Reliability Parameters

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 1

Reliability Parameters
Figure 1: Reliability parameters

 How to compute reliability parameters from assumed failure

rates?
Prepared by Dr. Osama Al-Jamal Reliability Parameters - 2

1
 Cycle-dependent Operations
 Reliability of a one-shot item (e.g., wooden match) is its
success probability under the specified environmental and
operating conditions, i.e.,
𝑆 𝑆
R = P(s) = =
𝑁 𝑆+𝐹
=1–f (a)
Where:
P(S) = success prob.
N = Number of sequential trials
S = Number of successes
F = Number of failures
f = Failure rate (failures per cycle)

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 3

Cycle-dependent Operations (cont.)

 Reliability of an item that functions through repeated cycles

is also its success probability.
 Example: Electric switch.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 4

2
 Cycle-dependent Operations (cont.)
Example (1):
A billing process that generates monthly statements has a failure
rate of 3.5 per 10 million cycles. What is the reliability of the
process?

Solution:
𝑓 = 3.5 𝑝𝑒𝑟 107 𝑐𝑦𝑐𝑙𝑒𝑠 = 3.5𝑥10−7 𝑝𝑒𝑟 𝑐𝑦𝑐𝑙𝑒

𝑅 = 1 − 𝑓 = 0.99999965

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 5

Time-dependent Operations
 When an item’s function is time-dependent, and it is needed
throughout a time interval (only specified time interval) its
reliability is known as mission reliability:

R = R(t)
Where,
R(t) = Reliability for the specified duration t.
t = specified mission time duration.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 6

3
 Time-dependent Operations (cont.)
 For continuous operating items, mission reliability is not the best
way to define or specify reliability requirements.
 More useful reliability parameters are the mean time between
failures (MTBF) and the availability.
 MTBF tells how frequently, on average, we can expect our item to
experience an outage.
 Availability is the proportion of the time that we can expect our
item to be operating.
 Mean down time (MDT) is the average time that it takes to return
to an operating state after an outage has occurred.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 7

Time-dependent Operations (cont.)

 Outage Rate (complement of availability) is the portion of the time
that the item is down.
Availability (A) = 1 – Outage Rate (OR) (b)
𝑀𝑇𝐵𝐹
A = (c)
𝑀𝑇𝐵𝐹+𝑀𝐷𝑇
𝑀𝐷𝑇
OR = (d)
𝑀𝑇𝐵𝐹+𝑀𝐷𝑇
 Mean Time Between Maintenance Actions (MTBMA) is the
average frequency of maintenance actions on the item, whether or
not these actions impose or are in response to an outage (i.e.,
preventive or corrective).
Item is operable
during maintenance
activities.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 8

4
 Time-dependent Operations (cont.)
 Mean Time To Restore (MTTR) refers to the average time
required for maintenance activities, regardless of whether the
item is operational or not during those activities.
 If the item is always out during maintenance, then
MTBMA = MTBF and MTTR = MDT

Item is inoperable during maintenance activities.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 9

Items in Standby

 The standby item is prepared and ready to be used when

needed, ensuring that it is in a functional and operational
condition.
 Operational readiness is the availability of a standby item
while in standby.
 Dependability is the prob. that a safety standby item will be
operable upon demand and then will remain operating,
without failure, throughout its required demand time.
Dependability = Availability × Mission reliability
D(t) = A × R(t) (e)

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 10

10

5
 Reliability as a function of time
 For a simple non-maintained system (non-recoverable system):
 Non-maintained means that no maintenance or repair activities are
performed on the system once it fails.
 In a non-recoverable system, once the system fails, it remains
inoperable until a replacement or repair is performed.

𝑁𝑠 (𝑡) 𝑁𝑠 (𝑡)
𝑅 𝑡 = =
𝑁𝑜 𝑁𝑠 (𝑡) + 𝑁𝑓 (𝑡)

𝑁𝑓 𝑡
𝑅 𝑡 = 1− = 1−𝑄 𝑡 (1)
𝑁0

Where,
N0 = number of identical items put into operation at time t = 0.
Ns(t) = numbers of survivors and failures at time t (components that have
not failed or experienced any issues until time t).
Nf(t) = numbers of failures at time t.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 11

11

Reliability as a function of time (cont.)

Fig. 2: Reliability versus time for a time-dependent

nonrecoverable system.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 12

12

6
 Reliability as a function of time (cont.)
 Failure density function f(t) is the density function of the Time to
Failure (TTF) distribution, it can be described by:
𝑑𝑅(𝑡)
𝑓 𝑡 =− (2)
𝑑𝑡

𝑑𝑅(𝑡)
 Derivation of 𝑓 𝑡 = − 𝑑𝑡
𝑇
𝑄 𝑇 = 𝐹 𝑇 = න 𝑓 𝑡 𝑑𝑡
0
𝑇 ∞
𝑅 𝑇 = 1 − 𝑄 𝑇 = 1 − න 𝑓 𝑡 𝑑𝑡 = න 𝑓 𝑡 𝑑𝑡
0 𝑇
𝑑𝑅(𝑡) 𝑑 ∞
= න 𝑓 𝑡 𝑑𝑡 = 𝑓 ∞ − 𝑓 𝑇 = 0 − 𝑓 𝑇 = −𝑓(𝑇)
𝑑𝑡 𝑑𝑡 𝑇

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 13

13

Reliability as a function of time (cont.)

Because
𝑁𝑓 (𝑡) 𝑑 1 𝑑
𝑄 𝑡 = ➔ 𝑓 𝑡 = 𝑄 𝑡 = 𝑁𝑓 (𝑡)
𝑁0 𝑑𝑡 𝑁0 𝑑𝑡

But as 𝑡 → ∞, 𝑑𝑁𝑓 ∞ → 0 , Thus, 𝑓 ∞ = 0

 Generally:
𝑑𝑅(𝑡)
𝑓 𝑡 =−
𝑑𝑡
𝑑𝑅 𝑡 = −𝑓 𝑡 𝑑𝑡
𝑡 𝑡
‫׬‬0 𝑑𝑅(𝑡) = − ‫׬‬0 𝑓 𝑡 𝑑𝑡 , since 𝑅 0 = 1
𝑡
𝑅 𝑡 = 1 − න 𝑓 𝑡 𝑑𝑡
0

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 14

14

7
 Reliability as a function of time (cont.)
 The failure density function - that is, the density function of the
TTF distribution - can be developed by differentiating F(t) with
respect to time:
𝑑𝐹 𝑡 𝑑𝑄 𝑡 1 𝑑𝑁𝑓 𝑡
𝑓 𝑡 = = =
𝑑𝑡 𝑑𝑡 𝑁0 𝑑𝑡

 This result is illustrated in Fig. 5, which is the derivative of Fig. 4.

Fig. 3 Fig. 4 Fig. 5

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 15

15

Reliability as a function of time (cont.)

 From above, it can be proved that the unreliability relates to the

failure density function through:
𝑇
𝑄 𝑇 = න 𝑓(𝑡) 𝑑𝑡
𝑡=0
 Consequently, reliability relates to the failure density function:
𝑇 ∞
𝑅 𝑇 = 1 − 𝑄 𝑇 = 1 − න 𝑓(𝑡) 𝑑𝑡 = න 𝑓(𝑡) 𝑑𝑡
𝑡=0 𝑡=𝑇
 By differentiating reliability, R(t), with respect to time, we obtain
the relationship:
𝑑𝑅(𝑡) 𝑑 ∞
= න 𝑓 𝑡 𝑑𝑡 = 𝑓 ∞ − 𝑓 𝑇 = 0 − 𝑓(𝑇)
𝑑𝑡 𝑑𝑡 𝑡=𝑇

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 16

16

8
 Reliability as a function of time (cont.)
 From which we conclude:
𝑑𝑅 𝑇 𝑑𝑅(𝑡)
𝑓 𝑇 =− , or, more generally, 𝑓 𝑡 = − (2)
𝑑𝑡 𝑑𝑡

 If failure rate is function of time 𝜆(𝑡), then reliability can be

defined by:
𝑡
R(t) = 𝑒 − ‫𝑡𝑑)𝑡(𝜆 𝑜׬‬ (3)

where, 𝜆(𝑡) is failure rate as function of time.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 17

17

Reliability as a function of time (cont.)

Derivation of 𝜆(𝑡)
 The instantaneous failure rate 𝜆(𝑡), also known as the hazard rate,
quantifies the rate at which failures happen at a particular moment, given
that the system has survived until that time.
 It is the rate at which items are failing that is, the ratio of the increase
(change) in failure quantity at a given time with respect to time T to the
quantity of survivors at time t:
𝑑𝑁𝑓 (𝑡)Τ𝑑𝑇 1 𝑑𝑅(𝑇)
𝜆 𝑡 = =− −𝑁0
𝑁𝑠 (𝑡) 𝑁𝑠 (𝑡) 𝑑𝑇

𝑁𝑠 (𝑡) 𝑁𝑓 𝑡
Where T is the variable time, and 𝑅 𝑡 = =1−
𝑁0 𝑁0

Thus, 𝑁𝑓 𝑡 = 𝑁0 1 − 𝑅 𝑡 ➔ 𝑑𝑁𝑓 𝑡 = −𝑁0 𝑑𝑅 𝑡

−𝑑𝑅 𝑡 1 𝑑𝑁𝑓 𝑡
Or, 𝑓 𝑡 = = ➔ 𝑑𝑁𝑓 𝑡 = −𝑁0 𝑑𝑅 𝑡
𝑑𝑡 𝑁0 𝑑𝑡

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 18

18

9
 Reliability as a function of time (cont.)
 Thus,
𝑁0 𝑑𝑅(𝑇) 1 𝑑𝑅(𝑇)
𝜆 𝑡 =− =−
𝑁𝑠 (𝑡) 𝑑𝑡 𝑅(𝑡) 𝑑𝑇
𝑑𝑅 𝑇
∴ 𝜆 𝑡 𝑑𝑇 = − = −𝑑 𝑙𝑛𝑅(𝑇)
𝑅 𝑡

 Integrating between the limits 0 and t,

𝑡
𝑇=𝑡
න 𝜆 𝑡 𝑑𝑇 = −𝑙𝑛𝑅 𝑇 ቚ 𝑇=0 = 𝑙𝑛𝑅 0 − 𝑙𝑛𝑅 𝑡
𝑇=0

= 𝑙𝑛 1.0 − 𝑙𝑛𝑅 𝑡 = −𝑙𝑛𝑅 𝑡

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 19

19

Reliability as a function of time (cont.)

𝑡
 Hence, 𝑅 𝑡 = 𝑒 − ‫׬‬0 𝜆 𝑇 𝑑𝑇

 To clarify the relationship between time-dependent reliability and

time that we intuitively developed in Fig. 2 (see Fig. 6).

Fig. 2 Fig. 6 Reliability as a function of

instantaneous failure rate

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 20

20

10
 Reliability as a function of time (cont.)

 For constant failure rate:

𝜆(𝑡) → 𝜆 , 𝑅 𝑡 = 𝑒 −𝜆𝑡 (5)

and
𝑑 𝑑 𝑑
𝑓 𝑡 =− 𝑅 𝑡 = − 𝑒 −𝜆𝑡 = −𝑒 𝜆𝑡 𝜆𝑡
𝑑𝑡 𝑑𝑡 𝑑𝑡

𝑓 𝑡 = −𝜆𝑒 −𝜆𝑡 (6)

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 21

21

Mean time to failure, MTTF

 MTTF is the expected value of time to failure (TTF) that is, the
mean of the TTF distribution.
∞
 Mean of continues distribution is μ = ‫=𝑥׬‬−∞ 𝑥𝑓 𝑥 𝑑𝑥

So, for the variable t which has no negative values, the mean is:
∞ ∞ 𝑑𝑅 𝑡 ∞
𝑀𝑇𝑇𝐹 = ‫=𝑡׬‬0 𝑡 𝑓 𝑡 𝑑𝑡 = − ‫=𝑡׬‬0 𝑡 𝑑𝑡
𝑑𝑡 = − ‫=𝑡׬‬0 𝑡 𝑑𝑅(𝑡)

But ‫ 𝑣𝑢 = 𝑣𝑑𝑢 ׬‬− ‫𝑢𝑑𝑣 ׬‬ , let u = t , du = dt

dv = dR(t) , v = R(t)
∞ ∞
𝑀𝑇𝑇𝐹 = − න 𝑡 𝑑𝑅 𝑡 = − 𝑡 𝑅 𝑡 ቤ ∞0 − න 𝑅 𝑡 𝑑𝑡
𝑡=𝑜 0

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 22

22

11
 Mean time to failure, MTTF
At t = 0 → R(0) = 1 and t = ∞ → R(∞) = 0
∞
Thus, 𝑀𝑇𝑇𝐹 = ‫=𝑇׬‬0 𝑅 𝑡 𝑑𝑡 (7)

 For constant failure rate:

∞ ∞ 𝑡
𝑀𝑇𝑇𝐹 = න 𝑅 𝑡 𝑑𝑡 = න 𝑒 − ‫𝑡𝑑 𝑡𝑑)𝑡(𝜆 𝑜׬‬
𝑡=𝑜 𝑡=0
∞
𝑀𝑇𝑇𝐹 = ‫׬‬0 𝑒 −𝜆𝑡 𝑑𝑡
1
Thus, 𝑀𝑇𝑇𝐹 = (8)
𝜆

 For restorable item Mean Time Between Failures (MTBF) is used.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 23

23

Reliability as a function of time (cont.)

Example (2):
A type of power supply used in a computer is assumed to have a
constant failure rate of 8.5 failures per million hours. What is its
reliability within the system for a 5000-h mission?

Solution:
𝜆 = 8.5𝑥10−6 /ℎ t = 5000 h
𝜆 𝑡 = 5000𝑥8.5𝑥10−6 = 0.0425
𝑅(𝑡) = 𝑒 −0.0425 = 0.958

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 24

24

12
 Reliability as a function of time (cont.)
Example (3):
Compute the MTBF for the power supply of Example 2.

Solution:

𝑀𝑇𝐵𝐹 = 1/𝜆 = 1/8.5 𝑥10−6/ℎ = 118,000 ℎ

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 25

25

Reliability as a function of time (cont.)

Example (4):
If the MDT of the power supply of Examples 2 and 3 is 2 h, find
its availability.

Solution:

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 26

26

13
 Reliability as a function of time (cont.)
Example (5):
An engineer approximates the reliability of a process by:
2
𝑡
𝑅 𝑡 = 1− , 0 ≤ 𝑡 ≤ 𝑡0
𝑡𝑜
0 , 𝑡 ≥ 𝑡0

1. Determine the failure rate (t).

2. Does the failure rate increase or decrease with time?
3. Determine the MTTF.

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 27

27

Reliability as a function of time (cont.)

Solution:
𝑑
1. 𝑓 𝑡 = − 𝑑𝑡 𝑅 𝑡
𝑑 𝑡 2
= − 𝑑𝑡 1 − 𝑡
0
2 𝑡
= 𝑡0
1− 𝑡0
, 0 ≤ 𝑡 ≤ 𝑡0
𝑡0 𝑡
Hence, 𝑅 𝑡 = 𝑒 − ‫׬‬0 𝜆 𝑡 𝑑𝑡
➔ 𝑙𝑛 𝑅 𝑡 = − ‫׬‬0 0 𝜆 𝑡 𝑑𝑡
𝑑𝑅 𝑡 1 1
𝜆 𝑡 =− 𝑑𝑡 𝑅 𝑡
= 𝑓(𝑡) × 𝑅(𝑡)

𝑓(𝑡) 2 2
𝜆 𝑡 = 𝑅(𝑡) = 𝑡 =𝑡 , 0 ≤ 𝑡 ≤ 𝑡0
𝑡0 (1−𝑡 ) 0 −𝑡
0

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 28

28

14
 Reliability as a function of time (cont.)
2
2. (t) increases from at t = 0 to  at t = to.
𝑡0
∞
3. 𝑀𝑇𝑇𝐹 = ‫׬‬0 𝑅 𝑡 𝑑𝑡

𝑡 2
=‫ ׬‬1− 𝑑𝑡 + ‫ ׬‬0 𝑑𝑡
𝑡𝑜

𝑡 𝑡 𝑡2 𝑡2 𝑡3
= ‫׬‬0 0 1 − 2 + 𝑑𝑡 = 𝑡 − + 2ฬ 𝑡00
𝑡0 𝑡02 𝑡0 𝑡0

𝑡0
= 𝑡0 − 𝑡0 +
3

𝑡0
=
3

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 29

29

Reliability as a function of time (cont.)

 Complex Systems: The more complex the system, the more it
appears to behave as though its failure rate is constant. For a
complex, restorable system:
1
𝑀𝑇𝐵𝐹 = (9)
𝜆

 For a system composed of N serial items:

𝑅𝑠𝑦𝑠 𝑡 = 𝑅1 𝑡 × 𝑅2 𝑡 × … × 𝑅𝑁 𝑡 = ς𝑁
𝑖=1 𝑅𝑖 𝑡 (10)

𝑅𝑠𝑦𝑠 𝑡 = 𝑒 −𝜆1 𝑡 × 𝑒 −𝜆2𝑡 × … × 𝑒 −𝜆𝑁 𝑡 = ς𝑁

𝑖=1 𝑒
−𝜆𝑖 𝑡 (11)

From (7) and (11):

∞ 1
𝑀𝑇𝐵𝐹 = ‫׬‬0 ( ς𝑁
𝑖=1 𝑒
−𝜆𝑖 𝑡 )𝑑𝑡 =
σ𝜆
(12)
𝑖

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 30

30

15
 Reliability as a function of time (cont.)
 For increasing failure rate, Weibull distribution is used.

 Weibull probability distribution function:

𝛽
𝑓 𝑡 = 𝛽𝜇 𝛽 𝑡𝛽−1 𝑒 −(𝜇𝑡) (13)

If (t) is increasing with time, then  > 1.0

If (t) is decreasing with time, then  < 1.0

If  is constant, then  = 1.0

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 31

31

Reliability as a function of time (cont.)

 For mechanical and electrical hardware, failure rates are either
constant or increasing.

 If  < 1, then a burn-in period is assumed where failure rate

decreases with time (may be in service function as a consequence
of continuous improvement process).

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 32

32

16
 The End

Prepared by Dr. Osama Al-Jamal Reliability Parameters - 33

33

17