Module 4

Boundary value problems in linear
elasticity

Learning Objectives
• formulate the general boundary value problem of linear elasticity in three dimensions

• understand the stress and displacement formulations as alternative solution approaches
to reduce the dimensionality of the general elasticity problem

• solve uniform states of strain and stress in three dimensions

• specialize the general problem to planar states of strain and stress

• understand the stress function formulation as a means to reduce the general problem
to a single differential equation.

• solve aerospace-relevant problems in plane strain and plane stress in cartesian and
cylindrical coordinates.

4.1 Summary of field equations
Readings: BC 3 Intro, Sadd 5.1

• Equations of equilibrium ( 3 equations, 6 unknowns ):

σji,j + fi = 0 (4.1)

• Compatibility ( 6 equations, 9 unknowns):

1  ∂ui ∂uj 
εij = + (4.2)
2 ∂xj ∂xi

77

78 MODULE 4. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY

u
∂Bu b

b

u
B
b

f ∂Bt

e3
t

e2
e1

Figure 4.1: Schematic of generic problem in linear elasticity

or alternatively the equations of strain compatibility (6 equations, 6 unknowns), see
Module 2.3
εij,kl + εkl,ij = εik,jl + εjl,ik (4.3)

• Constitutive Law (6 equations, 0 unknowns) :

σij = Cijkl εkl (4.4)
εij = Sijkl σkl (4.5)

In the specific case of linear isotropic elasticity:

σij = λεkk δij + 2µεij (4.6)
1
εij = [(1 + ν)σij − νσkk δij ] (4.7)
E

• Boundary conditions of two types:

– Traction or natural boundary conditions: For tractions ¯
t imposed on the portion
of the surface of the body ∂Bt :

ni σij = tj = t¯j (4.8)

– Displacement or essential boundary conditions: For displacements u ¯ imposed on
the portion of the surface of the body ∂Bu , this includes the supports for which
we have u¯ = 0:

ui = u¯i (4.9)

4.2. SOLUTION PROCEDURES 79

We observe that the general elasticity problem contains 15 unknown fields: displacements
(3), strains (6) and stresses (6); and 15 governing equations: equilibrium (3), pointwise
compatibility (6), and constitutive (6), in addition to suitable displacement and traction
boundary conditions. One can prove existence and uniqueness of the solution ( the fields:
ui (xj ), εij (xk ), σij (xk )) in B assuming the convexity of the strain energy function or the
positive definiteness of the stiffness tensor.
It can be shown that the system of equations has a solution (existence) which is unique
(uniqueness) providing that the bulk and shear moduli are positive:

E E
K= > 0, G = >0
3(1 − 2ν) 2(1 + ν)

which poses the following restrictions on the Poisson ratio:

−1 < ν < 0.5

4.2 Solution Procedures
The general problem of 3D elasticity is very difficult to solve analytically in general. The
first step in trying to tackle the solution of the general elasticity problem is to reduce the
system to fewer equations and unknowns by a process of elimination. Depending on the
primary unknown of the resulting equations we have the:

4.2.1 Displacement formulation
Readings: BC 3.1.1, Sadd 5.4

In this case, we try to eliminate the strains and stresses from the general problem and
seek a reduced set of equations involving only displacements as the primary unknowns. This
is useful when the displacements are specified everywhere on the boundary. The formula-
tion can be readily derived by first replacing the constitutive law, Equations (4.4) in the
equilibrium equations, Equations (4.1):

Cijkl εkl ,j
+ fi = 0

and then replacing the strains in terms of the displacements using the stress-strain relations,
Equations (4.2):

Cijkl uk,l ,j + fi = 0 (4.10)

These are the so-called Navier equations. Once the displacement field is found, the strains
follow from equation (4.2), and the stresses from equation (4.4).

Concept Question 4.2.1. Navier’s equations.

i.ji + µui.i h i (λ + µ)e. Show that in the case that the body forces are uniform or vanish the volumetric defor- mation e = εkk = uk.kj δij + µ(ui.k δij + µ(ui.80 MODULE 4.ii + µe.i (λ + µ)e.ji + µui.j + uj.2.2.ij ) + fi = λuk.ki + µ(ui. This can be slightly simplified to finally obtain: (λ + µ)uj.jji + fi.j + fi = λuk. If fi is zero or constant.k is harmonic.i = 0 and we obtain the sought result ∂ 2e e.jj + fi = 0 .ii + µui.i ).i to Navier’s equations) Solution: h i (λ + µ)uj.jj + uj. and then insert this in the equilibrium equations: 0 = σij.ii = 0.e. Concept Question 4.ii + fi. (Hint: apply the divergence operator ()i.jj + uj.ii = =0 ∂xi 2 . Harmonic volumetric deformation. its Laplacian vanishes identically: ∇2 e = e.ij ) + fi .i = 0 (λ + 2µ)e.jj + fi = 0 .jj + fi = 0 It can be observed that this is the component form of the vector equation: (λ + µ)∇ · (∇u) + µ∇2 u + f = 0. fi.i = 0 (λ + µ)e. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY Specialize the general Navier equations to the case of isotropic elasticity So- lution: The strategy is to replace the strain-displacement relations in the constitutive law for isotropic elasticity σij = λkk δij + 2µεij = λuk.i = 0.i + µui.jj + fi.

33 = 0.jj + fi = 0. u1. From the displacement field we can calculate u1 = C1 x21 x2 x3 . the Navier equations can be simplified to 2Guj. u2. C1 . u2. SOLUTION PROCEDURES 81 Concept Question 4. u1.k .2.jj + fi = 0.11 = u3.33 = 2C3 x1 x2 . C2 . f3 10Gx1 x2 E where G = 2(1+ν) and ν = 1/4. u2 = C2 x1 x22 x3 .jj + fi = 0 2(1 + ν)(1 − 2ν) 2(1 + ν) E E e.jj + fi = 0 2Ge.ji + ui.3.11 = u2.i + ui.ji + Gui.22 = 2C2 x1 x3 . which allows us to write the following system of equations      6 4 4 C1 6 4 6 4 C2  =  −2  . 2(1 + ν)(1 − 2ν) 2(1 + ν) where e = εkk = uk.i + Gui.2 = ε22 = 2C2 x1 x2 x3 . u3. Solution: By applying the Navier’s equations E E uj.3 = ε33 = 2C3 x1 x2 x3 . e = ε11 + ε22 + ε33 = 2(C1 + C2 + C3 )x1 x2 x3 . Assume displacements given by     u1 C1 x21 x2 x3 u = u2  = C2 x1 x22 x3  .4. and C3 allowing the displacement field u to be solution of the Navier equations.22 = 0. For this specific problem.1 = ε11 = 2C1 x1 x2 x3 . u1.33 = 0. Consider a problem with body forces given by     f1 −6Gx2 x3 f = f2  =  2Gx1 x3  .11 = 2C1 x2 x3 . u3 C3 x1 x2 x23 Determine the constants. Introducing the above expressions into the Navier’s equations. A solution to Navier’s equations. we obtain (6C1 + 4C2 + 4C3 − 6) Gx2 x3 = 0 (4C1 + 6C2 + 4C3 + 2) Gx1 x3 = 0 (4C1 + 4C2 + 6C3 + 10) Gx1 x2 = 0. 4 4 6 C3 −10 . u2.22 = u1. u3 = C3 x1 x2 x23 . u3. u3.2.

i.kl + εkl.11) | {z } | {z } | {z } | {z } εij εkl εik εjl Concept Question 4.3) The derivation is based on replacing the constitutive law. Beltrami-Michell’s equations.e.ik . but it is possible to find the six meaningful relationships by setting k = l.ij = εik.kl + (Sklmn σmn ).2 Stress formulation Readings: Sadd 5. To eliminate the displacements.2. instead of using the strain- displacement conditions to enforce compatibility.jl + (Silmn σmn ). which leads to the following displacement field     u1 27x21 x2 x3 1 u = u2  =  −x1 x22 x3  . C2 .ij = εik.ij = (Sikmn σmn ). Introducing the compliance expression for isotropic materials 1 εmn = [(1 + ν)σmn − νσpp δmn ] E .82 MODULE 4. 7 u3 −29x1 x2 x23 Practical solutions of Navier’s equations can be obtained for fairly complex elasticity problems via the introduction of displacement potential functions to further simplify the equations. Derive the Beltrami-Michell equations corresponding to isotropic elasticity (Hint: as a first step.ik .jl + εjl.4.3 In this case we attempt to eliminate the displacements and strains and obtain equations where the stress components are the only unknowns. it is more convenient to use the Saint Venant compatibility equations (4.kk + εkk.e. −29). i. the first step involves doing: (Sijmn σmn ).ik (4. C3 ) = 17 (27.jk + εjk.4) into these equations. and then use the equilibrium equations (4. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY The solution of this algebraic system is (C1 .1). εij. This is useful when the tractions are specified on the boundary. use the compliance form of the isotropic elasticity constitutive relations and replace them into the Saint-Venant strain compatibility equations. if you take it this far. I will explain some additional simplifications) Solution: The general expression for the stress compatibility conditions is given by εij. −1. 4. Equations (4.2.

i 1+ν 1+ν 1+ν 1+ν σii.2. the above expresion can be simplified to 1 ν σij.kk + σkk.kk − 2fi. E E and after a little algebra we can write ν σij. 1−ν 1−ν which allows us to write the Equation 4.jk − σjk.ij = 0.j − fj. .ij − νσpp.kk + σkk.jk − σjk.12 as 1 ν σij.kk = σpp.2.ik − νσpp.kk δij − 3σpp.ij − σik.ii = σpp.j − fj.ij − σpp.ij = − fk.i .4.k δij − fi. 1+ν 1+ν Using the equilibrium equations σij.kk = − fi.ij ] 1+ν ν ν σij. Consider the case of constant body forces.ik = − [σpp.k δij − fi.ik = σkk. the previous equation reduces to 1 3ν σii. Beltrami-Michell’s equations expanded.kk + σkk.kk + σii.i .ij − σik.ik δjk ] .jk δik ] + [(1 + ν)σjk.ij = σpp.ij δkk ] 1+ν ν σij.kk δij ] + [(1 + ν)σkk.kk + σkk.j + fi = 0. 1+ν 1−ν and for constant body forces they reduce to (1 + ν)σij.k .ij δkk ] = E E 1 1 [(1 + ν)σik.5.ik δjk − σpp.kk + σkk.jk − νσpp. (4.ij − σik. Expand the general Beltrami-Michell equations written in index form into the six independent scalar equations Solution: The Beltrami-Michell’s equations in index notation are written as 1 ν σij.jk δik + σpp.ij + σpp.j − fj.kk + σkk.ik = − [σpp.i 1+ν 1−ν Concept Question 4.ij = − fk. we obtain that 1 1 [(1 + ν)σij.kk δij .kk δij − σpp.i = − fk.kk − 2fi. SOLUTION PROCEDURES 83 into the compatibility expressions.kk = σii.12) 1+ν 1+ν For the case i = j.ij + σpp.i 1+ν 1+ν 1 3ν σii.jk − σjk.kk δij − fi.kk − νσpp.kk + σkk.kk + σkk.

2: Illustration of the Principle of superposition in linear elasticity .2. 2 3 Of the six non-vanishing equations obtained. As we will see in 2D applications. Once the stresses have been found. Combining these with the three equations of equilibrium provides the necessary six equations to solve for the six unknown stress components. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY ∂2 ∂ 2 I1 (1 + ν)∇2 σ11 + (σ11 + σ22 + σ33 ) = (1 + ν)∇2 σ11 + = 0 ∂x21 ∂x21 ∂2 ∂ 2 I1 (1 + ν)∇2 σ22 + (σ11 + σ22 + σ33 ) = (1 + ν)∇2 σ22 + = 0 ∂x22 ∂x22 ∂2 ∂ 2 I1 (1 + ν)∇2 σ33 + (σ11 + σ22 + σ33 ) = (1 + ν)∇2 σ33 + = 0 ∂x23 ∂x23 2 ∂2 2 ∂ 2 I1 (1 + ν)∇ σ12 + (σ11 + σ22 + σ33 ) = (1 + ν)∇ σ12 + = 0 ∂x1 ∂x2 ∂x1 ∂x2 ∂2 ∂ 2 I1 (1 + ν)∇2 σ13 + (σ11 + σ22 + σ33 ) = (1 + ν)∇2 σ13 + = 0 ∂x1 ∂x3 ∂x1 ∂x3 ∂2 ∂ 2 I1 (1 + ν)∇2 σ23 + (σ11 + σ22 + σ33 ) = (1 + ν)∇2 σ23 + = 0.5 The principle of superposition is a very useful tool in engineering problems described by linear equations. one can use the constitutive law to determine the strains. 4. the principle states that we can linearly combine known solutions of elasticity problems (corresponding to the same geometry). the Beltrami-Michell equations are still very difficult to solve. In simple words. ∂x2 ∂x3 ∂x2 ∂x3 ∂2 ∂ 2 ∂ 2 where ∇2 = ∂x21 + ∂x 2 + ∂x2 and I1 = σ11 + σ22 + σ33 is the first invariant of the stress tensor. only three represent independent equations (just as with Saint-Venant strain compatibility equations). and the strain-displacement relations to compute the displacements.84 MODULE 4. = α1 +α2 α 1 S1 + α 2 S2 S1 S2 Figure 4.3 Principle of superposition Readings: Sadd 5. see Figure 4. We will introduce the concept of stress functions to further reduce the equations.

4.g. Figure 4.5 Solution methods Readings: Sadd 5. and develop analytically tractable models to analyze the structure of interest. SAINT VENANT’S PRINCIPLE 85 4.7 .4 Saint Venant’s Principle Readings: Sadd 5. One can after perform a detailed analysis of the elastic field surround- ing the point of application of the load (e.3 provides an illustration of the idea. as it allows the possibility to idealize and simplify loading conditions when the details are complex or missing. 4. think of a riveted joint in an aluminum frame structure).3: Illustration of Saint Venant’s Principle tural mechanics. strain. it has been demonstrated analytically in a convincing manner.6 Saint Venant’s Principle states the following: The elastic fields (stres. Although the principle was stated in a rather intuitive way by Saint Venant. This principle is extremely useful in struc- P P P P P P 2 2 3 3 3 Figure 4. displacement) resulting from two different but statically equivalent loading conditions are approximately the same everywhere except in the vicinity of the point of application of the load.4.

2. we can assume that in each cross-section we have uniform tension. Solution: The bar is free of stresses at x3 = 0. Solution: For a given area A = bc. Consider a prismatic bar hanging under its own weight. Problems involving simple geometries and loading conditions which result in stress fields with uniform or linear spatial distributions can be attacked by simple calculus methods.5. From your understanding of the physical situation. and thus the stress is given by σ33 = ρgx3 .86 MODULE 4. which leads to σ11 = σ22 = σ12 = σ13 = σ23 = 0. 3. b A L e3 e2 Figure 4.e. make assumptions about the state of stress to simplify the differential equations of stresss equilibrium. if the cross-sectional dimensions of the bar are far smaller than its length. b  L and c  L. Example of Direct Integration Methods: Prismatic bar hanging vertically under its own weight.1. The body forces for this problem are f1 = f2 = 0. 1. i. which after integration becomes Z σ33 (x3 ) = ρgdx3 = ρgx3 + C. Concept Question 4.4: Prismatic bar hanging under its own weight.e. σ33 = f (x3 ) 6= 0. Figure 4. The bar has a cross section area A and length L. . Apply the relevant boundary conditions of the problem and obtain the resulting stress field (i. where ρ is the material mass density and g is the acceleration of gravity. we obtain σ33.5. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY 4. Integrate the resulting equation(s) in closed form Solution: After introducing the previous stresses into the equilibrium equations.3 = −f3 = ρg.4. then we can impose that σ33 (x3 ) = 0. It enables us to determine that C = 0. σij (xk )). f3 = −ρg.1 Direct Integration The general idea is to try to integrate the system of partial differential equations analytically.

2 = − .1 (x1 . Use the constitutive equations to compute the strain components.3 = .1 (x1 . By replacing Eqs.2 (x1 . we obtain ρg 2 u3 = x + g1 (x1 .13 and 4. (4. (4.19) 2E 3 where g1 (x1 . x2 ). SOLUTION METHODS 87 4.1 (x1 . x2 ) − x3 = g3.14) E ρgx3 ε33 = u3. x2 )x3 + g2 (x1 .20) u2 = −g1.15) E 1 ε23 = (u2. (4. (4.16 we know that u1.18) 2 5. E . x2 ).5.20 and 4. x2 ) − x3 = g2.1 = − . 4.16) 2 1 ε13 = (u1. (4. x2 ) = − . Solution: After the integration of Eq.3 + u3. x2 ).22 (x1 .2 (x1 . x2 ).4.1 and u2.17 and 4. x2 ) E  νρg  g1.19 leads to u1.14. (4.21 into Eqs.15. x2 ) is an arbitrary function arising from the integration. x2 )x3 + g2. 4. Solution: From Hooke’s law. x2 ) are arbitrary functions coming from the integration.3 = −g1.3 + u3.17) 2 1 ε12 = (u1.2 ) = 0.3 = −u3.21) where g2 (x1 . we find that νρgx3 −g1. x2 )x3 + g3 (x1 .1 (x1 . x2 ). x2 ) = − E νρgx3 −g1. x2 ) and g3 (x1 . we directly obtain: νρgx3 ε11 = u1.2 (x1 . u2.2 . which in combination with eq.1 ) = 0. E which can be rearranged as  νρg  g1.3 = −u3. x2 )x3 + g3. (4. Integrate the strain-displacement relations and apply boundary conditions to obtain the displacement field.2 + u2. (4.13) E νρgx3 ε22 = u2. x2 ).11 (x1 .3 = −g1. (4.1 ) = 0. 4.2 (x1 . 4.11 (x1 . From Eqs.22 (x1 . After integration we obtain u1 = −g1. 4.

.26) 2E 1 νρg 2 g1 (x1 .30) g2. x2 ) = 0. m2 (x1 ).12 (x1 .26. x2 ) = m3 (x2 ) + A3 . From Eqs. the expressions in parentheses and the right hand side must vanish (note that they only depend on x1 and x2 ). x2 ) = m1. x2 )x3 + g2.2 (x1 .1 (x1 .22 (x1 . (4. 4. x2 ) + g3.27 and 4. and considering that this expression must hold for any value of x3 . x2 ) = . (4. x2 ) = x + m1 (x2 )x1 + A1 . 4.32) g3 (x1 . A2 .11 (x1 . x2 ) = m4 (x1 ) + A4 . 4.31) By replacing Eqs.25) After integration we obtain νρg 2 g1 (x1 .1 (x1 . (4. (4.24) g3. This result leads to g2 (x1 .31.28 and 4.18.2 (x1 . x2 ) = 0. 4. whose only possible solution is m3 (x2 ) = A5 x2 . while A1 .2 (x2 ) = m2.27) 2E 2 g2 (x1 .21 we also know that u1. x2 ) = x + m2 (x1 )x2 + A2 . m4 (x1 ) are arbitrary functions.28) g3 (x1 .29 into Eq. x2 ) = 0. x2 ) + g3.88 MODULE 4. (4.33) From Eqs. m3 (x2 ). (4.23) E g2.2 + u2.30.29) where m1 (x2 ). and m4 (x1 ) = −A5 x1 . (4. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY As these equations must hold for any value of x3 .1 (x1 . which leads to νρg g1. x2 ) = 0. x2 ) = . we can derive g1. x2 ) = −A5 x1 + A4 .1 (x1 ) = 0.2 (x1 . (4.12 (x1 .12 (x1 .22) E νρg g1.1 = −2g1.20 and 4. x2 ) = A5 x2 + A3 .1 (x1 ) = 0 .2 (x2 ) + m4. 4. (4. (4. (4. A3 and A4 are arbitrary constants. x2 ) = 0. (4. 4. we can write m3. the following equa- tions must satisfy g1.

4. L) = u3 (0. E νρg u2 = − x2 x3 .  u3 = − 2E which reduces to utip ρg 2 3 = − 2E L at the centerline (x1 = x2 = 0). 0. Solution: The axial displacement at the tip (x3 = 0) of the prismatic bar is ρg  2 L − ν(x21 + x22 ) . E ρg  2 x3 − L2 + ν x21 + x22 . where A6 and A7 are arbitrary constants. 4. With the help of superposition. 0. It means that the one-dimensional analysis shows the same behavior that the centerline of the three-dimensional approach. 4.20.19.33 and 4. 4.  g1 (x1 . 0. we can write the following expression for g1 (x1 . E νρg u2 = − x2 x3 − A5 x1 − A7 x3 + A4 .34 allow us to write the general solution for the displacement field νρg u1 = − x1 x3 + A 5 x2 − A 6 x3 + A 3 . 4. . When the finite dimension of the cross section is considered. which allows us to compute the displacement as Z L Z L ρg L ρgL2 Z tip σ u = εdx = dx = − xdx = − .34) 2E The Eqs 4. L) = ω3 (0.  u3 = x3 + 2E 2E The 6 constants are determined by imposing 3 displacements (u1 (0.32. the vertical displacement away from the centerline is reduced by a factor proportional to Poisson’s ratio and the square of the distance from the centerline. x2 ) νρg 2 x1 + x22 + A6 x1 + A7 x2 + A8 . 0. 0. SOLUTION METHODS 89 and hence m1 (x2 ) = A6 and m2 (x1 ) = A7 . L) = ω2 (0. E ρg 2 νρg 2 x1 + x22 + A6 x1 + A7 x2 + A8 . L) = 0) and 3 rotations (ω1 (0. x2 ) = (4. 0.5. L) = 0) νρg u1 = − x1 x3 . and then can be calculated as σ = P/A = −ρgx. 0 0 E E 0 2E The displacements utip and utip 3 are equal. For the one-dimensional analysis we can write the weight of the bar as P = −ρAgx.21. The stress-strain relation satisfies σ = Eε.  u3 = 2E 6. The stress is uniform in each cross section. L) = u2 (0. Compute the axial displacement at the tip of the bar and compare with the solution obtained with the one-dimensional analysis.

but the idea is to adopt a “general form” of the variation of the assumed field. sometimes informed by some assumption about the physical response of the system. the stress satisfies σ = P/A = −ρgL. etc. We will consider two of the basic 2-D elasticity problem types: . Laplace.5. Compute the axial displacement at the tip of the bar and compare with the solution for a weightless bar subject to a point load P = −ρgAL.6 Two-dimensional Problems in Elasticity Readings: Sadd. 4. plates and shells). with unknown constants or functions of reduced dimension- ality. Chapter 7 In many cases of practical interest. It is usually difficult to use this approach to obtain solutions to problems of specific practical interest. After replacing the assumed functional form into the field equations the system can be reduced in number of unknowns and dimensionality. complex variables. beams. More complex situations require advanced analytical techniques and mathematical meth- ods including: power and Fourier series. integral transforms (Fourier. and theory of structural elements (trusses. This is the approach that leads to the very important theories of torsion of prismatic bars.90 MODULE 4.2 Inverse Method In this method. Since the stress-strain relation is given by σ = Eε.5. 0 0 E E 0 E We observe that utip is the double of utip 3 . Solution: The axial displacement at the tip (x3 = 0) and centerline (x1 = x2 = 0) of the prismatic bar is ρg 2 utip 3 = − L. the three-dimensional problem can be reduced to two dimensions. 2E When the bar is subject to a point load P = −ρgAL. one selects a particular displacement or stress field distribution which satisfy the field equations a priori and then investigates what physical situation they correspond to in terms of geometry. we can calculate the dis- placement as Z L Z L ρgL L ρgL2 Z tip σ u = εdx = dx = − dx = − . 4.3 Semi-inverse Method This is similar in concept to the inverse method. 4. We will discuss these theories in detail later in this class. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY 7. The reduced system can typically be integrated explicitly by the direct or analytical methods. Hankel). boundary conditions and body forces.

x2 .1 Plane Stress Readings: Sadd 7. z y σ21 σ22 a σ11 h x b x3 σ33 = 0 σ13 = 0 x2 h Figure 4. Applies to situations of in-plane stretching and shearing of thin slabs: where h  a.e. TWO-DIMENSIONAL PROBLEMS IN ELASTICITY 91 4. x3 ) can be found from the constitutive law and the strain-displacement relations. σ22 = σ22 (x1 . • the only present stress components are: σ11 . b. . and ∂ • the stresses are constant through the thickness. x2 ). i. The remaining (non- zero) unknowns ε33 (x1 . x2 .6. i. there are no out-of-plane stress components. x2 ) The problem can be reduced to 8 coupled partial differential equations with 8 unknowns which are a function of the position in the plane of the slab x1 .5. loads act in the plane of the slab and do no vary in the normal direction.6.e. ∂x3 = 0. (Why is w a function of x3 ?).5: Schematic of a Plane Stress Problem: The σi3 components on the free surfaces are zero (there are no external loads on the surface) We can then assume that: • σ33 = σ32 = σ31 = 0.2 As the name indicates. which implies: σ11 = σ11 (x1 . Figure 4. the only stress components are in a plane. x2 ) and σ12 = σ12 (x1 .4. σ22 and σ12 . x2 ) and w(x1 .

Strain-displacements relations reduce to: ∂u1 ∂u2 ∂u3 11 = . ∂x1 ∂x2 Constitutive equations reduce to: 1  11 = σ11 − νσ22 E 1  22 = σ22 − νσ11 E ν  ν 33 = − σ11 + σ22 = − (11 + 22 ) E 1−ν 1 (1 + ν) 12 = σ12 = σ12 2G E 23 = 0 13 = 0. ∂x1 ∂x2 . 2(1 + ν) 2(1 − ν) ∂x2 ∂x1 ∂x2 Beltrami-Michell equations (start from the Saint-Venant compatibility condition for 2D problems ε11. x2 ) = 0 ∂x1 ∂x2 ∂σ12 ∂σ22 + + f2 (x1 .6. x2 ) = 0. Solution: Assumption: σi3 = 0 ⇒ σ13 = σ23 = σ33 = 0. Equilibrium equations reduce to: ∂σ11 ∂σ12 + + f1 (x1 .11 = 2ε12. 33 = ∂x1 ∂x2 ∂x3 ∂u1 ∂u2 212 = + . BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY Concept Question 4.1. Do the same for the Navier and Beltrami-Michell equations. Obtain the full set of governing equations for plane stress by specializing the general 3-D elasticity problem based on the plane-stress assumptions.22 + ε22. Governing equations for plane stress.j + fi = 0 and introduce the strain-stress relationship):   E 2 E ∂ ∂u1 ∂u2 ∇ u1 + + + f1 = 0 2(1 + ν) 2(1 − ν) ∂x1 ∂x1 ∂x2   E 2 E ∂ ∂u1 ∂u2 ∇ u2 + + + f2 = 0.92 MODULE 4.12 and introduce the strain-stress relationship):   2 ∂f1 ∂f2 ∇ (σ11 + σ22 ) = −(1 + ν) + . ∂x2 ∂x1 Navier equations (start from the equilibrium equation σij. 22 = .

Governing equations for plane strain. In this case. We can then assume that: ∂(. We therefore need to only analyze the 2-D solution in a generic slice.6. as shown in the figure.4. ∂x1 ∂x2 Constitutive equations reduce to: 1+ν ε11 = [(1 − ν)σ11 − νσ22 ] E 1+ν ε22 = [(1 − ν)σ22 − νσ11 ] E νE ε33 = 0 ⇒ σ33 = ν (σ11 + σ22 ) = (ε11 + ε22 ) (1 + ν)(1 − 2ν) 1 (1 + ν) ε12 = σ12 = σ12 2G E ε23 = 0 ε23 = 0. Do the same for the Navier and Beltrami-Michell equations. the same stresses and strains must be experienced by any slice along the x3 axis. there cannot be any displacements or body forces in that direction.e.2 Plane Strain Readings: Sadd 7. Concept Question 4. x2 ) = 0. TWO-DIMENSIONAL PROBLEMS IN ELASTICITY 93 4.) u3 = 0. Solution: Assumption: εi3 = 0 ⇒ ε13 = ε23 = ε33 = 0. By symmetry with respect to the x3 axis.6. Equilibrium equations reduce to: ∂σ11 ∂σ12 + + f1 (x1 .1 Consider elasticity problems such as the one in Figure 4. x2 ) = 0 ∂x1 ∂x2 ∂σ12 ∂σ22 + + f2 (x1 . ε12 . Obtain the full set of governing equations for plane strain by specializing the general 3-D elasticity problem based on the plane-strain assumptions. and the only strain components are those “in” the plane: ε11 . . . ε22 .6.2. which are only a function of x1 and x2 .6. the loads applied are parallel to the x1 x2 plane and do not change along x3 . It is clear that the solution cannot vary along x3 . L is much larger than the transverse dimensions of the structure. i. =0 (4.35) ∂x3 From this we can conclude that: ε33 = ε13 = ε23 = 0.

BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY x2 . x3 x1 x2 x3 x1 Figure 4.94 MODULE 4.6: Schematic of a typical situation where plane-strain conditions apply and there is only need to analyze the solution for a generic slice with constrained displacements normal to it ..

results in two significant simplifications in our governing equations: Concept Question 4.5 A very useful technique in solving plane stress and plane strain problems is to introduce a scalar stress function φ(x1 .1. 2(1 + ν) 2(1 + ν)(1 − 2ν) ∂x2 ∂x1 ∂x2 Beltrami-Michell equations (start from the Saint-Venant compatibility condition for 2D problems ε11. Solution: For a system free of body forces. The specific dependence to be considered is: ∂ 2φ σ11 = (4.7 Airy stress function Readings: Sadd 7.38) ∂x1 ∂x2 This choice. AIRY STRESS FUNCTION 95 Strain-displacements relations reduce to: ∂u1 ∂u2 11 = .11 = 2ε12.4.j + fi = 0 and introduce the strain-stress relationship):   E 2 E ∂ ∂u1 ∂u2 ∇ u1 + + + f1 = 0 2(1 + ν) 2(1 + ν)(1 − 2ν) ∂x1 ∂x1 ∂x2   E 2 E ∂ ∂u1 ∂u2 ∇ u2 + + + f2 = 0.36) ∂x2 2 ∂ 2φ σ22 = (4. 22 = . x2 ) such that all the relevant unknown stress components are fully determined from this single scalar function.22 + ε22.12 and introduce the strain-stress relationship):   2 1 ∂f1 ∂f2 ∇ (σ11 + σ22 ) = − + .37) ∂x1 2 ∂ 2φ σ12 = − (4. the equilibrium equations reduce to σij. ∂x2 ∂x1 Navier equations (start from the equilibrium equation σij. 1 − ν ∂x1 ∂x2 4.7.j = 0. Show that the particular choice of stress function given automatically satisfies the stress equilibrium equations in 2D for the case of no body forces.7. although apparently arbitrary. 33 = 0 ∂x1 ∂x2 ∂u1 ∂u2 212 = + . . Airy stress function and equilibrium conditions.

2.2222 = 0 (4. and introducing the compliance form of the plane stress approach 1 11 = (σ11 − νσ22 ) E 1 22 = (σ22 − νσ11 ) E 1+ν 12 = σ12 . .12 .39) This can also be written using the ∇ operator (see Appendix ??) as:  2 4  ∂2 ∂2  ∇ φ=  +  φ (4.12 . (4. it is straightforward to verify that σ11.12 .2 = −φ.22 . we can rewrite the Equation 4.22 + σ22.1111 + 2φ.7.1122 + φ. Concept Question 4.2211 = −2(1 + ν)φ.1 + σ22. Obtain a scalar partial differential equation for 2D elasticity problems with no body forces whose only unknown is the stress function (Hint: replace the stresses in the Beltrami- Michell equations for 2D (plane strain or stress) with no body forces in terms of the Airy stress function.11 = 2ε12. σ12 = −φ.11 − νσ11.22 − νσ22.2 = φ. σ12 = −φ. σ22 = φ.1212 .41) Considering that σ11 = φ. The final result should be: φ.122 = 0 σ21. Two-dimensional biharmonic PDE.221 − φ.11 .1122 + φ.96 MODULE 4.22 + ε22.1111 − νφ.2222 − νφ.41 as φ.22 .11 .121 + φ.40) ∂x1 2 ∂x2 2  | {z } ∇2 Solution: Starting from the compatibility conditions for 2D problems ε11.1 + σ12.112 = 0. σ22 = φ.11 = 2(1 + ν)σ12.12 . E we arrive to σ11. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY By introducing σ11 = φ.

7. 4.1 A number of useful solutions to problems in rectangular domains can be obtained by adopting stress functions with polynomial distribution. . Consider the stress function φ = Ax32 .1 Problems in Cartesian Coordinates Readings: Sadd 8. σ12 (x1 . and is also expressed as ∇4 φ = ∇2 ∇2 φ = 0. Concept Question 4. 1.7: Pure bending of a beam.7. This last PDE is so-called biharmonic equation. e2 2h e1 M M 2L Figure 4.1212 + φ. Verify the stress-free boundary condition at x2 = ±h. NB: we arrive to the same equation using the compliance form of the plane strain ap- proach. ∂x1 ∂x2 It is straightforward to verify that at x2 = ±h the shear and normal stresses σ12 and σ22 vanish. Show that this stress function corresponds to a state of pure bending of a beam of height 2h and length 2L subject to a bending moment M as shown in the Figure 4. i.3. AIRY STRESS FUNCTION 97 which reduces to φ. σ22 (x1 . NB: these boundary conditions are satisfied exactly pointwise.7.e.2222 = 0. x2 = ±h) = 0.7. Pure bending of a beam.1111 + 2φ.4. x2 = ±h) = 0. Solution: The Airy stress function enables us to determine the stress components by applying the relations ∂ 2φ σ11 = = 6Ax2 ∂x22 ∂ 2φ σ22 = = 0 ∂x21 ∂ 2φ σ12 = − = 0.

Establish a relation between M and the stress distribution at the beam ends in integral form to fully define the stress field in terms of problem parameters. x2 ) = − x2 x1 − C 1 x2 + C 2 . 2 ∂x2 ∂x1 2Eh dx1 dx2 where g(x1 ) and f (x2 ) are arbitrary functions of integration. 2Eh3 4Eh3 4Eh3 1 . The last equation can be separated into two independent relations in x1 and x2 3M dg(x1 ) df (x2 ) − 3 x1 + = C1 . we obtain that 3M ∂u1 3M ε11 = − x2 = ⇒ u1 (x1 . u2 (x1 .98 MODULE 4. x2 )x2 dx2 = −M. Integrate the strain-displacement relations to obtain the displacement field. x2 ) = − x2 x1 + f (x2 ) 2Eh3 ∂x1 2Eh3 3M ∂u2 3M 2 ε22 = ν x2 = ⇒ u2 (x1 . σ11 (x1 ± L. x2 ) = ν x2 + x + C 1 x + C3 . x2 )x2 dx2 = −M = 4Ah3 ⇒ A = − 3 . and 3M 2 g(x1 ) = x + C1 x + C3 . 4Eh3 1 By replacing these functions in the expressions for displacement. x2 ) = ν x + g(x1 ) 2Eh3 ∂x2 4Eh3 2   1 ∂u1 ∂u2 3M dg(x1 ) df (x2 ) ε12 = 0 = + ⇒ − 3 x1 + + = 0. What does the undetermined function represent? Solution: By applying the strain-displacement conditions and the strain-stress relationships for plain stress. Solution: The following integral conditions must also be satisfied Z h Z h σ11 (x1 ± L. we obtain 3M 3M 2 3M 2 u1 (x1 . = −C1 . −h −h This last equation allows us to calculate the constant A in function of the parameters M and h: Z h M σ11 (x1 ± L. 2Eh dx1 dx2 which leads to f (x2 ) = −C1 x2 + C2 . 2h3 3. x2 )dx2 = 0. −h 4h and then the stress field becomes 3M σ12 = σ22 = 0. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY 2. σ11 = − x2 .

AIRY STRESS FUNCTION 99 The integration constants C1 . 0) = 0. x2 ) = (x + νx22 − L2 ). in one case. σθθ . and integrals where needed. x2 → r. σ12 → σrr .8 shows the components of the stress tensor in cartesian coordinates on an area element dx1 . C2 and C3 can be determined by imposing the conditions u1 (0. uθ ε11 . θ u1 . x2 ) = 0 and u2 (±L. 4. dx2 in cartesian coordinates at a point x1 .3 We now turn to the very important family of problems that can be represented in polar or cylindrical coordinates. .7.4. an area element dx1 . This requires the following steps: Coordinate transformations from cartesian to polar: this includes the transformation of derivatives (see Appendix ??). σ22 . 0) = 0 ⇒ C3 = − . 4Eh3 and allows us to write the displacement field as 3M u1 (x1 .2 Problems in Polar Coordinates Readings: Sadd 7. u2 → ur . ε22 . 8. x2 where displacement and stress components are represented (use arrows as necessary). εθθ . u2 (±L. x2 ) = 0 ⇒ C1 = C2 = 0. Polar and cartesian coordinates. dx2 .6. Field variable transformations from a cartesian to a polar description: x1 .7. Draw schematics of a planar domain with a representation of cartesian and polar coor- dinates with the same origin and 1. which fixes the rigid-body motions 3M L2 u1 (0. Solution: The Figure 4. x2 ) = − x1 x2 2Eh3 3M 2 u2 (x1 . σrθ Concept Question 4. εrθ σ11 . ε12 → εrr . 4Eh3 1 The book provides a general solution method using polynomial series and several more examples.4.7.

θ which represents the same location as before where polar displacement and stress components are represented (use arrows as necessary). e2 σθθ σrr σθr σrθ dr b rdθ (r. as the coordinate transformation at any point is a rotation from one to another orthogonal coordinate systems. . Transformation of the governing equations to polar coordinates The book provides a general and comprehensive derivation of the relevant expressions. in the other case. an area element dr. Here. Solution: The Figure 4. θ) e1 Figure 4. e2 σ22 σ21 dx2 σ12 σ11 b (x1 . x2 ) dx1 e1 Figure 4. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY 2. The derivations of the strain-displacement relations and stress equilibrium equations in polar coordinates are given in the Appendix ??. The constitutive laws do not change in polar coordinates if the material is isotropic.8: Stress components on an area element cartesian coordinates. rdθ. rdθ in polar coordinates at a point r. we derive the additional expressions needed.100 MODULE 4.9: Stress components on an area in polar coordinates.9 shows the components of the stress tensor in polar coordinates on an area element dr.

45) ∂y ∂r ∂y ∂θ ∂y (4.49) ∂x ∂x x 1 + xy x2 x2 + y 2 r    ∂θ 1 1 x cos θ = 2 = 2 = (4.r + 2 sin θ (cos θφ.θ − φ.xxyy + φ.θ + sin θφ.rθ + 2 φ.  2 ∂2  2 ∂ φ ∂ 2φ  4 2 2 ∂ ∇ φ=∇ ∇ φ= + + (4.7.47) ∂x ∂x 2 x2 + y 2 r ∂r y = = sin θ (4. i.53) r r .4.46) The underlined factors are the components of the Jacobian of the transformation between coordinate systems: ∂r ∂ p 2 1 1 x = x + y2 = p 2 x = = cos θ (4.42) ∂x2 ∂y 2 ∂x2 ∂y 2 = φ.48) ∂y r   ∂θ ∂  −1  y  1 −y −y sin θ = tan = 2 = = − (4. y) ∂φ(r. θ) ∂r ∂φ(r.51) ∂x ∂r ∂θ r   ∂φ ∂φ ∂φ cos θ = sin θ + (4. φ = φ(r. θ. θ) upon a transformation to polar coordinates: −1 y p   2 2 r = x + y . θ = tan x How to replace first cartesian derivatives with expressions in polar coordinates: start by using the chain rule to relate the partial derivatives of φ in their cartesian and polar descriptions: ∂φ(x. θ) and its corresponding ∇4 φ(r.44) ∂x ∂r ∂x ∂θ ∂x ∂φ(x.52) ∂y ∂r ∂θ r How to replace second cartesian derivatives with expressions in polar coordi- nates: ∂ 2φ    ∂() sin θ ∂() ∂φ sin θ ∂φ = cos θ − cos θ − = ∂x2 ∂r r ∂θ ∂r r ∂θ sin θ cos θ sin θ cos θ sin θ cos θ φ.xxxx + 2φ.50) ∂y 1+ y x r r x   ∂φ ∂φ ∂φ − sin θ = cos θ + (4.43) We seek to express φ as a function of r. θ) ∂r ∂φ(r. θ) ∂θ = + (4.e. y). y) ∂φ(r.rθ + r r r sin2 θ 1 φ.θθ ) (4.rr cos2 θ − φ.yyyy (4. θ) ∂θ = + (4. AIRY STRESS FUNCTION 101 Biharmonic operator in polar coordinates In cartesian coordinates: φ = φ(x.

θθ (4.xx + φ.102 MODULE 4.r + 2 ().58) r r r r In general.θ (4.r + 2 φ.r + 2 ().xx + (). σθθ .rr + φ.rr + φ.r + 2 φ.rr + ().rθ + (2 − 2) φ.θθ − φ.rr sin2 θ + φ. so that: .θθ (4.57) r r This allows us to write the biharmonic operator as:    4 2 2 1 1 1 1 ∇ φ = ∇ (∇ φ) = (). σrθ can be obtained by notic- ing that any point can be considered as the origin of the x-axis. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY ∂ 2φ    ∂() ∂() cos θ ∂φ ∂φ cos θ = sin θ + sin θ + = ∂y 2 ∂r ∂θ r ∂r ∂θ r sin θ cos θ sin θ cos θ sin θ cos θ φ.θ + | {z } | {z } r | {z } r2 =1 =0 =0 =1 =1 z }| { z }| { 2 2 2 2 (cos θ + sin θ) (cos θ + sin θ) φ.56) r r In general: 1 1 ∇2 () = ().55) r r2 The Laplacian can then be written as: 1 1 ∇2 φ = φ. Expressions for the polar stress components: σrr .yy = (cos2 θ + sin2 θ) φ.θθ (4.θθ φ.rr + (−2 + 2) φ.54) r r r2 Obtain an expression for the Laplacian in polar coordinates by adding up the second derivatives: sin θ cos θ sin θ cos θ φ.rr + ().θ + φ.rθ + r r r cos2 θ cos2 θ sin θ cos θ φ.xx + φ.yy = ().θθ (4.r + 2 φ.rθ − 2 φ.r + φ. it is not necessary to expand this expression.yy = φ.

∂ 2 φ .

.

1 ∂φ 1 ∂ 2φ 1 1 σrr = σxx .

= 2.

59) .r + 2 φ. = + 2 2 → σrr = φ.θθ (4.

θ=0 ∂y θ=0 r ∂r r ∂θ r r .

∂ 2 φ .

.

∂ 2φ σθθ = σyy .

= = → σθθ = φ.rr (4.60) .

∂x 2 θ=0 ∂r 2 .

θ=0 .

.

∂2φ .

To obtain σrθ = σxy .

= − ∂x∂y .

. we need to evaluate φ.xy in polar coordinates: .

θ − (φ.θ sin θ) r r r r (4.rr sin θ cos θ+ φ.rθ − φ.r cos θ)− 2 (φ. θ=0 θ=0      ∂ ∂φ ∂() sin θ ∂() ∂φ ∂φ cos θ = cos θ − sin θ + ∂x ∂y ∂r r ∂θ ∂r ∂θ r 2   cos θ 1 sin θ sin θ = φ.θθ cos θ−φ.61) .rθ sin θ+φ.

3 Axisymmetric stress distribution In this case.4. Then the second equilibrium equation is identically 0.63) r r 1 2σrθ σθθ.rr = 2 φ.65) dr r .rθ 2 − 3 φ. (4.r + σrθ. σrθ = 0 by symmetry (i.r r r r r   1 1 1 1 1 1 σrθ.r + 3 φ.r = φ.rθ 2 +φ.rr r r r r r r r r Adding the the left and right hand sides we note that all terms on the right with like colors cancel out and the first equilibrium equation in polar coordinates is satisfied.e.θ 3 r r r we note that all terms on the right with like colors cancel out and the second equilibrium equation in polar coordinates is satisfied.rr ).θ = −φ.7.7. 4. In addition. Applying the same procedure to the second equilibrium equation in polar coordinates.θ = φ. there cannot be any dependence of the field variables on θ and all derivatives with respect to θ should vanish.r r r r r 2σrθ 2 2 = −φ.62) r r Verify differential equations of stress equilibrium The differential equations of stress equilibrium in polar coordinates are (see Appendix ??): 1 σrr − σθθ σrr.θ 2 = −φ.r + φ. because of the independence on θ σrθ could only be constant.r + =0 (4.64) r r Verification:   1 1 1 1 2 1 σrr. but that would violate equilibrium).θ 2 = −φ.θθ = − 2 φ.θ 2 (4.θθ − φ.rθ + φ. We now consider problems with symmetry with respect to the z-axis.rθ + φ.r = −φ.r + 2 φ.θ + σrθ.θθ 3 r r r r .rrθ r r  r  1 1 1 1 1 2 σrθ.rθ + φ. AIRY STRESS FUNCTION 103 1 1 σrθ = −φ.rθ 2 +φ. The first equation becomes (body forces ignored): dσrr σrr − σθθ + =0 (4.θθ + 2 φ.rθθ 2 +φ.r + 2 φ.64): 1 1 1 σθθ = (φ.θθr r r .rrθ +φ.θθ − φ.θ + =0 (4.rr − 3 φ.θ r r .θ r r   σrr − σθθ 1 1 1 1 1 1 1 = φ.

φ000 = r3 φ. t0 (r) = 1r = e−t φ0 (r) = φ0 (t)t0 (r) = φ0 (t)e−t . then: φ = C1 + C2 t + C3 e2t + C4 te2t Replace t = log r φ(r) = C1 + C2 log r + C3 r2 + C4 r2 log r (4.66) r r r We obtain a single ordinary differential equation for φ(r) with variable coefficients. We will now con- sider examples of application of this general solution to specific problems. Then: (r4 − 4r3 + 4r2 )ert = 0 ⇐⇒ r2 (r2 − 4r + 4) = 0 (4.rr = − 2 + 2C3 + C4 (2 log r + 3) (4. .70) r r C2 σθθ = φ.104 MODULE 4. ()0 (r) = ()0 (t)e−t 0 φ00 (r) = φ0 (t)e−t (t)e−t = e−2t (−φ0 (t) + φ00 (t)) 0 φ000 (r) = e−2t (−φ0 (t) + φ00 (t)) (t)e−t = e−3t (2φ0 (t) − 3φ00 (t) + φ000 (t))  φIV (r) = e−3t −6φ0 (t) + 11φ00 (t) − 6φ000 (t) + φIV (t)  Multiply equation (4. The gen- eral solution may be obtained by first reducing the equation to one with constant coefficients by making the substitution r = et . This involves applying the particular boundary conditions of the case under consideration in terms of applied boundary loadings at specific locations (radii) including load free boundaries. r = 2 (both double roots). t = log r. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY The biharmonic equation becomes: d2   2  1 d d φ 1 dφ + + =0 dr2 r dr dr2 r dr     IV 1 0 1 00 1 000 1 1 0 1 00 φ + − 2φ − φ + φ + − 2φ + φ = 0 r r r r r r 2 1 1 1 1 1 1 φIV + 3 φ0 − 2 φ00 − 2 φ00 + φ000 + φ000 − 3 φ0 + 2 φ00 = 0 r r r r r r r 2 1 1 φIV + φ000 − 2 φ00 + 3 φ0 = 0 (4.68) The roots of this equation are: r = 0. φ00 = r2 φ.r = 2 + 2C3 + C4 (2 log r + 1) (4. φIV = r4 φ.71) r This constitutes the most general stress field for axisymmetric problems.67) To solve this.69) The stresses follow from: 1 C2 σrr = φ. assume φ = ert . r0 (t) = et . φ0 = rert = rφ.66) by r4 and replace these results to obtain:: r4 φIV + 2r3 φ000 − r2 φ00 + rφ = φIV (t) − 4φ000 (t) + 4φ00 (t) = 0 (4.

r = 0 is part of the domain and according to the general solution (4. Solution: The centripetal acceleration −ω 2 r acts in the radial direction and appears in the stress equilibrium equation in the radial direction as an inertia term ∂σrr σrr − σθθ + = ρar = −ρω 2 r.10: Cut-away of a Rolls-Royce RB211-535E4B triple-shaft turbofan used in B-757’s (left) and detail of intermediate pressure compressor stages (right) B-757’s have a diameter D = 0. Young’s modulus E = 500GP a. Estimate the minimum clearance between the blade tips and the encasing required to prevent contact. Solution: If there is no hole.7.5. the only possible solution (if there’s no body forces) is a constant state σrr = σθθ = 2C3 . Compute the maximum turbine rotation velocity at which the material first yields plastically (ignore the effect of the blades). A counterexample (well.4. Concept Question 4.7m and are made of a metallic alloy with mass density ρ = 6500kg · m−3 . Use the general stress solution to show that the only possible solution (assuming there are no body forces) is a state of uniform state σθθ = σrr = constant. In this case. AIRY STRESS FUNCTION 105 Concept Question 4.3 and yield stress σ0 = 600M P a. Poisson ratio ν = 0. The boring case: Consider the case where there is no hole at the origin.70) the stresses go to ∞ at that location unless C2 = C4 = 0. not really): no hole but body forces The compressor disks in a Rolls-Royce RB211-535E4B triple-shaft turbofan used in Figure 4. 1. 2. or: ∂r r ∂σrr r + σrr − σrθ + ρω 2 r2 = 0 ∂r .7.6.7.

σrθ = 0 everywhere and the equation of equilibrium in the circum- ferential direction gives that ∂σ∂θθθ = 0. why?) Since the radial equilibrium equation involves two unknowns. For this we will need to consider the strain-displacement and constitutive relations. we obtain:   E dur ur σrr = +ν 1 − ν 2 dr r   E ur dur σθθ = +ν 1 − ν2 r dr which replaced in the equilibrium equation gives the following ordinary differential equation for the radial displacement: 1 − ν2 2 3 r2 ur. the problem cannot be solved from equilibrium considerations exclusively (statically indeterminate).rr + rur. In this case. the constitutive law can be written as: E σrr = (εrr + νεθθ ) 1 − ν2 E σθθ = (εθθ + νεrr ) 1 − ν2 Combining them.106 MODULE 4. The approach sought is to obtain a Navier equation involving the radial displacement as the only unknown. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY Due to axial symmetry.72) r 8E The stresses follow as:   E dur ur σrr = +ν 1 − ν 2 dr r E E C2 3 + ν 2 2 = C1 − − ρω r 1−ν 1 + ν r2 8  E ur dur σθθ = 2 +ν 1−ν r dr E E C2 1 + 3ν 2 2 = C1 + − ρω r 1−ν 1 + ν r2 8 . E The general solution of this equation is (details discussed separately): C2 1 − ν 2 2 3 ur = C1 r + − ρΩ r (4.r − ur + ρω r = 0. The strain-displacement relations in this case reduce to: dur ur εrr = . εθθ = dr r We will assume plane stress conditions for each disk. σθθ = σθθ (r) (This actually assumes constant angular velocity.

we expect it to be zero). Both stress components are maximum at r = 0. we obtain the maximum angular velocity: s 8σy ωmax = (3 + ν)ρR2 In order to estimate the necessary clearance. we need to compute the radial displacement at r = R for ωmax : (1 − ν) σy umax (r = R) = 2 R ν+3 E Replacing the values for the problem we obtain: ωmax = 675 · s−1 ∼ 6450rpm . The stresses on the outer boundary at r = R (ignoring the blades) should go to zero: E 3+ν 2 2 σrr (r = R) = 0 = C1 − ρω R 1−ν 8 (1 − ν)(3 + ν) 2 2 ⇒ C1 = ρω R 8E The radial displacement is then given by 1−ν 2  ρω r (3 + ν)R2 − (1 + ν)r2  ur (r) = 8E and the dimensionless radial and hoop stresses are: σrr 3+ν  1 − rˆ2  2 2 = ρω R 8   σθθ 3+ν 1 + 3ν 2 = 1− rˆ ρω 2 R2 8 3+ν where rˆ = r/R. AIRY STRESS FUNCTION 107 Application of boundary conditions: We know that the displacement at r = 0 cannot go to ∞ (in fact. According to the von Mises criterion (to be discussed later in the class). the material will yield when the following combination of stress components or effective stress σ ef f reaches the yield stress: 1 q ef f 2 σ =√ (σrr − σθθ )2 + σθθ 2 + σrr 2 ρω 2 R2 p 2 = (7ν + 2ν + 7)ˆ r4 − 4(1 + ν)(3 + ν)ˆ r2 + (3 + ν)2 ≤ σy 8 The maximum effective stress is achieved at rˆ = 0 and its value: ef f 3+ν 2 2 σmax = ρω R 8 must be less than the yield stress. This implies C2 = 0.7.4. From this.

7. σrr (r = b) = −pe 2. Concept Question 4. Now let’s get back to the general solution and apply it to other problems of interest. Solution: The solution is obtained by setting C4 = 0 (the proof requires consideration of the displacements). Write down the boundary conditions pertinent to this case. Solution: The boundary conditions for this case are σrr (r = a) = −pi . We need to have a hole at r = 0 so that other solutions are possible. What strategies would you consider to solve this problem? The displacement at r = R is: umax = 0. and doing: .7.35 · 10−3 m about a third of a millimeter. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY This is somewhat low for a modern turbine.108 MODULE 4. Obtain the distribution of radial and hoop stresses by applying the boundary conditions to specialize the general solution in equation (4.70) to this problem. pe b pi a 1. Consider the case of a cylinder of internal radius a and external radius b subject to both internal and external pressure. where angular velocities of ∼ 10000rpms are common.

Sketch the solution σrr (r) and σθθ (r) normalized with pi for pe = 0 and b/a = 2/1 and verify your intuition on the stress field distribution Solution: As can be seen from the figure. AIRY STRESS FUNCTION 109 C2 σrr (r = a) = −pi = + 2C3 a2 C2 σrr (r = b) = −pe = 2 + 2C3 b a2 b 2   1 1 ⇒ pe − pi = − C 2 . the stresses in the plane do not change whether we are in plane stress or strain.4. 4. ⇒ C 2 = (pe − pi ) a2 b2 b 2 − a2 1 (pi a2 − pe b2 ) (−pe b2 + pi a2 ) = (b2 − a2 )C3 .7. How does the solution for the stresses depend on the elastic properties of the material? Solution: As expected from the biharmonic equation for no body forces. the stresses in the plane do not depend on the elastic constants. How would the solution change for plane strain conditions? Solution: As a consequence. However. ⇒ C3 = 2 2(b2 − a2 ) C2 σrr (r) = 2C3 + r2 pi a2 − pe b2 a2 b2 (pe − pi ) σrr (r) = + 2 2 b 2 − a2 r (b − a2 ) C2 σθθ (r) = 2C3 − r2 pi a2 − pe b2 a2 b2 (pe − pi ) σθθ (r) = − 2 2 b 2 − a2 r (b − a2 ) 3. the maximum . the out-of-plane stress σ33 is of course zero in plane stress and in plane strain it would be p i a2 − p e b 2 σ33 = ν(σrr + σθθ ) = 2ν = constant!! b 2 − a2 5.

Arizona. suffered rapid depressurization at 34. a Boeing 737-300. leading to an emergency landing at Yuma International Airport. but the hoop stress does not.110 MODULE 4. In this problem.400 ft (10. The radial component vanishes at the outer boundary as it should. σ22 = σ∞ R σ11 = σ∞ . 2011. Concept Question 4. we will attempt to compute the stress field around a stress-free hole of radius R subject to a hydrostatic plane stress state at a large distance compared to the radius of the hole r  R. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY stresses happen at the inner radius and decrease toward the boundary. WN812).8. On April 1.485 m) near Yuma.7. we will take our first incursion into the analysis of stresses around a rivet hole (in no way should it be construed that this would be an analysis relevant to the incident mentioned) As a first step. Southwest Airlines Flight 812 (SWA812. The Na- tional Transportation Board reported that a thorough investigation “revealed crack indica- tions at nine rivet holes in the lower rivet row of the lap joint”.

7.9. Compute the maximum hoop stress and its location. Concept Question 4. AIRY STRESS FUNCTION 111 1. What is the stress concentration factor for this type of remote loading on the hole? Solution: Clearly. Write the boundary conditions far from the hole in cartesian coordinates Solution: σ11 (x1 → ∞. e. x2 ) = σ∞ . Applying the first boundary condition: 1 σrr (r → ∞) = σ∞ = 2C3 . C3 6= 0.7.70) to this problem. 1. Solution: The general solution supports finite stresses at r → ∞ for C4 = 0 and C2 .9. → C3 = σ∞ 2 Applying the second boundary condition: C2 0= + σ∞ . x2 ) = 0. σ12 (x2 → ∞. say σ11 (r → ∞) = σ∞ . remote uniform loading in one direction. The main issue we have is that this case does not correspond to axisymmetric loading. σ22 (x1 → ∞. → C2 = −R2 σ∞ R2 and the solution is: "  2 # R σrr = σ∞ 1− (4. Consider an infinite plate with a small hole of radius a as shown in Figure 4.g.4. σrr (r = R) = 0 2.74) r 3. The goal is to determine the stress field around the whole. Obtain the distribution of radial and hoop stresses by applying the boundary conditions to specialize the general solution in equation (4.73) r "  2 # R σθθ = σ∞ 1+ (4. . the formulation in polar coordinates still proves advantageous in this case. x2 ) = 0.7. the maximum hoop stress take place at r = R and its value is: max σθθ = 2σ∞ The next step is to look at the case of asymmetric loading. Infinite plate with a hole under uniaxial stress. It turns out. Write down the boundary conditions pertinent to this case. Solution: The boundary conditions for this case are σrr (r → ∞) = σ∞ .

θ) = 0. Consider a point far from the hole (r → ∞. 4. σrr (a. while no condition is required for the tangential stress σθθ (r = a. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY a σ11 = σ∞ Figure 4. 2 2 . θ) = . 2 2 σ∞ σrθ (r → ∞. 2 2 σ∞ σ∞ σθθ (r → ∞. θ) = + cos 2θ + σ12 sin 2θ 2 2   σ11 + σ22 σ11 − σ22 σθθ (r → ∞. θ) = + cos 2θ. θ) = − cos 2θ − σ12 sin 2θ 2 2   σ11 − σ22 σrθ (r → ∞. 2 3. σθθ (r → ∞. σrθ (a. Looking at the boundary conditions at ∞ we observe that there is a part that cor- responds to a plate with a hole subject to a remote bi-axial loading state with the following boundary conditions: I σ∞ I σ∞ I I σrr (r → ∞. θ) = .112 MODULE 4. θ) = 0. σrθ (r = a. θ). Rotate the far-field stress state to polar coordinates using the transformation rules:   σ11 + σ22 σ11 − σ22 σrr (r → ∞. θ) = − sin 2θ.11: Infinite plate with a hole under uniaxial stress 2. θ). θ) = 0. Write down the bondary conditions of stress at r = a Solution: The radial and shear stresses vanish at r = a for any θ σrr (r = a. θ) = 0. θ) = − sin 2θ + σ12 cos 2θ 2 Solution: σ∞ σ∞ σrr (r → ∞. θ) = − cos 2θ.

AIRY STRESS FUNCTION 113 Write down the solution for this problem: Solution: From the previous problem: a2   I σ∞ σrr (r. θ) = f (r) cos 2θ Replace this form of the solution in the biharmonic equation to obtain the following:   4 IV 2 000 9 00 9 0 ∇ φ= f + f − 2 f + 3 f cos 2θ = 0. r . Finally. the ODE in parentheses must vanish 2 9 9 f IV + f 000 − 2 f 00 + 3 f 0 = 0 r r r 6. Obtain the solution for this ODE by first converting it to constant coefficients via the substitution r = et and then using the method of undetermined coefficients which starts by assuming a solution of the form f (t) = eαt .7. 5. θ) = 0. From the resulting Airy’s stress function   2 4 C4 φ(r. Solution: 7. r r r Solution: As this expression is valid for any value of θ. θ) = f (r) cos 2θ = C1 + C2 r + C3 r + 2 cos 2θ. as we need to solve the homogeneous biharmonic equation in polar coordinates: ∂ 4φ 2 ∂ 4φ 1 ∂ 4φ 2 ∂ 3φ 2 ∂ 3φ 1 ∂ 2φ 4 ∂ 2φ 1 ∂φ ∇4 φ = + + + − − + + .4. θ) = 1− 2 2 r a2   I σ∞ σθθ (r. The result should be: f (r) = C1 + C2 r2 + C3 r4 + C4 /r2 . replace t = log r to obtain the final solution to the variable coefficient ODE. θ) = 1+ 2 2 r I σrθ (r. The second part is not axisymmetric and therefore more involved. ∂r4 r2 ∂r2 ∂θ2 r4 ∂θ4 r ∂r3 r3 ∂r∂θ2 r2 ∂r2 r4 ∂θ2 r3 ∂r Due to the periodicity in θ the problem is greatly simplified and we can assume a solution of the form: φ(r.

θ) = − 1 + 3 4 cos 2θ. θ) = − σ2∞ sin 2θ. r r 8. r ∂θ r ∂r∂θ Solution:   II C1 C4 σrr (r.114 MODULE 4. θ) = 0 and σrθ (a. r r   II 2 C4 σθθ (r. Replace the values of the constants to obtain the expressions for the stress field for the second part of the problem: a2 a4   σ∞ σrr (r. θ) = + 2 2 r ∂r r ∂θ 2 II ∂ φ σθθ (r. the boundary conditions σrr (a. θ) = 2C2 + 12C3 r + 6 4 cos 2θ. shear. BOUNDARY VALUE PROBLEMS IN LINEAR ELASTICITY Write the expressions for the radial. The following results should be obtained C3 = 0 and C2 = −σ∞ /4. θ) = 2C2 + 6C3 r − 2 2 − 6 4 sin 2θ. and circumferential components of the stress field for the second part of the problem (II) using the definitions: II 1 ∂φ 1 ∂ 2φ σrr (r. θ) = 1 − 4 2 + 3 4 cos 2θ. 2 r r 4   σ∞ a σθθ (r. θ) = − 2C2 + 4 2 + 6 4 cos 2θ. θ) = −1 − 2 2 + 3 4 sin 2θ. 2C2 − 2 2 − 6 4 = 0. Use the boundary conditions at ∞ for the second part of the problem to obtain the values of the constants. θ) = 0 give: C1 C4 C1 C4 2C2 + 4 2 + 6 4 = 0. 2 r a2 a4   σ∞ σrθ (r. 9. C1 = σ∞ a2 /2 and C4 = −σ∞ a4 /4: Solution: II The shear boundary condition is: σrθ (r → ∞. θ) = 2 − . II II In addition. r   II 2 C1 C4 σrθ (r. which gives C3 = 0 and C2 = −σ∞ /4. θ) = ∂r2 II 1 ∂φ 1 ∂ 2 φ σrθ (r. 2 r r Solution: . a a a a which gives the values sought.

θ) = 1 + 2 − 1 + 3 4 cos 2θ . 2 r r a2 a4   σ∞ σrθ (r. Figure 4. . θ) to obtain the stress concentration factor of the hoop stress around the hole.12.7. Sketch the plot of σθθ (a. θ) = 1 − 2 + 1 − 4 2 + 3 4 cos 2θ . we can add the solutions to problems (I) and (II) and obtain the complete stress field: a2 a2 a4      σ∞ σrr (r.12: Stress concentration: infinite plate with a central hole subject to uniaxial stress.4. 2 r r r a2 a4      σ∞ σθθ (r. 2 r r 10. θ) = −1 − 2 2 + 3 4 sin 2θ . AIRY STRESS FUNCTION 115 Finally. Solution: The hoop stress is illustrated in Figure 4.