journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338

Available online at www.sciencedirect.com

www.elsevier.com/locate/jmbbm

Research Paper

An interface finite element model can be used

to predict healing outcome of bone fractures

J.A. Aliertaa, M.A. Pérezb, J.M. Garcı́a-Aznarb,n

a
Escuela Politécnica Superior del Ejército, Ministry of Defense, Madrid, Spain
b
Mechanical Engineering, Aragón Institute of Engineering Research (I3A), University of Zaragoza,
Ed. Betancourt-Campus Río Ebro, C/María de Luna, 50018 Zaragoza, Spain

ar t ic l e in f o abs tra ct

Article history: After fractures, bone can experience different potential outcomes: successful bone con-
Received 16 May 2013 solidation, non-union and bone failure. Although, there are a lot of factors that influence
Received in revised form fracture healing, experimental studies have shown that the interfragmentary movement
20 September 2013 (IFM) is one of the main regulators for the course of bone healing. In this sense,
Accepted 23 September 2013 computational models may help to improve the development of mechanical-based
Available online 8 October 2013 treatments for bone fracture healing. Hence, based on this fact, we propose a combined

Keywords: repair-failure mechanistic computational model to describe bone fracture healing. Despite

Bone fracture healing being a simple model, it is able to correctly estimate the time course evolution of the IFM

Finite element prediction compared to in vivo measurements under different mechanical conditions. Therefore, this

Design of fracture fixators mathematical approach is especially suitable for modeling the healing response of bone to

Interfragmentary movement fractures treated with different mechanical fixators, simulating realistic clinical conditions.
This model will be a useful tool to identify factors and define targets for patient specific
therapeutics interventions.
& 2013 Elsevier Ltd. All rights reserved.

1. Introduction The role of the mechanical stabilization on the design of

Fractures are a common orthopedic problem and the fracture
healing is a natural process that regenerates bone to its original
state and function. Several factors influence these bone healing
events, such as genetic, cellular and biochemical factors, blood
supply, neural and hormonal regulation, age, the type of
fracture interfragmentary motion and fracture geometry
(Einhorn, 2005; Goodship et al.,1993; Hadjiargyrou et al., 1998;
Jagodzinski and Krettek, 2007; Marsell and Einhorn, 2011).
However, the most common orthopedic treatments consist on
the mechanical stabilization of the bone fracture gap, regulating
the interfragmentary movement.
fracture fixators has evolved very much in the last years,
updating from the concept of “open reduction and internal
fixation (ORIF)” to “biological fixation” (Perren, 2002). In both
cases, differences are presented from a biological and mechan-
ical point of view. In fact, the concept of “biological fixation”
is based on the application of the fixator as a minimally
invasive percutaneous osteosynthesis (MIPO). So, the contact
of the implant with bone is reduced at maximum, avoiding the
damage to the blood supply. In addition to biological differ-
ences, there is also a different concept in the role of the
mechanical stabilization. In the “biological fixation” primary
ossification is avoided, promoting the occurrence of a secondary

n
Corresponding author. Tel.: þ34 976 76 27 96; fax: þ34 976 76 26 70.
E-mail address: jmgaraz@unizar.es (J.M. García-Aznar).

1751-6161/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jmbbm.2013.09.023
 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338 329

ossification mechanism that induces the formation of bone

callus (Manjubala et al., 2009; Vetter et al., 2010). Therefore, the 2. Material and methods
mechanical design concept is updated from an absolute stabi-
lity to a controlled mechanical instability that favors the The bone fracture gap was modeled through the incorpora-
formation of bone callus. Hence, the mechanical stabilization tion of interface elements that connected the two fracture
due to the stiffness of the fixator has acquired a relevant role for ends simulating the discontinuity in the displacement field
achieving a successful bone healing (Bishop et al., 2003; Chao between fragments. A mathematical model is here proposed
et al., 1989; Draper et al., 1997; Goodship et al., 1993; Richardson to simulate the temporal evolution of the separation between
et al., 1994; Wehner et al., 2011). Many studies have examined both fragments, regulated by the mechanical conditions
the role of different local mechanical conditions on bone existent in the fracture gap.
fracture healing: the influence of the magnitude of interfrag-
mentary movement (IFM) and the initial size of the fracture gap
(Claes et al., 1997; Gómez-Benito et al., 2011; Goodship and 2.1. Fracture gap/interface mechanical behavior
Kenwright, 1985), the stiffness of the fixation (Schell et al., 2005)
and the type of movement of the gap (Augat et al., 2003; Bishop To model the fracture gap behavior, 6-nodes and 8-nodes
et al., 2006). cohesive elements (Fig. 1) are used to connect the fracture
Computational models of the fracture healing process may ends (García-Aznar et al., 2009). The thickness of these
prove useful in the determination of optimal mechanical elements, which is the dimension of the gap fracture, is thin
treatments. In fact, several mechano-biological models based enough to consider it negligible with respect to the overall
on finite element simulations studied the influence of local dimensions of the bone fracture.
mechanical conditions on biological events that regulate the Accordingly, the behavior of these elements is directly
temporal and spatial evolution of the different ossification established in terms of one traction versus one separation
mechanisms that occur during healing (Andreykiv et al., 2008; law.
Checa and Prendergast, 2009; Isaksson et al., 2008, 2009a; The nominal traction stress vector, t, consists of three
Lacroix and Prendergast, 2002b; Loboa et al., 2001; Reina- components: tn, ts, tt, which represents the normal (tn) (along
Romo et al., 2011; Shefelbine et al., 2005; Simon et al., 2011;
Wehner et al., 2010). Most of these models have focused on
how different mechanical stimuli (such as: fluid flow, octahe-
dral shear strain, deviatoric strain, strain energy density, etc.)
are able to predict spatial pattern distribution of tissues
regulated by intramembranous and endochondral ossifica-
tion (Checa and Prendergast, 2009; Lacroix and Prendergast,
2002a; Wehner et al., 2010). Other models have combined
tissue differentiation rules with callus growth (Comiskey
et al., 2013; García-Aznar et al., 2007; Gómez-Benito et al.,
2005) or callus resorption (Ament and Hofer, 2000; Byrne
et al., 2011; Isaksson et al., 2009b; Lacroix and Prendergast,
2002b) depending on the temporal stage of bone healing that
they studied.
Although these mathematical models are very useful
to understand the fundamental cellular mechanisms that
locally regulate tissue differentiation and callus growth/
resorption, they are not really helpful for the simulation of
realistic bone fractures, where a whole-organ analysis is
required. In fact, these models have only analyzed simple
tranversal or obliques (Comiskey et al., 2013; Loboa et al.,
2001) bone fractures with two fragments. However, fractures
are much more complexes with very different shapes of the
fracture line, complicated anatomical locations and a differ-
ent number of fragments.
Therefore, the main aim of the present study is to propose
a phenomenological computational model able to simulate
the temporal recovery of mechanical properties of the frac-
ture zone during the healing process, which is regulated by
the mechanical stability. If this approach proves feasible, it
offers the possibility of using computer simulations in the
clinical treatment of complex fractures with multiple frag-
ments, complicated geometries and different anatomical
locations and in other orthopedic applications where bone Fig. 1 – Constitutive model: (a) cohesive elements, (b) shear
regeneration occurs. traction and (c) normal traction.
330 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338
the local 3-direction) and the two shear tractions (ts, tt) (along
the local 1- and 2-directions), respectively (Fig. 1). The
corresponding separations are denoted by δn, δs and δt. Thus,
the nominal strains can be defined as
δn δs δt
εn ¼ ; εs ¼ ; εt ¼ ð1Þ
h h h
where h is the original thickness of the cohesive element
(fracture gap thickness).
The purpose of the present study is to simulate the
temporal evolution of the bone healing process. Therefore,
to quantify the degree of healing, union degree, we use α as
the main state variable. This variable is normalized, corre-
sponding α¼ 1 to a totally successful bone healing and α¼0 to
a completely non-union or malunion.
The union degree α, may be also related with the so-called
damage variable, d (Moreo et al., 2007), through the following
expression:
d ¼ 1 α ð2Þ
The net effect of the bone healing process on the macroscopic
interface mechanical properties is what it is represented by
the growth of α. In addition, the model allows the possibility
of reducing the union degree α, which means mechanical
failure of the gap, because it is not able to bear the local
demands.
For each element of the fracture gap, the union degree,
α, can be defined as
K0i  α ¼ Ki i ¼ n; s; t ð3Þ
where K0i defines the linear stiffness in a completely healed
fracture gap along the direction i (the maximum fracture gap
stiffness available), and Ki is the current interface stiffness in
the i direction (Fig. 1b and c).
The three components of the nominal traction stress
vector, t, can be calculated, using the above defined variables,
through the following equation:
ti ¼ K0i  αεi i ¼ n; s; t ð4Þ
2.2. Mechanical damage: α decrease
If mechanical demands in the fracture gap are important, a
reduction of the degree of union may occur as consequence
of its mechanical failure.
For the shear traction, we assume the degree of union is
reduced by the following law (Fig. 1b):
( ðε  ε Þ
ðε0i  εci Þ ; if εαi oεi oεci
i ci
αi ¼ i ¼ s; t ð5Þ
0; if εi Zεci
where, ε0i maximum shear strain at the linear region for α¼ 1
(perfectly healed interface), εαi maximum shear strain at the
linear region for each value of α and εci the maximum shear
strains allowed. If εci are exceeded, the fracture gap would
have no stiffness at all.
For the compression traction (εn o0) we assume a linear
behavior with respect to the associated strain until εn reaches
the value  1, which means that the fracture ends are in
contact. If lower values of εn are reached the simulation will
be aborted because no penetration is allowed.
For the tensile traction (εn Z0) we assume an initially
linear behavior with respect to the associated normal strain,
followed by a linear decay (Fig. 1c):
( ðε  ε Þ
n cn
; if εαn oεn oεcn
αn ¼ ðε0n  εcn Þ ð6Þ
0; if εn Zεcn
where ε0n, εαn and εcn have an equivalent meaning that those
defined for shear traction.
In Fig. 1b and c the dashed line represents the behavior
law (X¼ε (strain); Y ¼t (stress)) of a completely healed bone
gap (α¼1). The continuous line represents the behavior law
(t–ε) of a particular bone gap through the healing process
(αo1).
In these figures, the parts of the graphs with negative
slope represent the Eq. (5) (Fig. 1b) and 6 (Fig. 1c). In this part
of the graph, α decreases (α_ o0) because the limit strain (εα) is
reached (for example, due to a very high load). If this fact
occurs during the healing process, the healing will be
impaired or at least delayed.
Therefore, we finally select as union degree, α, the mini-
mum value associated to each direction:
α ¼ min fαs ; αt ; αn g ð7Þ
2.3. Biological union: growth of union degree α
The union degree, α, can increase due to bone healing, and
therefore it is calculated as the addition of two contributions:
α ¼ αc þ αb ð8Þ
where αc reflects the recovery of the mechanical properties of
the fracture gap mainly due to the formation of cartilage, and
αb reflects the recovery of the mechanical properties of the
fracture gap due to bone formation.
In order to estimate the temporal evolution of α
ðα_ ¼ α_ c þ α_ b Þ, we propose a regulatory law based on three
different healing zones (adapted from Claes and Heigele, 1999)
which are established as  a function  compression, εn,
of ffi the
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and the shear strains εshear ¼ ε2s þ ε2t of each interface
element (Fig. 2a) with the following characteristics:
Zone 1 non-union: The compression and shear strains are so
high that there is a high interfragmentary movement (IFM)
which impairs a successful fracture healing. Only fibrous
tissue is formed and there is no change in αc nor in αb
ðα_ c ¼ α_ b ¼ 0Þ
Zone 2 cartilage formation: The compression and shear
strains are moderate, in the same way that the IFM. We
consider that in those elements, whose strains belong to this
zone, cartilage will be formed and therefore αc will grow.
However, these elements should be under this strain thresh-
old during a certain period of time (Mc) before the growing of
αc, this time represents the maturation time needed for the
cartilage to appear close to the fracture gap.
It is established that αc will have an exponential evolution
with the time (Fig. 2b), therefore once the maturation time for
cartilage (Mc) is reached, the value of αc will grow exponen-
tially, providing that the strains keep in zone 2, during a fixed
period of time, tc (cartilage formation time), until it reaches
the maximum value (αcmax). If during the healing process, αc is
 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338
growing and the strains reach the zone 3, αc will continue
growing with the same rate until the maximum value (αcmax).
Zone 3 bone formation: In this zone strains and IFM are
lower than in zone 2, as a result, bone is formed, so αb will
grow. We also define, Mb, as the maturation time for the
formation of bone. It is established that αb will have a linear
evolution with the time (Fig. 2c), therefore once the matura-
tion time for cartilage (Mb) is reached, the value of αb will
Fig. 2 – (a) Healing zones, (b) temporal evolution of αc and
(c) temporal evolution of αb.
Table 1 – Examples of model application.
Goal Experiment
Calibration process Compression
Shear
Evaluation of the model predictive capacity Compression
Combined load case
Comminuted fracture
331
grow linearly, providing that the strains keep in the zone 3,
until it reaches the maximum value (αbmax) at the total
healing time, th. If αb begins to grow because the conditions
for bone formation are achieved and αc is still growing, a
simultaneous growing of both contributions will happen.
According to the definition of α, its maximum value is 1
(corresponding to a totally successful bone healing), therefore
the limits for αc (αcmax) and αb (αbmax) must satisfy the
following equation:
αmax ¼ αb max þ αc max ¼ 1 ð9Þ
The model parameters values represented in Fig. 2 (Lsh sh
c , Lb ,
comp comp
Lc , Lb , αcmax, αbmax, Mc, Mb, tc, th) will be later defined by
tuning the model to experimental results.
The model was numerically implemented in an Abaqus
user subroutine (UMAT) and all the finite element analyses
(FEA) were carried out in ABAQUS v.6.11.
2.4. Examples of application for the calibration process
In order to determine the value of the parameters that define
the model, we made a quantitative comparison between
the numerical results (obtained from our model) and the
experimental ones found in the literature. With this idea in
mind, we simulated two animal experiments (compression
and shear) from the literature that are detailed below and
summarized in Table 1.
2.4.1. Compression experiments
To estimate the compression parameters (Lcomp
c , Lcomp
b , αcmax,
αbmax, Mc, Mb, tc, th), we simulated the compression experi-
ments performed by Claes et al. (1997). They determined the
influence of the size of the osteotomy gap and interfragmen-
tary strain on callus formation in the healing of the sheep
metatarsus (Claes et al., 1997).
Therefore, a cylinder of external diameter of 17 mm
was simulated (Fig. 3a). The fracture gap was subjected to
an external compression load of 330 N (Claes et al., 1995). The
gap size was 2 mm and the initial interfragmentary strain
(IFS¼IFM/h) was approximately 31%, that was achieved
with the use of a fixator, simulated by four linear springs
with a total stiffness of 348 N/mm. In order to avoid rigid-
body motions the nodes at the base of the model were fully
constrained.
Bone was considered as a linear elastic material for
all analyses where the mechanical properties are shown in
Table 2.
Bone Load condition Reference
Sheep metatarsus Compression Claes et al. (1997)
Sheep tibia Shear Bishop et al. (2006)
Sheep metatarsus Compression Claes et al. (1997)
Human tibia Combined Wehner et al. (2010)
Sheep metatarsus Combined –
332 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338
Fig. 3 – FE model for the simulation of (a) compression and (b) shear experiments in order to calibrate the corresponding
parameters of the model.
Table 2 – Material properties of the tissue considered for
all analyses.
Tissue type Elastic modulus in MPa Poisson's ratio
Cortical bone 10,000a 0.2a
Trabecular bone 200a 0.2a
a
Hernandez et al. (2001).
2.4.2. Shear experiments
In a similar way, it is necessary that the parameters of the
model allow also to simulate correctly the healing process
under shear conditions. However, no experiments in the
literature have been found in which, the time course of the
interfragmentary movement (IFM) during the healing is eval-
uated under a pure shear load case. Nevertheless Bishop et al.
(2006) performed experiments studying the healing of a
transverse tibial osteotomy in sheep where interfragmen-
tary torsional shear (torsion) was applied under carefully
controlled conditions and some data of this experiment are
used for the estimation of the rest of the parameters of our
model (Lsh sh
c , Lb ).
We simulated these experiments taking the geometry and
the loads from Bishop et al. (2006), (Fig. 3b). A cylinder of
external diameter of 20 mm was simulated. A torsional
rotation (torsion) was applied over a 2.4 mm osteotomy with
a maximum interfragmentary principal strain magnitude of
25%. This corresponded to a 7.21 rotation calculated in the
outer cortical radius of 10 mm. A load limit of 1670 N  mm
was imposed to allow interfragmentary strain to decrease
with increasing callus stiffness. In order to avoid rigid-body
motions the nodes at the base of the model were constrained.
Table 3 – Fixation stiffness for each experiment simu-
lated in the compression predictive simulation.
Experiment Stiffness (N/mm)
Initial gap ¼1 mm; IFS ¼ 7% 3300
Initial gap ¼1 mm; IFS ¼ 31% 764
Initial gap ¼2 mm; IFS ¼ 7% 1650
To calibrate the model two data sets in the related article
(Bishop et al., 2006) were used:
– The stiffness increased steadily during the experiment
with the load limit achieved within the fourth week.
– After eight weeks of healing, the bending stiffness was
approximately the 80% of the intact bone. The union
degree α, is the variable that represents the improvement
of the stiffness and then is directly related to the bending
stiffness.
2.5. Examples of application for the evaluation of the
model predictive capacity
In order to evaluate the predictive capacity of the model,
three additional experiments (compression, combined load
case and comminuted fracture) that are summarized in
Table 1 were simulated using the parameters previously
obtained in the calibration process.
2.5.1. Compression experiments
Three compression experiments, similar to the described
in the model calibration section (Fig. 3a), were performed by
 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338 333

Claes et al. (1997). The differences among them were the gap
dimension and the initial IFS. In fact, gaps of 1 mm and 2 mm
and initial IFS of 7% and 31% were fixed in these experiments.
We simulated these three experiments with the same FE
model used in the model calibration section, changing the
gap dimension and the stiffness of the fixator (Table 3). The
same loading and boundary conditions were used.
To simulate the experiment, we applied a vertical load of
700N (corresponding to a single leg stance of a patient of
approximately 70 kg (Wehner et al.,2010)) at the proximal
condyles of the tibia (Fig. 4b). Due to this load, a combined
load state (compression-shear) was generated in the fracture
gap zone. The nodes of the base were constrained so that, the
rigid-body motions are avoided.

2.5.2. Combined load case experiment

2.5.3. Comminuted fracture
We found no more data in the literature of pure shear
A complex fracture was simplified and simulated in a sheep
experiments of fracture healing than those used in the
metatarsus. The comminuted fracture modeled consisted of
calibration (Bishop et al., 2006). That fact prevented us from
an oblique fracture with three fragments (Fig. 5). The loads,
validating our model exclusively with this type of loading
fixation system and bone geometry were the same that those
condition.
used in the Section 2.4.1.
However, Wehner et al. (2010) performed a clinical experi-
Due to the complex geometry of the fracture zone the
ment in a human patient with a fracture in the diaphyseal
vertical load applied generated combined mechanical
tibia. A combined compression-shear load case was applied,
demands in the different fracture gaps.
and at several time points during healing, the IFM was
Four different simulations were made changing the stiff-
determined during a single leg stance. Thanks to these
ness of the external fixator in order to get the best fixation
available data, simulating this experiment we were able to
evaluate the potential of our model under a complex com-
bined load case.
A human tibia was scanned and the images stored in
Dicom format. MIMICS V. 15.0 was used to reconstruct the
geometry (Fig. 4a) through the segmentation process. After-
wards, using this software, an osteotomy was made in the
diaphyseal part. In addition, a unilateral external fixator was
modeled according to the specifications shown in Wehner
et al. (2010). This fixator, once imported in MIMICS, was
implanted in the tibia, and the assembly was meshed with
tetrahedral elements. We used ABAQUS to create cohesive
elements in the fracture gap. The final FE model had a total
number of nodes of 19,499 and a total number of elements of
91,558 (118 linear wedge cohesive elements and 91,440 linear
tetrahedral elements). The element size used was inside the
asymptotic region of convergence and represents a good
trade off between numerical accuracy and computational
cost (Fig. 4c).

Fig. 5 – FE model for the simulation of the comminuted

fracture.

1.6
Simulated
1.4
Experimental (Claes,1997)
1.2

1
IFM (mm)

0.8

0.6

0.4

0.2

0
0 10 20 30 40 50 60
Healing time (days)

Fig. 6 – Compression calibration results: evolution of the

Fig. 4 – Predictive simulation using combined load case: interfragmentary motion in the computational simulation
(a) geometry, (b) applied loads and (c) FE of the fracture compared with the experimental data in a 2-mm gap
gap zone. of a sheep (Claes et al., 1997).
334 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338

Table 4 – Model parameters.

Parameter Value

Mc—Maturation time for cartilage (days) (see Fig. 2b) 4a

Mb—Maturation time for bone (days) (see Fig. 2c) 10b
αcmax—Maximum value of αc (see Fig. 2b) 0.3b
αbmax—Maximum value of αb (see Fig. 2c) 0.7b
tc—Cartilage formation time (days) (see Fig. 2b) 36b
th—Total healing time (days) (see Fig. 2c) 60c
Lcomp
c —Compression strain limit for cartilage formation (see Fig. 2a) 0.6b
comp
Lb —Compression strain limit for bone formation (see Fig. 2a) 0.25b
Lsh
c —Shear strain limit for cartilage formation (see Fig. 2a) 0.25b
Lsh
b —Shear strain limit for bone formation (see Fig. 2a) 0.15b
K0i (i ¼n,s,t)—Initial linear stiffness (MPa) 50d
ε0i (i¼ n,s,t)—Maximum strain at the linear region 0.3b
εci (i¼n,s,t)—Maximum allowed strain 1b
a
Park et al. (1998).
b
Estimated.
c
Klein et al. (2004).
d
Bishop et al. (2003).

system for this fracture. The stiffness of the fixator used for results are shown in Table 4. Some of the model parameters
each simulation was 400, 448, 500 and 560 N/mm. come from previous studies of the literature and the rest are
estimated with the results of Sections 3.1.1and 3.1.2.

3. Results
3.2. Evaluation of the model predictive capacity—results

Here, in first step, we discuss the results of the calibration

3.2.1. Compression experiments results
process, and next, we also explain additional experiments to
The experimental results obtained by Claes et al. (1997) for
evaluate the predictive capacity of the model.
different fracture gap sizes and IFS have been represented in
Fig. 7, where we can also see the experimental range of
3.1. Calibration process results
variation. In addition, we include in this figure the numerical
results corresponding to the computational simulations.
3.1.1. Compression experiments results
As we can see, a good agreement is found between numerical
Evolution of the IFM obtained experimentally (Claes et al.,
and experimental results for all the studied conditions.
1997) and computationally has been represented in Fig. 6
for a 2-mm gap of a sheep. The experimental variability
3.2.2. Combined load case experiment results
has been also represented and the computational results
Experimental results from Wehner et al. (2010) obtained for a
are within previous range of variation. Initially, there is no
combined load case have been represented and compared
fracture healing, because the IFM is relatively high (1 mm),
with the computational predicted ones in Fig. 8. Although
as the simulation evolves, the IFM is reduced. After 40 days
there were no more experimental data in order to represent
of healing, the IFM has been reduced by a ratio of 80%
its range of variation, our computational results are very
(see Fig. 6).
close to the experimental ones.
The IFM compression is very high at the initial healing
3.1.2. Shear experiments results
stages, and as the healing process evolves the IFM is reduced
The temporal evolution of the union degree, α, of the fracture
both experimentally and computationally due to the com-
gap has been estimated under a torsional load case. We have
bined load case. However, in the case of shear loads, the
been able to correctly reproduce the experimental results
shear IFM slightly decreases from the first days until the last
obtained by Bishop et al. (2006). Experimentally the torsion
days of healing (Fig. 8).
load limit of 1670N  mm is achieved within the fourth week,
using our model, we estimate1 that this load limit is achieved
the 32nd day. 3.2.3. Comminuted fracture results
The temporal evolution of the vertical displacement of the
3.1.3. Parameters of the model fixation points has been represented for four different fixators
(Fig. 9). It can be observed that the stiffness value that leads
Using the results obtained in the calibration process, the
to better results is K ¼ 500 N/mm, because the final interfrag-
parameters that best fit the simulated and the experimental
mentary displacements are lower, which means better healing
1
We determine the torsion load limit as the resultant torsion conditions.
moment, which is computed by means of the integration of the We have also represented the temporal evolution of the
shear traction stress vector over the bone section. bone reaction (Fig. 10) i.e. the load that the bone is able to
 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338 335

0.6
0.25

0.5 Simulated
0.2 Simulated
Experimental (Claes,1997)
0.4 Gap=2 mm; Initial IFS=7%
Experimental (Claes,1997)

IFM (mm)
IFM (mm)

0.15 Gap=1mm; Initial IFS=7%

0.3

0.1
0.2

0.05 0.1

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Healing time (days) Healing time (days)

0.3

Simulated
0.25 Experimental (Claes,1997)
Gap=2 mm; Initial IFS=7%
0.2
IFM (mm)

0.15

0.1

0.05

0
0 10 20 30 40 50 60
Healing time (days)

Fig. 7 – Evolution of the interfragmentary motion in the computational simulation compared with the experimental data
(Claes et al., 1997): (a) gap¼ 1 mm; initial IFS¼ 7%, (b) gap¼ 1mm; initial IFS ¼31%, and (c) gap¼ 2 mm; initial IFS¼7%.

1.6 Experimental IFM compression

(Wehner et al., 2010)
1.4 If we compare the different stiffness able to heal a simple
Experimental IFM shear
(Wehner et al., 2010) transverse fracture (K¼ 348 N/mm see Section 2.4.1) and a
1.2 Simulated IFM compression
comminuted fracture (K¼ 500 N/mm) under the same loading
1 Simulated IFM shear conditions, we can observe that the second one requires a
IFM (mm)

0.8 higher stiffness to achieve successful healing.

0.6

0.4
4. Discussion
0.2

0 In this study we propose a coupled computational model able

0 10 20 30
Healing time (days)
Fig. 8 – Predicted time course of the interfragmentary
movement (IFM) due to compression and shear stresses
compared to in vivo measurements of the patients (Wehner
et al., 2010).
bear during the healing process. The best result is for a
stiffness of K ¼500 N/mm, however it can be seen that K ¼560
N/mm (more rigid fixation) gives better results in the first
healing stages probably due to the primary bone formation.
The more flexible fixation systems give poorer results prob-
ably because the strains are too high and the healing process
is not successful.
40 50 60 70
to simulate the normal and the impaired (mechanical failure)
bone healing regulated by the mechanical stability or the
interfragmentary movement in the fracture gap. Indeed, the
model predicted the temporal recovery/failure of mechanical
properties of the fracture gap during the healing. The aim of
this model is to identify the role of different and combined
degrees of stabilization for fracture fixation. The model here
presented is based on the main mechanisms that regulate the
biology of fracture healing. In fact, bone heals by either direct
intramembranous ossification or indirect fracture healing,
which consists of both intramembranous and endochondral
ossification (Marsell and Einhorn, 2011). Therefore, this model
combines both approaches, modeling both direct and indirect
ossification or equivalently intramembranous and endochon-
dral ossification.
336 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338
Fig. 9 – Comminuted fracture results: temporal evolution of the vertical displacement of the fixation points (see position
of these points in Fig. 5) for different fixation stiffness.
Fig. 10 – Temporal evolution of bone reaction (N) for different
values of stiffness of the fixator.
Primary healing (intramembranous ossification) requires
extremely rigid fixation systems and negligible spaces
between bone fragments. The quality of regenerated bone
in primary healing cannot be always guaranteed and there-
fore a new fracture may occur. In those parts of the fracture
gap in which, only primary healing occurs, we assume that
there will be only the temporal evolution of one part (αb) of
the union degree (α) (contribution of α mainly due to the bone
formation).
On contrast, secondary healing (endochondral ossifica-
tion) takes place under more flexible fixation that allows
a higher, but controlled, instability of the fracture gap region.
In this second step, bone formation is involved restoring the
mechanical stability and recovering the mechanical proper-
ties in the bone fracture gap. In any case, secondary healing is
the most common in clinical treatments, since the high
quality of the regenerated tissue and also the probability of
a successful healing is higher than in primary healing
(Marsell and Einhorn, 2011). Therefore, when secondary
healing occurs we assume a double coupled pathway: first
we consider that there will be evolution of αc (contribution of
α due mainly to the cartilage formation) secondly this
cartilage will be replaced by bone (increasing αb).
We have to keep in mind that our model is based on some
simplifications, which present relevant implications for the
conclusions and the future applications.
First, due to the cohesive elements used in its implemen-
tation, it can only be used with fractures where the dimen-
sion of the gap is thin enough to consider it negligible with
respect to the overall dimensions of the bone fracture. There-
fore, we have to keep in mind that this model is not valid for
large bone defects.
Secondly, different mechano-biological models (Andreykiv
et al., 2008; Checa and Prendergast, 2009; Isaksson et al., 2008,
2009a; Lacroix and Prendergast, 2002b; Loboa et al., 2001;
Reina-Romo et al., 2011; Shefelbine et al., 2005; Simon et al.,
2011; Wehner et al., 2010) have been proposed where a
detailed study of the whole callus is performed, analyzing
 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338
how different mechanical stimuli are spatially distributed in
the callus in order to understand the fundamental cellular
mechanisms that locally regulate tissue differentiation and
callus growth. However, in this work, we do not study how
the mechanical stimulus is distributed in the whole callus
dominium, but we take into account the temporal progres-
sion of the callus stiffness in the fracture gap regulated by the
mechanical environment in this gap.
Thirdly, we have considered the following time-dependent
assumptions for the temporal evolution of the union degree α:
(1) an exponential evolution of the stiffness during the
formation of the cartilaginous structure (αc) and (2) a linear
evolution of the gap rigidity due to bone formation (αb).
Both assumptions are based on previous experiments that
we observed in-vivo for sheep animals in tibia fractures
(Gómez-Benito et al., 2011).
Fourthly, we have estimated the parameters of the model
(calibration process, Section 2.4.) using experiments of bone
healing in sheep and, afterwards, we have used these para-
meters in the predictive simulations (Section 2.5.) to simulate
the experiments of bone healing both in human and sheep,
obtaining good agreement in both cases. However, in the
clinical application of this model to simulate fracture healing
in human fractures, a finest estimation of the parameters
should be done taking into account different aspects like sex,
age, weight, etc.
Finally, this model is only regulated by mechanical factors;
however, as it is well known that bone healing is also
regulated by other factors (chemical, genetic, biological etc.).
This model could be considered as a useful first approach
that could be improved in future works including these
additional factors.
Therefore, this work presents a simple model able to describe
the temporal evolution of the bone fracture mechanical proper-
ties under different mechanical stability conditions. In addition,
this model is very versatile for the simulation of realistic bone
fracture geometries (oblique, comminuted, spiral and com-
pound) allowing a whole-organ analysis. Therefore, this model
can be used to study and improve healing of bone fragments
stabilized by different fixation systems, such as, locking plates,
nails, external fixators and intramedullary screws.
Acknowledgments
The research leading to these results has received funding
from the (European Commission) Seventh Framework
Programme (FP7/2007-2013) under grant agreement nº286179.
r e f e r e n c e s
Ament, C., Hofer, E.P., 2000. A fuzzy logic model of fracture
healing. Journal of Biomechanics 33, 961–968.
Andreykiv, A., van Keulen, F.V., Prendergast, P.J., 2008. Simulation
of fracture healing incorporating mechanoregulation of tissue
differentiation and dispersal/proliferation of cells.
Biomechanics and Modeling in Mechanobiology 7, 443–461.
Augat, P., Burger, J., Schorlemmer, S., Henke, T., Peraus, M., Claes, L.,
2003. Shear movement at the fracture site delays healing in a
337
diaphyseal fracture model. Journal of Orthopaedic Research 21,
1011–1017.
Bishop, N.E., Schneider, E., Ito, K., 2003. An experimental two
degrees-of-freedom actuated external fixator for in vivo
investigation of fracture healing. Medical Engineering and
Physics 25, 335–340.
Bishop, N.E., Rhinjn, M.v., Tami, I., Corveleijn, R., Schneider, E.,
Ito, K., 2006. Shear does not necessarily inhibit bone healing.
Clinical Orthopaedics and Related Research 443, 307–314.
Byrne, D.P., Lacroix, D., Prendergast, P.J., 2011. Simulation of
fracture healing in the tibia: mechanoregulation of cell activity
using a lattice modeling approach. Journal of Orthopaedic
Research, October, 1496–1503.
Claes, L.E., Wilke, H.-J., Augat, P., Rübenacker, S., Margevicius, K.J.,
1995. Effect of dynamization on gap healing of diaphyseal
fractures under external fixation. Clinical Biomechanics 10 (5),
227–234.
Claes, L.E., Augat, P., Suger, G., Wilke., H.-J., 1997. Influence of size
and stability of the osteotomy gap on the success of fracture
healing. Journal of Orthopaedic Research 15, 577–584.
Claes, L.E., Heigele, C.A., 1999. Magnitudes of local stress and
strain along bony surfaces predict the course and type of
fracture healing. Journal of Biomechanics 32, 255–266.
Comiskey, D., MacDonald, B.J., McCartney, W.T., Synnott, K.,
O’Byrne, J., 2013. Predicting the external formation of callus
tissues in oblique bone fractures: idealised and clinical case
studies. Biomechanics and Modeling in Mechanobiology http:
//dxdoi.org/10.1007/s10237-012-0468-6.
Chao, E., Aro, H., DGLewallen, Kelly, P., 1989. The effect of rigidity
on fracture healing in external fixation. Clinical Orthopaedics
and Related Research 241, 24–35.
Checa, S., Prendergast, P., 2009. A mechanobiological model for
tissue differentiation that includes angiogenesis: a lattice-
based modeling approach. Annals of Biomedical Engineering
37 (1), 129–145.
Draper, E.R.C., Strachan, R.K., Hughes, S.P.F., Nicol, A.C., Paul, J.P.,
1997. The design and performance of an experimental external
fixator with variable axial stiffness and a compressive force
transducer. Medical Engineering and Physics 19 (8), 660–695.
Einhorn, T.A., 2005. The science of fracture healing. Journal of
Orthopaedic Trauma 19 (10), S4–S6.
García-Aznar, J.M., Kuiper, J.H., Gómez-Benito, M.J., Doblaré, M.,
Richardson, J.B., 2007. Computational simulation of fracture
healing: Influence of interfragmentary movement on the
callus growth. Journal of Biomechanics 40, 1467–1476.
García-Aznar, J.M., Pérez, M.A., Moreo, P., 2009. Modelling of
interfaces in biomechanics and mechanobiology. Computer
Modeling in Engineering & Sciences: CMES 48 (3), 271–301.
Gómez-Benito, M.J., García-Aznar, J.M., Kuiper, J.H., Doblaré, M.,
2005. Influence of fracture gap size on the pattern of long bone
healing: a computational study. Journal of Theoretical Biology
235, 105–119.
Gómez-Benito, M.J., González-Torres, L.A., Reina-Romo, E., Grasa, J.,
Seral, B., García-Aznar, J.M., 2011. Influence of high-frequency
cyclical stimulation on the bone fracture-healing process:
mathematical and experimental models. Philosophical
Transactions Series A, Mathematical, Physical, and Engineering
Sciences 369, 4278–4294.
Goodship, A.E., Kenwright, J., 1985. The influence of induced
micromovement upon the healing of experimental tibial
fractures. Journal of Bone and Joint Surgery (British Volume)
67-b (4), 650–655.
Goodship, A.E., Watkins, P.E., Rigby, H.S., Kenwright, J., 1993. The role
of fixator frame stiffness in the control of fracture healing. An
experimental study. Journal of Biomechanics 26 (9), 1027–1035.
Hadjiargyrou, M., McLeod, K., Ryaby, J., Rubin, C., 1998.
Enhancement of fracture healing by low intensity ultrasound.
Clinical Orthopaedics and Related Research 355 (S), 216–229.
338 journal of the mechanical behavior of biomedical materials 29 (2014) 328 –338
Hernandez, C.J., Beaupre, G.S., Keller, T.S., Carter, D.R., 2001. The
influence of bone volume fraction and ash fraction on bone
strength and modulus. Bone 29 (1), 74–78.
Isaksson, H., Donkelaar, C.C.v., Huiskes, R., Ito, K., 2008. A
mechano-regulatory bone-healing model incorporating cell-
phenotype specific activity. Journal of Theoretical Biology 252,
230–246.
Isaksson, H., Donkelaar, C.C.v., Ito, K., 2009a. Sensitivity of tissue
differentiation and bone healing predictions to tissue
properties. Journal of Biomechanics 42, 555–564.
Isaksson, H., Gröngröft, I., Wilson, W., Donkelaar, C.C.v.,
Rietbergen, B.v., Tami, A., Huiskes, R., Ito, K., 2009b.
Remodeling of fracture callus in mice is consistent with
mechanical loading and bone remodeling theory. Journal of
Orthopaedic Research 27 (5), 664–672.
Jagodzinski, M., Krettek, C., 2007. Effect of mechanical stability on
fracture healing – an update. Injury 38 (S1), S3–S10.
Klein, P., Opitz, M., Schell, H., Taylor, W.R., Heller, M.O., Kassi, J.-P.,
Kandziora, F., Duda, G.N., 2004. Comparison of unreamed nailing
and external fixation of tibial diastases – mechanical conditions
during healing and biological outcome. Journal of Orthopaedic
Research 22, 1072–1078.
Lacroix, D., Prendergast, P.J., 2002a. Three-dimensional
simulation of fracture repair in the human tibia. Computer
Methods in Biomechanics and Biomedical Engineering 5 (5),
369–376.
Lacroix, D., Prendergast, P.J., 2002b. A mechano-regulation model
for tissue differentiation during fracture healing: analysis of
gap size and loading. Journal of Biomechanics 35, 1163–1171.
Loboa, E.G., Beaupré, G.S., Carter, D.R., 2001. Mechanology of
initial pseudarthrosis formation with oblique fractures.
Journal of Orthopaedic Research 19, 1067–1072.
Manjubala, I., Liu, Y., Epari, D.R., Roschger, P., Schell, H., Fratzl, P.,
Duda, G.N., 2009. Spatial and temporal variations of
mechanical properties and mineral content of the external
callus during bone healing. Bone 45, 185–192.
Marsell, R., Einhorn, T., 2011. The biology of fracture healing.
Injury 42 (6), 551–555.
Moreo, P., Pérez, M.A., García-Aznar, J.M., Doblaré, M., 2007.
Modelling the mechanical behaviour of living bony interfaces.
Computer Methods in Applied Mechanics and Engineering
196, 3300–3314.
Park, S.-H., O’Connor, K., McKellop, H., Sarmiento, A., 1998. The
influence of active shear or compressive motion on fracture
healing. Journal of Bone and Joint Surgery (American Volume)
80-A (6), 868–878.
Perren, S.M., 2002. Evolution of the internal fixation of long bone
fractures. Journal of Bone and Joint Surgery (British Volume)
84-B (8), 1093–1110.
Reina-Romo, E., Gómez-Benito, M.J., Domínguez, J., Niemeyer, F.,
Wehner, T., Simon, U., Claes, L.E., 2011. Effect of the fixator
stiffness on the young regenerate bone after bone transport:
computational approach. Journal of Biomechanics 44, 917–923.
Richardson, J.B., Cunningham, J.L., Goodship, A.E., O’Connor, B.T.,
Kenwright, J., 1994. Measuring stiffness can define healing of
tibial fractures. Journal of Bone and Joint Surgery (British
Volume) 76-B (3), 389–394.
Schell, H., Epari, D.R., Kassi, J.P., Bragulla, H., Bail, H.J., Duda, G.N.,
2005. The course of bone healing is influenced by the initial
shear fixation stability. Journal of Orthopaedic Research 23,
1022–1028.
Shefelbine, S.J., Augat, P., Claes, L., Simon, U., 2005. Trabecular
bone fracture healing simulation with finite element analysis
and fuzzy logic. Journal of Biomechanics 38, 2440–2450.
Simon, U., Augat, P., Claes, L., Utz, M., 2011. A numerical model of
the fracture healing process that describes tissue
development and revascularisation. Computer Methods in
Biomechanics and Biomedical Engineering 14 (1), 79–93.
Vetter, A., Epari, D.R., Robin Seidel, H.S., Fratzl, P., Duda, G.N.,
Weinkamer, R., 2010. Temporal tissue patterns in bone
healing of sheep. Journal of Orthopaedic Research 28 (11),
1440–1447.
Wehner, T., Claes, L., Niemeyer, F., Nolte, D., Simon, U., 2010.
Influence of the fixation stability on the healing time—a
numerical study of a patient-specific fracture healing process.
Clinical Biomechanics 25, 606–612.
Wehner, T., Penzkofer, R., Augat, P., Claes, L., Simon, U., 2011.
Improvement of the shear fixation stability of intramedullary
nailing. Clinical Biomechanics 26, 147–151.