Issue 
A&A
Volume 519, September 2010



Article Number  A75  
Number of page(s)  16  
Section  Planets and planetary systems  
DOI  https://doi.org/10.1051/00046361/201014281  
Published online  15 September 2010 
Online Material
Appendix A: Solution of the DuhamelNeumann equation
We outline the main steps needed to solve the DuhamelNeumann equation in Eq. (18) for our work. The temperature field, which produces the thermal stresses, is assumed to have a linearized form where is given by Eq. (9). The uniqueness of the solution arises from (i) the regularity in the whole volume; and (ii) matching the free boundary conditions (given by Eq. (19)) at the surface r = R.
There are different ways in which we can decompose the
displacement vector
into sphericalharmonicstype
expansion (see, e.g., Thorne 1980, for an insightful review).
Here we use a decomposition into spheroidal and toroidal components traditionally used in
geophysical analyses (e.g., Kaula 1968; Bullen 1975). With this
approach, related to what Thorne (1980) calls purespin vector
harmonics, we have
with
The first component in Eq. (A.1), , is actually two separate spheroidal terms, that we call S1 and S2 in Sect. 3.1.2, characterized with radialprofile amplitudes U_{nk}(r) and V_{nk}(r). The second component in Eq. (A.1), , is the toroidal component.
The spheroidal character of our source (temperature) term the DuhamelNeumann equation implies two simplifications. First, the toroidal part of the displacement vector becomes negligible and we have W_{nk}=0. Second, we can restrict the summation over degrees n in Eq. (A.1) to the dipole and higherorder terms only, ignoring the monopole n=0. This is because the monopole part would correspond to purely radial temperature field, such as has been considered, for instance, in the previous works on our topic (e.g., Tambovtseva & Shestakova 1999; Shestakova & Tambovtseva 1997; Kuehrt 1984). Our temperature representation does not contain a nontrivial, purely radial profile^{} and the only viable free monopole term must have U_{00}=W_{00}=0to match the boundary conditions. Finally, we note that we also anticipated the Fourierdevelopment structure in Eq. (A.1) as it follows from the source ( development).
Substituting the spheroidalvector representation of into the DuhamelNeumann Eq. (18) we obtain the following system of equations for the radial profile of the amplitude functions U_{nk}(r) and V_{nk}(r) ():
and
Here we found it useful to separate the r and tdependences of the t_{nk}(r,t) amplitudes of the development in Eq. (9) and introduce pure radial parts T_{nk}(r) such that . These represent source terms in Eqs. (A.4) and (A.5).
While the solution of U_{nk}(r) and V_{nk}(r) is coupled by means of Eqs. (A.4) and (A.5), the fundamental implication of the DuhamelNeumann equation linearity is that amplitude terms of different degrees and orders in the spherical harmonics development as well as the different Fourier modes are not mixed and can be solved separately.
Once we obtain U_{nk}(r) and V_{nk}(r), we can readily compute
components of the corresponding stress tensor
arising from the displacement vector
field by using the Hook's law in Eq. (16).
Given its linearity, we thus again have
where . Projecting components of the stress tensor onto the orthonormal basis from Eq. (33), as outlined in Sect. 3.1, we obtain (for simplicity we dropped here the degree and orderindexes n and k)
The partial derivatives of the spherical functions Y_{nk} are computed using
=  
(A.13)  
=  (A.14) 
Equations (A.7) to (A.12) yield components of the stress tensor that explicitly depend on the displacement vector . The part , which explicitly depends on the temperature (see the generalized Hook's law in Eq. (16), should be added separately to the total stresstensor field. Because of the explicit analytical solution for , this is achieved at no computational expense.
Because of the linearity of the DuhamelNeumann equation, a general solution is expressed in terms of a linear superposition of (i) a solution of the homogeneous system; and (ii) a particular solution of the inhomogeneous system. The next two sections discuss the two cases separately.
A.1 Solution of the homogeneous DuhamelNeumann equation
Equations (A.4) and (A.5) with zero righthand sides represent the homogeneous DuhamelNeumann equation broken into parts corresponding to the individual spheroidal modes. Its solution is quite complicated, but may be significantly simplified in our case. This is because for the range of material parameters, sizes and rotation frequencies that apply for meteoroids we always have^{} . With these we may neglect the troublesome term in Eqs. (A.4) and (A.5). A major implication of this is then that the system of solutions of the homogeneous DuhamelNeumann equation become degenerate in the k (order) index of the sphericalharmonics representation.
Adopting the aforementioned approximation, the homogeneous system
of Eqs. (A.4) and (A.5) now has a form of
Euler equations. As such, it has a fundamental system of powerlaw
solutions
U^{i}_{nk}=Q_{i} r^{mi} and
V^{i}_{nk} = r^{mi} with
,
realvalued exponents m_{i} and amplitudes Q_{i}.
After a straightforward algebra, we obtain
U^{1}_{nk}(r)  =  (A.15)  
V^{1}_{nk}(r)  =  r^{n+1},  (A.16) 
U^{2}_{nk}(r)  =  (A.17)  
U^{3}_{nk}(r)  =  (A.18)  
U^{4}_{nk}(r)  =  (A.19)  
V^{4}_{nk}(r)  =  r^{n}.  (A.20) 
The last two modes, 3 and 4, diverge at the center r=0 and therefore must be excluded. We are thus left with the first two modes, 1 and 2, that produce the spheroidal modes and in Sect. 3.1.2 and whose associated stress field was given in Eqs. (44)(49) and Eqs. (50)(55). Obviously, they have also been used to obtain the stationary part of the stress field discussed in Sect. 3.1.1. We note that the second spheroidal mode represents a pure shear with no volumic changes (compression or expansion) because .
A.2 Particular solution of the DuhamelNeumann equation
We next find a particular solution of the inhomogeneous DuhamelNeumann equation with the thermal source . We divide this task into a discussion of the stationary case (k=0) and timedependent case (). In both cases, we again use the approximation of neglecting the terms in Eqs. (A.4) and (A.5).
A.2.1 Timeindependent part
The stationary temperature field is given by
(Eqs. (9) and (10)) and thus
.
We again search the fundamental
system of solutions in a powerlaw form
and
with some realvalued exponents
and
amplitudes
.
After a brief algebraic
derivation, we obtain
=  (A.21)  
=  (A.22) 
We note that this mode has the same radial profile as the spheroidal model found above.
A.2.2 Timedependent part
The timedependent temperature field is given by
with (Eqs. (9) and (11)) and thus
.
We assume that the particular
solution has a form
.
Substituting
this ansätz to the DuhamelNeumann equation
Eq. (18), we obtain
where we have suitably assumed that the arbitrary constant on the righthand side canceled the monopole (constant) temperature part. This is an inhomogeneous wave equation on a sphere that, however, takes a simple form because of the sphericalharmonic and Fourier structure of the source term on the righthand side. Assuming thus a separation
we obtain
and
where is the elastic Pwave velocity as above. Adopting again the approximation , we may neglect the first term in the denominator of Eq. (A.26). Translating this solution into the amplitudefunctions of the spheroidalfield representation (A.2), we finally obtain
=  (A.27)  
=  (A.28) 
The corresponding stress tensor is expressed by Eqs. (58)(63).
A.3 Complete expression of the thermal stress tensor
The complete solution of the DuhamelNeumann
equation is a linear combination of the freespheroidal modes
and
from Sect. A.1
and the particular mode
from Sect. A.2.
In the individual spherical harmonics modes, we have
,
where Q^{1}_{nk} and Q^{2}_{nk} are
some coefficients. We have to choose them to satisfy the surface
boundary condition (19), namely
at r=R. Here the total stress tensor is given by
or again in the spherical harmonics modes
The truly active and independent conditions are and , from which the two constants Q^{1}_{nk} and Q^{2}_{nk} follow. One easily checks that the third condition, , is always linearly dependent on (Eqs. (A.8) and (A.9)) and thus we do not need to consider it.
We were able to carry out all necessary algedraic manipulations and obtain a close form of the resulting formulae for the case of the stationary (zonal, k=0) part of the stress field. These are given in Eqs. (36)(39) (Sect. 3.1.1). In the case of the timedependent part of the stress field, the algebra is more involved and we could not reach as simple and compact results as for the timeindependent part. We thus confine ourselves to provide formulae for the stresstensor components of the individual components and those for the integration constants Q^{1}_{nk} and Q^{2}_{nk}(Sect. 3.1.2).
Current usage metrics show cumulative count of Article Views (fulltext article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 4896 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.