Lecture11:Strainpart2

GEOS655TectonicGeodesy
JeFreymueller

StrainandRota@onTensors
Wedescribedthedeforma@onasthesumoftwo
tensors,astraintensorandarota@ontensor.
1 ui u j
1 ui u j
ui x 0 + dx = ui x 0 +
+

dx j +
dx j
2 x j x i
2 x j x i

( )

( )

ui x 0 + dx = ui x 0 + ij dx j + ij dx j
u1

x1

u2
x1
u
3
x1

u1
x 2
u2
x 2
u3
x 2

u1
u1

x1
x 3
u2 1 u1 u2
=
+

x 3 2 x 2 x1
u3 1 u u
1
3

x 3 2 x + x

3
1

1 u1 u2
+

2 x 2 x1
u2
x 2
1 u2 u3
+

2 x 3 x 2

1 u1 u3
+
0

2 x 3 x1
1 u2 u3 1 u2 u1
+

2 x 3 x 2 2 x1 x 2
u3
1 u3 u1
2 x x
x 3

1
3

symmetric, strain

1 u1 u2 1 u1 u3

2 x 2 x1 2 x 3 x1
1 u2 u3
0

2 x 3 x 2
1 u3 u2

2 x 2 x 3

anti-symmetric, rotation

OK,SoWhatisaTensor,Anyway
Tensor,nottonsure!
Examplesoftensorsofvariousranks:
Rank0:scalar
Rank1:vector
Rank2:matrix
AtensorofrankN+1islikeasetoftensorsofrankN,like
youcanthinkofamatrixasasetofcolumnvectors

Thestresstensorwasactuallythersttensorthe
mathema@cswasdevelopedtodealwithstress.
Themathema@caldeni@onisbasedon
transforma@onproper@es.

AndWri@ngTheminIndexNota@on
Wecanwritethestraintensor
11 12 13

1 u u

ij = i + j = 21 22 23
2 x j x i
31 32 33

Thestraintensorissymmetric,soonly6components
areindependent.Symmetryrequiresthat:

21 = 12
31 = 13
23 = 32

Therota@ontensorisan@symmetricandhasonly3
0
12 13
independentcomponents:

1 u u

ij = i j = 12
0
23
2 x j x i
13 23 0

Es@ma@ngStrainandRota@on
fromGPSData

ThisispreYyeasy.Wecanes@mateallofthecomponentsof
thestrainandrota@ontensorsdirectlyfromtheGPSdata.
Wecanwriteequa@onsintermsofthe6independentstraintensor
componentsand3independentrota@ontensorcomponents
Orwecanwriteequa@onsintermsofthe9componentsofthe
displacementgradienttensor

Writethemo@ons(rela@vetoareferencesiteorreference
loca@on)intermsofdistancefromreferencesite:

( )

ui x 0 + dx ui x 0 = ij dx j + ij dx j
Herex0istheposi@onofthereferenceloca@on,anddxisthe
vectorfromreferenceloca@ontowherewehavedata

TheEqua@onsin2D
Theequa@onsbelowarewriYenoutforthe
2Dcase:
v x = Vx + xxx + xy y + y
v y = Vy + xyx + yy y x

For2Dstrain,wehave4parameters(3strain,1
rota@on)andweneed2siteswithhorizontal
data.
For3D,wewouldhave9parametersandwould
require3siteswith3Ddata.

Veloci@esRela@vetoEurasia

Chen et al. (2004), JGR

DisplacementandStrain
Displacements(orrates)areacombina@onof
rigidbodytransla@on,rota@onandinternal
deforma@on

v east v e,body 11
+ 1

=
v north v n,body 2 (12 )
= strain tensor components
= rotation

1
2

(12 + )x
22

(x, y) = position
v = velocity

UniformStrainRate

 0
1 0
1 0
=
+
0
0 1
0 0
1

DeformingBlockModel
BasedonGPSdatafrom44sites
Fourblocksmovingrela@vetoeachother
onmajorfaults,plusuniformstrain
Blockmo@onsarepredominantlystrikeslip

Modelswithspa@alvaria@onsinstraindo
nottsignicantlybeYerthanuniform
strain
Modelswithallslipconcentratedonafew
faultstworsethanthedeformingblock
model.

4.4+/-1.1

7.4+/-0.7

5.9+/-0.7

Chenetal.DeformingBlockModel

Rota@onTensor

1 u2 u1

2 x1 x 2
1 u3 u1
2 x x

1
3

In terms of our earlier

coordinate rotation matrix:

1 u1 u2 1 u1 u3

2 x 2 x1 2 x 3 x1
1 u2 u3
0

2 x 3 x 2
1 u3 u2

2 x 2 x 3

1 u2 u1

1
2

1 u3 u1
2 x x

1
3

1 u1 u2

2 x 2 x1
0
1 u3 u2

2 x 2 x 3

1 u1 u3

2 x 3 x1
1 u2 u3

= ( ij ) x x
2 x 3 x 2

Rota@onasaVector
Wecanrepresenttherota@ontensorasavector.
Why?
Rota@oncanbedescribedbyavector;thinkangular
velocityvector
Withonlythreeindependentcomponents,thenumberof
termsaddsup

Wecanwriteacrossproductoperatorusingthe
permuta@ontensor(actuallyatensorofrank3,
likeamatrixwith3dimensions): k = 12 eijk ij
e123 = e231 = e312 = 1

eijk = e213 = e321 = e132 = 1

eiii = 0

MoreonthePermuta@onTensor
Wecanwritethevectorcrossproductintermsofthis
permuta@ontensor

(a b) k = eijk aib j
Rememberthisrule?

Thepermuta@ontensorislikeashorthandforthatrule

Thepermuta@ontensorcanbewriYenintermsofthe
Kroneckerdelta(thisissome@mesuseful):

eijk eist = jskt jtks

andij
Wecanwriteintermsofu:
k = 12 eijk ij = 12 ekij ij

u
j
i
k = 12 12 ekij
12 ekij

x j
x i

ui 1 u j
1 1
k = 2 2 ekji
+ 2 ekij

Notice the permutation

u
1
k = 2 ekji i
These two terms are equal. Why?
x j
=

1
2

( u)

It is the curl of u

= i+
j+ k
x
y
z

ijand
Wecanturnthisaroundandwriteijintermsof.Start
with:

k = 12 eijk ij
Mul@plybothsidesbyemnk

emnkk = 12 emnkeijk ij
emnkk = 12 (minj mjni ) ij
emnkk = 12 (minj ij mjni ij )
emnkk = 12 ( mn nm )
emnkk = 12 ( mn + mn )
emnkk = mn
ij = eijk k

OneFinalManipula@on
LetsgobacktotheoriginaltermintheTaylorSeries
expansion:
)
u(rot
= ij dx j = eijk k dx j = eikj k dx j
i

u(rot ) = dx
Thismakesitabitmoreclearthatthisisarota@on.

StrainTensor
Axial Strains

u1

x1

1 u1 u2
+

2 x 2 x1
1 u1 u3
2 x + x

3
1

Shear Strains

1 u1 u2
+

2 x 2 x1
u2
x 2
1 u2 u3
+

2 x 3 x 2

1 u1 u3
+

2 x 3 x1
1 u2 u3
+

2 x 3 x 2
u3

x 3

ExamplesofStrains
uniaxial
strains

An example
Including shear strain

PrincipalAxesofStrain
1

Foranystrain,thereisa
coordinatesystemwhere
thestrainsareuniaxialno
shearstrain.
Calledtheprincipalaxesof
thestraintensor.
Strainsinthosedirec@onsare
theprincipalstrains.

Principalaxesarethe
eigenvectorsofthestrain
tensorandprincipalstrains
aretheeigenvalues.

StraininaPar@cularDirec@on
Letslookathowwecanusethestraintensortogetaline
[email protected]
examplein2D:
11 0

11 > 0
0 0
Ifwestartwithacircle,itwillbedeformedintoanellipse.

StraininaPar@cularDirec@on
L

11 0

11 > 0
0 0

ux
r
ux

ux = x11 = ( r cos )11

StraininaPar@cularDirec@on
Forastrainonlyinthexdirec@on,
11 0
2

>
0
L
/L
=

cos

11
11
0 0

Ifwedothesameforastraininonlyinthey
direc@on,wegetsomethingsimilar,

0 0
2

22 > 0 L /L = 22 sin
0 22

Ingeneral,wegetanequa@onforanellipse:
2

e = 11 cos + 22 sin + 12 sin2

rr

Supposewehavemul@plelines
Ifwemeasureline
lengthchangesfor
severallinesofdierent
orienta@ons,i
Ifwehavelinesofthree
ormoredierent
orienta@ons,wecan
es@matethethree
horizontalstrains.
2

(err ) i = 11 cos i + 22 sin i + 12 sin2 i

Supposewehavemul@plelines
Thisisthebasisfor
es@ma@ngstrainfrom
EDMnetworks
Thelinesdonthaveto
havethesameorigin.I
justdrewitthatwayfor
convenience.
Themorelinesofdierent
orienta@ons,thebeYer
thees@matebecomes
2

(err ) i = 11 cos i + 22 sin i + 12 sin2 i

ChenandFreymueller(2002)
Evaluatedstrainrates
forfourdierentnear
faultEDMnetworks
alongtheSanAndreas
fault
Allsitesweresocloseto
thefaultweassume
uniformstrainwithin
thenetwork

Es@matedStrainRateTensors

Es@ma@ngStrainfromLineLengths
Obviously,whatwelookedatisa2Dproblem.Weassumed
thestateofplanestrain,inwhichalldeforma@onis
horizontal.
Linelengthsorlinelengthchangesareinvariantunder
rota@ons,sowecanlearnnothingabouttherota@ontensor
fromthem
Butwealsocanes@matestrainwithoutworryingaboutrota@on

Wehavetoassumeuniformstrain,whichmightbethewrong
assump@on.
Analterna@vemightbetoassumesimpleshear(noaxial
strains),suchasforglacialoworsomestrikeslipfault
mo@onproblems.

StrainandAngleChanges
Ifwestrainanobject,lineswithintheobjectwill,in
general,rotate:

Butauniformstrain(uniformscalingupordownin
size)doesnotrotateanylines.Sothatmeanswe
cantes@mateallthreecomponentsofthestrain
tensorfromanglechanges.

Es@ma@ngStrainfromAngle
Changes
Wecantes@matethedilata@on().Rewriteour
threehorizontalstrainsthisway:

= 11 + 22
1 = 11 22
2 = 212

Dilatation
Pure shear with N-S contraction, E-W extension
Pure shear with NW-SE contraction, NE-SW extension
(or simple shear with displacement along x-axis)

Thegammasarecalledtheengineeringshearstrains
(theepsilonsarethetensorstrains).Thefactorof2is
fromhistoricalprac@ce.
Olderpapersonenusethegammas.Newerworkusually
usestensorstrains.Becarefulofthefactorof2!

Illustra@onsofEngineeringShears
1 = 11 22 > 0

2 = 212 > 0

WhatCanBeDeterminedFromthe
Data?

Thereisagenerallessonhere:

Youmightwanttoknowallcomponentsofthestrainand
rota@ontensors.
Whenyourdataonlydeterminesomepartsofyour
parameters,itmakessensetoreparameterizetheproblem
sothatyoudirectlyes@matewhatyourdatacandetermine
Thealterna@veistoputpossiblyarbitraryconstraintsonyour
parameters

Thediscussionofwhatyoucanlearnaboutaplaterota@on
fromasinglesiteisanotherexample.
Whatwearedoingisbreakingupourparametersintotwo
sets(determinedandimpossibletodetermine)thatare
(ideally)orthogonalinsomeway.

Moreexamplesofthis
IfyouhaveGPSdisplacement/velocitydata,youcan
determineallcomponentsofthedisplacement
gradienttensor(orstrain+rota@on)
Ifyouhavelinelengthchanges,youcandetermine
allcomponentsofstrain,butnothingaboutrota@on
Sodonttrytoes@matethedisplacementgradienttensor;
es@matethestraintensorinstead

Ifyouhaveanglechanges,youcantdetermine
rota@onordilata@on
Sodonttrytoes@matethestraintensor;es@matethe
gammasinstead.

AxisofMaximumShear
Oneotherthingyoucangetoutofthe
gammasistheaxisofmaximumshear
tan2 s =

s = 12 tan1 (1 2 )

= 12 + 22

Magnitude of maximum shear

If1=0,themaximumshearisalongthexaxis.
If2=0,themaximumshearis45fromthexaxis.

(Quan@ta@ve)Rota@onofaLine
Segment
x+dx

Denetherota@onvectorofthe
[email protected]
indexnota@onitis:

+d

eijk dx j d k
i =
dx n dx n

Becausei=xi+ui,
x

k
uk
d k =
dx m = km +
dx m
x m
x m

Note:curlydandstraightdare
Define the rotation vector

dx d
=
dx d
= sin

notthesame([email protected]
deriva@ves).
Alsothemagnitudesofdxandd
areequalforapurerota@on.

Total vs. Partial Derivative

The total derivative is

df f f dx f dy
= +
+
dt t x dt y dt

(Quan@ta@ve)Rota@onofaLine
Segment
x+dx

Denetherota@onvectorofthe
[email protected]
indexnota@onitis:

+d

eijk dx j d k
i =
dx n dx n

Becausei=xi+ui,
x

k
uk
d k =
dx m = km +
dx m
x m
x m

Note:curlydandstraightdare
Define the rotation vector

dx d
=
dx d
= sin

notthesame([email protected]
deriva@ves).
Alsothemagnitudesofdxandd
areequalforapurerota@on.

Rota@onofLineSegment
Subs@tu@ngthisingives:

1
uk
i =
eijk dx j km +
dx m
dx n dx n
x m

1
uk
i =
dx m
eijk dx j km dx m +
dx n dx n
x m

1
uk
i =
dx m
eijk dx j dx k + eijk dx j
dx n dx n
x m

This term is just x x x = 0

i =

dx j dx m
0 + eijk (km + km )]
[
dx n dx n

i = eijk (km + km ) dx j dx m

This term is the velocity

These are just unit

WhatDoesThisMean?
Herestheequa@on:

i = eijk (km + km ) dx j dx m
Thisequa@ontellsusthatalinesegmentcanrotatebecause
ofeitherastrainorarota@on.

Therota@onpartisobvious
Thestrainpartwesawearlier(graphically).Dependingonthevalues
ofthestraincomponents,therota@onmightbezeroornonzero.
Also,thetwotermsmightcancelout!