Reliability Modeling
Cold Standbys with perfect switching
If the switching system is 100% reliable, system reliability is
determined by the n components. Let Ti denote the time to
failure of component i (i = 1, 2, . . . , n) and T denote that of the
entire system.
n
T =∑ T i
i=1
Cold Standbys with perfect switching
Now let’s consider a simple case where the cold standby system is comprised of two
components. The system will survive time t if any of the following two events occurs:
The primary component (whose life is T1) does not fail in time t, i.e T1 ≥ t .
If the primary component fails at time T1 < t, the cold standby component (whose life is T2) continues the function
and does not fail in the remaining time (t − T1 ).
Probabilistically, the event is described by Pr [(T1 < t) and (T2 ≥ t − T1 )]
Since the above two events are mutually exclusive, the system reliability is
t
R (t)=R 1 (t )+∫ f 1 ( τ) . R 2 (t−τ)d τ where Ri and fi are, respectively, the reliability and pdf of component i.
0
Cold Standbys with imperfect switching
Now let’s consider a simple case where the cold standby system is comprised of two
components and the switch has a finite probability of failing. The system will survive
time t if any of the following two events occurs:
The primary component (whose life is T1) does not fail in time t, i.e T1 ≥ t .
If the primary component fails at time T1 < t, the switch gets actuated without failing and the cold standby
component (whose life is T2) continues the function and does not fail in the remaining time (t − T1 ).
R (t )=R 1 (t )+∫ f 1 ( τ ). R S ( τ). R 2 (t − τ)d τ
0
where Ri and fi are, respectively, the reliability and pdf of component i and RS is the reliability of the switch
Homework
Build a reliability model for a cold stand by system with three components
Also think about how you would do reliability calculations for standby systems in the presence of
1. Mixed population
2. Competing failure modes
3. Duty cycles
Warm vs Cold Standbys
When the pdf of the TTF of the standby component is the same in standby mode as it is in active
mode, then components are in a "hot standby" configuration. This is the same type of redundancy
as the simple parallel.
When the pdf of the TTF of the standby component is different in standby mode than in active
mode, then you have a "warm standby" configuration.
If the component cannot fail when in standby, then you have a "cold standby" configuration.
Warm Standbys with perfect switching
Standby Act ive
500 hr 500 hr
Standby mode: Weibull Active mode: Weibull
Shape para = 3 Shape para = 2.5
Scale para = 5000 hr Scale para = 3000 hr
500 3
−( )
5000
R sb (500)=e =0.999
Since Rsb(500) = RA(te) at 500 hr,
t e 2.5
−( )
0.999=e 3000 t e =3000[−ln(0.999)]1/ 2.5
te = 189.32 hr
Warm Standbys with perfect switching
t
R 2, A (t e +t − τ)
R (t )=R 1 (t )+∫ f 1 ( τ). R 2 , sb ( τ) . dτ
0 R 2 A (t e )
R1 is the reliability of the active component
f1 is the pdf of the active component
R2,sb is the reliability of the standby component when in standby mode
R2,A is the reliability of the standby component when in active mode
te is the equivalent operating time for the standby unit if it had been operating at an active mode.
Since R2,sb(τ) = R2,A(te)
t
R (t )=R 1 (t )+∫ f 1 ( τ). R 2 , A (t e +t − τ) d τ
0
te can be calculated by using the equality and inverting the reliability function R2,A(te)
Reliability Across Phases
Phase 1 Phase 2 Phase 3
500 hr 500 hr 1000 hr
The age transfer principle is crucial in understanding the concept of phases.
Components accumulate age (wear-out) as they go from phase to phase.
First it is important to note that a component may have a different failure distribution in different phases,
and may have accumulated a different level of damage in each phase. Thus, to correctly transfer the age of
such a component across phases, the principle of virtual age is used.
Reliability Across Phases
f (Stress)
Consider two stress levels S1 and S2
S 1 3000hr S 2 5000hr S 1 3000hr
Phase 1 Phase 2 Phase 3
500 hr 500 hr 1000 hr
Consider a new component (with b = 2) operating across the above three
phases. Determine its reliability, R(2000)
Reliability Across Phases
S1 3000hr S 2 5000hr S1 3000hr
Phase 1 Phase 2 Phase 3
500 hr 500 hr 1000 hr
2
500
R1 = R(500) e 3000
0.97
T = 3000{ - ln (0.929) }1/2 = 812 hr
2
t
R3 = R(1000 | 812)
R(500) phase1 R(t ) phase2 0.97 e 5000
= R(1812) / R(812)
T { ln R(T )}1/ b
T 5000{ ln(0.97)}1/ b 833hr
R1 x R2 x R3
2
833 500
5000
e
R2 = R(500 | 833) 833
2 0.958
5000
e
R1 x R2 = 0.929
Reliability Across Phases: Homework Example
A B A 2000hr B 2500hr C 2000hr
S1 S1 S1
D 2500hr
S1
A S2
2500hr BS 2 3000hr CS 2 2500hr D 3000hr
S2
C D A 2000hr B 2500hr C 2000hr
S3 S3 S3
D 2500hr
S3
bA 2, bB 2.5, βC 1.5, bD 2
Phase 1 Phase 2 Phase 3
500 hr 500 hr 1000 hr
Duty cycle for A and B is 70% while that for C and D is 100%.
Comp A and B have clocked 200 hrs of operation while comp C and
D have clocked 500 hrs of operation.
Reliability Across Phases with PM
Consider a system with just one component (with b = 2) operating across the above three phases.
Determine the system’s reliability at 2000 hr with and without the scheduled PMs (preventive component
replacement).
S1 3000hr S 2 5000hr S1 3000hr
X X Phase 3
Phase 1 Phase 2
500 hr 500 hr 1000 hr
PM at 700 hr PM at 1400 hr
Reliability Across Phases: Example
Consider a new component with two competing failure modes FM1 and FM2 whose time to occurrence is
Weibull distributed. The component is expected to operate across two phases of a mission where the for
phase2 is half of that for phase1. The distribution parameters of the failure modes are bFM1= 2, FM1= 2000
hr and bFM2= 1, FM2= 1500 hr for phase1 and 2 respectively. Determine the mission reliability.
What is the expected improvement in reliability if the component is preventively replaced at the end of
phase 1?
Phase 1 Phase 2
500 hr 500 hr
What will be the hazard rate function of the component ?
How will it vary across the phases with and without preventive maintenance?
Life Multipliers
MIL-HDBK-217 is very well known in military and commercial industries. It is probably the most
internationally recognized empirical prediction method, by far. The latest version is MIL-HDBK-
217F, which was released in 1991 and had two revisions in 1992 and 1995.
where:
πS is the stress factor According to the handbook, the failure rate
of a commercial ceramic capacitor of
πT is the temperature factor
0.00068 μF capacitance with 80%
πE is the environment factor
operation voltage, working under 30
πQ is the quality factor degrees ambient temperature and "ground
πA is the adjustment factor benign" environment is 0.0217x10-6 hours.
QA