Stochastic Structural Dynamics

Lecture-39

Problem solving session-3

Dr C S Manohar
Department of Civil Engineering
Professor of Structural Engineering
Indian Institute of Science
Bangalore 560 012 India
manohar@civil.iisc.ernet.in

1
Discussion on properties of processes with
Independent increments

2
Problem 35
Let X (t ) be a process with stationary independent increments;
assume t  0 & X  0   0. Show that
 X  t   t;
Var  X  t     2t
Var  X  t   X  s     2  t  s 
Cov  X  t  X  s     2 min  t , s 
Note: clearly, X  t  1   &  2  variance of X (t  1).

3
Let f (t )  X  t 
f (t )  X  t   X  0 
f t  s   X t  s   X 0
 X t  s   X  s   X  s   X 0
 X t  s   X  s   X  s   X  0
 X  t   X  0   X  s   X  0   stationary increments
 f (t )  f ( s )
We get the functional equation f (t  s )  f (t )  f ( s )
 f  t   ct is the solution.
Given f 1  X 1    c  
X  t   t
4
Let g (t )  Var  X  t  
g (t )  Var  X  t    Var  X  t   X  0  
g  t  s   Var  X  t  s   X  0  
 Var  X  t  s   X  s   X  s   X  0  
 Var  X  t  s   X  s    Var  X  s   X  0  
 Var  X  t   X  0    Var  X  s   X  0  
 stationary increments
 g t  s   g t   g  s 
 g  t   ct
g 1  Var  X 1    2  c   2
 Var  X  t     2t
5
Let t  s
Var  X  t    Var  X  t   X  0  
 Var  X  t   X  s   X  s   X  0  
 Var  X  t   X  s    Var  X  s   X  0  
=Var  X  t   X  s    Var  X  s  
 Var  X  t   X  s    Var  X  t    Var  X  s  
  2 t  s 

6
Var  X  t   X  s  

  X  t   X  s   X  t   X  s  
2

   
2
  X t   X t   X  s   X  s  
 

 X t   X t     X  s   X  s  
2 2


2  X  t   X  t   X  s   X  s  

 Var  X  t    Var  X  s    2COV  X  t  , X  s  

7
COV  X  t  , X  s   
1
2
 
-Var  X  t   X  s   +Var  X  t   +Var  X  s  

2
 t  s  (t  s )   2
s  assuming that t  s 
2
 COV  X  t  , X  s     2 min  t , s 

8
Problem 34
Let X (t ) be a stationary Gaussian random process with
zero mean and PSD function of the form
 X2  2 
S XX    exp - 2  ;-    
2  2 
•Determine the autocorrelation and cross correlation
functions of the processes X  t  and X  t 
Find the average rate of upcrossing of level 
Find the PDF of time for first crossing of level 
Find the average rate of peaks above level 
Find the expected fractional occupation time above
level  over a duration 0 to T
Find the PDF of extreme of X (t ) over duration 0 to T
9
Spectral moments
 X2  2 
S XX    exp - 2  ;-    
2  2 

 X2  2 
0   exp - 2 d    X
2

 2  2 

2
   2
2   exp - 2 d    X2  2
2 X

 2  2 

2
   2
4   exp - 2 d   3 X2  4
4 X

 2  2 
10
Autocorrelation function
 2
  
2
S XX    exp - 2  ;-    
X

2  2 

1  X2  2 
RXX     exp - 2  exp  i  d 
2  2  2 
  
2 2
= X exp  
2

 2 
d m n
RXX  
t  X t      1
m n n
Recall: X
d m  n
11
 dRXX   d 2 RXX  
X  t  X  t      ; X  t  X  t    
d d 2
dRXX    d 2 RXX  
X t  X t    
 ; X  t  X  t      ;
d d 2

XX  
 RXX  
3 2
d R d
X t  X t    
  ; X t  X t    

d 3
d 2
XX  
 RXX  
3 4
d R d
X t  X t     
  ; X t  X t    
 
d 3
d 4

12
   
2 2
RXX     X exp  
2

 2 
  
2 2
d
d
RXX     X exp  
2
   2
 
 2 
d2    
2 2

XX    
2
R    2
exp     2

d 2 X
 2 
   
2 2
  X exp  
2
  2
 
 2 
   
2 2
  X  exp  
2 2

 1 
2 2

 2 
13
d2    
2 2

d 2
R XX     2
X  2
exp    1  
2 2

 2 
d3    
2 2

d 3
R XX  
   2
X  2
exp    1 
  
2 2
  2 
 2 
   
2 2
  X  exp  
2 2
2 2
 
 2 
   
2 2
  X  exp  
2 2
3 2
   
4 3

 2 

14
d3    
2 2

d 3
R XX     2
X  2
exp    3 2
  4 
4 3

 2 
d4    
2 2

d 4
RXX     X  exp  
2 2

 3   4   
2 4 3 2
 
 2 
   
2 2
 X  exp  
2 2
 
3 2
 12 
4 2

 2 
   
2 2
  X  exp  
2 2
 
 3 
4 2
 4  
6 4
 3 2
 12 
4 2

 2 
   
2 2
  X  exp  
2 2
 
3 2
 15 
4 2
 4 
6 4

 2 

15
    
2 2
RXX     X exp  
2

 2 
   
2 2
d
d
RXX     X exp  
2

  
2

 2 
d2    
2 2

d 2
RXX     X  exp  
2 2

 1 
2 2

 2 
d3    
2 2

d 3
RXX     X  exp  
2 2

 3   4 
2 4 3

 2 
d4    
2 2

d 4
R XX     2
X  2
exp    3 2
 15 
4 2
 4 
6 4

 2 
16
 4
process
derivative
3

1
acf

-1

-2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
tau

17
Level crossing statistics
T
N   , 0, T    X  t    X  t     dt
0

Average rate of upcrossing of level 

1   1  2

n  ,t 

X t    X t     
   X
exp   
2 2 X  2  2
X 
 X  X 
 
2 X 2 X 2
  1  2
   1 a 2
 
n  ,t 

exp   2 
 exp   2 
;a 
2  2  X   2
2
 2  X

18
PDF of time for first crossing of level 
For high levels of crossings we can approximate
the number of times the level is crossed as a Poisson
random variable.
 T 
k

P  N   , 0, T   k   exp  T  ; k  0,1, 2,

k!
  1 a2  
  n   , t   exp   2 
;a 
  2  X
T f  First passage time
P T f  T   P  No points in 0 to T   P  N   , 0, T   0 
 T  1 a2 
PT f  t   1  exp  T   1  exp   exp   2 
;T  0
   2   19
Average rate of peaks above level 

1
 1   22 32 x 2   23 
m ,t   [ S exp  
2   x
3   
 2   1  2
2
2 2   2 S  1 2

 x 2     2x  
exp   2  1  erf  3  ]dx
 2 1    1 2 S




20
  X 2 t  X  t  X  t  X  t  X  t  
 
S X  t  X  t  X 2  t  X  t  X  t  
 

 X  t  X  t  X  t  X  t  X 2  t  
 1 0  2 
 
  X2  0  2 0   S  3 X6  6 1   2 
  2 0 3 4 

 12   X2
 22   2 X2
 32  3 X2  4

21
Expected fractional occupation time above level 
over a duration 0 to T
T
1
y   , T    U  X  t     dt
T 0

U  X  t      1   p  x; t dx
X


1  x2 
 1  exp   2 dx
 2 x  2 x 
1   
   erf     y   , T 
 2   x  
22
 PDF of extreme of X (t ) over duration 0 to T
PT f  t   1  exp  t 
pT f  t    exp  t  0  t  
  1 a2  
 exp   2 
;a 
2  2  X
PX m     1  P T f     T 

23
 FACTORS OF SAFETY
&
PROBABILITY OF FAILURE

24
Problem 35
In traditional engineering practice, uncertainties in
specifying loads and structural resistance are accounted
for by overestimating the loads and underestimating the
strucutral resistance. The factors by which these estimates
are obtained are calibrated against past experience with
existing stock of strucutres. It is of interest to relate this broad
principle with the probabilistic modeling of uncertainties.
To illustrate this let us consider an idealized situation in which
demands on the structure and supply of structural capacity
are modeled as a pair of mutually independent Gaussian
random variables.
25
The failure event is defined by exceedance of load effect over the
available capacity. If the tolerable level of probability of failure
is specified to be PF , determine the factors by which the expected
load and capacity are to be mutiplied.

Extend the discussion to the case when several loads act

on the structure (like, for example, dead load, live load,
thermal loads, etc.)

26
27
R  N (  R ,  R ); S  N (  S ,  S ); R  S
R S
R  ; S 
R S
Z  R  S  Z  N (  R   S ,  R2   S2 )
   
PF  1    R S

  R2   S2 
   
 R S
  1  PF
  R2   S2 

  R   S   R2   S2  1[1  PF ]
 R   S    R2   S2
If  is large, risk is small 28
   2 2
 R
;   0.75
S
R S
R  S R  S
 
 R2   S2   R   S 
   R   S    R   S
  R   R   S   S
  R   R  R   S   S  S
  R [1   R ]   S [1   S ]
 R [1   S ]
    Central Safety factor
 S [1   R ]
29
 Capacity 
reduction    1   R
factor

  1   S
Load factor

  R   S

1   1 1  PF   R   R  1   1 1  PF   S   S
   

30
Nominal factor of safety
RN  R 1  K R R   1   S  1  K R R 
   
S N  S 1  K S  S   1   R  1  K S  S 

 RN   S N
1   R 1   S
 ;  
1  K R R 1  K S S

31
Case of multiple loads

(Ref: A Haldar and S Mahadevan, 2000, Probability,

reliability, and statistical methods in engineering design,
John Wiley, NY)

•Structures need to be designed for more than

one loads
•It is unlikely that all the loads would act
simultaneously
•Load combination needs to be considered
•Dead load + live load
•Dead load + live load + wind
•Dead load + live load + earthquake
•Dead load + wind, etc.
32
 n
S   Si ; Si  N ( i ,  i ); i  1,2,  , n
i 1

 S  N (S , S )

Use formulation already developed.

This leads to a single load factor for S.
Not very useful.
We need different load factors γi for i=1,2,..,n.

33
R  S      2
R
2
S

  2 2
Let   R S
R S
  R   S    R   S 


  S    R     2
S1
2
S2    2
Sn 
34
  S21   S22     S2n
Let  nn 
 S1   S2     Sn


  R   S    R   nn  S1   S2     Sn 
  
  R   S1   S2     Sn


  R   nn  S1   S2     Sn 
  

 (1   R )  R   S1 1   nn S1    1  
S2 
nn S2  

  Sn 1   nn Sn 
   (1   R ) &  i  1   nn Si

Knowing PF and variability measures

we can find the load and Resistance factors. 35
Generalization:
Methods of structural reliability analysis

36
Problem 36
Figure (next slide) shows the pseudo-acceleration spectra
for a rocky site according to the IS 1893 (Part 1) : 2002
document. The PGA is taken to be 0.24g. It is of interest
to develop a random process model for the ground
acceleration that is compatible with this response spectrum.
It may be assumed that the ground acceleration can be
modeled as a zero mean, stationary Gaussian random
process. The duration of the acceleration can be taken
to be 30s and the given response spectra may be
interpreted as locus of the 84% percentile point and
damping may be taken to be 5%.
37
 6

4
Acceleration m/s 2

1
Pseudo-acceleration spectra as per
IS 1893-2002
0
0 100 200 300 400 500 600 700
Frequency rad/s

38
How to generate a response spectrum compatible with a given PSD?
x  2 nn x  n2 x   
 xg
xg  t   zero mean, stationary, Gaussian random process;

xg  t  ~ N 0, S gg   

X m  max x  t 
0t T

PX m    exp      T 

   2

   
 x
exp   2 
2 x  2 x 


 H   S gg  d  &
2
with   2
x



 H    2 S gg  d 
2
 2
x

39
For a given probability p, the corresponding  is given by
  x  2  
p  exp   exp   2  T 
 2 x  2 x  
1
  2 x 2 
    2 x ln  
2
ln  p   
   xT  
Let R n , n  be the given pseudo-acceleration response spectrum.
We interpret R n ,n  as the p - th percentile point.
1
  2 x 2

 R n , n    2 x ln  
2 2
ln  p    (typically p =84%)
  xT
n
  

40
How to generate a PSD compatible with a given
response spectrum?

x  2nn x  n2 x   
 xg
xg  t   zero mean, stationary, Gaussian random process;

xg  t  ~ N 0, S gg   

To a first approximation we assume


 H   S gg  d  
2
 
2
x

1
 2nn  H n  S gg n    2nn  S gg n 
2

 
2
2 nn2
 x
&  n
2 x
41
  S gg n   2  
 R n ,n    2
2 4
ln   ln  p   
2nn  nT
n 3
  
To a first approximation we thus get
 n R 2 n ,n 
S gg n   .
  1  
n  ln   ln  p   
   nT  

42
Steps
(1) Set iteration N=1
(2) Start with the initial guess on the PSD given by
 n R 2 n ,n 
S N  n  
  1  
n  ln   ln  p   
   nT  
(3)Evaluate  and  using
2
x
2
x


 H   S gg  d &
2
 
2
x
N




 H    2 S gg  d
2
 
2
x
N

 43
 1
  2 x 2
(4)Evaluate R n , n     2 x ln  
N 2 2
ln  p    .
  xT
n
  
(5)Obtain an improved estimate of PSD using
2
N 1  R ( ) 
S ( )  S ( )  N
N

 R ( ) 
(6)Stop iterations if the PSD function has converged;
if not go to step 3.

44
 0
10
Derived
Used
-1
10
PSD (m2/s 4)/(rad/s)

-2
10

-3
10

-4
10

-5
10
0 100 200 300 400 500 600 700
Frequency rad/s

45
 6
Target
Derived
5

4
Acceleration m/s 2

0
0 100 200 300 400 500 600 700
Frequency rad/s

46
 2
0
-2
0 5 10 15 20 25 30
2
0
-2
Acceleration m/s 2

0 5 10 15 20 25 30
2
0
-2
0 5 10 15 20 25 30
2
0
-2
0 5 10 15 20 25 30
2
0
-2
0 5 10 15 20 25 30
Time s

47
 0
10
Target
Simulation
-1
10
PSD (m2/s 4)/(rad/s)

-2
10

-3
10

-4
10

-5
10
0 100 200 300 400 500 600 700
Frequency rad/s

48
 7
Target
From derived PSD
6
From samples

5
Acceleration m/s 2

0
0 100 200 300 400 500 600 700
Frequency rad/s

49
Problem 37
Figure (next slide) shows the psd function of ground
acceleration which is modeled using Kanai-Tajimi's
approach  with  g  15rad/s,  g  0.6  .
Determine the pseudo-acceleration spectra compatible
with this psd function. It may be assumed that the ground
acceleration is a zero mean, stationary Gaussian random
process. The duration of the acceleration can be taken to be
30s and the target response spectra may be interpreted as the
locus of the 84% percentile point and damping may be
taken to be 5%.
50
 2

1.8

1.6

1.4 Kanai-Tajimi PSD function

PSD (m2/s 4)/(rad/s)

1.2

0.8

0.6

0.4

0.2

0
0 100 200 300 400 500 600
Frequency rad/s

51
 40

35

30
Acceleration m/s 2

25

20

15

10

0
0 100 200 300 400 500 600
Frequency rad/s

52
 2
Target
1.8 Derived

1.6

1.4
PSD (m2/s 4)/(rad/s)

1.2

0.8

0.6

0.4

0.2

0
0 100 200 300 400 500 600
Frequency rad/s

53
Discussion on outcrossing theory of random
Processes and applications to problems of
Load combination

54
Problem 38 : Load combination
Let Q  t  be a quasi-static load on a structure
(e.g., sustained live load). If we are interested in designing
the structure for this load, we can estimate
Qm  max Q  t  and use that is the design.
t

What happens if more than one loads act simultaneously?

Q  t   Q1  t   Q2  t 
max Q  t   max Q1  t   Q2  t    max Q1  t   max Q2  t 
t t t t

 Maximum of Q1  t  & Q2  t  do not reach simultaneously.

Consider the failure event of Q(t ) crossing a critical barrier   t 
and let N T   number of times the level   t  is crossed
during the interval 0 to T . Show that if Q1  t  and Q2  t  are independent,
PF  P0  E  N T   . Obtain an expression for E  N T   .
55
PF  P  Failure at t  0  N T   1
 P  Failure at t  0  P  N T   1 
P  Failure at t  0  N T   1
 P0  P  N T   1

 P0   P  N T   n 
n 1

 P0   nP  N T   n 
n 1

 P0  E  N T  
 PF  P0  E  N T  
56
Recall that in order to characterize the average rate
of crossing of a critical barrier by a random process,
we need the jpdf of the process and its derivative
at the same time instant.
Consider
Q  t   Q1  t   Q2  t 
Q  t   Q  t   Q  t 
1 2

U  Q2  t 
V  Q  t 
2

pQQUV
  q, q, u, v   pQ1Q1Q2Q  q  u, q  v, u, v 
 pQ1Q1  q  u, q  v  pQ2Q  u , v 
 
pQQ  q, q , u, v     p  q  u, q  v  p  u, v  dudv
Q1Q1 Q2Q
 
57
X t   RECALL

Y t 
Y  t   U  X  t    

Y  t 
Y  t   X  t    X  t    

Y  t 
Y  t   X  t    X  t    

N t 
T
N (T )   X  t    X  t    dt
0

58
N T   number of times the level   t  is crossed in  0, T 
with positive slope
Y (t )  U Q  t     t  
Y  t   Q  t     t    Q  t     t  
Z  t   Q  t     t    Q  t     t   U Q  t     t  
T
N T    Q  t     t    Q  t     t   U Q  t     t   dt
0

 
   0, t   E Q  t     t    Q  t     t   U Q  t     t  

59
  
   t   E Q  t     t    Q  t     t   U Q  t     t  
 
 
 
 

q    t     q    t   U  q    t   p
     QQ  q, q; t  dqdq

   q    t   pQQ   t  , q; t  dq
t 



 

   q    t      pQ1Q1   t   u , q  v  pQ2Q 2  u, v  dudv  dq

 t     
T
E  N T       t  dt
0

60
Remark
The evaluation of
   t 



 

   q    t      pQ1Q1   t   u , q  v  pQ2Q 2  u, v  dudv  dq

 t     
is possible for Gaussian models for loads. A general solution
is difficult to obtain.

61
Problem 39
Discussion of fatigue crack growth modeling under
random loads using fracture mechanics concepts

62
 Fracture mechanics based approaches

Basic assumption:
there exists a crack in the structural component.

Question:
Given the geometry of the crack, loads, boundary conditions,
can we say if the crack is likely to grow?

Parameters for measuring the potency of the crack

•Stress intensity factor

•Energy release rate
•J-integral
•Crack tip opening displacement
63
Stress near the cracktip in
Mode I (plane strain)
an infinite plate
 a   3 
 11  cos 1  sin sin 
2 r 2 2 2
 a   3 
 22  cos 1  sin sin 
2 r 2 2 2
 a  3
 33  sin cos
2 r 2 2
K
 ij  fij    
2 r
x2

  r ,  u1 
 a  r 
 2  cos

1  2  sin 2 
   2  2 
x1
 a  r   2 
u2    sin 2  2  cos
Mode I   2  2  2 
u3  0 64
Stress Intensity Factor (SIF) and Critical SIF
In the expressions for stress and displacement components
the quantities  and  a appear together.
Can we give a name to the quantitiy   a ?
Recall: EI , mv, 0.5mv 2 ,   x  ct , Reynold's number,
Froude's number 
Definition
K I  lim 2 r 22  r ,  0 
r 0

K I  Mode I stress intensity factor =  a .

Definition
Analogy
Crack propagates if K I  K Ic
Stress Yield stress
K Ic  critical stress intensity factor
SIF Critical SIF
Critical SIF is a material property. 65
Mode I (plane strain)
KI  3  
 11  cos 1  sin sin 
2 r 2 2 2
KI   3 
 22  cos 1  sin sin 
2 r 2 2 2
KI  3
 33  sin cos
2 r 2 2
KI  r   2 
u1 
  2  cos 2 1  2  sin 2 
   
KI  r   2 
u2    sin  2  2  cos 
  2  2 2
u3  0
66
 K max  K c

Region I Region II
Crack Stable crack
Initiation growth
da da
 C  K 
n
dn dn

Region III
Unstable crack
growth

log K 67
Model for Stage II crack growth
da
 f  K , K max , K min , K th , E , , ys , ult ,  i , ki 
dN
 i  environmental variables
(temperature, humidity, salinity, etc.,)
ki  other material or mechanics variables
(frequency of excitation, grain size, ...)
Dimensional analysis
da  K   K Ic K th  ys  ult 
2
  F  R, , , , ,  i  i ,  i ki 
dN  E   K K E E 

68
Paris - Erdogan model
da
 C  K  ; K  0; a  0   a0
m

dN
 da 
 log    log C  m log K
 dN 

Example a in m & K in Mpa m 
da
 6.80  10  K 
12 3.00
Ferrite pearlite steel :
dN

69
69
 Modeling of uncertainties
Sources
 Macro-properties of specimens
(geometry, dimensions, and
material properties may differ
between specimens).
 External loading.
 Inhomgenous microstructure.

Tests on identical specimens

 Behavior of crack length of identical specimens is random
 The crack length behavior is nonlinear in time
 The curves of different specimens intermingle. 70
Two approaches
 Treat constants appearing in the differential equation
for evolution of a as a fucntion of N as random variables.
da
 C  K  ; K  0; a  0   a0
m

dN

 Introduce random process models

da
 C  K  X  t  ; K  0; a  0   a0
m

dN
t
N 
2

71
Cumulative jump models
(Reference : K Sobczyk and B F Spencer Jr., 1992,
Random fatigue from data to theory, Academic Pr ess )
Define A  t ,    random process: length of the dominant
crack at time t.
   (sample point). To be suppressed in further description.
N t 
A(t )  A0  Y ;
i 1
i Yi  Ai

 A0  Initial crack length; sufficiently long to propagate;

could be random.
 N (t )= a counting process; homogeneous Poisson process;
counts the number of crack increments in 0 to t .
72
P  N  t   k   exp  0t 
 0t  k

; k  0,1, 2, , 
k!
Yi i 1  iid sequence of non-negative rvs with a


common pdf pY  y 
 N  t   Yi i 1


 P  A  t   a   PA  a; t  [PDF]
dPA  a; t 
 p A  a; t   [pdf]
da

73
 N t 
Let A  t   A0  A1  t  with A1  t   Y
i 1
i

Consider the moment generating function of A1  t  .

 N t  
exp   sA1   exp   s Yi  
 i 1  
 
  N t  
 
exp   s Yi  N  t   k P  N  t   k 
 i 1 
k 0  
 0t 
 k

 exp  sY  exp  0t 

k
 i
k 0 k!
74
  0t 
 k

exp   sA1    exp  sY  exp  0t 

k
i
k 0 k!

 0t  k

 G  s  exp  0t 

k

k 0 k!
Here G  s  is the moment generating function of Yi .
That is, G  s   exp   sY 
Let us assume pY  y    exp   y  ; y  0

 G s  ; s  0.
 s

0t  k 1
ak
 p A1  a; t   exp  0t   a   k ! k  1!
k 0
;a  0
75
  0t  a k
 k 1

p A  a; t   exp  0t   a   ;a  0
k  0 k ! k  1!
1


 0t 
exp  0t   a  I1  2 0 at  ; a  0
a
where I1    Bessel's function of the first order.
A  t   A0  A1  t   p A  a; t   p A1  a  A0 , t 

76
Model for life time
Let  be the critical crack length
(esimtated from the knowledge of K Ic ).
T  time required for A(t ) to reach the critical length  .
P T  t   P  A  t    
 PT  t  

0t  exp
1  a  A0
 0t   a  A0  I1  2 
0 a  A0 t da
0

It can be shown that

pT  t   0 exp  0t   a  A0   I 0  2 0 a  A0 t  ;0  t  
 

77
Estimation of system parameters
Model parameters: 0 associated with the process N  t  ;
 : associated with pY  y  .
Idea: derive these model parameters from laws such as
the Paris law. An approximate method to achieve this
would be to modify the Paris law to allow for randomness
in applied stress and system parameters.
da p
 C  K  ; a p  0   a0
m

dN

78
Let S (t ) be the stress field that is modeled as
a Gaussian, stationary random process.
What is meant by cycle?
d d dt 1 d
N  s t   
dN dt dN s dt
da p
 s C  S max  S min   a p  ; a p  0   a0
m

dt  
Interpret s as the average rate of peaks in S (t ).

79
s  average rate of zero crossing of S  t  .
1
 4 2
  S s   d  

1   
s 
2   2 
  S   d  
 
 
s


Interpretation of ΔK
Recall: K    a . Interpret   S max  S min  mean range.

S mr  S max  S min  2 S rms
2
1  
2

1

  22 2
  1   ;  n   S s   d 

n

 0 4  

 
m
da p m
 s C  a p2 S max  S min ; a p (0)  a0
m

dt 80
Interpretation of 0
If we take N (t )=number of peaks above a level s0 ,
then 0 becomes the average rate of peaks in S (t )
above level s0 . Select s0  fatigue limit of material
(that is the endurance limit).
1
1  4  2  4 
0  {  1    s0  
2   2     4   2
2  
 s 
  2 2   s0    0 2 }
   2 
 4 2 
81
H
o
w
t
o
f
i
n
d
?

Select  such that
t*
F      Ap  t   A  t   dt
2

is minimized.
Here t  time required by Ap  t  to reach  .
*

82