Solid mechanics is a branch of continuum mechanics that studies the behavior of solid materials under the action of forces, temperature changes, or other external agents. Elasticity is a branch of solid mechanics that refers to the ability of the body to return to its original size and shape after the forces causing deformation are removed. In this appendix the basic equations of the three-dimensional elasticity theory are developed at a material point in the body. A material point, or particle, is identified by its position in a rectangular Cartesian coordinate system (x1,x2,x3). The fundamental equations of elasticity consist of the geometry of deformation in article 1.1, the stresses and equilibrium in article 1.2, and the stress-deformation relations in article 1.3. The focus is on the classical linear elasticity theory in which the strains are small with respect to unity and the material is linear elastic. The basic equations are summarized in article 1.4 along with a description of the boundary value problems of elasticity.
Geometry of deformation
A continuous three-dimensional body occupies a closed region denoted by B0 in the reference state. Let every point of B0 be defined in a fixed rectangular Cartesian system of axes x1, x2, and x3. Let B denote the closed region of the body after it undergoes a deformation. The position vector of the point P0 in region B0 with respect to the origin is ⇀r=x1ˆi1+x2ˆi2+x3ˆi3,
where the unit vectors along the fixed axes are ˆi1, ˆi2, and ˆi3. The particle at P0:(x1,x2,x3) passes to point P:(y1,y2,y3) in region B, where coordinates (y1,y2,y3) are defined in the same fixed coordinate system. See figure [figA.1]. The position vector of point P referred to the same origin is
⇀R=y1ˆi1+y2ˆi2+y3ˆi3.
The deformation of the body is defined by the equations y1=y1(x1,x2,x3)y1=y2(x1,x2,x3)y3=y3(x1,x2,x3),
where x1, x2, and x3 are restricted to B0 and y1, y2, and y3 are restricted to B. In eq. ([eqA.3]) the yi, i=1,2,3, on the right-hand side denotes a function of three variables x1, x2, and x3, and yi on the left-hand side denotes the value of the function. Equation ([eqA.3]) defines the final location of the particle in B that is located at point P0 in B0. To prohibit the possibility that a particle at point P in region B maps to more than one point in region B0, or vice versa, it is required that there is a one-to-one correspondence between points in regions B0 and B. It follows that in region B eq. ([eqA.3]) has single-valued solutions: x1=x1(y1,y2,y3)x2=x2(y1,y2,y3)x3=x3(y1,y2,y3).
The functions defined in eqs. ([eqA.3]) and ([eqA.4]) are assumed to be continuous and differentiable in their respective variables. Continuity insures no fracture of the body results in the deformation. If we choose eq. ([eqA.3]) to describe the deformation of the body then x1, x2, and x3 are the independent variables, and the formulation is called the Lagrangian or the referential or material description. In the Lagrangian formulation we follow the particle originally at point P0:(x1,x2,x3) as the deformation proceeds. If we choose eq. ([eqA.4]) to describe the deformation of the body, then y1, y2, and y3 are the independent variables, and the formulation is called the Eulerian or spatial description. In the Eulerian formulation the same fixed spatial position y1, y2, y3 is occupied by different particles as the deformation proceeds. The Lagrangian description of the deformation is selected for the developments that follow in this appendix. The position vector of point P relative to point P0 is denoted by ⇀u and is called the displacement vector. Thus, ⇀u=⇀R−⇀r=(y1−x1)ˆi1+(y2−x2)ˆi2+(y3−x3)ˆi3.
Components of the displacement vector are u1(x1,x2,x3)=y1(x1,x2,x3)−x1u2(x1,x2,x3)=y2(x1,x2,x3)−x2u3(x1,x2,x3)=y3(x1,x2,x3)−x3.
Deformation is quantified by the change in distance between any two points in a body. Consider two infinitesimally close points P0 and Q0 in region B0 that pass to points P and Q, respectively, in region B. The differential line element ⌢P0Q0 in region B0 shown in figure [figA.2](a) passes to the differential line element ⌢PQ in region B shown in figure [figA.2](b). The differential position vector of line element ⌢P0Q0 is d⇀r=dx1ˆi1+dx2ˆi2+dx3ˆi3. The square of the length of d⇀r is given by ds2=d⇀r∙d⇀r=(dx1)2+(dx2)2+(dx3)2. The unit vector of point Q0 with respect to point P0 is given by ˆn=d⇀rds=dx1dsˆi1+dx2dsˆi2+dx3dsˆi3.
Write unit vector in eq. ([eqA.7]) as ˆn=n1ˆi1+n2ˆi2+n3ˆi3,
where ni=dxids, i=1,2,3. The differential position vector of ⌢PQ is d⇀R=dy1ˆi1+dy2ˆi2+dy3ˆi3, and the square of its length is dS2=d⇀R∙d⇀R=(dy1)2+(dy2)2+(dy3)2. Write the differential vector as d⇀R=dSˆN where the unit vector of point Q with respect to point P is ˆN=dy1dSˆi1+dy2dSˆi2+dy3dSˆi3=N1ˆi1+N2ˆi2+N3ˆi3
The total differentials (dy1,dy2,dy3) of the functions y1(x1,x2,x3), y2(x1,x2,x3), and y3(x1,x2,x3) are written in terms of the total differentials (dx1,dx2,dx3) by [dy1dy2dy3]=[∂y1∂x1∂y1∂x2∂y1∂x3∂y2∂x1∂y2∂x2∂y2∂x3∂y3∂x1∂y3∂x2∂y3∂x3][dx1dx2dx3].
The determinate of the 3X3 matrix in eq. ([eqA.3]) is J=det[∂y1∂x1∂y1∂x2∂y1∂x3∂y2∂x1∂y2∂x2∂y2∂x3∂y3∂x1∂y3∂x2∂y3∂x3],
where J is called the Jacobian. Equation ([eqA.3]) possesses a continuous single-valued solution satisfying the inverse ([eqA.4]) if and only if the Jacobian is positive for all points in region B0 (Batra, 2006). The strain of line element ⌢PQ is defined by εL=12[dS2−ds2ds2]=12[(dSds)2−1].
Use the chain rule to write the square of the length of line element ⌢PQ with respect to the square of the length of line element ⌢P0Q0 as dS2=[(dy1ds)2+(dy2ds)2+(dy2ds)2]ds2.
From eq. ([eqA.10]) use the chain rule again to write the differential dy1 as follows: dy1=∂y1∂x1dx1+∂y1∂x2dx2+∂y1∂x3dx3=[∂y1∂x1dx1ds+∂y1∂x2dx2ds+∂y1∂x3dx3ds]ds.
-1Substitute dxi/ds=ni from eq. ([eqA.8]), and substitute u1+x1 for coordinate y1 from eq. ([eqA.6]), into eq. ([eqA.14]) to get dy1ds=[(1+∂u1∂x1)n1+∂u1∂x2n2+∂u1∂x3n3].
Starting with differentials dy2 and dy3 from eq. ([eqA.10]), we perform similar manipulations leading to eq. ([eqA.15]) to find dy2ds=[∂u2∂x1n1+(1+∂u2∂x2)n2+∂u2∂x3n3]dy3ds=[∂u3∂x1n1+∂u3∂x2n2+(1+∂u3dx3)n3].
Substitute eq. ([eqA.13]) into eq. ([eqA.12]) to write the equivalent expression for the strain in eq. ([eqA.12]) as εL=12[(dy1ds)2+(dy2ds)2+(dy2ds)2−(n21+n22+n23)],
where the number one that appears in eq. ([eqA.12]) is replaced by n21+n22+n23. Substitute the results from eqs. ([eqA.15]) and ([eqA.16]) into eq. ([eqA.17]) and write the result as εL=ε11n21+ε12n1n2+ε13n1n3+ε21n2n1+ε22n22+ε23n2n3+ε31n3n1+ε32n3n2+ε33n23.
The expression ([eqA.18]) for the strain can be written in the matrix form εL=[n1n2n3][ε11ε12ε13ε21ε22ε23ε31ε32ε33][n1n2n3].
The coefficients in the expression for the strain are as follows: ε11=∂u1∂x1+12[(∂u1∂x1)2+(∂u2∂x1)2+(∂u3∂x1)2]ε22=∂u2∂x2+12[(∂u1∂x2)2+(∂u2∂x2)2+(∂u3∂x2)2]ε33=∂u3∂x3+12[(∂u1∂x3)2+(∂u2∂x3)2+(∂u3∂x3)2]ε12=ε21=12[∂u1∂x2+∂u2∂x1+∂u1∂x1∂u1∂x2+∂u2∂x1∂u2∂x2+∂u3∂x1∂u3∂x2]ε13=ε31=12[∂u1∂x3+∂u3∂x1+∂u1∂x1∂u1∂x3+∂u2∂x1∂u2∂x3+∂u3∂x1∂u3∂x3]ε23=ε32=12[∂u2∂x3+∂u3∂x2+∂u1∂x2∂u1∂x3+∂u2∂x2∂u2∂x3+∂u3∂x2∂u3∂x3].
Physical interpretation of strain components εij
For the line element parallel to the x1-axis in the reference configuration, the components of the unit vector are (n1,n2,n3)=(1,0,0), and from eq. ([eqA.18]) its strain is given by εL=ε11. For the line element parallel to the x2-axis, the components of the unit vector are (n1,n2,n3)=(0,1,0), and its strain is given by εL=ε22. For the line element parallel to the x3-axis its strain is εL=ε33. The physical interpretation of component ε12 is determined from the passing of line elements dx1ˆi1 and dx2ˆi2 in region B0 to directions (ˆN)1 and (ˆN)2, respectively, in region B. Unit vector ˆN=N1ˆi1+N2ˆi2+N3ˆi3 where the components Ni are given by Ni=dyidS=dyidsdsdS=dyids(1√1+2εL)=(∂yi∂x1n1+∂yi∂x2n2+∂yi∂x3n3)(1√1+2εL),i=1,2,3.
Take (n1,n2,n3)=(1,0,0) in eq. ([eqA.26]), then from eq. ([eqA.18]) εL=ε11. In the transition from region B0 to region B the unit vector ˆi1→(ˆN)1, where the unit vector (ˆN)1 is given by (ˆN)1=[∂y1∂x1ˆi1+∂y2∂x1ˆi2+∂y3∂x1ˆi3](1√1+2ε11).
Take (n1,n2,n3)=(0,1,0) in eq. ([eqA.26]), then εL=ε22. In the transition from region B0 to region B the unit vector ˆi2→(ˆN)2. The result for (ˆN)2 is (ˆN)2=[∂y1∂x2ˆi1+∂y2∂x2ˆi2+∂y3∂x2ˆi3](1√1+2ε22).
The scalar product of the two unit vectors (ˆN)1 and (ˆN)2 is equal to the cosine of the angle between them. Let this angle be denoted by π/2−θ12 such that, (ˆN)1∙(ˆN)2=cos(π2−θ12)=sinθ12.
Angle θ12 represents the reduction in the right angle. Substitute eq. ([eqA.27]) for (ˆN)1 and eq. ([eqA.28]) for (ˆN)2 in the left-hand side of eq. ([eqA.29]) to get sinθ12=[∂y1∂x1∂y1∂x2+∂y2∂x1∂y2∂x2+∂y3∂x1∂y3∂x2]1√1+2ε11√1+2ε22.
Next substitute yi=ui+xi, i=1,2,3, in the terms in the square brackets of the previous equation, and compare the result to eq. ([eqA.23]) to find sinθ12=2ε12√1+2ε11√1+2ε22.
Thus, the right angle between line elements dx1ˆi1 and dx2ˆi2 in region B0 is changed in the transition to region B in direct proportion to the strain component ε12. If ε12=0, then θ12=0, and the right angle is preserved in the deformed body. Similarly, the strain component ε13 is proportional to the change in the right angle between line elements dx1 and dx3 in the transition to the deformed body.
Engineering strain
Engineering strain is defined by εE=dS−dsds=dSds−1.
Substitute for dS/ds from eq. ([eqA.32]) into eq. ([eqA.12]) to get εL=εE+ε2E/2. Since the product of the direction cosines commute, eq. ([eqA.18]) is rewritten as εL=εE+12ε2E=ε11n21+ε22n22+ε33n23+γ12n1n2+γ13n1n3+γ23n2n3,
where the engineering shear strains are defined by γ12=ε12+ε21γ13=ε13+ε31γ23=ε23+ε32.
Linear strain-displacement relations
For many materials the strains are very small in the elastic range. A linear theory of deformation is characterized by the magnitude of the nine displacement gradients |∂ui∂xj|∼10−3. Hence, we neglect the quadratic terms in the displacement gradients with respect to the linear terms in the strain-displacement eqs. ([eqA.20]) to ([eqA.25]). The resulting linear strain-displacement relations are ε11=∂u1∂x1ε22=∂u2∂x2ε33=∂u3∂x3,γ12=∂u1∂x2+∂u2∂x1γ13=∂u1∂x3+∂u3∂x1γ23=∂u2∂x3+∂u3∂x2, andεE=ε11n21+ε22n22+ε33n23+γ12n1n2+γ13n1n3+γ23n2n3.
Transformation of the strains between two Cartesian coordinate systems
In the reference configuration B0 consider the two orthogonal Cartesian coordinate systems (x1,x2,x3) and (x′1,x′2,x′3), which have the same origin. In the (x1,x2,x3) system, or simply the xi-system, the corresponding unit vectors are (ˆi1,ˆi2,ˆi3). In the (x′1,x′2,x′3) system, or simply the x′i-system, the corresponding unit vectors are (ˆi′1,ˆi′2,ˆi′3). The position vector ⇀r of a point P0 in the three-dimensional space is the same if written in the xi-system or in the x′i-system. That is, ⇀r=x1ˆi1+x2ˆi2+x3ˆi3=x′1ˆi′1+x′2ˆi′2+x′3ˆi′3.
The linear form x1ˆi1+x2ˆi2+x3ˆi3 is said to remain invariant under the transformation of variables. Take the scalar product, or dot product, of eq. ([eqA.38]) with unit vector ˆi′1 to get x1ˆi1∙ˆi′1+x2ˆi2∙ˆi′1+x3ˆi3∙ˆi′1=x′1.
Define the nine direction cosines λij, i,j=1,2,3, by ˆi′i∙ˆij≡λij=cos(x′i,xj).
For example, the notation λ12=cos(x′1,x2) represents the cosine of the angle between the x′1-axis and the x2-axis, and the notation λ21=cos(x′2,x1) represents the cosine of the angle between the x′2-axis and the x1-axis. From the definition of λij in eq. ([eqA.39]) becomes x′1=λ11x1+λ12x2+λ13x3. If we take the dot product of eq. ([eqA.38]) with ˆi′2 and use the definition ([eqA.40]), then we find x′2=λ21x1+λ22x2+λ23x3. The relation between the x′i-coordinates and the xi-coordinates at point P0 is given by the linear transformation [x′1x′2x′3]=[λ11λ12λ13λ21λ22λ23λ31λ32λ33][x1x2x3].
The compact form of matrix eq. ([eqA.41]) is {x′}=[λ]{x}, where{x′}=[x′1x′2x′3][λ]=[λ11λ12λ13λ21λ22λ23λ31λ32λ33]{x}=[x1x2x3].
The unit vectors in the x′i-system are related to those in the xi-system by the same relation given in eq. ([eqA.41]): [ˆi′1ˆi2ˆi′3]=[λ11λ12λ13λ21λ22λ23λ31λ32λ33][ˆl1ˆl2ˆl3].
Consider the dot product ˆi′1∙ˆi′1=1, and from eq. ([eqA.44]) write it in terms of the unit vectors in the xi-system; i.e., 1=(λ11ˆi1+λ12ˆi2+λ13ˆi3)∙(λ11ˆi1+λ12ˆi2+λ13ˆi3).
Since unit vectors (ˆi1,ˆi2,ˆi3) are mutually perpendicular we find the relation 1=λ211+λ212+λ213.
Consider the dot product ˆi′1∙ˆi′2=0 and write ˆi′1 and ˆi′2 from eq. ([eqA.44]) to get 0=(λ11ˆi1+λ12ˆi2+λ13ˆi3)∙(λ21ˆi1+λ22ˆi2+λ23ˆi3).
Again (ˆi1,ˆi2,ˆi3) are mutually perpendicular, so we find the relation 0=λ11λ21+λ12λ22+λ13λ23.
We can proceed by performing the scalar products ˆi′1∙ˆi′3=0, ˆi′2∙ˆi′2=1, ˆi′2∙ˆi′3=0, and ˆi′3∙ˆi′3=1. Collectively we find the following relations between the direction cosines: 1=λ211+λ212+λ2130=λ11λ21+λ12λ22+λ13λ230=λ11λ31+λ12λ32+λ13λ33,1=λ221+λ222+λ2230=λ21λ31+λ22λ32+λ23λ33, and 1=λ231+λ232+λ233.
There are six relations in eq. ([eqA.49]) between the nine direction cosines. Hence, only three of the direction cosines are independent. We show some interesting properties of the direction cosine matrix [λ] beginning with the matrix product [λ][λ]T. The result is [λ][λ]T=[λ211+λ212+λ213λ11λ21+λ12λ22+λ13λ23λ11λ31+λ12λ32+λ13λ33λ11λ21+λ12λ22+λ13λ23λ221+λ222+λ223λ21λ31+λ22λ32+λ23λ33λ11λ31+λ12λ32+λ13λ33λ21λ31+λ22λ32+λ23λ33λ231+λ232+λ233].
Compare the elements of the matrix in eq. ([eqA.50]) to the relations in eq. ([eqA.49]) to find [λ][λ]T=[100010001]=[I].
Equation ([eqA.51]) shows that the matrix [λ] is an orthogonal matrix. That is, the inverse [λ]−1 is equal to its transpose [λ]T. Also det([λ][λ]T)=det[λ]det[λ]T, but det[λ]T=det[λ]. Hence. det([λ][λ]T)=(det[λ])2=(det[I])2=1.
The determinate of an orthogonal matrix is either 1 or −1. The inverse of eq. ([eqA.42]) is {x}=[λ]T{x′} which written in expanded form is [x1x2x3]=[λ11λ21λ31λ12λ22λ32λ13λ23λ33][x′1x′2x′3].
From eq. ([eqA.38]) the differential of the position vector in eq. ([eqA.38]) is d⇀r=dx1ˆi1+dx2ˆi2+dx3ˆi3=dx′1ˆi′1+dx′2ˆi′2+dx′3ˆi′3.
The square of the length of d⇀r is ds2=d⇀r∙d⇀r. The length is the same in the xi-system and the x′i-system, so ds2=(dx1)2+(dx2)2+(dx3)2=(dx′1)2+(dx′2)2+(dx′3)2.
Hence, the unit vector ˆn in both the xi-system and the x′i-system has the invariant forms ˆn=d⇀rds=dx1dsˆi1+dx2dsˆi2+dx3dsˆi3=dx′1dsˆi′1+dx′2dsˆi′2+dx′3dsˆi′3.
The derivatives of the coordinates in eq. ([eqA.56]) are obtained from eq. ([eqA.41]). These derivatives are [dx′1dsdx′2dsdx′3ds]=[λ11λ12λ13λ21λ22λ23λ31λ32λ33][dx1dsdx2dsdx3ds].
Let ni=dxids and n′i=dx′1ds, i=1,2,3. The matrix form of eq. ([eqA.57]) is {n′}=[λ]{n},
where we write the unit vector in matrix notation as {n}=[n1 n2 n3]T and {n′}=[n′1 n′2 n′3]T. The inverse of eq. ([eqA.58]) is {n}=[λ]T{n′}.
The strain components εij are functions in the variables (x1,x2,x3), and the strains ε′ij are functions in the variables (x′1,x′2,x′3). These strain components are the elements in the matrices [ε]=[ε11ε12ε13ε21ε22ε23ε31ε32ε33], and [ε′]=[ε′11ε′12ε′13ε′21ε′22ε′23ε′31ε′32ε′33].
Equations ([eqA.23]) to ([eqA.25]) show that the strain matrix [ε] is symmetric. The strain of line element ⌢P0Q0 must the same, or invariant, if computed in the xi-system or in the x′i-system. The expression for the strain ([eqA.19]) in matrix notation is εL={n}T[ε]{n}.
Substitute eq. ([eqA.59]) for the unit vector {n} into eq. ([eqA.61]) to get εL=([λ]T{n′})T[ε][λ]T{n′}, orεL={n′}T[λ][ε][λ]T{n′}.
For eq. ([eqA.62]) to be an invariant form of eq. ([eqA.61]) we conclude [ε′]=[λ][ε][λ]T.
Hence, εL={n′}T[ε′]{n′}.
Compare the forms of eq. ([eqA.61]) and eq. ([eqA.64]) to note their similarity. Also, the inverse transformation of eq. ([eqA.63]) is [ε]=[λ]T[ε′][λ].
The transpose of eq. ([eqA.63]) is [λ][ε]T[λ]T, but [ε]T=[ε], so matrix [ε′] is also symmetric. From eq. ([eqA.34]) the engineering shear strain γ12=ε12+ε21. But ε12=ε21, which means ε12=ε21=γ12/2. In terms of engineering shear strains ([eqA.34]), the strain transformation ([eqA.63]) is [ε′11γ′12/2γ′13/2γ′12/2ε′22γ′23/2γ′13/2γ′23/2ε′33]=[λ][ε11γ12/2γ13/2γ12/2ε22γ23/2γ13/2γ23/2ε33][λ]T.
Stress
The reference configuration of the body occupying region B0 is assumed to be the natural or unstressed state. External forces acting on the body cause a state of stress in the deformed configuration occupying region B, and it is in the configuration B where the study of stresses is to be carried out. However, for infinitesimal displacement gradients the point P0 in region B0 and point P in region B lie very close together so that we do not distinguish between them. For small deformation theory, the study of equilibrium at a point in a deformable body is performed in the reference configuration.
Consider a continuous deformable body acted on by external forces as shown in figure [figA.3](a). Due to the action of the external forces there will be internal forces acting between particles of the body. To examine the internal forces we pass a plane labeled mm through point P that is parallel to the x2x3 plane. Consider the free body to the left of the plane mm shown in figure [figA.3](b). Plane mm is divided into a large number of small areas, each Δx2 by Δx3. The internal forces acting on each of these areas varies in magnitude and in direction.
The internal force Δ⇀F acting at point P is a resultant of distributed force intensities acting over the area Δx2Δx3. Let ΔA1 denote the area Δx2Δx3. Force Δ⇀F represents the action exerted by the material outside the plane mm through area ΔA1 on the material inside the plane mm. Point forces do not occur in nature. Forces are always distributed throughout regions, which can have dimensions of length, area, or volume. (However, point forces are an essential concept in the mechanics of solid bodies.) Consequently, as ΔA1→0 the resultant of the distributed force intensities acting over ΔA1 vanishes (i.e, Δ⇀F→0). The stress vector or traction vector acting at point P is defined as ⇀T(ˆi1)=limΔA1→0(Δ⇀F/ΔA1),
where the unit normal to area ΔA1 is ˆi1. Now consider a rectangular parallelepiped with edges Δx1, Δx2, and Δx3 cut out of the body. It will have six separate plane surfaces, which enclose the volume containing point P. Identify a surface face in terms of the coordinate axis normal to the surface. A face is defined as a positive face when its outwardly directed normal vector points in the direction of the positive coordinate direction, and as a negative face when its outward normal vector points in the negative coordinate direction. The projection of the parallelepiped in the x1-x2 plane is shown figure [figA.4], where only the internal forces acting on the positive and negative faces normal to the x1-axis are explicitly shown. Not shown in the figure are the surface forces acting on the four lateral faces and the body force acting on the volume. Let Δx1→0 without changing the values of Δx2 and Δx3. In the limit the forces acting on the lateral surfaces and the body force vanish, and force equilibrium yields
l164pt
⇀T(−ˆi1)ΔA1+⇀T(ˆi1)ΔA1=0.
Since ΔA1>0, we find from equilibrium that the stress vector on the negative x1-face is equal to the negative of the stress vector on the positive x1-face; i.e., ⇀T(−ˆi1)=−⇀T(ˆi1).
To simplify the notation let ⇀T1=⇀T(ˆi1). Stress vectors acting on the positive x2-face and the positive x3-face are denoted by ⇀T2=⇀T(ˆi2) and ⇀T3=⇀T(ˆi3), respectively. Stress vectors acting on the negative x2-face and the negative x3-face are denoted by −⇀T2 and −⇀T3, respectively.
Define the stress components σij=⇀Ti∙ˆij, i,j=1,2,3. The first subscript on σij is associated with the direction normal to the face, and the second subscript is associated with the direction of the stress component. Thusthe stress vectors in terms of components are ⇀T1=σ11ˆi1+σ12ˆi2+σ13ˆi3⇀T2=σ21ˆi1+σ22ˆi2+σ23ˆi3⇀T3=σ31ˆi1+σ32ˆl2+σ33ˆl3 or equivalently [⇀T1⇀T2⇀T3]=[σ11σ12σ13σ21σ22σ23σ31σ32σ33][ˆl1ˆi2ˆi3].
Positive stress components acting on the positive faces of the rectangular parallelepiped are shown in figure [figA.5]. The stress components on the negative faces of the parallelepiped are equal and oppositely directed to those on the positive faces according to conditions like eq. ([eqA.69]). Hence, there are nine stress components at a point, not eighteen. We express the nine stress components at a point in the matrix form [σ]=[σ11σ12σ13σ21σ22σ23σ31σ32σ33].
The diagonal elements in the stress matrix ([eqA.71]) are the normal stresses, and the off-diagonal elements are the shear stresses. The nine stresses in matrix ([eqA.71]) are shown in figure [figA.5].
We pose the following question: Are the nine stress components at point P sufficient to determine the stresses on an arbitrarily orientated plane face through the point? To answer this question we consider equilibrium of a tetrahedron cut from the body at point P. The external surfaces of the tetrahedron shown in figure [figA.6](a) consist of three right triangles normal to the coordinate axes, and one oblique triangular area that is shaded in figure [figA.6]. For the surface with unit outward normal vector −ˆi1, the area is ΔA1=(Δx2Δx3)/2; for the surface with unit outward normal −ˆi2, the area is ΔA2=(Δx1Δx3)/2; and for the surface with unit outward normal vector −ˆi3 the area is ΔA3=(Δx1Δx2)/2. The area of the oblique surface is denoted by ΔAn, and its unit outward normal vector is ˆn. To calculate the area of the oblique face we use the fact that the cross product of two position vectors is equal to the area of a parallelogram formed by the vectors and in a direction normal to the plane of the parallelogram. Two of the vectors along the edges of the oblique face are −Δx1ˆi1+Δx2ˆi2 and −Δx1ˆi1+Δx3ˆi3, and the area of the parallelogram formed by these vectors is equal to 2An. Thus, 2ΔAnˆn=(−Δx1ˆi1+Δx2ˆi2)×(−Δx1ˆi1+Δx3ˆi3)=(Δx2Δx3ˆi1+Δx1Δx3ˆi2+Δx1Δx2ˆi3),
which simplifies to ΔAnˆn=ΔA1ˆi1+ΔA2ˆi2+ΔA3ˆi3.
From eq. ([eqA.73]) we find that area of the oblique face ΔAn=√(ΔA1)2+(ΔA2)2+(ΔA3)2, and the components of the unit normal vector are n1=ΔA1/ΔAnn2=ΔA2/ΔAnn3=ΔA3/ΔAn.
Equilibrium of the free body diagram in figure [figA.6](b) leads to ⇀T(ˆn)ΔAn+(−⇀T1ΔA1)+(−⇀T2ΔA2)+(−⇀T3ΔA3)+⇀B(dVol)=0,
where ⇀B is the body force vector per unit volume. The tetrahedron is also a triangular pyramid where ΔAn is the area of its triangular base, and the volume of the pyramid is hΔAn/3 where h is its height. Divide eq. ([eqA.75]) by ΔAn to get ⇀T(n)=⇀T1n1+⇀T2n2+⇀T3n3−⇀B(h/3).
It can be shown that the height in this case is given by h=Δx1n1=Δx2n2=Δx3n3. In the limit where Δx1→0, Δx2→0 and Δx3→0, the height h→0. Hence, in the limit the equilibrium equation is ⇀T(ˆn)=⇀T1n1+⇀T2n2+⇀T3n3.
The implication of eq. ([eqA.77]) is that the nine stress components σij at point P are sufficient to determine the traction, or stresses, on any face through the point.
Equilibrium differential equations
Consider the forces acting on a rectangular parallelepiped at point P. The free body diagram is shown in figure [figA.7]. The vector sum of forces is ⇀T1Δx2Δx3|x1+Δx1−⇀T1Δx2Δx3|x1+⇀T2Δx1Δx3|x2+Δx2−⇀T2Δx1Δx3|x2+⇀T3Δx1Δx2|x3+Δx3−⇀T3Δx1Δx2|x3+⇀BΔx1Δx2Δx3.
For small increments in Δxi, i=1,2,3, use the Taylor series representation of surface forces results in the equilibrium equation to get eq. ([eqA.78]) below. ∂∂x1(⇀T1Δx2Δx3)Δx1+∂∂x2(⇀T2Δx1Δx3)Δx2+∂∂x3(⇀T3Δx1Δx2)Δx3+⇀BΔx1Δx2Δx3+O((Δxi)4)=0.
Arrange the terms in eq. ([eqA.78]) to the form (∂⇀T1∂x1+∂⇀T2∂x2+∂⇀T3∂x3+⇀B)Δx1Δx2Δx3+O((Δxi)4)=0
Divide eq. ([eqA.79]) by the volume followed by the limit as Δx1Δx2Δx3→0 to get the vector differential equation of force equilibrium at point P as ∂⇀T1∂x1+∂⇀T2∂x2+∂⇀T3∂x3+⇀B=0.
Substitute eq. ([eqA.70]) for the traction vectors in eq. ([eqA.80]) to write the equilibrium differential equations in the xi-coordinate directions. In the order of x1, x2, x3 coordinate directions these equations are ∂σ11∂x1+∂σ21∂x2+∂σ31∂x3+B1=0∂σ12∂x1+∂σ22∂x2+∂σ32∂x3+B2=0∂σ13∂x1+∂σ23∂x2+∂σ33∂x3+B3=0.
Now consider moment equilibrium about the coordinate axes of the rectangular parallelepiped at point P. For moment equilibrium about the x1-axis refer to the free body diagram in figure [figA.8]. The moment arm from point P to the line of action of the normal force (σ22Δx1Δx3)x2+Δx2 acting on the positive x2-face is denoted by εΔx3, where ε is a small numerical value. Parameter ε is not known, but this will not matter in the end result. The moment arm from point P to the line of action of the shear force (σ23Δx1Δx3)x2+Δx2 acting on the positive x2 face is Δx2/2. Including all the forces shown in figure [figA.8], the sum of moments about the x1-axis through point P, counterclockwise positive, is −εΔx3(σ22Δx1Δx3)x2+Δx2+εΔx3(σ22Δx1Δx3)x2+Δx22(σ23Δx1Δx3)x2+Δx2+Δx22(σ23Δx1Δx3)x2+εΔx2(σ33Δx1Δx2)x3+Δx3−εΔ2(σ33Δx1Δx2)x3−Δx32(σ32Δx1Δx2)x3+Δx3−Δx32(σ32Δx1Δx2)x3=0.
Use the Taylor series to expand the forces acting on the positive coordinate faces with respect to the forces acting on the negative coordinate faces to get −εΔx3[∂σ22∂x2Δx2+O(Δx22)]Δx1Δx3+Δx22[2σ23+∂σ23∂x2Δx2+O(Δx22)]Δx1Δx3+εΔx2[∂σ33∂x3Δx3+O(Δx23)]Δx1Δx2−Δx32[2σ32+∂σ32∂x3Δx3+O(Δx23)]Δx1Δx2=0.
Expand eq. ([eqA.83]) in powers of Δxi to write it as (σ23−σ32)Δx1Δx2Δx3+ε[−∂σ22∂x2Δx1Δx2Δx23+∂σ33∂x3Δx1Δx22Δx3]+ H.O.T. =0,
where H.O.T. means higher order terms, that is, terms of quartic powers and higher in the increments in the coordinates. Notice the terms multiplied by ε are quartic powers of the increments in the coordinates. Division of eq. ([eqA.84]) by Δx1Δx2Δx3, followed by the limit of Δxi→0 leads the condition of moment equilibrium about the x1-axis that σ23−σ32=0. Moment equilibrium about the x2-axis leads to σ31−σ13=0, and moment equilibrium about the x3-axis leads to σ12−σ21=0. The equations of moment equilibrium at point P are σ12=σ21σ13=σ31σ23=σ32.
Hence, the stress matrix ([eqA.71]) is symmetric.
Transformation of stresses between two Cartesian coordinate systems
At point P coordinates (x′1,x′2,x′3) are linearly related to coordinates (x1,x2,x3) by eq. ([eqA.41]). The stress components σij are functions in the variables (x1,x2,x3), and the stresses σ′ij are functions in the variables(x′1,x′2,x′3). The stress vectors acting on the xi-faces are denoted by ⇀Ti, and those acting on the x′i-faces are denoted by ⇀T′i. These stress vectors are written in their respective coordinate systems by [⇀T1⇀T2⇀T3]=[σ11σ12σ13σ21σ22σ23σ31σ32σ33][ˆi1ˆi2ˆi3], and [⇀T′1⇀T′2⇀T′3]=[σ′11σ′12σ′13σ′21σ′22σ′23σ′31σ′32σ′33][ˆi′1ˆi′2ˆi′3].
In eq. ([eqA.86]) the stress matrices are [σ]=[σ11σ12σ13σ21σ22σ23σ31σ32σ33], and [σ′]=[σ′11σ′12σ′13σ′21σ′22σ′23σ′31σ′32σ′33].
Note that the stress matrix [σ] is symmetric by eq. ([eqA.85]). The stress transformation equations between the Cartesian coordinate systems (x1,x2,x3) and (x′1,x′2,x′3) are determined by selecting the unit normal in eq. ([eqA.77]) to be either ˆi′1, ˆi′2, or ˆi′3. First let ˆn=ˆi′1 such that ⇀T(n)=⇀T′1 in eq. ([eqA.77]). From eq. ([eqA.44]) we have ˆn=λ11ˆi1+λ12ˆi2+λ13ˆi3. Hence, eq. ([eqA.77]) becomes ⇀T′1=λ11⇀T1+λ12⇀T2+λ13⇀T3.
Second, let ˆn=ˆi′2 such that ⇀T(n)=⇀T′2. From eq. ([eqA.44]) we have ˆn=λ21ˆi1+λ22ˆi2+λ23ˆi3. Hence, eq. ([eqA.77]) becomes ⇀T′2=λ21⇀T1+λ22⇀T2+λ23⇀T3.
Third, let ˆn=ˆi′3 that ⇀T(n)=⇀T′3 and ˆn=λ31ˆi1+λ32ˆi2+λ33ˆi3. Hence, ⇀T′3=λ31⇀T1+λ32⇀T2+λ33⇀T3.
The three selections for the unit normal in eq. ([eqA.77]) relate the tractions acting on the x′i coordinate faces to the tractions acting on the xi-faces by [⇀T′1⇀T′2⇀T′3]=[λ11λ12λ13λ21λ22λ23λ31λ32λ33][⇀T1⇀T2⇀T3].
Substitute the expressions for the stress vectors from eq. ([eqA.86]) into eq. ([eqA.91]) to get [σ′11σ′12σ13σ′21σ′22σ23σ′31σ′32σ33][ˆi′1ˆi′2ˆi′3]=[λ11λ12λ13λ21λ22λ23λ31λ32λ33][σ11σ12σ13σ21σ22σ23σ31σ32σ33][ˆi1ˆi2ˆi3].
The inverse eq. ([eqA.44]) is [ˆi1ˆi2ˆi3]=[λ11λ21λ31λ12λ22λ32λ13λ23λ33][ˆi′1ˆi′2ˆi′3].
Substitute eq. ([eqA.93]) into the right-hand side of eq. ([eqA.92]) and rearrange the result to find ([σ′11σ′12σ′13σ′21σ′22σ′23σ′31σ′32σ′33]−[λ11λ12λ13λ21λ22λ23λ31λ32λ33][σ11σ12σ13σ21σ22σ23σ31σ32σ33][λ11λ21λ31λ12λ22λ32λ13λ23λ33])[ˆi′1ˆi′2ˆi′3]=[000].
To satisfy eq. ([eqA.94]) we find that the stress components σ′ij in the x′i-system are related to the stress components σij in the xi-system by [σ′11σ′12σ13σ′21σ′22σ′23σ′31σ′32σ′33]=[λ11λ12λ13λ21λ22λ23λ31λ32λ33][σ11σ12σ13σ21σ22σ23σ31σ32σ33][λ11λ21λ31λ12λ22λ32λ13λ23λ33].
Equation ([eqA.95]) in compact form is [σ′]=[λ][σ][λ]T.
Pre-multiply eq. ([eqA.96]) by [λ]T, post-multiply it by [λ], and note that [λ]T[λ]=[λ][λ]T=[I] to find the inverse transformation [σ]=[λ]T[σ′][λ].
The transpose of eq. ([eqA.96]) is [λ][σ]T[λ]T, but [σ]T=[σ], so the stress matrix [σ′] is also symmetric. Comparing the strain transformation eq. ([eqA.63]) to the stress transformation eq. ([eqA.96]), it is clear that the transformation of strains εij is the same form as the transformation of the stresses σij.
Cartesian tensors
A tensor is a system of numbers or functions, whose components obey a certain law of transformation when the independent variables undergo a linear transformation. If the independent variables are the rectangular cartesian systems xi and x′i transforming by the linear relations given by {x′}=[λ]{x} at point P, then the systems obeying certain laws of transformation are called Cartesian tensors.
Definition.A system of order two may be defined to have nine components εij in xi and nine components ε′ij in x′i. If [ε′ij]=[λ][εij][λ]T
then the functions εij and ε′ij are the components in their respective variables of a second order Cartesian tensor. Similarly, functions σij and σ′ij are the components in their respective variables of a second order Cartesian tensor.
Linear elastic material law
To this point in the study of the mechanics of a solid body we have eighteen unknown functions of the Cartesian coordinates x1, x2, and x3. These are the three displacements u1, u2, and u3, the six strains ε11, ε22, ε33, γ23, γ31, and γ12, and nine stresses σ11, σ12, σ13, σ21, σ22, σ23, σ31, σ32, and σ33. There are twelve equations relating these unknowns; the six strain-displacement equations ([eqA.35]) and ([eqA.36]), and the six equilibrium equations ([eqA.81]) and ([eqA.85]). Therefore we need six more equations to get the number of unknowns equal to number of equations. The additional six equations come from the relations between the strains and the stresses, which express the material law. This relation between strains and stresses for different materials is established by material characterization tests on standard test specimens.
Solid bodies that can instantly recover their original size and shape when the forces producing the deformation are removed are called perfectly elastic. The elastic limit is defined as the greatest stress that can be applied without resulting in permanent strain on release of the stress. Elasticity is applicable to any body provided the stresses do not exceed the elastic limit. For many solid bodies there is a region where the stress is very nearly proportional to strain. The proportional limit is defined as the greatest stress for which the stress is still proportional to the strain. Both the elastic limit and proportional limit cannot be precisely determined from test data since they are defined by the limiting cases of no permanent deformation and no deviation from linearity. In practice the definition of the yield strength of a material is used to determine the limit of elastic behavior. See article [sec4.2] on page for a discussion on yield criteria.
First law of thermodynamics
The theoretical basis for an elastic material law is the first law of thermodynamics applied to an arbitrary infinitesimal rectangular parallelepiped isolated from the body. We assume the deformation process is adiabatic. That is, no heat is lost or gained in the body during the deformation. An alternate definition of elastic behavior is that the work expended in the transition from the reference state to the final deformed state is independent of the manner in which the process proceeds. The first law of thermodynamics states that work done on the rectangular parallelepiped is equal to the change in internal energy of the material contained in the parallelepiped. In elasticity the internal energy is called the strain energy. The incremental work of the tractions and body force acting on the parallelepiped are formulated in terms of incremental displacements from the equilibrium state. These incremental displacement functions are denoted by δui(x1,x2,x3), i=1,2,3, and to be kinematically admissible they are continuous and single-valued in the independent variables. In addition, functions δui are assumed to be infinitesimal in magnitude. The total displacement at point P is ˜ui=ui+δui, where ui are the displacements components in the equilibrium state.
R158.14pt
The distinction between δ\textit{u}1 and \textit{du}1.
In one dimension we define u1(x1) as the displacement function of a particle originally at coordinate x1 in the reference configuration of the body (region B0). The definition of incremental work necessitates consideration of the incremental displacement of a particle in the body. The distinction between δu1 and du1 is illustrated in figure [figA.9]. The incremental displacement δu1 is at a fixed value of the independent variable x1, and the differential du1 is the change in displacement with respect to the change in the independent variable x1. In the formulation of incremental work we interpret δu1 as the infinitesimal change in a the displacement of one particle identified by its coordinate in region B0. The interpretation of du1 is the relative displacement between two particles, one originally at x1+dx1 and the other originally at x1 in region B0. The symbol δ is called a variational operator, and the symbol d or the symbol ∂ are a differential operators.
For an adiabatic deformation process the first law of thermodynamics for the material in the rectangular parallelepiped of figure [figA.7] is ⇀T1Δx2Δx3∙δ⇀u|x1+Δx1+(−⇀T1Δx2Δx3∙δ⇀u)|x1+⇀T2Δx1Δx3∙δ⇀u|x2+Δx2+(−⇀T2Δx1Δx3∙δ⇀u)|x2+⇀T3Δx1Δx2∙δ⇀u|x3+Δx3+(−⇀T3Δx1Δx2∙δ⇀u)|x3+⇀BΔx1Δx2Δx3∙δ⇀u=δUΔx1Δx2Δx3,
where the variation in the displacement vector is δ⇀u=δu1ˆi1+δu2ˆi2+δu3ˆi3, and the variation the strain energy per unit volume, or the strain energy density, is δU. These variations are at fixed values of the independent variables xi. Expand the tractions acting on the faces of the rectangular parallelepiped at point P in a Taylor series keeping only those terms to the first degree in the differentials Δxi to get ∂(⇀T1∙δ⇀u)∂x1Δx1Δx2Δx3+∂(⇀T2∙δ⇀u)∂x2Δx1Δx2Δx3+∂(⇀T3∙δ⇀u)∂x3Δx1Δx2Δx3+⇀BΔx1Δx2Δx3∙δ⇀u=δUΔx1Δx2Δx3.
Divide eq. ([eqA.100]) by the volume Δx1Δx2Δx3 to get (∂⇀T1∂x1+∂⇀T2∂x2+∂⇀T3∂x3+⇀B)∙δ⇀u+⇀T1∙∂∂x1(δ⇀u)+T2∙∂∂x2(δ⇀u)+˙T3∙∂∂x3(δ⇀u)=δU.
The first term on the left-hand side vanishes via equilibrium eq. ([eqA.80]). Hence, ([eqA.101]) reduces to ⇀T1∙∂∂x1(δ⇀u)+⇀T2∙∂∂x2(δ⇀u)+⇀T3∙∂∂x3(δ⇀u)=δU.
Consider the term ⇀T1∙∂∂x1(δ⇀u)=[σ11ˆi1+σ12ˆi2+σ13ˆi3]∙[∂∂x1(δu1)ˆi1+∂∂x1(δu2)ˆi2+∂∂x1(δu3)ˆi3].
The variation in the derivative of a function is defined by δ(∂ui∂xj)≡∂˜ui∂xj−∂ui∂xj, i,j=1,2,3.
Substitute ˜ui=ui+δui into eq. ([eqA.104]) to get δ(∂ui∂xj)=∂∂xj(ui+δui)−∂ui∂xj=∂∂xj(δui).
Hence, the variation of the derivative of a function is equal to the derivative of the variation in the function. That is, δ(∂ui∂xj)=∂∂xj(δui).
Employing the result of eq. ([eqA.105]) in eq. ([eqA.103]) we write the latter as ⇀T1∙∂∂x1(δ⇀u)=[σ11ˆi1+σ12ˆi2+σ13ˆi3]∙[δ(∂u1∂x1)ˆi1+δ(∂u2∂x1)ˆi2+δ(∂u3∂x1)ˆi3]=σ11δ(∂u1∂x1)+σ12δ(∂u2∂x1)+σ13δ(∂u3∂x1).
Similarly the remaining terms on the left-hand side of eq. ([eqA.102]) can be evaluated as was done starting with eq. ([eqA.103]). The result is σ11δ(∂u1∂x1)+σ22δ(∂u2∂x2)+σ33δ(∂u3∂x3)∗+σ23[δ(∂u2∂x3)+δ(∂u3∂x2)]+σ31[δ(∂u1∂x3)+δ(∂u3∂x1)]+σ12[δ(∂u1∂x2)+δ(∂u2∂x1)]=δU.
The variations in the strains are determined from the linear strain-displacement relations ([eqA.35]) and ([eqA.36]) by letting ui→ui+δui, The variations in the strains are δε11=δ∂u1∂x1δε22=δ(∂u2∂x2)…δγ12=δ(∂u1∂x2)+δ(∂u2∂x1)
Thus eq. ([eqA.107]) is σ11δε11+σ22δε22+σ33δε33+σ23δγ23+σ31δγ31+σ12δγ12=δU.
In elasticity the strain energy density function is defined to be a function of the strain components; i.e., U=U(ε11,ε22,ε33,γ23,γ31,γ12). The strain energy density function depends on the physical properties of the material. The variation in the strain energy for the material in the rectangular parallelepiped as the strains are varied is determined from the series U(ε11+δε11,ε22+δε22,…,γ12+δγ12)=U(ε11,ε22,…,γ12)+∂U∂ε11δε11+∂U∂ε22δε22+∂U∂ε33δε33+⋯+∂U∂y12δγ12+H.O.T.,
where H.O.T. are higher order terms that contain quadratic and higher powers in the varied strain components. The change in strain energy is given by ΔU=U(ε11+δε11,ε22+δε22,…,γ12+δγ12)−U(ε11,ε22,…,γ12)=δU+H.O.T.
For infinitesimal variations in the strains ΔU∼δU, where δU=∂U∂ε11δε11+∂U∂ε22δε22+∂U∂ε33δε33+∂U∂γ23δγ23+∂U∂γ31δγ31+∂U∂γ12δγ12.
Compare eqs. ([eqA.108]) and ([eqA.109]) to identify σ11=∂U∂ε11σ22=∂U∂ε22σ33=∂U∂ε33σ23=∂U∂γ23σ31=∂U∂γ31σ12=∂U∂γ12.
For a reversible, adiabatic deformation process it is proved (Fung, 1965) that the strain energy density exists and has the property that its derivative with respect to a strain component equals the corresponding stress component as is illustrated by eq. ([eqA.110]).
To simplify further developments of the material law, we use the facts that the strain and stress matrices are symmetric. Introduce the following shorthand notation for the stresses and strains: σ1=σ11σ2=σ22σ3=σ33σ4=σ23σ5=σ31σ6=σ12, andε1=ε11ε2=ε22ε3=ε33ε4=γ23ε5=γ31ε6=γ12.
In the single subscript notation the stress and strain matrices are [σ]=[σ1σ6σ5σ6σ2σ4σ5σ4σ3], and [ε]=[ε1ε6/2ε5/2ε6/2ε2ε4/2ε5/2ε4/2ε3].
For the adiabatic condition we have shown that there is a scalar function U(εi), i=1,2,…,6, with the property σi=∂U∂εi
An elastic material is also defined by postulating that a scalar function exists such that its derivative with respect to a strain component determines the corresponding stress component. Now consider the stress-strain law in a general thermal environment, not limited by the adiabatic condition. Expand the strain energy density function in a power series of the strains. That is, U=6∑i=1Siεi+126∑i=16∑j=1Sijεiεj+…,
where coefficients Si and Sij are functions of the temperature. In eq. ([eqA.115]) the strain energy is assumed to vanish when all the strain components are zero. Employing the properties given in eq. ([eqA.114]) we get σk=∂U∂εk=Sk+6∑j=112(Skj+Sjk)εj, k=1,2,…,6.
Note that when all strain components equal zero, eq. ([eqA.116]) yields σk=Sk. Non-zero stresses can occur in a state of vanishing strains when there is a change in temperature. Let the change in temperature from the reference state be denoted by T−T0. Assume the linear relation Sk=−βk(T−T0),
where the βk are thermal moduli. For k=1 and k=2, eq. ([eqA.116]) expands to σ1=S1+S11ε1+12(S12+S21)ε2+⋯+12(S16+S61)ε6σ2=S2+12(S21+S12)ε1+S22ε2+⋯+12(S26+S62)ε6.
Clearly, 12(S12+S21)=12(S21+S12), so in eq. ([eqA.118]) we can take S12=S21 without changing the stress-strain relation. By implication Sij=Sji. The full expression for the linear elastic material law is [σ1σ2σ3σ4σ5σ6]=−[β1β2β3β4β5β6](T−T0)+[S11S12S13S14S15S16S21S22S23S24S25S26S31S32S33S34S35S36S41S42S43S44S45S46S51S52S53S54S55S56S61S62S63S64S65S66]⏟[S][ε1ε2ε3ε4ε5ε6].
The 6X6 elasticity matrix [S] is symmetric, which means there are twenty-one independent elastic constants, and there are six independent thermal moduli. Equation ([eqA.119]) is the material law for an anisotropic material where the number of independent elastic constants is determined by the existence of the strain energy density function and the symmetry of the strain and stress tensors. Equation ([eqA.119]) is called the Duhamel-Neumann form of Hooke’s law.
Material symmetry
Consider a monoclinic material for which the x1x2 plane at a point P is a plane of elastic symmetry. This means that the elastic constants at point P have the same values for a pair of Cartesian coordinate systems that are mirror images of one another in the elastic plane. The elastic constants Sij are invariant under the reflection coordinate transformation x′1=x1, x′2=x2, and x′3=−x3. The direction cosines matrix ([eqA.41]) for this reflection transformation is [λ]=[10001000−1].
Consider the material law for σ1 in the xi-system from eq. ([eqA.119]). It is σ1=−β1(T−T0)+S11ε1+S12ε2+S13ε3+S14ε4+S15ε5+S16ε6.
The material law for σ′1 in the x′i-system is written as σ′1=−β1(T−T0)+S11ε′1+S12ε′2+S13ε′3+S14ε′4+S15ε′5+S16ε′6.
From strain and stress transformations in eq. ([eqA.66]) and eq. ([eqA.95]), respectively, we find σ′i=σiε′i=εii=1,2,3,6σ′4=−σ4σ′5=−σ5ε′4=−ε4ε′5=−ε5.
Substitute the stress and strain transformation relations ([eqA.123]) into eq. ([eqA.122]) to find σ1=−β1(T−T0)+S11ε1+S12ε2+S13ε3−S14ε4−S15ε5+S16σ6.
Comparison of eq. ([eqA.124]) to eq. ([eqA.121]) requires that S14=0 and S15=0 if the material law for σ1 is to be the same with respect to the plane of elastic symmetry. Constructing the material law for σ′2 and following a similar procedure used for the σ′1 material law leads to S24=0 and S25=0. Constructing for the material law for σ′3 leads to S34=0 and S35=0, constructing the material law for σ′4 leads to β4=0 and S46=0, and constructing the material law for σ′5 leads to β5=0 and S56=0. The material law for the x1x2 plane of elastic symmetry is [σ1σ2σ3σ4σ5σ6]=−[β1β2β300β6](T−T0)+[S11S12S1300S16S21S22S2300S26S31S32S3300S36000S44S450000S54S550S61S62S6300S66][ε1ε2ε3ε4ε5ε6].
There are thirteen independent elastic constants Sij, and four independent thermal moduli. Equation ([eqA.125]) is the material law for a monoclinic material.
Certain elastic constants in eq. ([eqA.125]) vanish if in addition to the x1x2 plane of elastic symmetry the x2x3 plane is a plane of elastic symmetry. The reflection coordinate transformation is x′1=−x1, x′2=x2, and x′3=x3. The direction cosines matrix ([eqA.41]) for this reflection transformation is [λ]=[−100010001].
Consider the material law for σ1 in the xi-system from eq. ([eqA.125]). It is σ1=−β1(T−T0)+S11ε1+S12ε2+S13ε3+S16ε6.
The material law for σ′1 in the x′i-system is written as σ′1=−β1(T−T0)+S11ε′1+S12ε′2+S13ε′3+S16ε′6.
From strain and stress transformations in eq. ([eqA.66]) and eq. ([eqA.95]), respectively, we find σ′i=σ1ε′i=εii=1,2,3,4σ′5=−σ5σ′6=−σ6ε′5=−ε5ε′6=−ε6.
Substitute the transformations for the stress and strain from eq. ([eqA.129]) into eq. ([eqA.128]) to get σ1=−β1(T−T0)+S11ε1+S12ε2+S13ε3−S16ε6.
Comparison of eq. ([eqA.130]) and eq. ([eqA.127]) leads to S16=0. Also, following the same procedure for the equation starting with σ′2 leads to S26=0, and starting with the equation for σ′3 leads to S36=0. The material law for σ′4=S44ε′4+S45ε′5 transforms by eq. ([eqA.129]) to σ4=S44ε4−S45ε5. Hence S45=0. Finally, the material law for σ′6=−β6(T−T0)+S66ε′6 transforms to −σ6=−β6(T−T0)−S66ε6. Hence β6=0. The material law for two orthogonal elastic planes of symmetry is [σ1σ2σ3σ4σ5σ6]=−[β1β2β3000](T−T0)+[S11S12S13000S21S22S23000S31S32S33000000S44000000S55000000S66][ε1ε2ε3ε4ε5ε6].
If we additonally impose that the x3x1 plane is a plane of elastic symmetry, this reflection transformation does not change the results given in eq. ([eqA.131]). An orthotropic material has three mutually orthogonal planes of elastic symmetry, nine independent elastic constants Sij, and three independent thermal moduli. Equation ([eqA.131]) is the material law for an orthotropic material (e.g., wood).
The material properties are independent of direction for an isotropic material. Starting with the orthotropic material law ([eqA.131]) consider the following sequence of rotations from the (x1,x2,x3) coordinates to the (x′1,x′2,x′3) coordinates.
90∘ rotation about the x1-axis. The direction cosine matrix [λ]=[1000010−10].
90∘ rotation about the x3-axis. The direction cosine matrix [λ]=[010−100001].
45∘ rotation about the x3-axis. The direction cosine matrix [λ]=[1/√21/√20−1/√21/√20001].
For the material law to be invariant from the first rotation we find S12=S13β3=β2S33=S22S55=S66.
For the material law to be invariant from the second rotation we find β2=β1S12=S23S11=S22S44=S55.
For the third rotation, the material law for σ′1 in the (x′1,x′2,x′3) coordinates is σ′1=−β1(T−T0)+S11ε′1+S12ε′2+S12ε′3,
since S13=S12. The stress and the strain transformation relations are σ′1=(σ1+σ2)/2+σ6ε′1=(ε1+ε2+ε6)/2ε′2=(ε1+ε2−ε6)/2ε′3=ε3
Substitute the transformations in eq. ([eqA.135]) into eq. ([eqA.134]) to get (σ1+σ2)/2+σ6=−β1(T−T0)+12(S11+S12)ε1+12(S11+S12)ε2+S12ε3+12(S11−S12)ε6.
In the (x1,x2,x3) coordinates the formulation of the quantity (σ1+σ2)/2+σ6 is (σ1+σ2)/2+σ6=−β1(T−T0)+12(S11+S12)ε1+12(S12+S11)ε2+S12ε3+S44ε6.
For eqs. ([eqA.136]) and ([eqA.137]) to be identical we find 12(S11−S12)=S44.
For an isotropic material there are two independent elastic constants and one thermal modulus. Let S12=λ and S44=G, where λ and G are called Lame’s elastic constants. From eq. ([eqA.138]) S11=λ+2G. The isotropic material law is [σ1σ2σ3σ4σ5σ6]=−[βββ000](T−T0)+[λ+2Gλλ000λλ+2Gλ000λλλ+2G000000G000000G000000G][ε1ε2ε3ε5ε6].
Boresi (1965) gives the strain energy density for an isotropic material as U=λ(ε1+ε2+ε3)2/2+G(ε21+ε22+ε23+ε242+ε252+ε262)−β(ε1+ε2+ε3)(T−T0).
The isotropic material law ([eqA.139]) is retrieved from eq. ([eqA.140]) using the relation in eq. ([eqA.114]). The strain-stress form of eq. ([eqA.139]) is [ε1ε2ε3ε5ε6]=[C11C12C12000C12C11C12000C12C12C11000000G−1000000G−1000000G−1][σ1σ2σ3σ4σ5σ6]+[C11+C12+C12C12+C11+C12C12+C12+C11000]β(T−T0),
where C11=λ+G3λG+2G2C12=−λ2(3λG+2G2).
Consider a uniaxial tension test conducted on a circular cylindrical bar made of an isotropic, homogenous material at the reference state temperature. The test apparatus is configured such that the applied axial normal force divided by the cross-sectional area of the bar is equal to the normal stress σ1, and the remaining stresses σ2=σ3=σ4=σ5=σ6=0. The normal strains (ε1,ε2,ε3) are monitored and plotted with respect to the applied normal stress σ1. In the linear elastic range of the test data the following relationships are established: ε1=σ1/E, ε2=ε3=−νε1=−ν(σ1/E), where E denotes Young’s modulus, or the modulus of elasticity, and ν denotes Poisson’s ratio. The tension test results correspond to the first column of the 6X6 compliance matrix ([eqA.141]). Hence, C11=1/E and C12=−v/E. Substitute the values for C11 and C12 from the tensile test into eq. ([eqA.142]) to get λ+G3λG+2G2=1E−λ2(3λG+2G2)=−νE.
Solve eq. ([eqA.143]) to get Young’s modulus and Poisson’s ratio in terms of λ and G: E=3λG+2G2λ+Gν=λ2(λ+G).
It can be shown from the two expressions in eq. ([eqA.144]) that the Lame’s elastic constants in terms of E and v are G=E2(1+ν)λ=νE(1−2ν)(1+ν).
Also note that (C11+C12+C12)β=[(1−2ν)/E]β=α, where α is the linear coefficient of thermal expansion. In shear tests of an isotropic, homogeneous material, Lame’s elastic constant G is called the shear modulus of the material. In terms of engineering constants the material law for a homogenous and isotropic material is ε11=1E[σ11−vσ22−vσ33]+α(T−T0)ε22=1E[−vσ11+σ22−vσ33]+α(T−T0)ε33=1E[−vσ11−vσ22+σ33]+α(T−T0), and γ23=σ23/Gγ31=σ31/Gγ12=σ12/G.
The strain energy for an isotropic material given by eq. ([eqA.140]) is positive for any state of strain if E1+ν>0andE1−2ν>0.
To satisfy eq. ([eqA.147]) the range of Poisson’s ratio is −1<ν<1/2, and modulus E>0. However, materials with a negative Poisson’s ratio are theoretically possible but they are rare.
Summary and the boundary value problems of elasticity
At a point P:(x1,x2,x3) in the body, the unknown functions are
three displacements u1(x1,x2,x3), u2(x1,x2,x3), and u3(x1,x2,x3),
six strains ε11,ε22,ε33,γ23,γ31, and γ12,
nine stresses σ11,σ22,σ33,σ12,σ13,σ23,σ21,σ31, and σ32.
The total number of unknown functions is eighteen.
The number of equations are
The total number of equations is eighteen.
Let the boundary surface of region B0 be denoted by S. On the surface S we can prescribe the displacements and/or the tractions. In eq. ([eqA.77]) let ˆn be the unit outward normal to the surface S. We write eq. ([eqA.77]) in the form ⇀T(ˆn)=T(n)1ˆi1+T(n)2ˆi2+T(n)3ˆi3=⇀T1n1+⇀T2n2+⇀T3n3,
where T(n)1, T(n)2, and T(n)3 are the Cartesian components of the traction vector acting on the surface. We use eq. ([eqA.70]) to determine that these traction components are related to the stresses by T(n)j=n1σ1j+n2σ2j+n3σ3j,j=1,2,3.
On the portion of the surface where tractions are prescribed we have the boundary conditions T(n)j=n1σ1j+n2σ2j+n3σ3j=prescribed functions.
Boundary value problem 1: Determine the distribution of stress and displacement in the interior of an elastic body in equilibrium when the body forces Bi(xj) are prescribed and the distribution of the surface tractions T(n)j are prescribed.
Boundary value problem 2: Determine the distribution of stress and displacement in the interior of an elastic body in equilibrium when the body forces Bi(xj) are prescribed and the displacements ui of points on the surface of the body are prescribed functions.
Boundary value problem 3, or the mixed boundary value problem: Determine the distribution of the stress and displacement of an elastic body in equilibrium when body forces are prescribed, and the distribution of surface tractions are prescribed on surface Sσ and displacements are prescribed on surface Su. That is, surface S is separated into parts Sσ and Su.
In boundary value problem 1, the prescribed body forces and prescribed surface tractions must satisfy overall equilibrium of the body.
Batra, R., C. Elements of Continuum Mechanics. Reston, VA: American Institute of Aeronautics and Astronautics, Inc., 2006, p. 37.
Boresi, A. P. Elasticity in Engineering Mechanics. Englewood Cliffs, NJ: Prentice-Hall, Inc., 1965, p. 115.
Fung, Y. C. Foundations of Solid Mechanics. Englewood Cliffs, NJ: Prentice-Hall, Inc., 1965, pp. 347 & 348.