356 lines
5.2 KiB
Plaintext
356 lines
5.2 KiB
Plaintext
Compendium of vector analysis with applications to continuum mechanics
|
||
|
||
compiled by Valery P. Dmitriyev
|
||
Lomonosov University P.O.Box 160, Moscow 117574, Russia
|
||
e-mail: dmitr@cc.nifhi.ac.ru
|
||
|
||
1. Connection between integration and differentiation
|
||
|
||
Gauss-Ostrogradsky theorem
|
||
|
||
We transform the volume integral into a surface one:
|
||
|
||
+
|
||
|
||
| ( ) ∫∂iPdV
|
||
( ) V
|
||
|
||
= ∫∂iPdxidx jdxk V
|
||
|
||
=
|
||
|
||
∫ dx jdxk
|
||
S(V )
|
||
|
||
x
|
||
i −
|
||
x
|
||
i
|
||
|
||
x j ,xk x j ,xk
|
||
|
||
P
|
||
|
||
=
|
||
|
||
( ( ) ) ( ( ) ) =
|
||
|
||
∫ dx jdxk
|
||
S(V )
|
||
|
||
|
||
|
||
P
|
||
|
||
xi+
|
||
|
||
x j ,xk
|
||
|
||
,x j ,xk
|
||
|
||
− P xi−
|
||
|
||
x j ,xk
|
||
|
||
,x j ,xk
|
||
|
||
|
||
|
||
=
|
||
|
||
= ∫cos θe+xtdSP − ∫cosθi−ntdSP = ∫cos θextdSP = ∫n⋅eiPdS
|
||
|
||
S+
|
||
|
||
S−
|
||
|
||
S
|
||
|
||
Here the following denotations and relations were used:
|
||
( ) P is a multivariate function P xi,x j ,xk , ∂i = ∂ / ∂xi , V volume,
|
||
S surface, ei a basis vector, ei ⋅e j = /ij , n the external normal to the element dS of closed surface with
|
||
|
||
dx jdxk = n⋅ei dS , n⋅ei = cosθ .
|
||
|
||
Thus
|
||
|
||
∫ ∂ i PdV
|
||
V
|
||
|
||
= S(∫V )Pn⋅eidS
|
||
|
||
(1.1)
|
||
|
||
Using formula (1.1), the definitions below can be transformed into coordinate representation.
|
||
|
||
1
|
||
|
||
Gradient
|
||
|
||
( ) S(∫VP) ndS
|
||
|
||
=
|
||
|
||
∫ n⋅ei
|
||
S(V )
|
||
|
||
ei PdS
|
||
|
||
= ∫∂iPeidV V
|
||
|
||
where summation over recurrent index is implied throughout. By definition
|
||
|
||
gradP = ∇P = ∂iPei
|
||
|
||
Divergence
|
||
|
||
( ) S(∫VA)
|
||
|
||
⋅
|
||
|
||
ndS
|
||
|
||
=
|
||
|
||
S
|
||
|
||
∫
|
||
(V
|
||
|
||
n⋅ei
|
||
)
|
||
|
||
AidS = ∫∂i AidV V
|
||
|
||
(1.2)
|
||
|
||
By definition
|
||
|
||
di#A = ∇⋅A = ∂i Ai
|
||
|
||
Curl
|
||
|
||
( ) ∫n× AdS = ∫ n⋅ei ei × Aje jdS = ∫∂i Ajei×e jdV (1.3)
|
||
|
||
S(V )
|
||
|
||
S(V )
|
||
|
||
V
|
||
|
||
By definition
|
||
|
||
curlA = ∇× A = ∂i Ajei ×e j
|
||
|
||
Stokes theorem follows from (1.3) if we take for the volume a right cylinder
|
||
|
||
with the height h → 0 . Then the surface integrals over the top and bottom areas
|
||
|
||
mutually compensate each other. Next we consider the triad of orthogonal unit
|
||
|
||
vectors
|
||
|
||
m, n, 2
|
||
|
||
where m is the normal to the top base and n the normal to the lateral face
|
||
|
||
2 =m×n
|
||
|
||
Multiplying the left-hand side of (1.3) by m gives
|
||
|
||
∫m⋅n×AdS = ∫(m×n)⋅AdS = ∫ 2 ⋅AdS = h∫ 2⋅Adl
|
||
|
||
lateral
|
||
|
||
lateral
|
||
|
||
lateral
|
||
|
||
l
|
||
|
||
where 2 is the tangent to the line. Multiplying the right-hand side of (1.3) by m gives
|
||
|
||
h∫m⋅ curlAdS
|
||
|
||
S
|
||
|
||
where m is the normal to the surface. Now, equating both sides, we come to the
|
||
|
||
formula sought for
|
||
|
||
∫ 2⋅Adl = ∫m⋅ curlAdS
|
||
|
||
l
|
||
|
||
S
|
||
|
||
The Stokes theorem is easily generalized to a nonplanar surface (applying to it
|
||
|
||
Ampere's theorem). In this event, the surface is approximated by a polytope.
|
||
|
||
Then mutual compensation of the line integrals on common borders is used.
|
||
|
||
2
|
||
|
||
2. Elements of continuum mechanics
|
||
( ) A medium is characterized by the volume density ρ x,t and the flow ( ) velocity u x,t .
|
||
|
||
Continuity equation
|
||
|
||
The mass balance in a closed volume is given by
|
||
|
||
∂t ∫ ρdV + ∫ ρu⋅ndS = 0
|
||
|
||
V
|
||
|
||
S(V )
|
||
|
||
where ∂t = ∂ / ∂t . We get from (1.2)
|
||
∫ ρu⋅ndS = ∫∂i(ρui )dV
|
||
|
||
Thereof the continuity equations follows
|
||
( ) ∂t ρ + ∂i ρui = 0
|
||
Stress tensor
|
||
We consider the force df on the element dS of surface in the medium and are interested in its dependence on normal n to the surface
|
||
df (n)
|
||
where
|
||
df(−n)= −df(n)
|
||
|
||
With this purpose the total force on a closed surface is calculated. We have for the force equilibrium at the coordinate tetrahedron
|
||
df (n)+ df (n1)+ df(n2)+ df (n3)= 0
|
||
|
||
where the normals are taken to be external to the surface
|
||
|
||
( ) ( ) ( ) n1 = −sign n⋅e1 e1 , n2 = −sign n⋅e2 e2 , n3 = −sign n⋅e3 e3
|
||
|
||
Thence
|
||
|
||
df (n)= sign(n⋅e j )df (e j )
|
||
|
||
(2.1)
|
||
|
||
3
|
||
|
||
( ) The force density 1 n is defined by
|
||
|
||
Insofar as
|
||
|
||
df = 1dS
|
||
|
||
we have for (2.1)
|
||
|
||
dS j= n⋅e j dS
|
||
|
||
df (n)= sign(n⋅e j )1(e j )dS j = sign(n⋅e j ) n⋅e j 1(e j )dS = n⋅e j1(e j )dS
|
||
|
||
i.e.
|
||
1(n)= n⋅e j1(e j )
|
||
|
||
( ) = n⋅e jei1i e j
|
||
|
||
( ) The latter means that 1 n possesses the tensor property. The elements of the stress
|
||
|
||
tensor are defined by
|
||
|
||
( ) σij =σi e j
|
||
|
||
Now, using (1.2), the force on a closed surface can be computed as a volume integral
|
||
|
||
( ) ∫1ndS = ∫1e j e j ⋅ndS = ∫∂ j1 e j dV
|
||
|
||
(2.2)
|
||
|
||
V
|
||
|
||
Euler equation
|
||
|
||
The momentum balance is given by the relation
|
||
|
||
∂t ∫ ρudV + ∫(ρu)u⋅ndS = ∫1dS
|
||
|
||
(2.3)
|
||
|
||
V
|
||
|
||
S(V )
|
||
|
||
S(V )
|
||
|
||
We have for the second term by (1.2)
|
||
( ) ∫(ρu)u⋅ndS = ∫(ρu)u je j ⋅ndS = ∫∂ j ρu ju dV
|
||
|
||
Using also (2.2) gives for (2.3)
|
||
|
||
∂t (ρu)+ ∂ j (ρu ju)= ∂ j1(e j )
|
||
|
||
or
|
||
|
||
( ) ρ∂tu + ρu j∂ ju = ∂ j1 e j
|
||
|
||
(2.4)
|
||
|
||
4
|
||
|
||
Hydrodynamics
|
||
|
||
The stress tensor in a fluid is defined from the pressure as
|
||
|
||
That gives for (2.4)
|
||
|
||
σ ij = − pδij ρ∂tui + ρu j∂ jui + ∂ j p = 0
|
||
|
||
Elasticity
|
||
|
||
( ) The solid-like medium is characterized by the displacement s x,t . For small
|
||
|
||
displacements
|
||
|
||
u=∂ts
|
||
|
||
and the quadratic terms in the left-hand part of (2.4) can be dropped. For an isotropic homogeneous medium the stress tensor is determined from the Hooke's law as
|
||
( ) ( ) σ i e j = λδij∂k sk + µ ∂is j +∂ jsi
|
||
|
||
where λ and µ are the elastic constants. That gives
|
||
|
||
( ) ( ) ( ) ∂ jσi e j
|
||
|
||
= λ∂i∂k sk + µ ∂i∂ js j +∂2j si
|
||
|
||
=
|
||
|
||
λ+µ
|
||
|
||
∂i∂
|
||
|
||
j
|
||
|
||
s
|
||
|
||
j
|
||
|
||
+
|
||
|
||
µ∂
|
||
|
||
2 j
|
||
|
||
si
|
||
|
||
and
|
||
∂ j1(e j )= (λ+µ)graddi# s + µ∇2s
|
||
|
||
= (λ+2µ)∇2s + (λ+µ)curlcurls
|
||
|
||
= λ graddi# s − µ curlcurls
|
||
|
||
where graddi# = ∇2 + curlcurl was used. Substituting it to (2.4) we get finally
|
||
|
||
Lame equation
|
||
|
||
( ) ρ∂t2s = λ+µ graddi# s + µ∇2s
|
||
|
||
where ρ is constant.
|
||
|
||
5
|
||
|