Reliability and Quality

Control

POLITECNICO DI MILANO

Lesson: Reliability Analysis in the Design Phase

Part II
Loredana Cristaldi
 References POLITECNICO DI MILANO

• A. Birolini, “Reliability Engineering- Theory and Practice”, Springer,

2010
• M. Catelani, L. Cristaldi, M. Lazzaroni, L. Peretto, P. Rinaldi,
“Reliability engineering: basic concept and application in ICT”,
Springer, 2011
• R. Bellinton, R.N. Allan: "Reliability Evaluation of Engineering
Systems"- Plenum Press, NY, 1983.
• K. Dehnad: "Quality control, Robust design and the Taguchi
method", Wadsworth & Brooks, 1989.
• R. Billinton “Power System Reliability Evaluation” Gordon and
Breach, Science Publishers, New York 1974

Reliability and Quality Control 2 /38

 References POLITECNICO DI MILANO

For this lesson

• Slides
• Chapter 2 and Annex A6 of:
 A. Birolini, “Reliability Engineering: Theory and Practice”,
Springer.
• Chapter 3 of:
 M. Catelani, L. Cristaldi, M. Lazzaroni, L. Peretto, P. Rinaldi,
“Reliability engineering: basic concept and application in ICT”,
Springer, 2011.
• Gamma function:
 HASTINGS, C., JR., Approximations for Digital Computers,
PRINCETON UNIV. PRESS (1955).

Reliability and Quality Control 3 /38

 Concept of redundancy POLITECNICO DI MILANO

• Redundancy is the existence of more than one means (in

a item) for performing the required function
• We speak of Pseudo Redundancy when a redundant
element fulfill only a part of the required function
• Redundancy does not imply a duplication of hardware
 Software level
 Time redundancy
• Redundancy is represented with a parallel of block on the
RBD
 Not necessarily in the hardware

Reliability and Quality Control 4 /38

 Concept of redundancy POLITECNICO DI MILANO

• From the operating point of view redundancy can be

classified in Active, Warm and Standby:

• Active: Redundant elements are subjected from the

beginning to the same load as operating elements
 Load sharing is not possible when we consider independent
elements.

• Warm: Redundant element are subjected to a lower load

until one of the operating element fails
 Load sharing is present
 The failure rate in reserve state is lower than in operation
Reliability and Quality Control 5 /38
 Concept of redundancy POLITECNICO DI MILANO

• Standby: Redundant elements are subjected to no load

until one of the operating element fails
 Load sharing is possible
 The failure rate in reserve state is assumed to be zero

Reliability and Quality Control 6 /38

 RBD : k-out-of-n configuration POLITECNICO DI MILANO

• A parallel model consist of n element in active

redundancy.
• A k-out-of-n is a structure where n element are in active
redundancy and k are necessary to perform the required
function and the remaining n-k are in reserve
 Elements have the same λ E1
 1≤k<n
 Ex.: steel cable formed of n strands
Ei
that can withstand foreseen stress
if at least k numbers of strands are intact
En

SYSTEM

Reliability and Quality Control 7 /38

 RBD : k-out-of-n configuration POLITECNICO DI MILANO

• To calculate the reliability of this configuration we use

binomial distribution
• The generic element of the system can assume only two
conditions:
 correct functioning → p → R(t)
 in failure → (1-p) → (1-R(t))
• Indicating with R(t) the reliability of the generic element,
the reliability of the system RS(t) can be expressed as:

n n
RS (t ) = ∑   (R(t )) (1 − R(t ))
i n −i

i =k  i 

Reliability and Quality Control 8 /38

 RBD : k-out-of-n configuration POLITECNICO DI MILANO

n n!
  =
 i  i!(n − i )!

n
n
RS (t ) = ∑   (R(t )) i (1 − R(t )) n−i
i =k  i 

n
( )( )
n
n −i
RS (t ) = ∑   e -λt i
1− e -λt

i =k  i 

Reliability and Quality Control 9 /38

RBD : k-out-of-n configuration POLITECNICO DI MILANO

n
( )( )
n
n −i
RS (t ) = ∑
  e -λt i
1 − e -λt
i =k  i 

MTTFS = ∫0
+∞
( )
 n  n  -λt
∑   e
i
(1 − e -λt
) n −i 
 dt
 i =k  i  

Reliability and Quality Control 10 /38

 RBD : k-out-of-n example POLITECNICO DI MILANO

 3  −i λ t
( )  3  − 2 λt
( )  3  − 3 λt
3
3−i
RS (t ) = ∑   e 1 − e −λt =  e 1− e − λt
+   e
i =2 i  2  3

RS (t ) = 3e −2 λt
- 2e −3 λt
⇒
( )
RS 10 = 3e 4 − 2⋅ λ⋅10 4
- 2e −3⋅ λ⋅10 4
≅ 0,833

Reliability and Quality Control 11 /38

RBD : k-out-of-n example POLITECNICO DI MILANO

RS (t ) = 3e −2 λt
- 2e −3 λt

MTTF = ∫ (3e − 2 λt
- 2e −3 λt
)dt =
−∞

3 2
− = 27778 h ⇒ 3 years
2 ⋅ λ 3⋅ λ

Reliability and Quality Control 12 /38

RBD : k-out-of-n example POLITECNICO DI MILANO

RS (t ) = 3e −2 λt
- 2e −3 λt
MTTF ≅ 27778 h

(
RS t = 27778 = 3e
*
) −2 λ⋅t *
- 2e −3 λ⋅t *
= 0,402

Reliability and Quality Control 13 /38

RBD : k-out-of-n example POLITECNICO DI MILANO

Configuration Reliability model MTTFS

Single e − λt 1/ λ
a
element
b 1-out-of-2 2e − λt − e −2λt 9 / (6 λ)

c 1-out-of-3 3e − λt − 3e −2λt + e −3λt 11 / (6 λ)

d 2-out-of-3 3e −2λt − 2e −3λt 5 / (6 λ)

Reliability and Quality Control 14 /38

 RBD : k-out-of-n example POLITECNICO DI MILANO

R(t) 1

Configuration 0.9

a Single element 0.8

b 1-out-of-2 0.7
cc
c 1-out-of-3 0.6

d 2-out-of-3 0.5 bb

0.4
aa
0.3

0.2

0.1 dd

0
0 50 100 150 200 250
1/ λ 2/ λ

Reliability and Quality Control 15 /38

 Residual reliability POLITECNICO DI MILANO

− λt −1 1
R(t ) = e → R( MTTF ) = e ≅ 0,368 MTTF =
λ

RS (t ) = 2 ⋅ e
−λ t −2⋅λ t 3
−e MTTFS =
2⋅λ

RS (MTTF ) = 2 ⋅ e
−λ 3 −2⋅λ 3
2⋅λ
−e 2⋅λ
≅ 0,397

Reliability and Quality Control 16 /38

 Residual reliability POLITECNICO DI MILANO

− λt −1 1
R(t ) = e → R( MTTF ) = e ≅ 0,368 MTTF =
λ

RS (t ) = 3 ⋅ e
−2 λ t −3⋅λ t
− 2⋅e MTTFS =
5
6⋅λ

RS (MTTFs ) = 3 ⋅ e
−2 λ 5 −3⋅λ 5
6⋅λ
− 2⋅e 6⋅λ
≅ 0,402
Single element

RS ( λ ) = 3 ⋅ e
−2 λ 1 −3⋅λ 1
1 λ
− 2⋅e λ
≅ 0,306
Reliability and Quality Control 17 /38
Weibull Distribution POLITECNICO DI MILANO

 t ≥γ
β
 t −γ 
− 
e  η 
R(t ) = 
 t ≤γ
1

Reliability and Quality Control 18 /38

Weibull Distribution POLITECNICO DI MILANO

γ =0
1
λ=
η

− (λt )β
R(t ) = e

f (t ) = β ⋅ λ ⋅ (λ ⋅ t ) − (λt )β
β −1
⋅e
Reliability and Quality Control 19 /38
 Weibull Distribution POLITECNICO DI MILANO

− (λt )β
R(t ) = e
Reliability

λt

Reliability and Quality Control 20 /38

Weibull Distribution POLITECNICO DI MILANO

h(t ) = β ⋅ λ ⋅ (λ ⋅ t )
β −1

Reliability and Quality Control 21 /38

 Weibull Distribution POLITECNICO DI MILANO

h(t ) = β ⋅ λ ⋅ (λ ⋅ t )
β −1

λ*

β=2

λt
Reliability and Quality Control 22 /38
Weibull Distribution POLITECNICO DI MILANO

γ ∞
 1
MTTF = ∫ R(t )dt + ∫ R(t )dt = γ + η ⋅ Γ1 + 
0 γ  β
=0

Γ(1 + x ) = x ⋅ Γ(x ) Γ(x ) = ∫ e − r ⋅ r x −1dr

0

Reliability and Quality Control 23 /38

Weibull Distribution POLITECNICO DI MILANO

β
  1 
− Γ  1+  
R(MTTF ) = e   β 

Reliability and Quality Control 24 /38

Weibull Distribution POLITECNICO DI MILANO

7
Γ(x ) ≈ + ∑ bk x
1 k

x k =0

Reliability and Quality Control 25 /38

 Weibull Distribution POLITECNICO DI MILANO

1 7
Coeff. Values Γ(x ) ≈ + ∑ bk x k

B0 -0,577191652 x k =0
B1 0,988205891
B2 -0,897056937
B3 0,918206857
B4 -0,756704078
B5 0,482199394
B6 -0,193527818
B7 0,035868342

Reliability and Quality Control 26 /38

 Weibull Distribution POLITECNICO DI MILANO

7
Γ(x ) ≈ + ∑ bk x
1
Γ(1 + x ) = x ⋅ Γ(x )
k

x k =0
β
β β 1 7
1 
k 
  1
− Γ  1+

 
1 1
− ⋅Γ 

  − ⋅ β + bk ⋅
∑  
β  k =0  β 
R(MTTF ) = e   β β β 

=e 
=e   

 1 7 k β
1  
 β ∑
− 1+ ⋅ bk ⋅ 
 
R(MTTF ) = e  k =0 β

Reliability and Quality Control 27 /38

Weibull Distribution POLITECNICO DI MILANO

 1 7 k β
1  
 β ∑
− 1+ ⋅ bk ⋅ 
 
R(MTTF ) = e  k =0 β
R(MTTF)

β
Reliability and Quality Control 28 /38
Weibull Distribution POLITECNICO DI MILANO

 s +t  β  s  β 
−    −  
  η   η  
R( s + t s) = e

Reliability and Quality Control 29 /38

Weibull Distribution POLITECNICO DI MILANO

s +t
h ( x )dx
R( s + t s ) = e ∫s
−

s +t
∫s h ( x )dx

Reliability and Quality Control 30 /38

 Weibull Distribution & DBA POLITECNICO DI MILANO

System characterized by a series structure (we consider only

two item):

h(t ) = β ⋅ λ A ⋅ (λ A ⋅ t ) + β ⋅ λB ⋅ (λB ⋅ t ) = β ⋅ λS ⋅ (λS ⋅ t )

β −1 β −1 β −1

With:
(λs ) β
= (λ A ) + (λB )
β β

Reliability and Quality Control 31 /38

Weibull Distribution & DBA POLITECNICO DI MILANO

− (λS t )β
R(t ) = e

1  1
MTTF = ⋅ Γ1 + 
λs  β 

Reliability and Quality Control 32 /38

Weibull Distribution & DBA POLITECNICO DI MILANO

1
MTTFS =
β β
 1   1 
β   +  
 MTTFA   MTTFB 

β=1

1
MTTFS =
 1   1 
  +  
 MTTFA   MTTFB 
Reliability and Quality Control 33 /38
 Weibull & DBA: example POLITECNICO DI MILANO

1 1
MTBFS = MTBFS =
 1   1  β β
  +    1   1 
β 
 MTTFA   MTTFB   MTTF  +  MTTF 
 A   B 

λS = λ A + λ B =
1
MTBFS =
(10 −3
)
+ 10 − 2 ⋅ h −1 = 0,011 h −1 3
 1   1 
  +
3


3

 1000   100 
1
MTBFS = ⋅ h = 90,91 h 1
0,011 MTBFS = ⋅ h = 99,967 h
0,0100033

Reliability and Quality Control 34 /38

Allocation of redundancy: example POLITECNICO DI MILANO

E1 Ei En

E1 Ei En
Ei

Reliability and Quality Control 35 /38

Allocation of redundancy: example POLITECNICO DI MILANO

RS (t ) = (2 − R1 (t )) ⋅ Rseries (t )

E = 0.4 E = 0.7 E3 = 0.9

1 2

E = 0 .4

Reliability and Quality Control 36 /38

Mixed configuration: example POLITECNICO DI MILANO

E 1 (λ1 ) E 2 (λ2 ) E 3 (λ3 )

E 6 (λ6 ) E 7 (λ7 )
E 4 (λ4 ) E 5 (λ5 )

Reliability and Quality Control 37 /38

Mixed configuration: example POLITECNICO DI MILANO

E 1 (λ1 ) E 2 (λ2 ) E 3 (λ3 )

E 6 (λ6 ) E 7 (λ7 )
E 4 (λ4 ) E 5 (λ5 )

− ( λ1 + λ2 + λ3 + λ6 + λ7 ) t − ( λ4 + λ5 + λ6 + λ7 ) t
R(t) = e +e
− ( λ1 + λ2 + λ3 + λ4 + λ5 + λ6 + λ7 ) t
−e
1 1
MTTF = +
(λ1 + λ2 + λ3 + λ6 + λ7 ) (λ4 + λ5 + λ6 + λ7 )
1
−
(λ1 + λ2 + λ3 + λ4 + λ5 + λ6 + λ7 )
Reliability and Quality Control 38 /38