rates of departure .
Therefore, if m1 m2, m3 and m4 are defined as the mean duration of
states 1-4 respectively, then,
In case of identical components,
Cycle time between individual states
The cycle time T is defined as the reciprocal of the frequency of
encounter f. A cycle time for each individual state can therefore be
deduced from the frequency of encountering this state. This value of
cycle time represents the mean time between entering (or departing
from) a given state to next entering (or next departing from) the same
state. In the case of the two component system and considering the
case of identical components,
4
Frequency of encountering cumulated states
There may be several states of the system which have a similar impact
on the system behaviour. States leading to the same system outcome
can be combined or cumulated to give, for example, the system up
state and system down state.
The probability of residing in one of these cumulated states can be
evaluated by simply summating the mutually exclusive probabilities
of each appropriate state. A similar technique can be used to evaluate
the frequency of encountering cumulated states.
We consider the combination of states 3 and 4. We define the
cumulative frequency of combining states 3 and 4 as f34.
The transition frequencies to be included in f34 must include the
frequencies of all transitions that leave and enter the combined state
(3 and 4) but must ignore all transition frequencies that occur between
states 3 and 4 since these do not represent transitions out of the
combined state (3 and 4). Therefore,
Substituting for P3 and P4,
Similarly,
f12 = frequency of encountering the combined state (1 and 2)
5
It is seen from above that f12 and f34 are identical. This is to be
expected since the system has been effectively reduced to two
cumulated states (1 and 2) and (3 and 4), in which case the frequency
of encountering each must be the same. The effective state space
diagram of this system is shown in the Figure below.
For a series system,
Pup = P1
Pdown = P2 + P3 + P4
Therefore f(up)=f1
f(down) = f2 + f3 + f4 - (frequency of encounters between 2 and 3,
between 3 and 4, between 2 and 4).
This system has been effectively reduced to a two state system, i.e.,
state 1 and cumulated states (2,3,4).
Parallel redundant system
Here,
Pup = P1 + P2 + P3
Pdown = P4
Therefore f(down) = f4
f(up) = f1 + f2 + f3 - (frequency of encounters between 1 and 2 and 3)
( ) ( ) ( )
6
The Generation Model:
Examples
A simple numerical example is considered, the system data of which
is described in Table -I. The probability of being in a working state is
0.96 and the probability of being out of service is 0.04.
Draw the three-unit state space diagram and develop the generation
system model using the loss of capacity method.
Solution:
If each unit can exist in two states, then there are 2n states where n
is number of units. Figure -1 represents all possible states for a three
unit state space diagram and the transition modes from one state to
another.
The rate of departure from one state is the sum of individual rates of
departure to other states. For example, for state 5, the rate of departure
from state 5 is the sum of rates of departures: µ2 + µ1 + λ3.
The state probability and state frequency is calculated in Table B. The
frequency of encountering certain state is the probability of being in
the state multiplied by the rate of departure from the state. These rates
are given in Table A and then used in the last column of Table B.
Calculation for State-3 in Table-A:
Capacity out = 20 MW [i.e. 2nd unit of 20 MW is in ‘down’ state and
unit-1 (20 MW) and unit-3 (40 MW) is in ‘up’ state]
Therefore state probability (p3) = 0.96*0.04*0.96= 0.036864
7
State frequency (f3) = p3 * (rate of departure from state-3)
= p3* (λ1 + µ2 + λ3) = 0.036864 * (0.02 + 0.48 + 0.02) = 0.0191693
Table A contains a number of identical capacity states that can be
combined by summing obtained information (in our example, we can
sum data from the states 2 and 3, 4 and 5, and 6 and 7). The results
are shown in Table B.
Table B gives the probability and the frequency of having a given
level of capacity forced out of service and of the complementary level
of capacity in service.
Fig. 1: Three-unit state space diagram
8
References
1. Reliability Evaluation of Engineering Systems – Concepts and Techniques, Roy Billinton and
Roland Allan, 2nd Edition, Springer India Pvt. Ltd. New Delhi.
2. Reliability Evaluation of Power Systems, Roy Billinton and Roland Allan, 2nd Edition, Springer
India Pvt. Ltd. New Delhi.