Journal of Biomechanics 34 (2001) 471–479

A theoretical model of the effect of continuum

damage on a bone adaptation model
S. Ramtani*, M. Zidi
Laboratoire de Me´canique Physique/CNRS ESA 7052, Faculté des Sciences et Technologie, Universite´ Paris Val de Marne, 61, avenue du ge´ne´ral De
Gaulle, 94010 Creteil Cedex, France
Accepted 8 November 2000

Abstract

Throughout life, bone is continuously turning over by the well-regulated processes of bone formation and resorption. Everyday
activities damage bone, and this damage is normally repaired in a continuous remodelling process. When an imbalance in this
remodelling process occurs, bones may become more susceptible to fracture. This paper is devoted to a theoretical modelling of the
competition between damage and internal remodelling in bones. The general theory of adaptive damaged-elastic materials proposed
here as a model for the physiological process of damaged-bone remodelling follows the general framework of continuum
thermodynamics where new damaged-bone remodelling law and associated thermodynamical restrictions are stated, and specialized
to the case of small strain in isothermal processes. An attempt is also made to derive: (a) the damage force (adaptive damage energy
release rate ) which controls the microcracks propagation and arrest, and (b) the damage rule by introducing damage thresholds and
loading/unloading conditions. # 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Bone adaptation; Damage mechanics; Adaptive energy release rate; Continuum thermodynamics

1. Introduction biochemically regulated processes influenced by me-

The first hypothesis about the dependence between
the form of bones and the load they carry was argued by
Galileo in 1638 (Ascenzi, 1993). The nature of this
dependence was first described by Wolff (1986), who
stated that every change in the form or function of a living
bone is followed by adaptive changes in its internal
architecture and its external shape. Thus, several
mechanisms have been proposed for the transduction
of mechanical loads to the remodelling response
(Treharne, 1981), including: (1) piezoelectric or stream-
ing potentials; (2) mechanical fatigue microdamage; (3)
extracellular fluid pressure effects on bone cells; (4)
direct loads on bone membranes; and (5) alterations in
mineral solubility due to stress. Experimental data can
be found to support each of the above-mentioned
mechanism. However, it is well known that bone
tissue growth, maintenance, and degeneration are
*Corresponding author.
E-mail address: ramtani@univ-paris12.fr (S. Ramtani).
chanical function.
By the use of laser scanning confocal microscopy or
transmitted light microscopy techniques, tiny micro-
cracks were observed in bone (Burr et al., 1985; Burr
and Hooser, 1995; Lee, 1997; Zioupos, 1996): (a)
dispersed throughout the volume under strain, (b) in
close proximity, but (c) isolated from each other and
apparently impervious to each other’s presence. These
microcracks have been implicated in physiological
phenomena including stress fractures (Burr et al.,
1990), remodelling (Mori and Burr, 1993) and adapta-
tion (Prendergast and Taylor, 1994; Lee et al., 1998).
For example, the study of Lee et al. (1998) has shown
that overloaded bone adapts by both forming new bone
at surfaces and by remodelling the existing cortex.
Microcracks were found in control sheep, which
suggests that an amount of damage is present at
equilibrium. The process of adaptation to altered
applied load is preceded by a significant rise in the
incidence of microcracks adjacent to sites of bone
modelling and remodelling, which is consistent with
microdamage being a stimulus for adaptation.

0021-9290/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 1 - 9 2 9 0 ( 0 0 ) 0 0 2 1 5 - 3
472 S. Ramtani, M. Zidi / Journal of Biomechanics 34 (2001) 471–479
Despite the fact that damage accumulation has been
demonstrated in living bone and postulated as a
stimulus to the bone modelling and remodelling
response, a relationship between microdamage and
adaptive bone-remodelling was only tentatively pro-
posed. Frost (1960) offers the idea that microdamage
may trigger a remodelling response via the canalicular
network surrounding the osteocytes. According to this
theory, there is a physiological strain range where bone
is in equilibrium, determined by metabolic factors alone.
The lower limit of this range is given by a ‘‘remodelling’’
process while the upper limit is given by a ‘‘modelling’’
one. The author suggested that these processes are
stimulated if the rate of microdamage formation is due
to unphysiological strains. Chamay and Tschantz (1972)
have observed damage in vivo as ‘plastic slip lines’ or
collagen bundle ‘wrinkling’ in the area where remodel-
ling by bone deposition was most extensive. Carter and
Hayes (1977) have hypothesized that fatigue damage
acts to produce a bone maintenance stimulus. This
hypothesis was developed by Prendergast and Taylor
(1994) to predict the time-course of bone adaptation, by
accounting for the accumulative nature of the damage
with the concept of a continuous damage rate. Recently,
Prendergast and Huiskes (1996a) have proposed a
microstructural finite element analysis to explore the
relationship between damage formation and local strain
of osteocyte-containing lacunae for various types of
damage. Their study was based on the hypothesis that, if
the microstructural changes are substantial enough, and
the osteocytes do function as strain sensors then damage
would stimulate bone remodelling using the osteocyte-
sensor mechanism. The three microdamage mechanisms
tested by the authors are the following: (a) collagen
bundle break-down, which occur when the collagen
fibrils separate from the supporting hydroxyapatite
crystals, (b) the interlamellar debonding and (c) the
microcracking in the cement-line. Their results have
given theoretical support to the experimental studies
that have shown a correlation between microdamage
and the initiation of resorption as the first step in bone
remodelling.
2. Basic and fundamental assumptions
To construct a general framework for a phenomen-
ological description of both strain- and damage-induced
bone remodelling process, we invoke the following
assumptions (Cowin and Hegedus, 1976): (a) The
porosity of the bone matrix is affected by the ambient
long-term strain history of the bone, by the addition of
mass or the removal of mass from the bone matrix and
by the presence of the microcracks in the bone matrix. (b)
The transfer of mass occurs as the result of a
biochemical reaction which is mediated by the bone
cells. Energy and entropy may also be transferred to or
from the matrix structure by these chemical reactions.
(c) The extracellular fluid is always in contact with the
blood plasma. The plasma supplies the materials
necessary for the synthesis of bone matrix. (d) The
characteristic time of chemical reaction is several orders
of magnitude greater than the characteristic time
associated with a complete perfusion of the blood
plasma in the bone, hence any excess of heat generated
by the chemical reactions is quickly carried out by the
circulation.
2.1. Strain- and damage-induced remodelling
From theoretical point of view, it makes sense to
consider strain as the basic mechanical signal, because it
is: (a) the immediate local effect of external bone
loading, and (b) a primary and directly measurable
physical quantity (and not the stress which is as well
defined but not directly measurable quantity). Thus,
Justus and Luft (1970) demonstrated that straining bone
will increase the calcium concentration in the interstitial
fluid. They showed that this was due to a change in the
solubility of the hydroxyapatite crystals and also
demonstrated this mechano-chemical effect with inor-
ganic crystals in vitro.
As pointed by Huiskes (1997), assuming microcracks
development as the stimulus for bone remodelling is
very tempting for a number of reasons: (a) the
continuous physiological process of resorption and
deposition makes no sense other than as a repair
process for microdamage, (b) bone repair is a well-
known biological phenomenon; any other assumption
for a stimulus effect requires the introduction of
biological mechanisms of which the actual existence is
unknown, and (c) the development of microdamage is
automatically related to the dynamic loading history,
which is not the case for the other possibilities
mentioned above.
2.2. Kinematics
The motion of material point is described in the
Cartesian system by yi ¼ y~i ðxk ; tÞ where y~i ðxk ; tÞ gives
the location of the material particle xk at time t:
14i; k43, i and k are indices labelling the three different
coordinate directions in the spatial and reference
configurations, respectively. The deformation function
yi ¼ y~i ðxk ; tÞ is assumed to be one to one, continuous
and invertible. The velocity field of the continuum at
time t, is given by vi ¼ @yi =@t. The deformation gradient
Fik , the local volume change J and the velocity gradient
Lij are, respectively, defined by
@yi
Fik ¼ ; ð1aÞ
@xk
 S. Ramtani, M. Zidi / Journal of Biomechanics 34 (2001) 471–479
J ¼ det F ð1bÞ
and
@vi
Lij ¼ : ð1cÞ
@xj
For one, we will consider the bulk density r of the
damaged porous structure to be expressible as
r ¼ gf; ð2Þ
where g is the density of the material that composes the
bone matrix and f is the volume fraction of that
material present (current solid volume fraction). Both f
and g can be considered as field variables by the same
arguments as employed in Goodman and Cowin (1972).
The internal remodelling variable was initially intro-
duced by assuming: (a) the bone as a porous elastic
solid, and (b) the normal adaptive process that occur in
bone remodelling as strain controlled mass deposition
and resorption processes which modify the porosity of
the porous elastic solid. If x denotes the volume fraction
of the matrix material in an unstrained reference
configuration, the physical meaning of the variable x is
taken from Cowin and Hegedus (1976) : one might
imagine a block of porous elastic material with the four
points, the vertices of a tetrahedron, marked on the
block for the purpose of measuring the strain. When the
porosity changes, material is added or taken away from
the pores, but if the material is deformation-free it
remains so and the distance between the four vertices
marked on the block do not change. Thus x can change
while the zero-strain reference state remains the same.
Secondly, damage can be described as the local loss of
mechanical integrity or the ‘state of partial cracking’ due
to nucleation and growth of microcracks (Kachanov,
1980; Krajcinovic et al., 1987; Krajcinovic and Lemai-
tre, 1987; Chaboche, 1988, Ramtani et al., 1992). In
most of the studies concerning the development of
damage in bone (as measured by a reduction in elastic
modulus) during the fatigue tests, the damage is
represented by a scalar variable d (Zioupos et al.,
1996; Prendergast and Huiskes, 1996a).
By analogy with the effective strain or stress concept
(Kachanov, 1958; Chaboche, 1988, Ju, 1989), we
introduce the effective volume fraction of the matrix
structure in an unstrained reference state as xd , and by
the use of the intrinsic incompressibility condition of the
damaged solid phase (Suh and Spilker, 1994), we have
xd ¼ ð1  d Þx ¼ Jf: ð3Þ
d
Then, a formal definition of x can be made in terms of
the volume fraction f associated with the strain-state
characterized by the local volume change J. In other
words, x and f are the same but viewed from the
effective free deformation reference state and the current
deformed state, respectively. Note that both x and xd
473
may change without changing the reference state for
strain if we assume that, at constant temperature and
zero body force, there exists a unique zero-strain
reference state for all values of x (Cowin and Hegedus,
1976). Thus, by combining Eqs. (2) and (3), the bulk
density r of the damaged porous structure is stated as
gx
r ¼ ð1  d Þ : ð4Þ
J
3. Field equations
We consider an arbitrary fixed surface S enclosing a
volume O of the porous matrix structure. Because mass
is being added to the body, the traditional transport
theorem has been modified as follows (Cowin and
Hegedus, 1976; Coussy, 1995):
Z Z
D
gfg dO ¼ ðmg þ gfg_ Þ dO; ð5Þ
Dt O O
where both the derivative indicated and the super-
imposed dot are the material or substantial time
derivatives, g denotes an arbitrary field quantity, and
m is the rate at which mass per unit volume is added to
or removed from the porous matrix structure (m arises
from chemical reactions). If the differentiation indicated
in Eq. (2) is performed in the case where g is chosen as
the identity function (i.e., g ¼ 1) and the result written
as a field equation, we have

gf þgfLkk ¼ m ð6Þ
which is the local statement of balance of mass for the
porous matrix.
When Eq. (4) is substituted into the statement of mass
balance (6), a new relation between the rate of change of
the remodelling variable x, _ the mass deposition or
resorption m and the damage rate d_ is stated as
(Ramtani and Zidi, 1999)
m xd_
x_ ¼ J þ ; ð7Þ
gð 1  d Þ ð 1  d Þ
which can be viewed as an extension of an earlier
proposal (Cowin and Hegedus, 1976) by introducing the
damage variable.
Firstly, note that an increase of the damage level in
the bone material leads to an increase in the remodelling
rate, which is in accordance with Prendergast and
Taylor (1994), and Prendergast and Huiskes (1996a),
who have demonstrated that damage-stimulated resorp-
tion could be a local manifestation of strain-adaptive
remodelling. Secondly, the proposed remodelling rule
follows the idea that more than one control system may
be operative in bone mass regulation (Carter, 1984). In
particular, it has been suggested that the remodelling
stimulus is not exclusively damage or strain but rather
474 S. Ramtani, M. Zidi / Journal of Biomechanics 34 (2001) 471–479

that both operate together in a process of maintaining equivalent to the following two balance equations in
mass and avoiding failure (Prendergast and Huiskes, differential form:
1995). @Tki
From Eq. (7), it becomes clear that the remodelling þ gffi þ pi ¼ gf_vi ; ð11aÞ
@yk
rate can be cast in two terms: (a) the first one is linked to
the elastic opening of the existing (or strain-induced) @Qk
microcracks, and (b) the second is due to the damage gfU_  gfr þ  Tki Lik  h ¼ 0; ð11bÞ
@yk
growth (coupled term). On the other hand, if we look for
the case where the bone material is submitted to loads which are customary and, respectively, the local version
(or strains) such as no further damage takes place, the of balance of linear momentum for the porous matrix
damage growth do not contribute to the remodelling and the local energy equation for the porous matrix.
process (i.e., d_  0 and d  1) and leads to
J m
x_ ¼ ð8Þ 4. Second axiom of thermodynamics
ð1  dÞ g
which is similar to the one obtained by Cowin and Consider an entropy production inequality of the
Hegedus (1976) with d ¼ 0 and corresponds probably to form (Coleman and Gurtin, 1967; Malvern, 1969;
the case where the bone turnover rate is sufficient to Germain, 1973; Bowen, 1976; Cowin and Hegedus,
mend the microdamage as it is formed. In order to show 1976)
Z Z Z
that there exist situations where bone fractures are never D r Q k nk
repaired, let us investigate the singular case where d_ 6¼ 0 gfZ dO5 gf dO  dS
Dt O O y @O y
and d  1. When imposed on Eq. (7), the following Z !
remodelling rate is obtained by x_ !/, and this yielding h~
þ þ Zm dO; ð12Þ
to infinity is incompatible with the well-known fact that O y
the time scale of the remodelling processes is of the order
where the scalar Z is the specific entropy, y is the local
of months or possibly years.
temperature and h~ is defined in the same fashion as h
The system to which Eqs. (9) and (10) apply is the
above. The distinction between h and h~ is made to
porous matrix structure while the effect of the internal
indicate that all the energy transferred from the
perfusant is accounted for by transfer terms in each
perfusant to the matrix need not contribute to the
equation. We postulate, respectively, a balance of
entropy production. By a known standard argument
momentum and a balance of energy at time t in the
(Coleman and Noll, 1963), the local form of the entropy
form (Coleman and Gurtin, 1967; Malvern, 1969;
inequality is
Germain, 1973; Bowen, 1976 ; Cowin and Hegedus,
 Qk @y
1976) gf Zy_ þ C þ Tki Lik  þ h * 50; ð13Þ
Z Z Z y @yk
D
gfvi dO ¼ Tij nj dS þ gffi dO
Dt O @O O where h * ¼ h  h~ and C ¼ U  Zy is the Helmholtz free
Z
energy density. However, there are several other
þ ðpi þ mvi Þ dO; ð9Þ specialized approaches that leads to the same results
O
Z   Z (Cowin, 1999).
D 1
gf vi vi þ U dO ¼ gfðr þ fi vi Þ dO
Dt O 2 O
Z
5. Constitutive assumptions and thermodynamical
þ ðnk Tki vi  ni Qi Þ dS restrictions
Z@O   
1
þ pi vi þ m vi vi þ U þ h dO; ð10Þ A thermodynamic process for the composite bone
O 2
material under discussion is a set of scalar functions
where vi , Tki , fi , U, r and Qi are respectively an (Coleman and Gurtin, 1967; Germain, 1973; Bowen,
arbitrary velocity field of the continuum at time t, 1976; Cowin and Hegedus, 1976) from which the free
the Cauchy stress tensor, the specific body force, the energy is chosen as (Miehe, 1995)
specific internal energy, the specific heat supply function ~ ðy; Fik ; xÞ:
C ¼ ð1  d ÞC ð14Þ
per unit time and the heat flux vector. Besides, pi is the
force the perfusant applies to the porous matrix Taking the time derivative of the free energy
structure and h represent any energy transfer due to function (14), substituting into (13), and making use of
the local interaction between the matrix and the standard arguments (Coleman and Gurtin, 1967;
perfusant. Under suitable smoothness assumptions the Germain, 1973; Bowen, 1976; Cowin and Hegedus,
balance equations (9) and (10) in integral form are 1976), we obtain the following restrictions which
 S. Ramtani, M. Zidi / Journal of Biomechanics 34 (2001) 471–479 475

corresponds to the caloric equation and the behavior al., 1987; Chaboche, 1988; Prendergast and Taylor,
law for the damaged bone: 1994), or derived directly by means of damage criterion,
@C damage consistency condition and loading/unloading
Z¼ ; ð15aÞ conditions (Chaboche, 1988; Ju, 1989, Ramtani et al.,
@y
 ~  1992).
gx @ C Eqs. (1a), (3), (7), (15b) and (19) and the damage
Tij ¼ ð1  d Þ2 Fjk : ð15bÞ
J @Fik evolution law (to be specified) can be viewed as a system
From the fundamental assumptions, we have supposed of 21 scalar equations for the 21 scalar unknowns Tij ,
that, at constant temperature and zero body force, there Fik , y~i , r, x and d.
exists a unique zero-strain reference state for all values
of x which can now be written as
7. Small-strain approximation
Tij ðy0 ; dik ; x; d Þ ¼ 0; ð16Þ
where y0 and x are, respectively, the fixed temperature To set the stage for the small-strain approximation,
and an arbitrary value of the remodelling variable. the independent variables x and Fik are changed to e and
By substituting Eq. (15) back into Eq. (13) and Eij where e is a measure of the change in volume fraction
making use of Coleman and Noll (1963) arguments, from a reference volume fraction x0 (Hegedus and
the reduced entropy inequality becomes Cowin, 1976),
 ~
@C Qk @y e ¼ x  x0 : ð22Þ
x ð1  d Þm  þ h * 50 ð17Þ
@x y @yk The new functions are now denoted by superposed bars,
and

 e; Eij ; d ¼ m
m ~ ðx; Fik ; d Þ; ð23aÞ
 ~ 
gx @C ~ _

 ð1  d Þ x  C d50 ð18Þ  e; Eij ¼ C
C ~ ðx; Fik Þ; ð23bÞ
J @x
Note that the entropy inequality (17) has been stated by where the linearized strain tensor E has the following
Cowin and Hegedus (1976) for the ‘‘undamaged’’ bone representation in terms of displacement gradient
material and that (18), can be viewed as a new

Eij ¼ 12 ui;j þ uj;i : ð24Þ
thermodynamical restriction for the damage dissipation
of the adaptive-elastic bone material. The constitutive equation for the stress takes the form
  
2 g ð x0 þ e Þ @C
sij ¼ ð1  d Þ Fik Fjl ð25Þ
½detð1 þ 2EÞ 1=2 @Ekl
6. A summary of the theory
and Eq. (7) can be reduced as
We now summarize the equations that constitute the 1 n
 o
proposed theory in the quasi-static and isothermal e_ ¼ F e; Eij ; d þ ðx0 þ eÞd_ ; ð26Þ
ð1  d Þ
processes. These equations are the deformation gradient
given by Eq. (1a), balance of mass in the form (7), the where



balance of linear momentum (11a) as  e; Eij ; d ¼ 1m
F  e; Eij ; d ½detð1 þ 2EÞ 1=2 : ð27Þ
@ski g
þ gffi ¼ 0; ð19Þ
@yk Based on substantial physical justification (Scott, 1957;

where the inertia term (quasi-static assumption) and the
Bonfield and Li,
1966),
 the approximations for F
perfusant force pi have been neglected, the volume 
e; Eij ; d and C e; Eij are obtained by neglecting the
fraction of the reference configuration x defined by terms of order 3 in a Taylor series expansion,
Eq. (3), the behavior law (15b). The quantities m and C  ðe; Eik ; d Þ ffi aðe; d Þ þ Aij ðe; d ÞEij
F
are functions of Fik ; x and d,
þ 12Bijkm ðe; d ÞEij Ekm ; ð28aÞ
m¼m
~ ðFik ; x; d Þ; ð20aÞ
~ ðFik ; xÞ; C  ðe; 0Þ þ 1 Cijkm ðeÞEij Ekm ;
 ðe; Eik Þ ffi C ð28bÞ
C ¼ ð1  d ÞC ð20bÞ 2g
~ ðFik ; xÞ satisfy the condition
and it is required that C where
@C~
  ðe; 0; d Þ;
aðe; d Þ ¼ F ð29aÞ
dij ; x ¼ 0; ð21Þ
@Fik

@F
and the damage evolution law which can be evaluated Aij ðe; d Þ ¼ ðe; 0; d Þ; ð29bÞ
by the use of the remaining life approach (Krajcinovic et @Eij
476 S. Ramtani, M. Zidi / Journal of Biomechanics 34 (2001) 471–479

@2F  8. Characterization of damage

Bijkm ðe; d Þ ¼ ðe; 0; d Þ; ð29cÞ
@Eij @Ekm
8.1. Damage force
@ C 2
Cijkm ðeÞ ¼ g ðe; 0Þ: ð29dÞ This subsection is dealing with the derivation of the
@Eij @Ekm
damage force by the use of the damage-dissipative
Thus, when Eq. (28b) is substituted into expression (25) restriction (18), which is rewritten as
for the stress and terms of second order in the   
@C  _
displacement gradients are neglected, we find that  ð x0 þ e Þ  C d50 ð36Þ
@e
sij ¼ ð1  d Þ2 gðx0 þ eÞCijkl ðeÞEkl ð30Þ and with the aid of Eqs. (28b) and (32), it becomes
 
and it is a modified form of Hooke’s law in that the  @ 
proportionality between stress and strain is dependent Cðe; 0Þ  ðx0 þ eÞ Cðe; 0Þ
@e
on the volume fraction of the material present.  
x0 1 0
The expression for the remodelling rate, e, _ is obtained þ Cijkl  Cijkl
2gðx0 þ eÞ
from combining Eqs. (26) and (28a) and neglecting   
terms of order 2 in the displacement gradient, thus 1 0 1 _
þ x C þ eCijkl Eij Ekl d50: ð37Þ
2gðx0 þ eÞ 0 ijkl
1 h i
e_ ¼ aðe; d Þ þ Aij ðe; d ÞEij þ ðx0 þ eÞd_ : ð31Þ However, one can choose the free energy function such
ð1  d Þ
that Eq. (37) reduces to
An important prediction of the theory of small-strain     
 ðe; 0Þ þ 1 0 1 _
adaptive elasticity can be inferred from expression (31) C x0 Cijkl þ eCijkl Eij Ekl d50
for the remodelling rate in the presence of microcracks. 2gðx0 þ eÞ
The functions Cijkl ðeÞ, aðe; d Þ and Aij ðe; d Þ characterize ð38aÞ
the damaged material properties and there are very few which is rewritten in the following form
experimental data on these functions. Thus, we have
supposed that they are damage-independent ones. We _
Yd50 ð38bÞ
can make a polynomial approximation of these func- with Y stated as
tions as in Hegedus and Cowin (1976) and as first   
 ðe; 0Þ þ 1 0 1
approximation, we propose Y ¼ C x0 Cijkl þ eCijkl
2gðx0 þ eÞ
1 n 0 1
o
Cijkl ðeÞ ¼ x0 Cijkl þ eCijkl ; ð32aÞ ~
x0 þ e  ðe; Eik Þ ¼ @ C:
Eij Ekl ¼ C ð38cÞ
@d
aðe; d Þ ¼ aðeÞ ¼ a0 þ a1 e þ a2 e2 ; ð32bÞ As Y is a quadratic function (i.e. C  ðe; 0Þ ¼ 0), this
leads to well-known damage growth condition
Aij ðe; d Þ ¼ Aij ðeÞ ¼ A0ij þ A1ij e: ð32cÞ
_
d50 ð38dÞ
With the above approximations, Eqs. (30) and (31) are
rewritten in more suitable form as Therefore, we conclude that the undamaged energy
n o function Y ¼ C  ðe; Eik Þ is the thermodynamic force
sij ¼ ð1  d Þ2 x0 Cijkl
0 1
þ eCijkl Ekl ; ð33Þ (adaptive damage energy release rate) conjugate to the
  damage variable d, which controls the microcracks
e_ ¼ a c  2be þ e2 ð34Þ propagation and arrest. Note that the proposed
adaptive energy release rate reduces to the well-known
with scalar damage force C0 ðEik Þ when one sets x0 ¼ 1 and
a2 e ¼ 0 (Chaboche, 1988).
a¼ ; ð35aÞ
ð1  d Þ
 8.2. Damage mechanism
a1 þ d_ þ A1ij Eij
b¼ ; ð35bÞ In order to complete the proposed constitutive
a2
model, a simple isotropic adaptive-elastic damage
a0 þ x0 d_ þ A0ij Eij mechanism is characterized in this section to describe
c¼ ; ð35cÞ the possible progressive degradation of mechanical
a2
properties of bone material during the remodelling
0 1
and where Cijkl , Cijkl , a0 , a1 , a2 , A0ij and A1ij are constants process. For that, we propose to employ the adaptive
representing the material properties. energy release rate to characterize the damage loading/
 S. Ramtani, M. Zidi / Journal of Biomechanics 34 (2001) 471–479 477

unloading conditions. For convenience, we define the As a summary, the basic system of equations for the
notation as proposed theory consist of the equilibrium equation (19)
 ðe; Eik Þ:
Y * ¼ jY j ¼ C ð39Þ where, to terms of first order in strain, the bulk density r
given by (4) is rewritten as
The state of damage in the material is
then characterized

by means of damage criterion ‘ Yt* ; kt with the r ¼ ð1  d Þgðx0 þ eÞð1  Ekk Þ; ð45Þ
following functional form:

 the stress–strain relation (30), the remodelling rate
‘ Yt* ; kt  Yt*  kt 40; t 2 Rþ : ð40Þ
equation (31), the strain-displacement relations (24)
Here, the subscript t refers to the value at current time, and the damage rate equation (41). This is a system of 17
and kt signifies damage threshold (energy barrier) at scalar equations involving the 17 scalar unknowns, Tij ,
current time t (i.e. the radius of the damage surface). If Eij , ui , e and d.
k0 denotes the initial damage threshold before any
loading is applied, condition (40) then states that
damage in the material is initiated when the damage
9. Conclusion
energy release rate Yt* exceeds the initial damage
threshold k0 . Note that the above energy-based damage
From biomechanical point of view, there is a
criterion is fundamentally linked to the history of the
great need that an attempt be made to put some
bone’s loading.
framework in which theoretical statement can be
Furthermore, initial microdamage can (i) also pre-
made about the damage effect on the bone adaptation
exist even if the bone material is free deformation, (ii)
process. To achieve this goal, we have used continuum
participate in the remodelling process as an elastic
thermodynamics and stated some new results which
opening, and (iii) can propagate only if the new damage
can be viewed as: (a) an extension of the theory of
threshold is reached. To describe the damage growth
adaptive elasticity, presented by Cowin and colleagues
and the expansion of damage surfaces, it is necessary to
(Cowin and Hegedus, 1976; Hegedus and Cowin,
specify the equations of evolution for d and k. In order
1976), by introducing the presence of the microcracks,
to describe the microdamage (microcrack) growth
(b) and as a logical following of the mathematical
without considering the microcracks interactions, we
model proposed by Prendergast and Taylor (1994).
define the general evolution rule as (Chaboche, 1988; Ju,
The essential purpose of the present work is to
1989; Ramtani et al., 1992; Miehe, 1995)

 demonstrate one possible way for modelling the
d_ ¼ lH
_ Yt* ; dt ; k0 k_t ¼ l;_ ð41Þ damage interaction with remodelling process in bones
_
where l50 is a damage consistency parameter which and we can summarize the main points of this work as
defines damage loading/unloading conditions according follows:
to the Kuhn–Tucker relations: 1. The proposed remodelling rule follows the idea
that: (a) more than one control system may be operative
_
l50; ð42aÞ in bone mass regulation (Carter, 1984), (b) the

 remodelling stimulus is not exclusively damage or strain
‘ Yt* ; kt 40; ð42bÞ
but rather that both operate together in a process of

 maintaining mass and avoiding failure (Prendergast and
_ Yt* ; kt ¼ 0:
l‘ ð42cÞ
Huiskes, 1995). From this study, it becomes clear that
In order to derive the damage consistency
parameter,
 let the remodelling rate can be cast in two terms where the
us consider two possible events: (a) if ‘ Yt* ; kt 40, the first one is associated with the elastic opening of the
damage criterion is not satisfied and by condition (42c) existing microcracks, and the second is due to the
we have l_ ¼ 0; hence, the damage rule (41) implies that damage growth. Besides, the associated state laws and
d_ ¼ 0 and no further damage takes place, (b) if l50, _
thermodynamical restrictions (dissipative inequalities)
that is further damage

is taking place, condition (42c) are also stated for the general theory.
now implies that ‘ Yt* ; kt ¼ 0. In this event the value of 2. We have investigated, by the use of an approxima-
_
l50 is determined by the ‘‘damage consistency con- tion scheme involving series expansion, the possible
dition’’; i.e. small-strain formulation of damaged-bone remodelling



‘ Yt* ; kt ¼ ‘_ Yt* ; kt ¼ 0 ð43aÞ theory. In particular, we have derived the basic system
of governing equations and identified, under some
) l_ ¼ Y_ t* : ð43bÞ suitable assumptions, the damage force for driving the
So that kt is given by the expression damage evolution. In particular, the new concept of
  adaptive damage energy release rate is introduced. The
kt ¼ max k0 ; max Ys : * ð44Þ local damage initiation and propagation are linked to
s2½1;t
the damage force, which is checked against the debond-
478 S. Ramtani, M. Zidi / Journal of Biomechanics 34 (2001) 471–479
ing energy (current damage threshold) required for
microcrack growth.
This work has as primary goals the increased under-
standing and quantitative description of the functional
adaptation of bone material, and ultimately the predic-
tion of that functional adaptation. The process to be
followed to achieve these goals will involve a combina-
tion of this theory, experiments (Scott,1957; Bonfield
and Li,, 1966), stability analysis (Zidi and Ramtani,
1999) and numerical implementation of the proposed
model in finite element code (Hart et al., 1984;
Prendergast and Huiskes, 1996b). Since this model has
provided a general basis for investigating damaged-bone
adaptation, our hope is that this theory will be applied
and lead, for example, to a better understanding (needed
by the orthopaedic surgeon who places the metallic pin
in the bone) of bone remodelling induced by a medullary
pin (Cowin and Van Buskirk, 1978). What is required
first, is the experimental identification of all constants
representing the adaptive material properties and the
derivation of the exact form of the damage evolution
during the adaptation process.
References
Ascenzi, A., 1993. Biomechanics and Galileo Galilei. Journal of
Biomechanics 26, 95–100.
Bonfield, W., Li, C.H., 1966. Deformation and fracture of bone.
Journal of Applied Physics 37, 869–875.
Bowen, R.M., 1976. Theory of mixtures. In: Eringen, A.C. (Ed.),
Continuum Physics, Vol. 3. Academic Press, New York, pp. 1–127.
Burr, D.B., Hooser, M., 1995. Alterations to the en bloc basic fuchsin
staining protocol for the demonstration of microdamage produced
in vivo. Bone 17, 431–433.
Burr, D.B., Martin, R.B., Schaffler, M.B., Radin, E.L., 1985. Bone
remodelling in response to in vivo fatigue microdamage. Journal of
Biomechanics 18, 189–200.
Burr, D.B., Milgrom, C., Boyd, R.D., Higgins, W.L., Radinel, G.,
1990. Experimental stress fractures of the tibia-biological and
mechanical aetiology in rabbits. Journal of Bone and Joint Surgery
72B, 370–375.
Carter, D.R., 1984. Mechanical loading histories and cortical bone
remodelling. Calcified Tissue International 36, S19–S24.
Carter, D.R., Hayes, W.C., 1977. Compact bone fatigue damage-I.
Residual strength and stiffness. Journal of Biomechanics 10, 325–
337.
Chaboche, J.L., 1988. Continuum damage mechanics. Part I, General
concepts, Part II, Damage growth, crack initiation and crack
growth. Journal of Applied Mechanics 55, 233–247.
Chamay, A., Tschantz, P., 1972. Mechanical influences on bone
remodelling: experimental research on Wolf’s law. Journal of
Biomechanics 5, 173–180.
Coleman, B.D., Gurtin, M., 1967. Thermodynamics with internal
variables. Journal of Chemical Physics 47, 597–613.
Coleman, B.D., Noll, W., 1963. The thermodynamics of elastic
materials with heat conduction. Archive for Rational Mechanics
and Analysis 13, 167–178.
Coussy, O., 1995. Mechanics of porous media. Wiley, Chichester.
Cowin, S.C., 1999. Bone poroelasticity. Journal of Biomechanics 32,
217–238.
Cowin, S.C., Hegedus, D.M., 1976. Bone remodelling I: theory of
adaptive elasticity. Journal of Elasticity. 6, 313–325.
Cowin, S.C., Van Buskirk, W.C., 1978. Internal bone remodelling
induced by a medullary pin. Journal of Biomechanics 11, 269–275.
Frost, H.M., 1960. Presence of microscopic cracks in vivo in bone.
Bulletin of Henry ford Hospital 8, 25–35.
Germain, P., 1973. Cours de mécanique des milieux continus. Masson,
Paris.
Goodman, M.A., Cowin, S.C., 1972. A continuum theory for granular
materials. Archieve Rational Mechanics and Analysis 44, 249–266.
Hart, R.T., Davy, D.T., Heiple, K.G., 1984. A computational method
for stress analysis of adaptive elastic materials with a view toward
applications in strain induced bone remodelling. Journal of
Biomechanical Engineering. 106, 342–350.
Hegedus, D.M., Cowin, S.C., 1976. Bone remodelling II : Small strain
adaptive elasticity. Journal of Elasticity. 6, 337–352.
Huiskes, R., 1997. In: Simulation of Self-Organization and Functional
Adaptation in Bone. Springer, Berlin, pp. 299–319.
Ju, J. W., 1989. On energy-based coupled elastoplastic damage
theories: constitutive modeling and computational aspects. Inter-
national Journal of Solids and Structures 25(7), 803–833.
Justus, R., Luft, J.H., 1970. A mechanochemical hypothesis for bone
remodelling induced by mechanical stress. Calcified Tissue
Research 5, 222–235.
Kachanov, M., 1980. Continuum model of medium with cracks.
Journal of Engineering Mechanical Division 106, 1039–1051.
Kachanov, L.M., 1958. Time of the rupture process under creep
conditional. Isv. Akad. Ssr. Qtd. Tekh. Nauk 8, 26–31.
Krajcinovic, D., Lemaitre, J., 1987. Introduction to continuum
damage mechanics. Theory and applications. CISM courses and
lectures, No. 295, Sringe, Wien.
Krajcinovic, D., Trafimow, J., Sumarac, D., 1987. Simple constitutive
model for cortical bone. Journal of Biomechanics 20, 779–784.
Lee, T. C., Noelke, L., McMahon, G. T., Mulville, J. P., Taylor, D.,
1998. Functional adaptation in bone. In: Pedersen, P., Bendsoe,
MP. (Eds.), Synthesis in BioSolid Mechanics. Kluwer Academic
Publishers, Dordrecht.
Lee, T.C., 1997. Detection and accumulation of microdamage in bone.
M.D. Thesis, University of Dublin.
Malvern, L.E., 1969. Introduction to the Mechanics of a Continuum
Medium. Prentice-Hall, Inc., Englewood Cliffs, NJ.
Miehe, C., 1995. Discontinuous and continuous damage evolution in
Ogden-type large-strain elastic materials. European Journal of
Mechanics, A/Solids. 14 (5), 697–720.
Mori, S., Burr, D.B., 1993. Increased intracortical remodelling
following fatigue damage. Bone 14, 103–109.
Prendergast, P. J., Huiskes, R., 1995. Mathematical modeling of
microdamage in bone remodelling and adaptation. In: Odgaard,
A., Weinans, H. (Eds.), Bone Structure and Remodelling, Recent
Advances in Human Biology, World Scientific, Singapore, 2, pp.
213–223.
Prendergast, P.J., Huiskes, R., 1996a. Microdamage and osteocyte-
lacuna strain in bone: A microstructurale finite element analysis.
Journal of Biomechanical Engineering. 118, 240–246.
Prendergast, P.J., Huiskes, R., 1996b. Finite element analysis of
fibrous tissue morphogenesis } A study of the osteogenic index
using a biphasic approach. Mechanics of Composite Materials 32,
209–218.
Prendergast, P.J., Taylor, D., 1994. Prediction of bone adaptation
using damage accumulation. Journal of Biomechanics 27, 1067–
1076.
Ramtani, S., Berthaud, Y., Mazars, J., 1992. Orthotropic behavior of
concrete with directional aspects: modelling and experiments.
Journal of Nuclear Engineering and Design 133, 97–111.
 S. Ramtani, M. Zidi / Journal of Biomechanics 34 (2001) 471–479
Ramtani, S., Zidi, M., 1999. Damaged bone remodelling theory:
thermodynamical approach. Mechanics Research Communications
26, 701–708.
Scott, J.H., 1957. The mechanical basis of bone formation. Journal of
Bone and Joint Surgery 39B, 134–144.
Suh, J.K., Spilker, R.L., 1994. Indentation analysis of biphasic
articular cartilage: nonlinear phenomena under finite deformation.
Journal of Biomechanical Engineering 116, 1–9.
Treharne, R.W., 1981. Review of Wolff’s law and it’s proposed means
of operation. Orthopaedics Review 10, 35–47.
479
Wolff, J.L., 1986. The Law of Bone Remodelling. Springer, Berlin.
Zidi, M., Ramtani, S., 1999. Bone remodelling theory applied to the
study of n unit-elements model. Journal of Biomechanics 32,
743–747.
Zioupos, P., Wang, X.T., Currey, J.D., 1996. Experimental
and theoretical quantification of the development of damage
in fatigue tests of bone and antler. Journal of Biomechanics 29,
989–1002.