Some elements of nonlinear elasticity of growing materials

Michael H. Köpf

January 8, 2015
A very good introduction to nonlinear elasticity can be found in the book by Fu and Ogden [FO01]. A necessary starter to dive into elasticity theory is certainly the classical textbook by Landau and Lifshitz [LL70].

In the following, the component notation ci = Aijbj with Einstein summation over repeated indices is preferred over the notation c = Ab, because the latter has to be translated to the former for every actual calculation, anyway.

1 Basic definitions

The elastic deformation can quantified by comparison of the deformed state B to a reference state A. We denote the coordinates in the reference frame A by xi(A) and use Xi(B) for the coordinates in the current state B1 . Further, we introduce a mapping F that relates the coordinates of identical material points in A to the corresponding coordinates in B (see Fig. 1).


PIC

Figure 1: We distinguish two configurations: The reference state A and the elastically deformed state B. The mapping F relates both configurations to each other.


In order to simplify our notation, we introduce the shorthands i(A) := ∂∕∂xi(A) and i(B) := ∂∕∂xi(B) for the partial derivatives with respect to the coordinates in both frames. We define the deformation gradient tensor F by

Fij = ∂(A )x(B)
      j  i
(1)

and introduce the symbol JF = |F| for its determinant. The deformation of the solid is quantified by the strain ω defined as 2

ω  = F  F  - δ .
 ij   ki kj   ij
(2)

Nanson’s formula In order to relate the oriented area element dAi(A) in frame A to dAi(B) in frame B, we consider infinitesimal volume elements in both frames A and B and use dxi(B) = Fijdxj(A):

   (B)       (A)       (A)  (A )
dV   = JF dV   = JF dAi  dxi
(3)

and

  (B)     (B)  (B)    (B)    (A)
dV   = dA i dxi  = dAi  Fijdxj  .
(4)

From these two formulas we conclude

JFdAi(A) = dA k(B)F ki
JF( - 1)
 F ildAi(A) = dA l(B)
JF(    )
 F- 1 ilni(A)dA(A) = n i(B)dA(B). (5)
Analogously, we find
 (A )  (A )   -1    (B)  (B)
ni dA    = JF Fkink dA   .
(6)

Helpful relations A few non-obvious identities can be found using Nanson’s formula and very simple geometric arguments. Consider an arbitrary closed surface δω of some volume ω in the reference frame A.

   ∫         (A )  ∫              (B)
0 =    dA(A)ni  =    dA (B)J-F1Fkink
    ∂ω             ∂Ω
(7)

Since we can choose the surface ∂ω arbitrarily, we find

 (B)( -1  )
∂k   JF Fki = 0.
(8)

In other words: in the sum k(B) and JF-1Fki commute. Analogously, we can show that

 (A)(   ( -1) )
∂k   JF F    ki = 0.
(9)

Three different stress tensors The surface force per unit area on a vector area element dA(B) in the current frame B is expressed by the Cauchy stress Σij. Using relation (5) we can translate this to a force per unit area on a vector area element dA(A) in the reference frame A.

    (B)     ( -1)      (A)  (A)      (A )  (A )
Σijnj  = JF--F- -kjΣijnk  dA   = σijnj dA   ,
         ◟   =:◝ σ◜ik   ◞
(10)

where we have defined the first Piola-Kirchhoff stress σ. The inverse relation is

Σij = J-F1Fjkσik.
(11)

Note, that unlike the Cauchy stress Σ, the first Piola-Kirchhoff stress σ is not symmetric.

Both ΣdA(B) and σdA(A) are force vectors in the current frame B. Using F, we can map it to the reference configuration A and obtain

( -1)
 F ijΣjknk(B)dA(B) =    ( -1)  ( -1)
J◟F--F---ij◝F◜----lk-Σkj◞=: silnl(A)dA(A)
= ( - 1)
 F---ijσjk
◟   ◝◜   ◞=: siknk(A)dA(A)
= sijnj(A)dA(A) (12)
where we defined the second Piola-Kirchhoff stress s. The inverse relations are
Σij = J -1FilFjkskl and  σij = Fikskj.
      F
(13)

Note, that j(B)Σij on the one hand and j(A)σij and jsij on the other hand are force densities with respect to different volumina and therefore differ by a factor of JF. This can be easily seen by noting that

        ∫
F (tiot) =   dV (B)∂(jA)σij
         ω
(14)

and

       ∫               ∫
F(itot)=    dV(B)∂j(A)Σij =   dV(B)JF∂(jB)σij.
        Ω               ω
(15)

Relation of the Piola-Kirchhoff stresses to the elastic free energy In Hookean elasticity, where, the Cauchy stress is the derivative of the elastic free energy density per unit volume Lel with respect to the linear strain ωij(lin) = ui,j + uj,i:

Σij = 2-∂L(eliln).
       ∂ωij
(16)

In nonlinear elasticity, the second Piola-Kirchhoff stress takes the role of the Cauchy stress:

      ∂L
sij = 2--el.
      ∂ ωij
(17)

The first Piola-Kirchhoff stress is given by the derivative of Lel with respect to the deformation gradient tensor Fij:

σij = ∂Lel
∂Fij = ∂Lel
∂ωkl∂ωkl
∂Fij = 1
2skl(Fikδlj + Filδkj) = 1
2(skj + sjk)Fik
= Fikskj. (18)
In the last step, we have used the symmetry of the second Piola-Kirchhoff stress s.

2 Elasticity and growth

When the considered material is growing, for example because it is a biological tissue consisting of proliferating cells, one has to seperate the deformation due to growth from elastic deformation. This is basically the distinction between growth of a spring (that might be caused by uniform thermal expansion) versus extension of the spring by external forces. To make this distinction quantitative, seperate the deformation gradient tensor F into a product of the two tensors G and A as proposed by Rodriguez et al. [RHM94] yielding

Fij = AikGkj,
(19)

where the growth tensor G describes the deformation due to growth and A describes the elastic part of the deformation3 . We can then distinguish not only between the reference state A and the current deformed state B, but also define the virtual state V, that describes the grown but otherwise undeformed material (see Fig. 2).


PIC

Figure 2: If we have growth and elasticity, we distinguish three configurations: The reference state A, the virtual configuration of a stress-free grown state V, and the actual grown and elastically deformed state B. The mappings G, A, and F relate the three configurations to each other.


Analogously to the case without growth, we use the shorthands i(A) := ∂∕∂xi(A), i(B) := ∂∕∂xi(B), and i(V) := ∂∕∂xi(V) for the partial derivatives with respect to the different coordinate frames and define JF = |F|, JG = |G|, and JA = |A|. Elastic deformation is deformation from V to B, only. Therefore, we can for the moment forget about the frame A and consider V the reference state. We can then define the virtual Piola-Kirchhoff stress σ(V) that translates the force per unit area in the frame B to a force per unit area in the frame V as

 (V)  -∂Lel
σ   = ∂Aij .
(20)

Since the physical force is the same, no matter which frame we choose, the following two equalities have to hold:

Σijn(Bj)dA (B) = σijn(jA)dA(A) = σ(iVj) n (Vj)dA(V).
(21)

Using the transformation of the spatial derivative4 j(V) = (G-1)kjk(A) one can use Nanson’s formula (5) applied to V, B, and G to find

 (V)  (V)     (  -1)  (A)  (A)
ni dA    = JG G    ijnj  dA   .
(22)

When we insert this result into eq. (21), we obtain

   (V)( -1)   (A)  (A)      (A )  (A)
J◟Gσij◝◜G---kj◞nk  dA   = σijnj dA   ,
    = σik
(23)

and have thus established the relation between the Piola-Kirchhoff stresses σij and σij(V).

References

[Cow04]    Stephen C. Cowin. Tissue growth and remodeling. Annu. Rev. Biomed. Eng., 6(1):77–107, July 2004.

[FO01]    Y. B. Fu and R. W. Ogden. Nonlinear Elasticity: Theory and Applications. Cambridge University Press, Cambridge, 2001.

[LL70]    L. D. Landau and E. M. Lifshitz. Theory of Elasticity. Pergamon Press, New York, 2nd edition, 1970.

[RHM94]   Edward K. Rodriguez, Anne Hoger, and Andrew D. McCulloch. Stress-dependent finite growth in soft elastic tissues. Journal of Biomechanics, 27(4):455–467, April 1994.