Escolar Documentos
Profissional Documentos
Cultura Documentos
Abstract—The basic fatigue crack growth is important consideration in structural design and constrained operations of safety in critical
structural component. The basic law used in this study has been modified for analyzing the fracture mechanics of structures under random
loading. In any mechanical system the behavior is modeled by differential equations deterministic and stochastic in nature. We give Paris law in
random loading and differential equation governing such real time situations.
Keywords— Fatigue crack growth, Paris Law, Random loading, Stochastic differential equation, Paris Erdogan model.
__________________________________________________*****_________________________________________________
A. Paris Law These laws are based on fixed stress level fatigue
experiments (constant amplitude cyclic loadings). Bell and
In general, the fatigue crack growth can be Wolfmann discussed the Eller‘s crack closure model based
expressed as on effective stress range concept 𝐾𝑒𝑓𝑓 = 𝑆𝑚𝑎𝑥 − 𝑆𝑐 𝜋𝐿
dL = F L, S, C, T, s; t dN …….. (1) where Sc is the crack closure stress. Since 𝑆𝑚𝑎𝑥 − 𝑆𝑐 =
𝑆𝑚𝑎 𝑥 1 − 𝐶𝑓 where 𝐶𝑓 is the closure factor, the proposed
Where S, C and T are the qualities indicated above and s fatigue crack growth equation is
denotes all other parameters important in fatigue problems
𝑛
(chemical properties, internal stresses etc.). L denotes the 𝑑𝐿 ∆𝑆
=𝐴 1 − 𝐶𝑓 𝜋𝐿
length of a dominant crack and N denotes the number of 𝑑𝑁 1−𝑅
cycles corresponding to the crack length L. But the 𝑛
1−𝐶𝑓
information about the influence of each of these parameter = ∆𝑘 ……… (7)
1−𝑅
are insufficient.
The analysis of experimental data, revealed that 𝐶𝑓 can be
Recent studies of fatigue crack growth in elastic expressed as
materials have shown that the stress intensity factor K is the
𝑞
primary quantity for characterizing the fatigue growth rate. 𝐶𝑓 = 𝑎 + 𝑏 1 + 𝑅 …….. (8)
Where ∆𝐾 = 𝐾𝑚𝑎𝑥 − 𝐾𝑚𝑖𝑛 and𝐾 = 𝑆 𝜋𝐿. Also C and n are Where 𝐶𝑓𝑜 and 𝐶𝑓−1 are the values of 𝐶𝑓 at 𝑅 = 0 and
assumed constants for a given material. The above Paris law 𝑅 = −1 respectively and q is a positive constant.
(2) has essential deficiencies as the constants n and c
depends on many factors and their numerical values deviate Since (7) is of the form (4), we can use (4) as a
greatly from experiment to experiment. base for further analysis.
As the stress ratio 𝑅 = 𝑆𝑚𝑖𝑛 /𝑆𝑚𝑎𝑥 was recognised Consider a situation when the structural element is subjected
to have sufficient influence on fatigue crack growth (2) can to time varying random loading.
be replaced by
We assumed that, the material considered is homogeneous,
𝑑𝐿 linearly elastic for which the stress intensity factor is
= 𝐹(𝑅, ∆𝐾) ……….. (3)
𝑑𝑁 expressible in terms of the basic parameters of the material
and applied stress. While there may be a number of cracks
The influence of R on fatigue crack length depends
in a specimen it is assumed that the final damage is due to
on the structure and mechanical properties of the material
the growth of the dominant crack. The configuration of a
considered. The experiments show that the effect of mean
dominant crack depends on one quantity – the length L
stress under the tensile stresses results on increasing of
(t).Randomness in crack evolution is caused by random
fatigue crack growth rate. Thus (3), can be expressed as
applied stress and some other uncertainties relevant to
𝑑𝐿 fatigue.
= 𝐶𝑔 𝑅 (∆𝐾)𝑛 ……… (4)
𝑑𝑁
Also,
Brock and Schifne (1963) generalized (4) in the form
Let 𝑆 𝑡 be a stochastic process characterizing the
𝑑𝐿 1 2
=𝐶 3
(∆𝐾) ………… (5) random applied stress and 𝑀𝑠 𝑡 𝑟𝑠 𝑡1 , 𝑡2 denote
𝑑𝑁 1−𝑅
respectively the mean and autocorrelation function for the
148
IJRITCC | February 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 6 Issue: 2 147 - 153
______________________________________________________________________________________
process. If 𝑆 𝑡 is a stationary process then𝑀𝑠 𝑡 = 𝑀𝑠 , a In which 𝐾𝑟𝑚𝑠 , 𝑄 and 𝜇 𝑡 characterize the effect of stress
constant which is independent of time. intensity range, stress ratio and frequency content
respectively. The correlation properties of random applied
𝑟𝑠 𝑡1 , 𝑡2 = 𝑟𝑠 𝑡2 − 𝑡1 and stress are introduced though 𝜇 𝑡 .Then,
𝑆𝑟𝑚𝑠 𝑡 = 𝑆𝑜 = 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑑𝐿 𝑛
= 𝐶𝜇 𝑡 𝑔 𝑄 𝐾𝑟𝑚𝑠 ; 𝐿 𝑡0 = 𝐿𝑜 ….. (15)
𝑑𝑁
Recent studies strongly suggest that ∆𝐾 should be
Using (9). The above equation (15) can be written as
replaced by the root mean square 𝐾𝑟𝑚𝑠 of the stress intensity
factor. 𝑑𝐿
= 𝑓 𝑡 𝐿𝑝 𝑡 ,
𝑑𝑡
i.e.𝐾𝑟𝑚𝑠 = 𝑆𝑟𝑚𝑠 𝜋𝐿 …….. (9)
𝑓 𝑡 = 𝐴𝜇 𝑡 𝑔 𝑄 𝑡 𝑠 𝑛 𝑟𝑚𝑠 𝑡
The mean stress effect is modified by the new ratio.
<𝑆>
𝐿 𝑡0 = 𝐿𝑜 , 𝑃=𝑛 2 , 𝐴 = 𝐶 𝜋.
𝑄= ……… (10)
𝑆𝑟𝑚𝑠
𝜏 𝜏
𝐿
The further modification is concerned with the cycle which = 𝑓 𝑡 𝑑𝑡
𝑡𝑜 𝐿𝑝 𝑡 𝑡𝑜
plays a conventional role in fatigue analysis. In the case of
𝜏
random we replace N by 𝜂 𝑡 where the relation between 𝐿 −𝑝 +1 𝑡
i.e. = 𝜙 𝜏 ……….(16)
number of cycles N and time‗t‘ is random and 𝜂 𝑡 is a 1−𝑝 𝑡𝑜
point stochastic process. Assuming that 𝜂 𝑡 characterizes a 𝜏
number of local maxima of a stress process in the interval Where,𝜙 𝜏 = 𝑡𝑜
𝑓 𝑠 𝑑𝑠
𝑡0 , 𝜏 the quantity 𝑛 𝑡 is defined as
1
i.e. 𝐿1−𝑝 𝜏 − 𝐿0 1−𝑝 = 𝜙 𝜏
𝑇 1−𝑝
𝑛 𝑡 = 𝑛 𝑠 𝑑𝑠
𝑡𝑜 𝐿1−𝑝 𝑡 = 𝐿0 1−𝑝 − 𝑝 − 1 𝜙 𝜏
𝑇 1
Also < 𝑛 𝑡 > = 𝑡𝑜
𝜇 𝑠 𝑑𝑠, 𝐿 𝑡 = 1 ………. (17)
1 𝑝 −1
𝑝 −1
− 𝑝−1 𝜙 𝜏
𝐿𝑜
𝜇 𝑡 =< 𝑛 𝑡 >………..(11)
The explosion time τ is obtained from the equation
So 𝜇 𝑡 describes the average number of cycles of a stress
process per unit time. 1
𝑝−1 𝜙 𝜏 = ………….. (18)
𝐿𝑜 𝑝 −1
Thus the relation between the cyclic and temporal
description can be put forth as If S 𝑡 is stationary then𝑄 𝑡 = 𝑄𝑂 ; 𝑆𝑟𝑚𝑠 𝑡 =
𝑆𝑜 ;𝜇 𝑡 = 𝜇𝑜 ;𝑓 𝑡 = 𝑓𝑜 and we have
𝑑𝐿 𝑑𝐿
= 𝜇 𝑡 …….. (12) 1
𝑑𝑁 𝑑𝑁
𝜏 = 𝑡𝑜 + ………. (19)
𝑓𝑜 𝑝−1 𝐿𝑜 𝑝 −1
If the process 𝑆 𝑡 is sufficiently regular (if atleast twice
mean square differentiable) then the expected number of For the non-stationary case, assume that the time 𝑡 ∗ to reach
maxima per unit time of 𝑆 𝑡 above a certain levels 𝑆𝑜 is, the crack size 𝐿∗ as
0 ∞ 1
𝜇 𝑡; 𝑠𝑜 = − 𝑑𝑠 " 𝑆 " 𝑃 𝑠, 0, 𝑠 " , 𝑡 𝑑𝑠 ……… (13) 𝜙 𝑡∗ = 𝐿∗ 1 − 𝑝 − 𝐿𝑜 1−𝑝 …….. (20)
−∞ 𝑠𝑜 1−𝑝
Where 𝑃 𝑠, 𝑠 ′ , 𝑠 " , 𝑡 is the joint density function of And for the stationary case
𝑠 𝑡 , 𝑠 ′ 𝑡 , 𝑠 " 𝑡 at time t.
𝐿𝑜 1−𝑝 −𝐿 ∗1−𝑝
𝑡 ∗ = 𝑡𝑜 + ………. (21)
𝑓 𝑝−1
The average crack growth rate under random
loading takes the modified form. The modified growth equation gives a satisfactory
𝑑𝐿
picture for the case of stationary loading when both the
= 𝜇 𝑡 𝐹 𝑄, 𝐾𝑟𝑚𝑠 …….. (14) stress range (characterized by 𝑆𝑟𝑚𝑠 ) and expected frequency
𝑑𝑁
𝜇 are constant. For the non- stationary case, the
149
IJRITCC | February 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 6 Issue: 2 147 - 153
______________________________________________________________________________________
determination of 𝜇 𝑡 in analytical form possess certain
difficulties.
B. Illustration 01
2𝜋
1
<𝑆 >= 𝑎 𝑠𝑖𝑛 𝑊𝑡 + 𝑋 𝑑𝑋 = 0
0 2𝜋
𝑟𝑠 𝑡1 , 𝑡2 = 𝑟𝑠 𝑡, 𝑡 + 𝜏
2𝜋 1
= 0
𝑎 𝑠𝑖𝑛 𝑊𝑡 + 𝑋 𝑎 𝑠𝑖𝑛 𝑊𝑡 + 𝑊𝜏 + 𝑋 dX
2𝜋
𝑎2
= cos 𝑤𝜏
2 III. STOCHASTIC CRACK GROWTH EQUATION
𝑎 w
𝑆𝑟𝑚𝑠 = , 𝑄 = 0, 𝜇 𝑡 = 𝜇𝑜 = We improve the model for random fatigue by accounting for
2 2π
randomness of other factors provoking fatigue.
The function f (t) takes the form
The model proposed is
𝑤 𝑎 𝑛
𝑓 𝑡 =𝐴 𝑔 0 dL
= Cμ t g Q K rms n
X t, γ ………. (22)
2𝜋 2 dt
Fatigue cracks in a pavement layer are caused by If 𝑋 𝑡, 𝛾 = 𝑚𝑥 + 𝑋 𝑡, 𝛾 where Mx is the average value of
the combination of repetitive strains and apparent reduction 𝑋 𝑡, 𝛾
of tensile strength caused by fatigue of the layer material.
𝑑𝐿 n
This type of failure occurs when the pavement has been a = Cμ t g Q K rms 𝑚𝑥 + 𝑋 𝑡, 𝛾 …….. (23)
𝑑𝑡
stress to the limit of its fatigue life by repetitive axle load
applications.It is often called alligator cracking. The fatigue The above equation can be rewritten as
cracking is often caused by high moisture content, poor sub
𝑑𝐿 n
grade or some other local problems that can be repaired = Cμ t 𝑚𝑥 g Q K rms +
𝑑𝑡
without major reconstruction. In these instances the poor Cμ t g Q K rms n
𝑋 𝑡, 𝛾 …….. (24)
material is removed and replaced with good material and
drainage improved. If the failed area is extensive one typical It is convenient to write the stochastic growth model
repair strategy is to place a HMA overlay over the entire (23),(24) in the form
surface. Investigation should involve determination of the
𝑑𝐿
thickness of layers and the material quality so that a suitable = 𝑎 𝐿, 𝑡 + 𝜎 𝐿. 𝑡 𝑋 𝑡, 𝛾 ……… (25)
𝑑𝑡
solution is selected.
𝐿 𝑡𝑜 = 𝐿𝑜
Crack in pavement under varied stress with the help of
illustration 01 , we can illustrate the fatigue crack growth in Or
the pavement is illustrated given in the figure.
150
IJRITCC | February 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 6 Issue: 2 147 - 153
______________________________________________________________________________________
𝑑𝐿 (28) looks like a differential equation, it is really G formal
= 𝑓1 𝑡 𝐿𝑝 𝑡 + 𝑓2 𝑡 𝐿𝑝 𝑡 𝑋 𝑡, 𝛾 ……….. (26)
𝑑𝑡
record of symbols since 𝜉 𝑡, 𝛾 is an abstraction and not a
𝐿 𝑡𝑜 = 𝐿𝑜 physical process. Equation (25) with (28) does not define a
stochastic process 𝐿 𝑡, 𝛾 yet, it is a pre-equation.
𝑑𝑎
In the study of crack propogation in materials, most Now = 𝑘𝑑𝑛
𝑎𝛽 2
of the researchers take into consideration the paris-Erdogan
equation. 𝑎 −𝛽 2 +1
i.e. = 𝑘𝑛 + 𝐶
−𝛽 2 +1
𝑑𝑎 We have the initial condition 𝑎 0 = 𝑎𝑜 .
𝑚
= 𝐶 ∆𝐾 2
𝑑𝑁 𝑎 2−𝛽 2
=𝐶
2−𝛽 𝑜
For the rate of fatigue growth under homogeneous
cyclic stressing which was attained from experimental 2 − 𝛽 2−𝛽 2
2
𝑑𝑎 𝑎 = 𝑘𝑛 + 𝑎 2−𝛽 2
results as the linear regression of 𝑙𝑜𝑔 on 𝑙𝑜𝑔 ∆𝐾 2 2−𝛽 𝑜
𝑑𝑁
where ‗a‘ is the crack length. ‗N‘ is the number of cycles 2−𝛽
and k is the range of stress intensity factor at the crack tip. 𝑎 𝑛 = 𝑎𝑜 2−𝛽 2
+ 𝑘𝑛2 2−𝛽
= 𝑘 𝑎𝑜 ……. (35)
2
We consider the paris-Erdogan model from two points of We have given the behavior of crack growth under random
view namely. loading using continuous stochastic models of stress
152
IJRITCC | February 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 6 Issue: 2 147 - 153
______________________________________________________________________________________
behavior and stochastic differential equation model of
fatigue behavior. Illustration is given to explain these
modifications in the basic law under random loading.
References
[1] Arnold. L, ―Stochastic differential equations: theory and
Applications‖, Wiley, New York, 1974.
[2] Bogdanoff,J.L., ―Probability models of fatigue crack
growth II‖, Engg. Fracture Mech., Vol.20, pp.255-
270,1984.
[3] Doob,J.L.,‖Stochastic Processes‖,John Wiley, New
York,1953.
[4] Hand book on Causes and prevention of cracks in
buildings, Indian Standards Institution (SCIP), New Delhi.
[5] Keith ortiz, ―Stochastic modelling of fatigue crack growth‖,
Engg. Fracture Mech., Vol.29,No.3, pp.317-334, 1988.
[6] Kozin.F and Bogdanoff. J.L ,‖A critical analysis of some
probabilistic models of fatigue crack growth‖, Engg
Fracture Mechanics,Vol.14, pp.59-89, 1981
[7] Lin,Y.K.,‖Probability Theory of Structural Dynamics‖, Mc.
Graw Hill, New York, p50,1967.
[8] Oh,K.P.,‖A diffusion model for fatigue crack growth‖, Proc
Roy Soc Series A Vol367 pp47-58 1979.
[9] Oh,K.P.,‖A.,‖ A weakest- link model for the prediction of
fatigue crack growth‖, Trans ASME,J Engg Mater
Technology Vol.100,1978.
[10] Oh,K.P.,‖Prediction of fatigue life under random loading –
a diffusion model‖, International Journal of Fatigue, Vol.2,
pp.99-104, 1980
[11] Sobczyk. K,‖Markowian models for fatigue
accumulation‖,Mecanique.J.L. Theor. Appl.,Numero
spec.pp147-160,1982.
[12] Sobczyk.K, ―Modelling of random fatigue crack growth‖,
Engg. Fracture Mechanics, Vol.24,No.4, pp.609-623, 1986.
153
IJRITCC | February 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________