No Slide Title

Download Report

Transcript No Slide Title

FRACTALS IN MECHANICS OF MATERIALS

Martin Ostoja-Starzewski

Department of Mechanical Science & Engineering Institute for Condensed Matter Theory, and Beckman Institute University of Illinois at Urbana-Champaign [Support: NSF CMMI-1030940]

Morphogenesis of Fractals at Elastic-Plastic Transitions

(a) Does the elastic-plastic transition display a fractal character?

- if so, estimate the fractal dimension of plastic grains

(b) How does yielding develop in material body?

- study evolution of fractal dimension

(c) What kind of correlation exists between fractal dimension and material properties?

- are the responses among different materials universal?

2

… homogeneous elastic-plastic solid

d

ij

' 

d

ij

' 2

G d

 

d

K d

    

f

ij d

ii

,

d

 3 

d

ii

3 3

… inhomogeneous elastic-plastic solid

d

ij

' 

d

ij

' 2

G d

 

d

K d

    

f

ij d

ii

,

d

 3 

d

ii

3 4

… a heterogeneous elastic-plastic solid

5

evolution of plastic set’s fractal dimension (estimated by box method)

10 9 8 7 6 5 4 3 2 1 0 -5 -4.5

-4 -3.5

-3 ln r -2.5

-2 -1.5

-1 -0.5

10 7 6 9 8 2 1 0 -5 5 4 3 -4.5

-4 -3.5

-3 ln r -2.5

-2 -1.5

-1 -0.5

D

  log(

N r

) 10 9 8 7 6 5 4 3 2 1 0 -5 -4.5

5 4 3 10 7 6 9 8 2 1 0 -5 -4.5

-4 -3.5

-3 ln r -2.5

-2 -1.5

-1 -4 -3.5

-3 ln r -2.5

-2 -1.5

-1 -0.5

-0.5

6

stress-strain responses fractal dimension-strain responses [

ASME J. Appl. Mech.

, 2010;

Proc. R. Soc. Lond. A

, 2010;

JoMMS

, 2011;

Phil. Mag

. 2012] 7

massively parallel simulation of elastic-plastic grains evolving in a 100 x 100 x 100 system 8

massively parallel simulation of plastic grains evolving in a 100 x 100 x 100 system

D

  log(

N r

) by box-counting 9

1.205

x 10 8 1.2

1.195

1.19

1.185

1.18

1.175

1.17

0 0.1

0.2

0.3

0.4

0.5

Plastic strain 0.6

0.7

Displacement BC Traction BC Homogeneous 0.8

0.9

x 10 -5 1 stress-strain responses 3 2.95

2.9

2.85

2.8

2.75

2.7

2.65

2.6

0 0.1

0.2

0.3

0.4

0.5

Plastic Strain 0.6

0.7

Displacement BC Traction BC Homogeneous 0.8

0.9

x 10 -5 1 fractal dimension-strain responses 10

Universal properties at phase transitions

• Define order parameters (by analogy to condensed matter physics):

reduced von Mises stress: s

:  

s

E p

 

s p

 

reduced fractal dimension: d

:  (

E p

s

/ /

E

) 

p E

3 

D

3   

s E p E

p

  • Postulate a power-law

scaling function

:

d

Motivated by observation:

s

0

0

m

11

Scaling function for different materials -3 -3.5

-4 -4.5

-5 -5.5

-6 -6.5

-7 -7.5

-6.5

-6 -5.5

-4.5

-4 3D 2D -3.5

-5 Log s Large difference between

m

and

a

:

m=2.12/4.07, log(a)=5.78/12.98 (2D/3D);

2d slower than 3d: lower

m

and

a

12

Functionally Graded Material

• Properties varying spatially • Heterogeneous material system • Examples in nature: wood, bone, sea shells,… • Composites…

FGM structure

• Random medium B     } •

P(Black, White) = {0.5,0.5}

• Conventional model: smooth

E

(

x

), n =const micromechanically invalid

Edge Plots

Interfacial fractal dimension

R

N D D R

 

X n Log

10

Log

10 2    

Beta function fit of

D

(

x

)

f

(

x

) 

x

  1 (1 

x

)  1

B

(  ,  )

C α = 1.692 ;β = 1.692

C = 0.8129

B

(

x

,

y

)   (

z

)   (

x

)  (

y

)  (

x

y

)   0 

t z

 1

e

t dt

1 0 

t x

 1 (1 

t

)

y

 1

dt

Material Properties

Property

Young's Modulus Poisson’s Ratio

Value

104 0.3

Units

GPa • FGM system under consideration: Ti-TiB • Ti rich – left half of domain • TiB rich – right half of domain Yield Strength Density 482.633

4512 MPa Kg/mm 3

Material properties of commercially pure Titanium (A70) at room temperature Property

Young's modulus Poisson’s ratio

Value

370 0.4

Units

Gpa density 4630 Kg/mm 3

Material properties of Titanium Monoboride (TiB) at room temperature

   (0) 

h

 • •

σ(0) = 510 MPa h = 833 MPa

Determine mesoscale properties from Hill-Mandel condition: (for Cauchy materials)

 :    :    

B

 (

t

  

n

)  (

u

  

x

)

dS

 0 Uniform boundary conditions:

displacement (Dirichlet) b.c.

• traction (Neumann) b.c.

x

 

B

u

  

x t

  

n

• displacement-traction b.c.

(mixed-orthogonal)

(

t

  

n

)

(

u

  

x

)

0

19

Displacement Boundary Conditions

Domain size 50

• UKBC applied to FGM

u

  0 

x

x

B

• Pure shear loading by  0 11   0 22  0  0 12   0 21  

Domain size 100

Evolution of Plastic Grains (UKBC) – mesh size 200

800 600 400 200 0 0

Stress v/s Strain (Ti)

0,02 0,04 0,06

Strain

0,08 0,1 1200 1000 800 600 400 200 0 0

Stress v/s Strain (TiB)

0,02 0,04 0,06

Strain

Traction Boundary Conditions

Domain size 50

• USBC applied to the model according to,

t

  0 

n

x

B

• Pure Shear loading by,  11 0   0 22  0  0 12   0 21  

Domain size 100

Evolution of Plastic Grains (USBC) – mesh size 200

800 600 400 200 0 0

Stress v/s Strain (Ti)

0,02 0,04 0,06

Strain

0,08 0,1 1200 1000 800 600 400 200 0 0

Stress v/s Strain (TiB)

0,02 0,04 0,06

Strain

Stress v/s Strain

Fineness: 50 Fineness: 100 Fineness: 200

Fractal Dimension (D) using Box Counting

D = 1.9739 (UKBC) fineness:200 D = 1.7981 (USBC) fineness:200

.

.

.

Fractal Geometry of Nature

, B.B. Mandelbrot

Fractals Everywhere

, M.F. Barnsley

Fractals (Physics of Solids and Liquids)

, J. Feder

?

Continuum Mechanics

(assumes smooth fields)

Can continuum mechanics be generalized so as to handle field problems of materials with fractal geometries?

Mass in a fractal geometric structure

W

obeys a power law 

L D

,

D

 3 [V.E. Tarasov,

Ann. Phys

., 2005] Use a fractional integral to represent mass in a fractal region )  

W

R

dV D

 

W

R

c D

R

dV

3

Mass in a fractal geometric structure

W

obeys a power law 

L D

,

D

 3 [V.E. Tarasov,

Ann. Phys

., 2005] Use a fractional integral to represent mass in a fractal

W

 R 3 )  

W

R

dV D

 

W

R

c D

R

dV

3

in

R 3 

dimensional regularization

( , ) 

R

D

 3 2 3 

D

 (3 / 2)  (

D

/ 2)

Green-Gauss theorem  

W

v n

dA d

 

W c

 3 1  where

dS d

c

2  ,  2   

c

2 

dV D

c

3  ,  3 

v

dV D

surface fractal dimension of

W

mass fractal dimension of

W

dimensional regularization

( , ) 

R

d

 2  2 2 

D

Fractal continuity equation for

W d

 

D

D k v k

Fractal linear and angular momentum equations 

d

 

D v k

 

f k

 

l D

kl

kl

 

kl

Fractional equation of energy balance 

d

 

D u

   

v

 

D k q k

Fractional equation of 2 nd 0 

T

D s

   

ij

law (C-D inequality)      

D u

(

i

  ,

j

)  

ij

non-fractal medium

(D=3, d=2)

D

ij

  

T

,

k T q k

recover conventional forms

 two generalized operators 

D k f

c

3    

x k

 

dt D c

2 

f

:  

f

t

   : 

c

3    

k k

f

x k c

2  where  

c

3 

c

2   

R

d D

2 

D d

1     

d D

/2 /2      

R

d

 2 2 2  

d

d

/2 

R

D

 3 2 3   

D D

   /2   

c

 3 1  ,    

 two generalized operators 

D k f

c

3    

x k

 

dt D c

2 

f

:  

f

t

   : 

c

3    

k k

f

x k c

2  where  

c

3 

c

2   

R

d D

2 

D d

1     

d D

/2 /2      

R

d

 2 2 2  

d

d

/2 

R

D

 3 2 3   

D D

   /2   

c

 3 1  ,     

D k fg D k g cg k f

holds 

k D fg

f

k D g

 

k D f

Fluid in a porous fractal medium

Formulation of wave propagation equation by application of dimensional regularization

 2

p

t

2     

c

2

D

 3  

R

2 

D

 2  3 

R

2  2

p

[V. Tarasov,

Ann. Phys

. (2005), “Fractional hydrodynamic equations for fractal media”] Solved analytically and numerically in [

ZAMP

isotropic fractal media”] (2011), “On the wave propagation in

cortical fold = sulcus smooth area = gyrus CerebroSpinal Fluid The surface of the brain, where the highest level of thinking takes place contains a large number of folds. Hence, a human (the most intellectually advanced “animal”) has the most folded surface of the brain:

D

= 2.73 – 2.79. 34

Natural Fractals in Human Body

• Systems – respiratory – lymphatic – nervous – circulatory

• • • • • Increase the surface area for absorption and transfer Self-similar (easily constructed) “Packing efficiency” (lungs, small intestines, etc) Decrease speed of fluids or air Minimize material used to form structures

Drawbacks of Tarasov’s formulation 1.

Usual fractional derivative (Riemann-Liouville) of a constant  0 2.

The mechanics-type derivation of wave equations yields a different result from the variational-type derivation 3.

The 3d wave equation  2

p

t

2     

c

2

D

 3  

R

2 

D

 2    3 

R

2  2

p

does not reduce to the 1d wave equation

c D x

1 ( , )  2

p

t

2   

x

  2

v c D x

1 ( , ) 

p

x

  V.E. Tarasov, Continuous medium model for fractal media, Phys. Lett. 336 (2005) V.E. Tarasov, Fractional hydrodynamic equations for fractal media,

Ann. Phys

.

318

(2005) V.E. Tarasov,

Fractional Dynamics

, Springer (2010)

Formulation via product measures…

Mass in an anisotropic fractal: , 3 )

x

1

l

10

x

2

l

20 

l k

0 characteristic length in

x k x

3

l

30     3 

k

fractal dimension along

x k

In other directions the fractal dimension is not necessarily the sum of projected fractal dimension… (Falconer, 2003): "

Many fractals encountered in practice are not actually products, but are product-like

."  expect

D

  2   3

Formulation via product measures…

power law relation w.r.t. each coordinate use a fractional integral with

product measure

and length measurement in each coordinate

Vector calculus on anisotropic fractals

fractal derivative

(

fractal gradient

) operator 

D

e

D k

 or 

D k

 

c

1 1

e

k

= base vectors   

x k

  

fractal divergence

of a vector field div

f

 

D

f

or 

D k f k

 1

c

( ) 1

k

f k

x k fractal curl

operator of a vector field curl

f

 

D

f

or

e jki

D k f i

e jki

1

c

( ) 1

k

f

x k i

four fundamental identities of vector calculus 39

divergence of curl of a vector field 1 

f

f

e jki

D k f i

c

1 

x j

 

e jki c

1 1 

f i

x k

  

e jki c

1 1

c

1 1

x

f i x k

 0 curl of gradient of a scalar field  

e ijk D j D k

 ) 

e ijk c

1 1  

x j

 

c

1 1   

x k

  

e jki c

1 1

c

1 1

x

f i x k

 0 divergence of gradient of a vector field 

D k

 

c

1 1 

j

 

c

1 1   

x j

  

c

1 1 fractal

Laplacian

   

c

1 ,

j

  curl of curl operating on a vector field 

e prj

r D

(

e jki

r D f i

)   

r D f p

40

divergence of curl of a vector field 1 

f

f

e jki

D k f i

c

1 

x j

 

e jki c

1 1 

f i

x k

  

e jki c

1 1

c

1 1

x

f i x k

 0 curl of gradient of a scalar field  

e ijk D j D k

 ) 

e ijk c

1 1  

x j

 

c

1 1   

x k

  

e jki c

1 1

c

1 1

x

f i x k

 0 divergence of gradient of a vector field 

D k

 

c

1 1 

j

 

c

1 1   

x j

  

c

1 1 fractal

Laplacian

   

c

1 ,

j

  curl of curl operating on a vector field 

e prj

r D

(

e jki

r D f i

)   

r D f p

Helmholtz decomposition holds: with

F

U

V

curl

U

0

, # div

V

 0 # 41

Stokes theorem  

W

( 

D

  

W n e c

1 1   

W n e f f c

1

n

dS d

  

W n e

D j i dS dS d

2     

W

W n e

1

n e f c

1

f c

1  

c

2

d

dS

2

dS

2  

S d

n

f

dS d

 

l

f

d

l

 1 42

volume coefficient

surface coefficient

associated with the surface …adopt a modified Riemann-Liouville fractional integral to express the

linear coefficient

[Jumarie, 2005, 2008]:

Fractional integral

dimensional regularization

fractal derivative : [

ZAMP

2009;

Proc. R. Soc. A

, 2009;

J. Elast

, 2011]

1.

D

is the “inverse” operator of fractional integrals: and 2. rule of “term-by-term” differentiation is satisfied: 3. operation on any constant is zero: Note: usual fractional derivative (Riemann-Liouville) of a constant fractional generalization of Reynolds transport theorem:

fractal effects are present between the resolutions (upper and lower cut-offs)

l

and

L

in a fractal RVE

random

Apollonian packing

47 Elastodynamics of Fractal Micropolar Solids – H. Joumaa & M. Ostoja-Starzewski

random 2d

Cantor set

48

Express Cauchy stress via fractional integral, and strain via fractal derivative … balance law of linear momentum in fractal solids e.g. a linear elastic solid:

Conservation of linear and angular momenta in fractal media … using G-G theorem and Reynold’s transport theorem, we localize to: In general, due to anisotropy of a fractal surface force and surface couple in the fractal setting are specified via fractional integrals

 1   1

micropolar continuum

so the energy balance can be written with [

Int. J. Eng. Sci.,

2011]

1d wave equation

Mechanical approach: Hooke’s law  

E

 

u

c

1  1  ,

x

u

Ec

1  1  ,

x

With conventional strain definition With our strain definition    1

c u

,

x

 

u

,

x

Variational approach: 

u

  1

Ec u

1 ,

xx

u

Ec

1  1 

c u

1  1 ,

x

 ,

x T

 1 2   2

u dl D

 1 2   2

u c dx

1

U

 1 2

E

  2

dl D

 1 2

E

  2

c dx

1  

Ldt

   

T

  0 52

 (1) (2)

c c u

1 1 3

2d anti-plane wave equation

   

c

1 (2)  

u

3,1

c

1 (1)   ,1 

c

1 (1)  

u

3,2

c

1 (2)   ,2    0 

u i

 

c

1 1   

u c

1

3d wave s

   ,

j

  

c

1 1   

u c

1    ,

i I

Timoshenko beam

 0

Aw

 

D x

  

D x EI

D x

EI

  

D x

A

 

D x

 

D x w

A w

    0 

D x w

      mechanical approach is consistent with the variational approach 53

• Constitutive law 

ij

C ijkl

kl

ij

C ijkl

kl

 

ij ij

     

ij ij

     

ji ji

   

kk

 

kk ij ij

• Stiffness constants

C ijkl C ijkl

  

jk

  

jk il il

  

jl ik

  

ij kl

  

jl ik

ij kl

 ,   ,  ,  ,  Lamé’s constants micropolar constants • Kinematics – – – – strain: 

ij

 

i D u j

e kij

curvature: gradient: 

ij

 

i D

 

i

 1 product measure:

g i j

k g j

 

x i g i

i

i

x i

D i

 1

g

  

i g i

54 • Momentum balance – linear 

u i

 

j D

ji

I

angular 

i

 

D j

ji

 

f i

e ijk

jk g j

in  T  

Im i

in  T 

u i

 

I

i

e ijk

   

e ijk g j

   

j j D u i

   

D j u k g j

  

e ijk

i g j

 

Governing Equations

j i D u j

e ijk

D j

    

j D u k g k

       

e ijk

i g k

  

Im i j

 

k g j

 

j D

i

       

k g i j i D

j

   

f i

• Characteristics – reproduction of Eringen’s work in absence of fractal effects – limitation to box shaped domains that can be contained in Cartesian system – fractal dimensions in three directions 55

Numerical Solution

•  

Finite element formulation

g u w d

3

i i

         

g u w

3

i g g i j

   ,

j

g u w

3

g g i j

 

d

         

g u w

3

i

 

g g j j

,

j

g u w

3

g g j j

  

d

 •   

u w d i i

   inertia term   

u w g g i j

Resulting equation

M

   U = 0

d

    

u w g g j j d

  0 first elastic term

x 2 L

1

L

2  1  1

L

3  6

L 3

second elastic term

D

1 

D

2  1 ln 8 2 ln 3

D

3  ln 2 ln 3

L 1 x

1

L 2 x 3

Electromagnetism on Anisotropic Fractal Media

Charge conservation on anisotropic fractals  

W dS d

  

W

dV D

 global form 

W

local form

J

dV D D

 

d dt

W

 

t

 

dV D

i i

in the respective Cartesian direction as well as the spatial resolution. = 1,2,3) All the formulas may be evaluated by

c

( ) 1

k

 

k

 

l k

l k

0

x k

 

k

 1 ,

k

 1, 2, 3, (no sum) 57

Electromagnetism on Anisotropic Fractal Media

Charge conservation on anisotropic fractals  

W dS d

  

W

dV D

 global form 

W

local form

J

dV D D

 

d dt

W

 

t

 

dV D

Ohm’s law for anisotropic fractals:

J E

or

J i

 

ij E j

… by analogy to elastic media where Hooke's law is unchanged when going from non-fractal to fractal media that result ensured the consistency of the Newtonian and Lagrangian Hamiltonian approaches to the derivation of governing equations 58

Electromagnetism on Anisotropic Fractal Media

Faraday’s law

d dt

A

dS d

  

l

E

d

l

i

 0   

t B k

e kji

D j E i

  

t B k

e kji

1

c

( ) 1

j E

0

  

t

B

 

D

E

Ohm’s law for anisotropic fractals:

J E

or

J i

 

ij E j

… by analogy to elastic media where Hooke's law is unchanged when going from non-fractal to fractal media that result ensured the consistency of the Newtonian and Lagrangian Hamiltonian approaches to the derivation of governing equations 59

Electromagnetism on Anisotropic Fractal Media

Ampère's law 

S

dS d

 

l

H

d

l

 1  

S

n

 ( 

D

H

)

dS d

  

J

 

D

t

 

D

H

dS d

 0 or 1  0

J

  

E

t

c

2 

D

0

where

C

D

t

… subject to constraints:

c

 1/  0  0 Gauss law for magnetism Gauss law 

D

B

 0, 

D

E

 0, # 60

Electromagnetism on Anisotropic Fractal Media L

  0   1 2 Derivation from variational principle  

L

dV dt

 0

i i

 1 2 

A i

 with  

E t i

  the same set of equations as before 

D

1 

B

 

D

0

0

D

1

c

 

E

t

 

D

  1 0

J i

  

E

0 4 

c

J

0

  [Z

AMP

2012] 61

In fractal porous media:

• • • • • • when the surface and volume fractal dimensions (

d

and

D

) become integers (2 and 3, resp.), or when

R

falls outside [

l, L

], all the equations revert back to well-known forms of conventional continuum mechanics of non-fractal media Beltrami-Michell reciprocity theorem, uniqueness extension to thermomechanics with internal variables wave equations derived from variational principle are the same as those from mechanical approach [

IJES

2011, Z

AMP

2009] modified Reynolds stress for turbulence … • •

Study of wave propagation in isotropic fractal media:

isotropic fractal model that emulates fluid material elastodynamics in anisotropic fractal, linear elastic solids