Appendixes
Appendix 1: Projection of Area
Consider the projection of the area having the unit normal vector n onto the surface having the normal vector m in Fig. A.1. Now, suppose the plane (abcd in Fig. A.1) which contains the unit normal vectors m and n. Then, consider the line ef obtained by cutting the area having the unit normal vector n by this plane. Further, divide the area having the unit normal vector n to the narrow bands perpendicular to this line and their projections onto the surface having the normal vector m. The lengths of projected bands are same as the those of the original bands but the projected width db are obtained by multiplying the scalar product of the unit normal vectors, i.e. m • n to the original widths db . Eventually, the projected area da is related to the original area da as follows:
da = m • nda (A.1)
a e
n b
db f m d
db= nm• db c
Fig. A.1 Projection of area
422 Appendixes
Appendix 2: Proof of ∂(FjA/J)/∂x j = 0
∂ FjA 1 ∂(∂x /∂X ) ∂x ∂J J = j A J − j ∂x J2 ∂x ∂X ∂x j ⎧ j A j ⎫ ∂ ∂ ∂ ⎨ ∂ε x1 x2 x3 ⎬ ∂(∂x /∂X ) ∂ ∂ ∂ ∂x PQR ∂X ∂X ∂X = 1 j A ε x1 x2 x3 − j P Q R 2 PQR J ⎩ ∂x j ∂XP ∂XQ ∂XR ∂XA ∂x j ⎭ ∂ 2 ∂ 2 ∂ 2 ∂ ∂ ∂ = 1 x1 + x2 + x3 ε x1 x2 x3 J2 ∂X ∂x ∂X ∂x ∂X ∂x PQR ∂X ∂X ∂X A 1 A 2 A 3 P Q R ∂ ∂ 2 ∂ ∂ ∂ ∂ ∂ 2 ∂ − ε x1 x1 x2 x3 + x2 x1 x2 x3 PQR ∂X ∂X ∂x ∂X ∂X ∂X ∂X ∂X ∂x ∂X A P 1 Q R A P Q 2 R ∂ ∂ ∂ ∂ 2 + x3 x1 x2 x3 ∂X ∂X ∂X ∂X ∂x A P Q R 3 ∂ 2 ∂ ∂ ∂ ∂ ∂ 2 ∂ ∂ = 1 ε x1 x1 x2 x3 + x1 x2 x2 x3 J2 PQR ∂X ∂x ∂X ∂X ∂X ∂X ∂X ∂x ∂X ∂X A 1 P Q R P A 2 Q R ∂ ∂ ∂ 2 ∂ + x1 x2 x3 x3 ∂X ∂X ∂X ∂x ∂X P Q A 3 R ∂ 2 ∂ ∂ ∂ ∂ ∂ 2 ∂ ∂ − x1 x1 x2 x3 + x2 x2 x1 x3 ∂X ∂x ∂X ∂X ∂X ∂X ∂X ∂x ∂X ∂X P 1 A Q R A Q 2 P R ∂ ∂ ∂ 2 ∂ + x1 x2 x3 x3 ∂X ∂X ∂X ∂x ∂X P Q R 3 A ∂ 2 ∂ ∂ ∂ ∂ 2 ∂ ∂ ∂ ∂ 2 = 1 ε x1 x2 x3 + x1 x2 x3 + x1 x2 x3 J2 PQR ∂X ∂X ∂X ∂X ∂X ∂X ∂X ∂X ∂X ∂X ∂X ∂X A P Q R P A Q R P Q A R ∂ 2x ∂x ∂x ∂x ∂ 2x ∂x ∂x ∂x ∂ 2x − 1 2 3 + 1 2 3 + 1 2 3 = 0 ∂XP∂XA ∂XQ ∂XR ∂XP ∂XQ∂XA ∂XR ∂XP ∂XQ ∂XR∂XA (A.2)
Appendix 3: Covariant and Contravariant Base Vectors and Components
Consider the general curvilinear coordinate system (θ 1,θ 2,θ 3) with the primary base i {ai} and the locally defined coordinate system (θ1,θ2,θ3) with the reciprocal base {a }. The infinitesimal line-element dx is described from Eq. (1.35) in these coordinate sys- tems as follows: ⎧ ⎪ ∂x ⎨ dθ i = dθ ia ∂θi i dx = (A.3) ⎩⎪ i dθ ai Appendixes 423 with i i dθ = dx • a , dθi = dx • ai (A.4)
∂x ∂θi a = , ai = (A.5) i ∂θi ∂x which satisfies Eq. (1.34), i.e.
j ∂x ∂θ j a • a j = • = δ (A.6) i ∂θi ∂x i ∗ ∗ ∗ Now, consider the another curvilinear coordinate system (θ 1,θ 2,θ 3) with the ∗ ∗ ∗ ∗ base {a } and the locally defined coordinate system (θ ,θ ,θ ) with the reciprocal ∗i 1 2 3 base {a i}. The following coordinate transformation rules hold for the base vectors by the chain rule of differentiation. ∗ ∗ ∗ ∂x ∂x ∂θ j ∂θ j ∂θ j ∗ ∂θ j ∂θr a = = = a , a = a = a = δ ra (A.7) i ∂θ∗i ∂θ j ∂θ∗i ∂θ∗i j i ∂θi j ∂θi ∂θ∗ j r i r
∗ ∗ ∗ ∂θ i ∂θi ∗ ∂θi ∂θ j a i = a j, ai = a j = ar = δ iar (A.8) ∂θ j ∂θ∗ j ∂θ∗ j ∂θr r The vector v is described by ⎧ j ⎪ ∗ ∗ ∗i ∂θ ⎪ via = v ia = v a ⎨ i i ∂θ∗i j v = (A.9) ⎪ ∗ ⎩⎪ ∗ ∗ ∗ ∂θ i v ai = v a i = v a j i i i ∂θ j from which one has the transformation rules: ⎫ i ∗i ∂θ ∗ ∗ ∂θ ⎪ vi = v j, v i = v j ⎪ ∂θ∗ j ∂θ j ⎬ (A.10) ∗ j j ⎪ ∂θ ∗ ∗ ∂θ ⎪ v = v , v = v ⎭ i ∂θi j i ∂θ∗i j
In the analogous way, one has the following transformation rules for the second-order tensor. ⎫ ∗i ∗ j i j ∗ ∂θ ∂θ ∂θ ∂θ ∗rs ⎪ t ij = trs, tij = t ⎪ ∂θr ∂θs ∂θ∗r ∂θ∗s ⎪ ⎪ ⎪ ∗i s i ∗s ⎪ ∗ ∂θ ∂θ ∂θ ∂θ ∗ ⎪ t i = tr , ti = t r ⎪ • j ∂θr ∂θ∗ j •s • j ∂θ∗r ∂θ j •s ⎬ (A.11) r ∗ j ∗r j ⎪ ∗• j ∂θ ∂θ • • j ∂θ ∂θ ∗ • s ⎪ i = s, = r ⎪ t ∗i s Tr ti i ∗s t ⎪ ∂θ ∂θ ∂θ ∂θ ⎪ ⎪ ∗ ∗ ⎪ ∗ ∂θr ∂θs ∂θ r ∂θ s ∗ ⎭⎪ t = t , t = t ij ∂θ∗i ∂θ∗ j rs ij ∂θi ∂θ j rs 424 Appendixes
It can be recognized from Eqs. (A.7), (A.8), (A.10) and (A.11) that the transforma- tion rules of the base vectors and the components of vector and tensor are classified into the type with the subscript and the other type with the superscript. The partial-derivative ∗ operator ∂θi or ∂θ i is placed in the denominator and the numerator for the base vec- i tor or component denoted by ()i and () , respectively, in their transformation rules. The former and the latter are referred to as the “covariant”andthe“contravariant”, respectively, base vector, component and description of vector and tensor.
Appendix 4: Euler’s Theorem for Homogeneous Function
The homogeneous function of degree n is defined to fulfill the relation
n f (ax1, ax2, ···, axm)=a f (x1, x2, ···, xm) (A.12) for the variables x1, x2,···, xm, letting a denote an arbitrary scalar constant. Then, con- sider the homogeneous function given by the polynomial expression:
s ni ni i ( , , ···, )= 1 2 ··· nm f x1 x2 xm ∑ ci x1 x2 xm (A.13) i=1 where s is the number of terms of polynomial expression and ci are constants, provided to fulfill m i = ∑ n j n for each i (A.14) j=1 Eq. (A.13) leads to
m s m s ∂ f (x , x , ···, x ) ni ni ni −1 i ni ni i 1 2 m x = c ni x 1 x 2 ···x j ···xnm x = n c x 1 x 2 ···xnm ∑ ∂ j ∑ ∑ i j 1 2 j m j ∑ i 1 2 m j=1 x j i=1 j=1 i=1
Then, it holds that m ∂ ( , , ···, ) f x1 x2 xm = ( , , ···, ) ∑ ∂ x j nf x1 x2 xm (A.15) j=1 x j which is called the Euler’s theorem for homogeneous function. For the simple example (m =3,n =4,s =3):
f (x, y, z)=αx4 + βx3y + γx2yz
Eq. (A.15) is confirmed as follows:
∂ f ∂ f ∂ f x + y + z =(4αx3 + 3βx2y + 2γxyz)x +(βx3 + γx2z)y + γx2y • z = 4 f ∂x ∂y ∂z Eq. (A.15) yields Eq. (6.33) for the yield function (n =1). Appendixes 425
Appendix 5: Normal Vector of Surface
The quantity (∂ f (t)/∂t): dt is regarded as the scalar product of the vectors ∂ f (t)/∂t and dt in the nine-dimensional space (t11,t12,t33,···,t31,t13) . Here, it holds that ⎧ ⎨ > 0: dt is directed outward-normal to surface ∂ f (t) : dt = 0: dt is directed tangential to surface (A.16) ∂t ⎩ < 0: dt is directed inward-normal to surface Therefore, ∂ f (t)/∂t is interpreted to be the vector designating the outward-normal of the surface. This fact holds also for the yield surface f (σ )=F.
Appendix 6: Relationships of Material Constants in lnv − ln p and e − ln p Linear Relations
The following relation holds from Eqs. (2.131), (11.3) and (11.14) for pe = 0, provided εe = ( + εe) that Eq. (2.131) holds for elastic volumetric strain, i.e. v ln 1 v .
p κ p −κ˜ ln = ln 1 − ln (A.17) p0 1 + e0 p0 from which one has κ ( − p ) ln 1 + ln κ = 1 e0 p0 ˜ p (A.18) −ln p0 ∼ It follows from Eq. (A.18) for infinitesimal deformation under p = p0 that κ − 1 1 + e0 p κ p κ p ln(1 − ln ) 1 − ln 1 + e p 1 + e p κ lim κ˜ = lim 0 0 = lim 0 0 = (A.19) → → p → + p p0 p p0 −ln p p0 − 1 1 e0 p0 p resulting in κ κ˜ =∼ (A.20) 1 + e0
Further, substituting Eqs. (11.3) and (11.4) into Eq. (2.131), i.e. εv = ln(1 + εv),itfol- lows that
p p κ p κ p λ p −κ˜ ln − (λ˜ − κ˜)ln y = ln 1 − ln + ln y − ln y p0 py0 1 + e0 p0 1 + e0 py0 1 + e0 py0 426 Appendixes i.e. p p κ p p λ p κ˜ y − − + y − − y ln ln ln 1 + ln ln + ln λ˜ = py0 p0 1 e0 py0 p0 1 e0 py0 p ln y py0 from which one has λ λ˜ = lim + (A.21) p → p0 1 e0 py → py0
Based on Eqs. (A.19) and (A.21), λ˜ and κ˜ may be given by λ κ λ˜ = , κ˜ = (A.22) 1 + e0 1 + e0 which can be calculated from a plenty of data on λ and κ accumulated in the past. Then, the analysis would be improved over the finite deformation by using Eq. (11.4) with Eq. (A.22) instead of Eq. (11.15) or (11.17). Obviously, it is more appropriate to find the relations of λ˜ and κ˜ to λ and κ which are applicable over the whole range of pressure in relevant analysis. Needless to say, one has to determine the material parameters λ˜ and κ˜ directly from test data for soils without the data of λ and κ or for the case that an accurate formulation is required. Here, note that the curve fitting of lnv − ln p linear relation to test data is easier than the fitting of the e−ln p linear relation to test data because real soil behavior is far nearer to the former than the latter.
Appendix 7: Derivation of Eq. (11.22)
Differentiation of Eq. (11.18) under the condition f (σ ) = const. leads to
∂η ∂η (η ) + (η ) m + m σ g m dp pg m ∂ dp ∂σ d P σ 1 = g(η )dp+ pg(η ) − dp+ dσ = 0 m m p2M pM from it holds that σ g(η ) − g (η ) dσ m pM m g(η ) = = M m − η (A.23) dp 1 g(η ) m g(η ) m M m Considering dσ /dp =0atηm = 1 in Eq. (A.23), one has Eq. (11.22). Appendixes 427
Appendix 8: Convexity of Two-Dimensional Curve
When the curve is described by the polar coordinates (r, θ) asshowninFig.A.2,the following relation holds rdθ tanα = (A.24) dr where α is the angle measured from the radius vector to the tangent line in the anti- clockwise direction. Eq. (A.24) is rewritten as
r cotα = (A.25) r where ( ) designates the first order differentiation with respect to θ. The equation of the tangent line at (r, θ) of the curve r = r(θ) is described by the following equation by using the current coordinates (R, Θ) on the tangent line.
Rcos[Θ −{θ − (π/2 − α)}]=rcos(π/2 − α) which is rewritten as 1 1 −Rsin(Θ − θ − α)=rsinα → = − Rsin(Θ − θ − α) rsinα
1 1 1 → = cos(Θ − θ) − cotα sin(Θ − θ) R r r Substituting Eq. (A.25) to this equation and noting (1/r) = −r/r2, one has the relation
Curve y α
r r R d d α rd θ r dθ T
a θ n d g r e n
t
l i α n − e π/2 θ Θ Θθπ−−{(/2)} −α 0 x
Fig. A.2 Curve in the polar coordinate (r, θ) 428 Appendixes
1 1 1 = cos(Θ − θ)+ sin(Θ − θ), (A.26) R r r
Equation (A.26) is rewritten by applying the Taylor expansion to cosϑ and sinϑ as
1 1 1 1 1 = cosϑ + sinϑ = 1 − ϑ 2 + ··· R(Θ) r(θ) r(θ) r(θ) 2 + 1 ϑ − 1ϑ 3 + ··· r(θ) 6 1 1 1 1 = + ϑ − ϑ 2 + ··· (A.27) r(θ) r(θ) 2 r(θ) where ϑ ≡ Θ − θ. On the other hand, the radius r(Θ) (Θ = θ + ϑ)ofthecurveis described by the Taylor expansion as follows:
1 1 1 1 1 = +( ) ϑ + ϑ 2 + ··· (A.28) r(Θ) r(θ) r(θ) 2 r(θ)
Eqs. (A.27) and (A.28) lead to