Contact mechanics and elements of tribology
Lecture 5.
Computational Contact Mechanics
Vladislav A. Yastrebov
MINES ParisTech, PSL Research University, Centre des Mat´eriaux, CNRS UMR 7633, Evry, France
@ Centre des Mat´eriaux February 8, 2016
Outline
Introduction
Governing equations
Optimization methods
Resolution algorithm
Examples
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Aircraft’s engine GP 7200 www.safran-group.com
[1] M. W. R. Savage J. Eng. Gas Turb. Power, 134:012501 (2012)
V.A. Yastrebov 3/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
High speed train TGVwww.sncf.com
Wilde/ANSYSwildeanalysis.co.uk
V.A. Yastrebov 4/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Bearings www.skf.com
[1] F. Massi, J. Rocchi, A. Culla, Y. Berthier Mech. Syst. Signal Pr., 24:1068-1080 (2010)
V.A. Yastrebov 5/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Helical gearwww.tpg.com.tw
www.mscsoftware.com
V.A. Yastrebov 6/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Assembled breaking system www.brembo.com
www.mechanicalengineeringblog.com
V.A. Yastrebov 7/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Tire-road contactwww.michelin.com
[1] M. Brinkmeier, U. Nackenhorst, S. Petersen, O. von Estorff, J. Sound Vib., 309:20-39 (2008)
V.A. Yastrebov 8/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Deep drawingwww.thomasnet.com
[1] G. Rousselier, F. Barlat, J. W. Yoon Int. J. Plasticity, 25:2383-2409 (2009)
V.A. Yastrebov 9/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Crash-testwww.porsche.com
[1] O. Klyavin, A. Michailov, A. Borovkov www.fea.ru
V.A. Yastrebov 10/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Human articulations www.sportssupplements.net
J. A. Weiss, University of Utah Musculoskeletal Research Laboratories
V.A. Yastrebov 11/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Sand duneswww.en.wikipedia.org
E. Azema et al, LMGC90
V.A. Yastrebov 12/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Damage at electric contact zone www.taicaan.com
Simulation of electric current www.comsol.com
V.A. Yastrebov 13/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
San-Andreas fault, by M. Rightmire www.sciencedude.ocregister.com
[1] J.D. Garaud, L. Fleitout, G. Cailletaud Colloque CSMA (2009)
V.A. Yastrebov 14/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Drill Bit toolRobitRocktools; extraction of geothermal energy (SINTEF,NTNU)
[1] T. Saksala, Int. J. Numer. Anal. Meth.
Geomech. (2012)
V.A. Yastrebov 15/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Impact crater, Arizona www.MrEclipse.cometmaps.google.com
Simulation of formation of Copernicus crater Yue Z., Johnson B. C., et al. Projectile remnants in central peaks of lunar impact
craters. Nature Geo 6 (2013)
V.A. Yastrebov 16/75
Industrial and natural contact problems
1
Assembled parts, e.g. engines
2
Railroad contacts
3
Gears and bearings
4
Breaking systems
5
Tire-road contact
6
Metal forming
7
Crash tests
8
Biomechanics
9
Granular materials
10
Electric contacts
11
Tectonic motions
12
Deep drilling
13
Impact and fragmentation
14
etc.
Impact crater, Arizona www.MrEclipse.cometmaps.google.com
Simulation of formation of Copernicus crater Yue Z., Johnson B. C., et al. Projectile remnants in central peaks of lunar impact
craters. Nature Geo 6 (2013)
V.A. Yastrebov 17/75
Equilibrium and contact conditions
Balance of momentum
∇ · = + σ f
v= 0 in Ω
1,2σ = · n = t
0on Γ
fu = u
0on Γ
u? on Γ
cFrictionless contact conditions (intuitive)
1
No penetration
2
No adhesion
3
No shear transfer
u f u
f 1
2
c1 c2
Equilibrium and contact conditions
Balance of momentum
∇ · = + σ f
v= 0 in Ω
1,2σ = · n = t
0on Γ
fu = u
0on Γ
u? on Γ
cFrictionless contact conditions (intuitive)
1
No penetration
2
No adhesion
3
No shear transfer
n v
u f u
f 1
2 c1
c2
Equilibrium and contact conditions
Balance of momentum
∇ · = + σ f
v= 0 in Ω
1,2σ = · n = t
0on Γ
fu = u
0on Γ
u? on Γ
cFrictionless contact conditions (intuitive)
1
No penetration
2
No adhesion
3
No shear transfer
u f u
f 1
2 c1
c2
Equilibrium and contact conditions
Balance of momentum
∇ · = + σ f
v= 0 in Ω
1,2σ = · n = t
0on Γ
fu = u
0on Γ
u? on Γ
cFrictionless contact conditions (intuitive)
1
No penetration
2
No adhesion
3
No shear transfer
u f u
f 1
2 c1
c2
Equilibrium and contact conditions
Balance of momentum
∇ · = + σ f
v= 0 in Ω
1,2σ = · n = t
0on Γ
fu = u
0on Γ
u? on Γ
cFrictionless contact conditions (intuitive)
1
No penetration
2
No adhesion
3
No shear transfer
u f u
f 1
2 c1
c2
Gap function
Gap function g
gap = – penetration asymmetric function defined for
• separation g > 0
• contact g = 0
• penetration g < 0 governs normal contact
Master and slave split
Gap function is determined for all slave points with respect to the master surface
g=0 g<0
g>0
penetration non-contact
contact
n n n
Gap between a slave point and a master surface
V.A. Yastrebov 23/75
Gap function
Gap function g
gap = – penetration asymmetric function defined for
• separation g > 0
• contact g = 0
• penetration g < 0 governs normal contact
Master and slave split
Gap function is determined for all slave points with respect to the master surface
Normal gap
g
n= n · h
r
s− ρ(ξ
π) i , n is a unit normal vector, r
sslave point, ρ(ξ
π) projection point at master surface
g=0 g<0
g>0
penetration non-contact
contact
n n n
Gap between a slave point and a master surface
n
( (
r
sDefinition of the normal gap
Consider existence and uniqueness
V.A. Yastrebov 24/75
Frictionless or normal contact conditions
No penetration
Always non-negative gap g ≥ 0
No adhesion
Always non-positive contact pressure σ
∗n≤ 0
Complementary condition
Either zero gap and non-zero pressure, or non-zero gap and zero pressure
g σ
n= 0
No shear transfer (automatically) σ
∗∗t= 0
σ∗n=(σ=·n)·n=σ=: (n⊗n) σ∗∗t =σ=·n−σnn=n·σ=·
=I−n⊗n
g
n 0
Scheme explaining normal
contact conditions
Frictionless or normal contact conditions
No penetration
Always non-negative gap g ≥ 0
No adhesion
Always non-positive contact pressure σ
∗n≤ 0
Complementary condition
Either zero gap and non-zero pressure, or non-zero gap and zero pressure
g σ
n= 0
No shear transfer (automatically) σ
∗∗t= 0
σ∗n=(σ=·n)·n=σ=: (n⊗n) σ∗∗t =σ=·n−σnn=n·σ=·
=I−n⊗n
g = 0,
n< 0 co nt ac t
non-contact
restricted regions
g
n
g > 0,
n= 0 0
Improved scheme explaining
normal contact conditions
Frictionless or normal contact conditions
In mechanics:
Normal contact conditions
≡
Frictionless contact conditions
≡
Hertz
1-Signorini
[2]conditions
≡
Hertz
1-Signorini
[2]-Moreau
[3]conditions also known in optimization theory as Karush
[4]-Kuhn
[5]-Tucker
[6]conditions
g = 0,
n< 0 co nt ac t
non-contact
restricted regions
g
n
g > 0,
n= 0 0
Improved scheme explaining normal contact conditions
g ≥ 0, σ
n≤ 0, gσ
n= 0
1Heinrich Rudolf Hertz(1857–1894) a German physicist who first formulated and solved the frictionless contact problem between elastic ellipsoidal bodies.
2Antonio Signorini(1888–1963) an Italian mathematical physicist who gave a general and rigorous mathematical formulation of contact constraints.
3Jean Jacques Moreau(1923) a French mathematician who formulated a non-convex optimization problem based on these conditions and introduced pseudo-potentials in contact mechanics.
4William Karush(1917–1997),5Harold William Kuhn(1925) American mathematicians, 6Albert William Tucker(1905–1995) a Canadian mathematician.
Relative sliding
Recall:
• Convective coordinate in parent space ξ
i∈ ( − 1; 1)
• Mapping to real space
ρ(ξ
1, ξ
2, t) =
8
X
i=1
N
i(ξ
1, ξ
2)ρ
i(t)
1 5 26 3 7 4
8
2
3
4 1
6 5
7 8
( )
e
3e
2e
1*1 * 2}
} ,
*1 * ,2 1
2
1 -1
-1 1
Relative sliding
Recall:
• Convective coordinate in parent space ξ
i∈ ( − 1; 1)
• Mapping to real space
ρ(ξ
1, ξ
2, t) =
8
X
i=1
N
i(ξ
1, ξ
2)ρ
i(t)
1 5 26 3 7 4
8
2
3
4 1
6 5
7 8
( )
e
3e
2e
1*1 * 2}
} ,
*1 * ,2 1
2
1 -1
-1 1
Tangential slip velocity v
tmust take into account:
• only tangential component
• relative rigid body motion
• master’s deformation
v
t= ∂ ρ
∂ξ
1ξ ˙
1+ ∂ ρ
∂ξ
2ξ ˙
2where ∂ρ/∂ξ
iare the tangent vectors of the local basis and ξ
iare the convec- tive coordinates.
Relative slip between a slave point and a
deformable master surface
Relative sliding: example
Consider a one-dimensional example:
P is a projection of A on segment BC.
x
P= ξx
C+ (1 − ξ)x
B(1)
Velocity of the projection point
˙
x
P= ξ x ˙
C+ (1 − ξ) ˙ x
B| {z }
∂xP
∂t
+ (x
C− x
B) ˙ ξ
| {z }
∂xP
∂ξξ˙
Substract the velocity of point x
Pfor fixed ξ
v
t= x ˙
P−
∂xP∂t
= (x
C− x
B) ˙ ξ =
∂ξ∂xξ ˙
Compute tangential slip increment
∆g
n+1t=
∂x∂ξξn
(ξ
n+1− ξ
n)
A
B C
x P
Example of a one-dimensional
relative slip
Relative sliding: example
Consider a one-dimensional example:
P is a projection of A on segment BC.
x
P= ξx
C+ (1 − ξ)x
B(1)
Velocity of the projection point
˙
x
P= ξ x ˙
C+ (1 − ξ) ˙ x
B| {z }
∂xP
∂t
+ (x
C− x
B) ˙ ξ
| {z }
∂xP
∂ξξ˙
Substract the velocity of point x
Pfor fixed ξ
v
t= x ˙
P−
∂xP∂t
= (x
C− x
B) ˙ ξ =
∂ξ∂xξ ˙
Compute tangential slip increment
∆g
n+1t=
∂x∂ξξn
(ξ
n+1− ξ
n)
A
B C
x P
Example of a one-dimensional relative slip
Ship-river analogy
Relative sliding: example
Consider a one-dimensional example:
P is a projection of A on segment BC.
x
P= ξx
C+ (1 − ξ)x
B(1)
Velocity of the projection point
˙
x
P= ξ x ˙
C+ (1 − ξ) ˙ x
B| {z }
∂xP
∂t
+ (x
C− x
B) ˙ ξ
| {z }
∂xP
∂ξξ˙
Substract the velocity of point x
Pfor fixed ξ
v
t= x ˙
P−
∂xP∂t
= (x
C− x
B) ˙ ξ =
∂ξ∂xξ ˙
Compute tangential slip increment
∆g
n+1t=
∂x∂ξξn
(ξ
n+1− ξ
n)
A
B C
x P
Example of a one-dimensional relative slip
Ship-river analogy
Relative sliding: example
Consider a one-dimensional example:
P is a projection of A on segment BC.
x
P= ξx
C+ (1 − ξ)x
B(1)
Velocity of the projection point
˙
x
P= ξ x ˙
C+ (1 − ξ) ˙ x
B| {z }
∂xP
∂t
+ (x
C− x
B) ˙ ξ
| {z }
∂xP
∂ξξ˙
Substract the velocity of point x
Pfor fixed ξ
v
t= x ˙
P−
∂xP∂t
= (x
C− x
B) ˙ ξ =
∂ξ∂xξ ˙
Compute tangential slip increment
∆g
n+1t=
∂x∂ξξn
(ξ
n+1− ξ
n)
A
B C
x P
Example of a one-dimensional relative slip
Ship-river analogy Li derivative: the change of a vector field along the change of
another vector field
Amontons-Coulomb’s friction
No contact g > 0, σ
n= 0 Stick | v
t| = 0
Inside slip surface / Coulomb’s cone
f = | σ
t| − µ | σ
n| < 0 Slip | v
t| > 0
On slip surface / Coulomb’s cone
f = | σ
t| − µ | σ
n| = 0 Complementary condition
Either zero velocity and negative slip criterion, or non-zero velocity and zero slip criterion
| v
t|
| σ
t| − µ | σ
n|
= 0
n t
v
t tn 1
0 0
Scheme explaining frictional contact
conditions
Amontons-Coulomb’s friction
No contact g > 0, σ
n= 0 Stick | v
t| = 0
Inside slip surface / Coulomb’s cone
f = | σ
t| − µ | σ
n| < 0 Slip | v
t| > 0
On slip surface / Coulomb’s cone
f = | σ
t| − µ | σ
n| = 0 Complementary condition
Either zero velocity and negative slip criterion, or non-zero velocity and zero slip criterion
| v
t|
| σ
t| − µ | σ
n|
= 0
st ic k
n
restricted regions slip
t
v
tstick
t
n 1
restricted
region
0 0
slip
Improved scheme explaining
frictional contact conditions
Amontons-Coulomb’s friction
No contact g > 0, σ
n= 0 Stick | v
t| = 0
Inside slip surface / Coulomb’s cone
f = | σ
t| − µ | σ
n| < 0 Slip | v
t| > 0
On slip surface / Coulomb’s cone
f = | σ
t| − µ | σ
n| = 0 Complementary condition
Either zero velocity and negative slip criterion, or non-zero velocity and zero slip criterion
| v
t|
| σ
t| − µ | σ
n|
= 0
Scheme of 2D frictional contact
t2
t1 n
v
t nn
t2
t1
stick
slip slip
stick
Scheme of 3D frictional contact
| v
t| ≥ 0, | σ
t| − µ | σ
n| ≤ 0, | v
t|
| σ
t| − µ | σ
n|
= 0
More friction laws
• Static criteria
stick
t
0 n
stick
t
0 n
slip slip
stick
t
0 n
slip max t
stick
t
0 n
max t
slip
(a) (b) (c) (d)
(a) Tresca (b) Amontons-Coulomb (c) Coulomb-Orowan (d) Shaw
• Kinetic criteria
stick
n
slip
t
vt
0
s
k
stick
n
slip
t
vt
0
s
k
(a) (b)
stick
n slip
t
vt
0
s
(c)
stick
n
slip
t
0
s
k
(d)
log( +v )0 gt
(a,b) velocity weakening (c) velocity weakening-strengthening (d) Linear slip weakening
• µ
sstatic and µ
kkinetic coefficients of friction.
Rate and state friction and regularization
• Rate and state friction law
Rate v
t= | v
t| – relative slip velocity State θ – ≈ internal time
Dieterich–Ruina–Perrin (1979, 83, 95) Frictional resistance
σ
ct= | σ
n| µ
s+ bθ + a ln(v
t/v
0) Evolution of the state variable
θ ˙ = −
vtL
h θ + ln
vt v0
i
• Prakash-Clifton friction law (1992,2000)
Viscous type evolution of frictional resistance σ
tσ ˙
t= −
vtL
(σ
t+ µσ
n)
0 0.1 0.2 0.3 0.4 0.5
Slip velocity
Slip velocity
0.2 0.3 0.4 0.5 0.6 0.7 0.8
500 600 700 800 900 1000 1100 1200
Frictional resistance
Time Resistance
Rate and state friction law
0 100 200 300 400 500
Contact pressure
Contact pressure
0 100 200 300 400 500
500 600 700 800 900 1000 1100 1200
Frictional resistance
Time Resistance
Prakash-Clifton regularization
Rate and state friction and regularization
• Rate and state friction law
0 0.1 0.2 0.3 0.4 0.5
Slip velocity
Slip velocity
0.2 0.3 0.4 0.5 0.6 0.7 0.8
500 600 700 800 900 1000 1100 1200
Frictional resistance
Time Resistance
Rate and state friction and regularization
• Rate and state friction law
Rate v
t= | v
t| – relative slip velocity State θ – ≈ internal time
Dieterich–Ruina–Perrin (1979, 83, 95) Frictional resistance
σ
ct= | σ
n| µ
s+ bθ + a ln(v
t/v
0) Evolution of the state variable
θ ˙ = −
vtL
h θ + ln
vt v0
i
• Prakash-Clifton friction law (1992,2000)
Viscous type evolution of frictional resistance σ
tσ ˙
t= −
vtL
(σ
t+ µσ
n)
0 0.1 0.2 0.3 0.4 0.5
Slip velocity
Slip velocity
0.2 0.3 0.4 0.5 0.6 0.7 0.8
500 600 700 800 900 1000 1100 1200
Frictional resistance
Time Resistance
Rate and state friction law
0 100 200 300 400 500
Contact pressure
Contact pressure
0 100 200 300 400 500
500 600 700 800 900 1000 1100 1200
Frictional resistance
Time Resistance
Prakash-Clifton regularization
Rate and state friction and regularization
• Prakash-Clifton friction law (1992,2000)
0 100 200 300 400 500
Contact pressure
Contact pressure
0 100 200 300 400 500
500 600 700 800 900 1000 1100 1200
Frictional resistance
Time Resistance
Rate and state friction and regularization
• Prakash-Clifton friction law (1992,2000)
0 100 200 300 400 500
Contact pressure
Time Contact pressure
0 0.5 1 1.5 2
500 600 700 800 900 1000 1100 1200
Normalized frictional resistance
Time Normalized resistance
From strong to weak form
• Balance of momentum and boundary conditions
∇ · = + σ f
v= 0 in Ω = Ω
1∪ Ω
2+ B.C.
u f u
f 1
2 c1 c2
Two solids in contact
From strong to weak form
• Balance of momentum and boundary conditions
∇ · = + σ f
v= 0 in Ω = Ω
1∪ Ω
2+ B.C.
• Balance of virtual works Z
∂Ω
n · σ = · δu dΓ + Z
Ω
h f
v· δu − σ = ·· δ ∇ u i dΩ = 0
u f u
f 1
2 c1 c2
Two solids in contact
From strong to weak form
• Balance of momentum and boundary conditions
∇ · = + σ f
v= 0 in Ω = Ω
1∪ Ω
2+ B.C.
• Balance of virtual works Z
∂Ω
n · σ = · δu dΓ = Z
Ω
h f
v· δu − = σ ·· δ ∇ u i dΩ = 0
Z
Γc1
n · = σ · δρ dΓ
c1+ Z
Γc2
ν · σ = · δr dΓ
2c+ Z
Γf
σ
0· δu dΓ
fn v
u f u
f 1
2 c1
c2
Two solids in contact
From strong to weak form
• Balance of momentum and boundary conditions
∇ · = + σ f
v= 0 in Ω = Ω
1∪ Ω
2+ B.C.
• Balance of virtual works Z
∂Ω
n · = σ · δu dΓ ⇒ Z
Ω
h f
v· δu − σ = ·· δ ∇ u i dΩ = 0
Z
Γc1
n · = σ · δρ dΓ
c1+ Z
Γc2
ν · = σ · δr dΓ
c2= Z
Γf
σ
0· δu dΓ
f= Z
Γc1
n · = σ · δ(ρ − r) dΓ
c1= Z
Γc1
σ
nδg
n+ ∼ σ
Ttδ ∼ ξ dΓ
1cn v
u f u
f 1
2 c1
c2
Two solids in contact
From strong to weak form
• Balance of momentum and boundary conditions
∇ · = + σ f
v= 0 in Ω = Ω
1∪ Ω
2+ B.C.
• Balance of virtual works Z
∂Ω
n · = σ · δu dΓ ⇒ Z
Ω
h f
v· δu − σ = ·· δ ∇ u i dΩ = 0
Z
Γc1
n · = σ · δρ dΓ
c1+ Z
Γc2
ν · = σ · δr dΓ
c2= Z
Γf
σ
0· δu dΓ
f= Z
Γc1
n · = σ · δ(ρ − r) dΓ
c1= Z
Γc1
σ
nδg
n+ ∼ σ
Ttδ ∼ ξ dΓ
1cn v
u f u
f 1
2 c1
c2
Two solids in contact
Z
Ω
σ = · · δ ∇ u dΩ + Z
Γc1
σ
nδg
n+ ∼ σ
Ttδ ∼ ξ dΓ
c1| {z } Contact term
= Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ
From strong to weak form
• Balance of momentum and boundary conditions
∇ · = + σ f
v= 0 in Ω = Ω
1∪ Ω
2+ B.C.
• Balance of virtual works Z
Ω
σ = · · δ ∇ u dΩ
| {z }
Change of the internal energy
+ Z
Γc1
σ
nδg
n+ ∼ σ
Ttδ ∼ ξ dΓ
c1| {z } Contact term
=
Z
Γf
σ
0· δu dΓ
| {z }
Virtual work of external forces
+ Z
Ω
f
v· δu dΩ
| {z }
Virtual work of volume forces
n v
u f u
f 1
2 c1
c2
Two solids in contact
• Functional space
u ∈ H
1(Ω) Hilbert space of the first order
and u satisfy boundary and contact conditions.
Towards variational inequality
Contact term Z
Γc1
σ
nδg
n+ ∼ σ
Ttδ ξ ∼ dΓ
c1Z
Γc1
σ
nδg
ndΓ
1c≤ 0
Contact configuration
σnδgn=0, σn≤0Z
Ω
= σ · · δ ∇ u dΩ + Z
Γc1
∼ σ
Ttδ ∼ ξ dΓ
c1
≥
Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ,
K = n
δu ∈ H
1(Ω)
δu = 0 on Γ
u, g
n(u + δu) ≥ 0 on Γ
co
Towards variational inequality
Contact term Z
Γc1
σ
nδg
n+ ∼ σ
Ttδ ξ ∼ dΓ
c1Z
Γc1
σ
nδg
ndΓ
1c≤ 0
Virtual change of the configuration Z
Ω
= σ · · δ ∇ u dΩ + Z
Γc1
∼ σ
Ttδ ∼ ξ dΓ
c1
≥
Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ,
K = n
δu ∈ H
1(Ω)
δu = 0 on Γ
u, g
n(u + δu) ≥ 0 on Γ
co
Towards variational inequality
Contact term Z
Γc1
σ
nδg
n+ ∼ σ
Ttδ ξ ∼ dΓ
c1Z
Γc1
σ
nδg
ndΓ
1c≤ 0
Normal term in separation
δgn>0Z
Ω
= σ · · δ ∇ u dΩ + Z
Γc1
∼ σ
Ttδ ∼ ξ dΓ
c1
≥
Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ,
K = n
δu ∈ H
1(Ω)
δu = 0 on Γ
u, g
n(u + δu) ≥ 0 on Γ
co
Towards variational inequality
Contact term Z
Γc1
σ
nδg
n+ ∼ σ
Ttδ ξ ∼ dΓ
c1Z
Γc1
σ
nδg
ndΓ
1c≤ 0
Normal term in sliding
δgn=0Z
Ω
= σ · · δ ∇ u dΩ + Z
Γc1
∼ σ
Ttδ ∼ ξ dΓ
c1
≥
Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ,
K = n
δu ∈ H
1(Ω)
δu = 0 on Γ
u, g
n(u + δu) ≥ 0 on Γ
co
Back to variational equality (unconstrained)
• Constrained minimization problem Z
Ω
= σ · · δ ∇ u dΩ + Z
Γc1
∼ σ
Ttδ ∼ ξ dΓ
c1
≥
Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ, K = n
δu ∈ H
1(Ω)
δu = 0 on Γ
u, g
n(u + δu) ≥ 0 on Γ
co
• Use optimization theory to convert to Z
Ω
σ = · · δ ∇ u dΩ + Z
Γ1c
C(σ
n, σ
t, g
n, ∼ ξ , δu)
| {z } Contact term
∗dΓ
1c= Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ,
Unconstrained functional space K = n
δu ∈ H
1(Ω)
δu = 0 on Γ
uo
Contact term
∗is defined on the potential contact zone Γ
1c.
Optimization methods: recall
Functional to be minimized F(x) under constraint g(x) ≥ 0 Penalty method
Lagrange multipliers method
Augmented Lagrangian method
Optimization methods: recall
Functional to be minimized F(x) under constraint g(x) ≥ 0 Penalty method
• New functional
F
p(x) = F(x) + − g(x)
2= F(x) +
0, if g(x) ≥ 0 non-contact g
2(x), if g(x) < 0 contact where is the penalty parameter.
• Stationary point must satisfy
∇ F
p(x) = ∇ F(x) + 2 − g(x) ∇ g(x) = 0
• Solution tends to the precise solution as → ∞
Lagrange multipliers method Augmented Lagrangian method
Macaulay bracketshxi=
x, ifx≥0 0, otherwise
Optimization methods: recall
Functional to be minimized F(x) under constraint g(x) ≥ 0
Penalty method F
p(x) = F(x) + − g(x)
2Lagrange multipliers method
• New functional called Lagrangian L(x, λ) = F(x) + λg(x)
• Saddle point problem min
xmax
λ
{ L(x, λ) } −→ x
∗←− min
g(x)≥0
{ F(x) }
• Stationary point
∇
x,λL =
"
∇
xF(x) + λ ∇
xg(x) g(x)
#
= 0 need to verify λ ≤ 0
Augmented Lagrangian method
Macaulay bracketshxi=
x, ifx≥0 0, otherwise
Optimization methods: recall
Functional to be minimized F(x) under constraint g(x) ≥ 0
Penalty method F
p(x) = F(x) + − g(x)
2Lagrange multipliers method L(x, λ) = F(x) + λg(x)
Augmented Lagrangian method
[Hestnes 1969], [Powell 1969], [Glowinski & Le Tallec 1989], [Alart & Curnier 1991], [Simo & Laursen 1992]
• New functional, augmented Lagrangian
L
a(x, λ) = F(x) +
λg(x) + g
2(x) , if λ + 2g(x) ≥ 0, contact
−
14
λ
2, if λ + 2g(x) < 0, non-contact
• Stationary point
∇
x,λL
a=
∇
xF(x) + λ ∇
xg(x) + 2g(x) ∇ g(x) g(x)
= 0, if contact
∇
xF(x)
−
λ
= 0, if non-contact
Macaulay bracketshxi=
x, ifx≥0 0, otherwise
Optimization methods: example
Functional : f (x) = x
2+ 2x + 1 Constrain : g(x) = x ≥ 0
Solution : x
∗= 0
Optimization methods: example
Functional : f (x) = x
2+ 2x + 1 Constrain : g(x) = x ≥ 0
Solution : x
∗= 0
Penalty method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Penalty method
F
p(x) = F(x) + − g(x)
2Penalty method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Penalty method
F
p(x) = F(x) + − g(x)
2= 0
Penalty method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Penalty method
F
p(x) = F(x) + − g(x)
2= 1
Penalty method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Penalty method
F
p(x) = F(x) + − g(x)
2= 10
Penalty method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Penalty method
F
p(x) = F(x) + − g(x)
2= 50
Penalty method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Penalty method
F
p(x) = F(x) + − g(x)
2Advantages ,
simple physical interpretation simple implementation
no additional degrees of freedom
“mathematically” smooth functional
Drawbacks /
practically non-smooth functional
solution is not exact:
too small penalty →
large penetration
too large penalty →
ill-conditioning of the
tangent matrix
user has to choose penalty
properly or automatically and/or
adapt during convergence
Lagrange multipliers method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Lagrange multipliers method
L(x, λ) = F(x) + λg(x) → Saddle point → min
x
max
λ
L(x, λ) Need to check that λ ≤ 0
-2 -1 0 1 2
-3 -2 -1 0 1
X λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
Lagrange multipliers method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Lagrange multipliers method
L(x, λ) = F(x) + λg(x) → Saddle point → min
x
max
λ
L(x, λ) Need to check that λ ≤ 0
Advantages ,
exact solution
no adjustable parameters
Drawbacks /
Lagrangian is not smooth
additional degrees of freedom
not fully unconstrained: λ ≤ 0
Augmented Lagrangian method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Augmented Lagrangian method
L
a(x, λ) = F(x) +
λg(x) + g
2(x) , if λ + 2g(x) ≥ 0, contact
−
14
λ
2, if λ + 2g(x) < 0, non-contact
-2 -1 0 1 2
-3 -2 -1 0 1
X λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
Yellow line separates contact and non-contact regions
Augmented Lagrangian method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Augmented Lagrangian method
L
a(x, λ) = F(x) +
λg(x) + g
2(x) , if λ + 2g(x) ≥ 0, contact
−
14
λ
2, if λ + 2g(x) < 0, non-contact
-2 -1 0 1 2
-3 -2 -1 0 1
X λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
Yellow line separates contact and non-contact regions
Augmented Lagrangian method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Augmented Lagrangian method
L
a(x, λ) = F(x) +
λg(x) + g
2(x) , if λ + 2g(x) ≥ 0, contact
−
14
λ
2, if λ + 2g(x) < 0, non-contact
-2 -1 0 1 2
-3 -2 -1 0 1
X λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
-2 -1 0 1 2
-3 -2 -1 0 1
X
λ
Yellow line separates contact and non-contact regions
Augmented Lagrangian method: example
F(x) = x
2+ 2x + 1, g(x) = x ≥ 0, x
∗= 0 Augmented Lagrangian method
L
a(x, λ) = F(x) +
λg(x) + g
2(x) , if λ + 2g(x) ≥ 0, contact
−
14
λ
2, if λ + 2g(x) < 0, non-contact
Advantages ,
exact solution smooth functional (!) fully unconstrained
Drawbacks /
additional degrees of freedom
quite sensitive to parameter
need to adjust during
convergence
Application to contact problems: weak form
Z
Ω
= σ · · δ ∇ u dΩ + Z
Γ1c
C
|{z}
Contact term dΓ
1c=
Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ,
K = n
δu ∈ H
1(Ω)
δu = 0 on Γ
uo
Penalty method
Pressure: σ
n= g
n, Shear: σ
t=
g
t, if stick | σ
t| < µ | σ
n| µg
nδg
t
/ | δg
t
| , if slip | σ
t| = µ | σ
n|
Contact term C = C(g
n, g
t
, δg
n, δg
t
) = σ
nδg
n+ σ
t· δg
t
Application to contact problems: weak form
Z
Ω
= σ · · δ ∇ u dΩ + Z
Γ1c
C
|{z}
Contact term dΓ
1c=
Z
Γf
σ
0· δu dΓ + Z
Ω
f
v· δu dΩ,
K = n
δu ∈ H
1(Ω)
δu = 0 on Γ
uo
Augmented Lagrangian method
Contact term C = C(g
n, g
t
, λ
n, λ
t, δg
n, δg
t
, δλ
n, δλ
t)
C=
−1
λnδλn−λt·δλt
, if non-contactλn+gn≥0
λˆnδgn+gnδλn+λˆt·δg
t+g
t
·δλˆt, if stick|ˆλt| ≤µ|σˆn|
λˆnδgn+gnδλn+µσˆn−µσˆn
λˆt
|λˆt|
·δg
t
−1
λt+µσˆn
λˆt
|λˆt|
·δλt, if slip|ˆλt| ≥µ|σˆn|
where λ ˆ
n= λ
n+ g
nand λ ˆ
t= λ
t+ g
t
.
Friction . . . .
Friction . . . .
The scream
Friction: Return mapping algorithm
Return mapping algorithm in 2D
t i+1
trial
f
t itrial ti+1
in
1
g
t*ig
t*i+1g
ti titi+1
s
tii+1n
i+1
f 0 <
if 0 <
s
tis
ig
ti+1g
ts
i t i ig
t*g
t ig
t ig
t ig
t*i ini+1n
As in plasticity[1]
[1] Simo J.C. and Hughes T.J.. Computational inelasticity. Springer (2006)
Friction: Return mapping algorithm
Return mapping algorithm in 2D
n
1
trial ti+1 ti+1 ti
in i+1n ni
i+1n
trial ti+1 ti
t t
1
in i
n i+1n i+1n
n
St
Analogy with non-associated plastic flow[2]
[2] Curnier A. A theory of friction. International Journal of Solids and Structures 20 (1984)
Friction: Return mapping algorithm
Return mapping algorithm in 3D
(
trial
sit
i+1
(
trial
y y
i+1
i+1
i
i+1 i
e
1e
2e
1e
2i
i
i i i
i+1
* i
i
(a) (b)
(d) (c)
i
x x
x x
e
1e
2e
1e
2triali+1
stif
g
g g
g g
g
g g g
g g g
g
g g
g g g g
f
si
si si
si si
t t t
t
t t
t t
t t
t t
t
t
t t t
t t t t
t
t t t
t t
t t
trial
i+1 i
i
i
i i
i
i
s
i+1 i
i+1 i+1 i+1
i+1
i+1
i+1
i+1 i
( (
n n
n n
n n
n
*
*
*
* i i
y y
A sligthly more messy thing
Application to contact problems: linearization
• Non-linear equation
R(u, f ) = 0
• Contains δg
n, δg
• Use Newton-Raphson method
t• Initial state at step i
R(u
i, f
i) = 0
• Should be also satisfied at step i + 1
R(u
i+1, f
i+1) = R(u
i+ δu, f
i+1) = 0
• Linearize
R(u
i+ δu, f
i+1) = R(u
i, f
i+1) + ∂R(u)
∂u δu = 0
• Finally
δu = −
" ∂R(u)
∂u
#
−1| {z } contains ∆δg
n, ∆δg
t