You are on page 1of 21

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329654233

Finite Element Analysis in Dentistry

Chapter · December 2018


DOI: 10.1142/9789813225688_0003

CITATIONS READS

3 559

6 authors, including:

Josete B C Meira Pavel Capetillo


University of São Paulo University of São Paulo
68 PUBLICATIONS   1,163 CITATIONS    5 PUBLICATIONS   72 CITATIONS   

SEE PROFILE SEE PROFILE

Marina G. Roscoe Paolo M Cattaneo


University of São Paulo University of Melbourne
38 PUBLICATIONS   357 CITATIONS    107 PUBLICATIONS   2,308 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Non-carious cervical lesions - oral health View project

Amalgam View project

All content following this page was uploaded by Rafael Yagüe Ballester on 02 September 2019.

The user has requested enhancement of the downloaded file.


CHAPTER 3

FINITE ELEMENT ANALYSIS IN DENTISTRY

Josete B. C. Meira1; Alice N. Jikihara1; Pavel Capetillo1; Marina G. Roscoe2; Paolo M.


Cattaneo3; Rafael Y. Ballester1
1 - Department of Biomaterials and Oral Biology, University of São Paulo
São Paulo, SP, Brazil
E-mail: jo@usp.br
2 - Department of Orthodontics and Restorative Dentistry, University of Guarulhos
São Paulo, SP, Brazil
3 - Section of Orthodontics, Department of Dentistry, Faculty of Health Science, Aarhus
University, Aarhus, Denmark

Finite element analysis (FEA) presents a wide range of application in Dentistry.


FEA models can precisely calculate the material stress in conditions of geometry
and boundaries that can properly represent the clinical reality. However, as any
research model, the FEA ones requires some simplifications to be feasible. The
researcher challenge is to distinguish between the necessary simplifications and
the misrepresentative ones. This chapter presents some information to help FEA
users to improve their FEA result interpretation. In addition, it will help readers
non familiar with this technique to understand the power of this tool, yet being
able to critically analyzing the published FEA studies.

1 Introduction
The experimental tests explored in the previous chapter enable the mechanical
characterization of dental materials and the comparison between analogous
materials in standard conditions. Once the mechanical characteristics are set, it is
possible to go forward in predicting the material behavior a “quasi-real” scenario
by using finite element analysis (FEA). The real shape of test object and actual
loading application can be represented in the FEA model [1, 2]. FEA has been used
to provide a better understanding of experimental tests [3], still this chapter will
focus on the “clinical scenario” studies.
In the last decade, there was a large increase in the number of FEA papers in
Dentistry. It is expected that, in this large pool of articles, we can find studies with
important errors in model generation and/or with fundamental flaws in the result
interpretation. The great challenge for the readers is to recognize the studies with

1
2 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

major faults and critically evaluate the limitations of the generated conclusions.
Still, it is important to remember that finite element models are always simplified
representations of the reality. Experimental models are also simplifications of the
reality. It is possible to say that all models present some limitations. Therefore, the
researcher challenge is to distinguish between the necessary or required
simplifications from the misrepresentative ones, either for experimental or FEA
models.

2 How the Model Geometry is Defined and Obtained?

2.1 Two or Three-dimensional Models?

A consistent finite element analysis can be developed either with a simple two-
dimensional (2D) geometry or with a more complex three-dimensional (3D)
patient-based model. Between these two extremes, there is a wide range of
geometric complexity that can be used for modeling. Frequently, models of teeth
with 2D geometry are criticized due to errors generated by limitations inherent to
2D models (see section 2.2). However, planar representations are still well
accepted in studies involving the simulation of complex phenomena, such as crack
propagation [4], and fatigue [5]. Two-dimensional models are also recommended
in the early stages of a simulation. The strategy is to start the investigation with
simple models and then gradually increase the degree of complexity throughout
the study.
The construction of a 2D geometry is easily performed with the CAD features of
the finite element software. In patient-based models, the geometry is obtained
using modern methods of image acquisition (see section 2.3). These models have
become increasingly common, due to technological advances in medical imaging
technology and computation fields. Equipment and technology, previously
restricted to selected research groups, have become more widespread in the dental
clinics and in dental academia. However, it is important to notice that the results
obtained in a model based on precise measurements of a single patient might be
more limited in respect to the generalization of their findings than those obtained
in anatomically simplified models with average measurements of a population. In
addition, excess of anatomical details can create "noise" that may compromises or
conceal the relevant results.
Characterization of dental biomaterials by Finite element analysis 3

2.2 Limitations of 2D models

In 2D FEA models, the geometry is drawn in two dimensions, but the software
considers that the real object is in fact a 3D entity. The plane models (plane stress
and plane strain) assume that the object has the same geometry along the non-
represented axis (z-axis), while the axisymmetric models assume a 360 revolution
around the symmetry axis (Fig. 1). Furthermore, the plane stress model considers
that the thickness in z direction is very small when compared to x and y
dimensions, which makes the stress in z insignificant. It justifies the simplification
of z = 0 used for these models. On the other hand, the plane strain models consider
that the thickness in z is very large, which makes the strain in z insignificant.
Therefore, z = 0.
Because of these geometric assumptions, both plane models are not able to
represent the tensile hoop stresses generated in the real root due to the intrusion of
an intraradicular post for an example [6]. However, axisymmetric models can
satisfactorily represent this wedge effect, since they reproduce the real structural
stiffness of the root. The main limitation of this model is its inability to simulate
clinically relevant oblique loading conditions.

Fig. 1. Two-dimensional models of a root with an intraradicular post. In plane stress and plane strain
models no hoop tensile stress is generated, due to the interpretation of the software regarding the real
object geometry. In axisymmetric model, the hoop tensile stress is present.
4 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

2.3 Strategies for geometry acquirement

The anatomic geometry can be obtained by different approaches. In this chapter,


we focused in three geometry sources: literature data, 3D scanners, and computer
tomography (CT). Regardless the image data source, the anatomic model is
converted in a parameterized model by the Computer Aided Design (CAD)
software and finally it is converted in a discretized model by the FEA software
(Fig. 2).

Fig. 2. Overview of different approaches used for FEA models construction

2.3.1 Geometry based on literature data

The geometric shape and measurement data of anatomic structure is collected from
literature. A well-known source for dental anatomy parameters is available in
Wheeler’s Dental Anatomy, Physiology, and Occlusion [7]. The book presents, in
a grid of 1 mm2, accurate figures of tooth external contours from five standard
aspects: labial, lingual (or palatal), mesial, distal and occlusal (or incisal). Tables
with average values of key measurements of each type of tooth are also offered:
length of crown, length of root, mesiodistal diameter of crown, mesiodistal
diameter of crown at cervix, buccolingual diameter of crown, buccolingual
diameter of crown at the cervix, curvature of cementoenamel junction and
curvature of cementoenamel junction on distal. The authors also dedicate a chapter
Characterization of dental biomaterials by Finite element analysis 5

to pulp chambers and root canal, in which the internal anatomy of the teeth is
presented in details.
Drawing the anatomic model from this data requires high ability of the operator
with the CAD software. First, the outer and inner curves are drawn, then the outer
and inner surfaces are created from the curves and finally the solids are created,
based on the surfaces. During this process, it is interesting to divide the model in
strategic planes (for example, buccolingual, or mesiodistal) to facilitate the results
viewing. It is also important to avoid gaps and interpenetration of solids.
Otherwise, the operator will have problems to mesh the model after transferring it
to the FEA software.

2.3.2 Geometry based on 3D scanning

In the last decade, the CAD-CAM technology has become more and more
widespread in Dentistry, thus dentists and prosthetic technicians had the possibility
to familiarize with 3D scanners. There are three types of 3D scanners: laser, LED
light and contact. In general, resolution (number of dots per unit area measured by
the device), trueness (degree of closeness to the true value), precision (deviation
of measurements from the mean value), and accuracy (sum of trueness and
precision) does not depend directly on the type of scanner. Thus, different types
may have the same resolution and accuracy, depending on specific technical
characteristics of the appliance.
The accuracy of contact scanners is dependent on the capacity of the scanning tip
to reach the desired surface. Therefore, it can present limitations in copying some
details. Moreover, the object must necessarily be coupled to a support, so it is
restricted to extraoral scanning. A recent study [8] evaluated some parameters of
different types of extraoral 3D scanners and found values between 34.3 points/mm2
and 299.8 points/mm2 for the resolution, and 29 μm to 46 μm for trueness.
The data are typically recorded in a STL filea, a format original from the 3D-
printing technology, in which only the 3D-surface geometry is described in a
triangular mesh. Generally, the STL mesh guides the construction of the geometric
model, but not the FEA mesh, since it does not have sufficient quality, and/or is
characterized by triangles with compromised edges.
The acquisition of geometry by 3D scanning is limited to the external surfaces of
the object. In order to obtain the internal geometry of interest for FEA models it is
necessary to use another method of image acquisition, such as drawing from

a The STL has several after-the-fact backronyms such as "Standard Triangle Language" and
"Standard Tessellation Language".
6 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

literature data, as in item 2.3.1, or a combination with computed tomography data


[9].

2.3.3 Geometry based on computed tomography

There are two types of computed tomography (CT) that have been used for clinical
application in Dentistry: cone-beam-CT (CBCT) and fan-beam-CT (FBCT).
CBCT is more used as a complementary diagnostic test for guiding dental
treatment, especially in implantology. It allows a considerably decrease of the
radiation dose when compared to FBCT, reducing the patient exposure to ionizing
radiation. When details of tooth geometry are important for the FEA model, the
anatomical volume view has to be limited to the region corresponded to few teeth.
FBCT is usually indicated as a medical complementary test, and it can offer high-
resolution for a whole skull model, due to its higher radiation power.
The resolution of CTs is low when compared to that from micro-CTs (µCT). While
the voxel from CT is around 0.3 x 0.3 x 0.3 mm3, that from µCT is in the range of
1–50 μm. Therefore, µ-tomography is preferred for low scale-models (see section
2.4). However, micro-CT is not feasible for large size models due to dimensional
limitation of the equipment, in which only small objects can be scanned, and by
the fact that the radiation level is much higher, when compared to clinical
tomography. Therefore, these limits the use of µCT as an “in vivo” investigation
tool, at least for humans. Only geometries of extracted teeth models or from human
autopsy, excised bone samples, or small animal models can be scanned with this
strategy.
An important feature of CTs is that they can provide not only the external contours,
but also the internal geometry of the anatomical structures in a non-destructive and
non-invasive way. The method applies a segmentation process, based on
differences on pixel grey level values in the image, which is dependent on the
radiopacity of the tissue.
Recently, some specific software have been created and improved to convert
medical images of DICOMb files into 3D FEA models [10-12]. They offer intuitive
tools to provide fast and consistent image segmentation. Furthermore, its finite
element mesh preparation workflow allows the user to obtain high-quality
volumetric meshes that are more easily converted in finite element mesh by the
FEA software.

b DICOM is the abbreviation for Digital Imaging and Communications in Medicine and correspond
to a series of standardization of medical images in order to create a common language between
professionals and between medical imaging devices.
Characterization of dental biomaterials by Finite element analysis 7

2.4 Represented structures and model scale

The decision about the structures that should be represented depends on the
purpose of the study. Generally, all essential structures should be represented and
well discretized (see section 3). However, sometimes it is not so easy to define
which structures are essential for a specific study. If non-essential structures are
represented (or if the mesh density is exaggerated in regions of less interest - see
section 3), there will be no gains, but an unnecessary computational cost.
Let us give an example: for studying the stress polymerization of a class II
composite restoration, it is not useful to represent the whole root. Representing just
the cervical portion of the root ensures that the boundary restriction remains
sufficiently far away from the region of interest [13, 14].
If the researcher is unsure about the extension of the root that should be
represented, some preliminary exploratory analysis are recommended. The
strategy to be used in this process is similar to that used for obtaining the
convergence curve (see section 3): run a series of models with different root length
representation, and then analyze the selected result parameter at the region of
interest as a function of root extension. The ideal model will be the one in which a
minimal root length is represented for independent results.
Another important aspect related to the discretization process is that the model
must represent only compatible size structure. When representing a thin structure,
the suitable element will be very small, because the structure should have at least
three layers of elements. Therefore, if a complete large structure is represented in
the same model, the number of elements would be excessively high, which may
turns it unfeasible. What is the solution? Avoid representing structures of very
different order of magnitude in a single model.
For didactic and practical purposes, this chapter divided FEA models for dentistry
into three scale levels, according to the size of the represented structures (Fig. 3).
Still, sometimes it is impossible or unproductive to represent all the structures of
interest in a single model due to the computational cost. The multiscale modeling
represents a solution for those cases.
8 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

Fig. 3. Didactic classification of FEA model scales: Level A – simulation of a whole maxilla, jaw or
skull represented (example from Lui et al., 2015 [12]); Level B – simulation of a tooth or few teeth
(example from Benazzi et al., 2013 [15]); Level C – simulation of micro or nanostructure of a material
(example from Junior et al, 2012 [16]).

2.4.1 Level A

This scale refers to FEA models in which the whole maxilla, jaw or skull are
represented. This type of model is becoming more frequent due to the facilities in
acquisition of anatomic geometry from DICOM tomography files (see section
2.3.3) and increased advances in computer science. In this scale, the tooth is
usually represented as a single body, as if it was composed of a single material [10,
11, 17, 18]. In some cases, the distinction between cortical and trabecular bone is
not represented [12, 19, 20]. In general, only thick structures like tooth, bone, and
implants are represented in these models. Yet, in some studies the periodontal
ligament (of 0.2 to 0.3 mm thick) is also simulated [10, 11, 19, 20].

2.4.2 Level B

This scale refers to FEA models in which a tooth (or part of it) or a group of teeth
are represented. In this scale, the tooth is represented with more details, including
enamel, dentin and pulp. Depending on the study focus, restorations [21, 22],
cavities [23, 24] and intraradicular posts [25] are also represented. The bone is
Characterization of dental biomaterials by Finite element analysis 9

sometimes represented distinguishing cortical from trabecular [26] and sometimes


only the cortical layer is considered. Contrary to level A models, thinner structures
as 0.3 mm, such as periodontal ligament, are generally represented [26].

2.4.3 Level C

This scale refers to FEA models in which the micro or nanostructure of a material
or a multilayer adhesion is represented. The large increase of articles using this
scale in the last decade can be attributed to the possibility of obtaining
experimentally the mechanical properties of the materials in micro or nano scales
from tests like nanoindentation, atomic force microscopy, or micropillar
compression experiments.
This scale has been well explored in studies of adhesive interfaces in which
microscopic structures, such as intertubular dentin, hybrid layer, adhesive layer,
adhesive tags and smear plugs, are represented [16, 27]. Parameters such as
thickness of the layer of dentin infiltrated by adhesive, thickness of the adhesive
layer, etc. are based on images achieved from scanning electron microscopy. Yet,
the structures have been represented by primitive geometries (such as rectangles,
triangles, parallelepipeds, cylinders cones, spheres), ignoring the precise irregular
shapes disclosed by microscopy.

2.4.4 Multi-scale

In a multi-scale analysis, two or more models, with different scale levels, are
related. The results of the first one are used as an input for the next scale level in
two ways: upscaling or downscaling. The upscaling approach has been used in
studies of mechanical characterization of enamel [28-31] and bone [32], while the
downscaling approach has been used in studies of adhesion and dental trauma [33].

3 Mesh quality and mesh convergence


It is well known that a finer mesh results in a more accurate solution. However,
using the finest mesh in the whole model is usually a bad strategy, because
unnecessary calculation will be performed, producing a useless computational
cost. The mesh convergence study enables the acquisition of a mesh that
satisfactorily balances accuracy and rationalization of computing resources.
10 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

3.1 Reducing the element size

Generally, the mesh is refined by reducing the element size. A good approach in
FEA simulation is to start the study with a coarse mesh. Although this mesh gives
an inaccurate solution, it is very useful in the early stages of the analysis, when
some model verifications are required (e.g. check of the applied loads and
constraints). After these verifications, the mesh is systematically refined and the
convergence curve is created.
In order to obtain the convergence curve, a selected result parameter at the region
of interest is plotted as a function of a measure of “mesh density” (Fig. 4), such as
the number of elements in the whole model or in the region of interest. Although
the decrease of element size is somehow an indicator of increase in mesh density,
the element size should not be used in the x-axis of the convergence curve. The
convergence is achieved at the ideal mesh density point, in which the result does
not change significantly with the increase of the refinement. This is also a
guarantee that unnecessary calculation will not be performed.

Fig. 4. Mesh convergence curve

In some cases, each increase of mesh density in the convergence curve is achieved
by splitting all the elements in the model. However, the local mesh refinement is
preferable instead of refining the whole mesh model. In that case, it is important
to create transition regions to gradate the element size from coarse to fine meshes.
The elements away from regions of interest can be considerably larger than those
in regions of interest, without jeopardizing the analysis accuracy.
Characterization of dental biomaterials by Finite element analysis 11

Fig. 5. Mesh convergence by reducing element size: generalized and localized refinement.

3.2 Increasing element order

The mesh can also be refined by increasing the element order. In this case, the
geometry and the number of elements in the mesh are not modified. It continues
the same, but with a higher order element. For example, if a first order element
(like a linear triangular one, which has three nodes) has been used in the coarse
model, a second order element (like a parabolic triangular element, which has six
12 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

nodes) can be chosen for the refined model. This is an easy and fast way to refine,
because no remeshing is needed. However, as every element and region of the
model receives the same increase in calculating power, the computational costs
increase faster than with the local refinement, previously presented. This technique
is interesting for a quick evaluation of a “final model” in which the convergence
test has not been performed (Fig. 6).

Fig. 6. Left: comparison between the results of first order element (tria 3) with a second
order element (tria 6). Right: Mesh convergence curve of different refinement technique.

4 Issues related to properties of biological materials


Usually, the material properties data used in FEA models are obtained from
published experimental studies. Researchers are faced with the problem of large
dispersion of data recorded for the biological structures. There are two main
reasons for this dispersion: 1) biological materials are not homogeneous and
standardized, since they varies as a function of individual aspects (genetic, age,
gender, etc.); and 2) the methods used to obtain the mechanical properties of
biological materials are not as standardized as those of materials for engineering
analysis.
Dispersion motivates some researchers to perform their own experimental tests in
order to obtain the required material properties inputs for FEA models [34]. This
approach does not solve the dispersion problem, as the new result will be one more
in the spectrum of the available data. Nevertheless, it is interesting because the
authors are not limited to published data and more specific and accurate input for
the particular conditions of the study can be obtained. On the other hand, an
approach that is usually not encouraged by journal reviewers is to use properties
Characterization of dental biomaterials by Finite element analysis 13

from other FEA studies. Although that would be not critical for some cases, it can
propagate inaccuracies.
Many FEA studies represent biological materials as isotropic, linear, and elastic.
In these cases, only two parameters are necessary: the modulus of elasticity and
the Poisson's ratio. However, biological materials are usually more complex, and
does not behave as an isotropic, linear, and elastic material [35]. On the other hand,
the more complex the material properties to be modelled, the higher the risk for
inaccuracy or misinterpretation of the results.
Non-linear elastic and viscoelastic models have frequently used for periodontal
ligament. This thin soft connective tissue that links tooth to bone is composed by
two basic components: the collagen fibers (which are arranged in different
directions along the root contour) and the fluid interstitial matrix. This histologic
arrangement allows the PDL to perform as both a shock absorber and a load
transmitter from tooth to bone.
Experimental studies have shown that periodontal ligament acts as a viscoelastic,
anisotropic and heterogeneous material [36]. Therefore, the way it behaves during
loading and unloading conditions varies depending both on the velocity of force
application as well as from its duration. In FEA studies, the PDL have been
simulated using distinct constitutive models, which have been organized into two
groups: 1) models for instantaneous or short-term loading cases; and 2) models for
delayed or long-term loading cases. The first group contains different elastic
constitutive laws, like linear-elastic, hiperelastic, or non-linear[37]. The second
group includes viscoelastic, poro-elastic [38], and multiphase models.

5 Boundary conditions and Loading


The boundary conditions define the loading and the constraints applied to the
model (usually to the nodes) and, therefore, determine the FEA output. In general,
in a 2D model, each free node has 4 degrees of freedom (2 translations and 2
rotations), while in a 3D model each free node has 6 degrees of freedom (3
translations and 3 rotations).
Constraints represent imposition of displacements on a finite element model,
which might be null to enable the software to reach a numerical solution in the
analysis. These restrictions of displacements concern rotations and translations
around X, Y, and, for the 3D models, Z-axes.
Loadings can be applied using different methods, such as point loads, pressure,
distributed loads across a specific area, as well as loads simulated by the contact
of the antagonist tooth. Regardless of the loading method, high magnitude of
stresses is always expected around the loaded nodes. Still, these are not realistic
14 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

stress concentrations, thus in all finite element models, loading and constraints
should remain sufficiently far from the region of interest to ensure accuracy of the
results.

6 Interpretation of FEA output


Interpretation of the results is a very critical stage of biomechanics FEA
simulation. It requires a thorough knowledge of the FEA methodology, of the
related mechanical concepts, and of the biological phenomenon under study.
Integrating these knowledge is not an easy task. Usually requires time, practice,
reflection, and patience. Therefore, the learning curve in this step is usually more
prolonged than that of the previous steps. It takes longer for the beginner to acquire
autonomy to choose the sound failure criterion and to describe the generated data.
An additional difficulty appears because the failure, while usually evident in
experimental tests, in finite element analysis it is mostly not directly discernable.
For example, in a FEA simulation of a tooth restored with a post and a metal-
ceramic crown, the stress distribution map obtained by the finite element analysis
is not enough for determining the risk of root fracture. It is necessary to compare
the stress values generated in each structure with the correspondent strength of its
material.

6.1 The failure criteria

In a simplified approach, we can say that the fracture of brittle materials is


governed by the magnitude of tensile stresses. Therefore, the maximum principal
stress (1) is frequently used as the failure criteria for these materials, since it
represents the maximum tensile stress. In addition, one important caution when
analyzing these cases is checking the compressive stresses, especially the
maximum compressive stresses, given by the minimum principal stress criteria
(3). The compressive stress promotes a protective effect in the brittle material
because it hinders the propagation of cracks that grow in the perpendicular
direction to the compression. Regarding to ductile materials, they usually fail
before the fracture, because the plastic deformation that appears beyond the yield
strength tends to jeopardize the object function. As the plastic deformation is
mainly governed by the shear stress, Tresca or Von Mises criterion are generally
used to assess the risk of failure of ductile materials.
In biomechanical simulations, the choice of the failure criterion is based not only
on the nature of the material under study (brittle or ductile), but also on the
Characterization of dental biomaterials by Finite element analysis 15

biological mechanisms involved in the failure process. Furthermore, the chosen


failure criteria result must be consistent with clinical outcome of published
Randomized Controlled Trials (RCT). Therefore, specified failure criteria have
been created in the FEA studies in Dentistry and Medicine, such as algorithms for
bone remodeling [39, 40].

6.2 The biological mechanism

The orthodontically induced inflammatory root resorption (OIIRR) treatment is a


very didactic example to demonstrate the importance of integrating the theoretical
concepts of failure criteria with the biological mechanisms involved in the
phenomenon under study. Although the resorption occurs in the root, it is not
linked with the stress supported by dentin or cement, as suggested in some studies
[41-43]. It is well described in literature that the biological mechanisms associated
with OIIRR is the obstruction of PDL vessels in over-compressed areas [44-49].
Therefore, the failure criteria need to take into account what happens in PDL that
favors the blood stasis. The minimal principal stress or the compressive values of
hydrostatic or radial stresses have been used to predict OIIRR by finite element
analysis [50-53] (Fig. 7). It seems coherent that if these compressive stresses
exceed the capillary hydrostatic pressure (4.7 kPa), the risk of obstruction and
resorption is high [50-52].

Fig. 7. Compressive stresses that have been used to predict the obstruction of PDL vessels: minimum
principal stress, hydrostatic stress and radial stress. The coordinate system in the radial stress image
exhibits the orientation of radial (rr), tangential (tt) and hoop () stresses.
16 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

6.3 Importance of results presentation

FEA software usually present results as a map of the distribution of the response
of interest (pressure, strain, displacement, etc.), in which each color represents a
range of values. Although the programs automatically set the scale, it is interesting
to the user to make some manual adjustments to improve the viewing and
interpretation of the outcomes. For example, when displaying the map of the
maximum principal stress (1), it is interesting to set the minimum value to zero,
to easily differentiate the regions with positive 1 values from those with negative
values. Likewise, for the minimum principal stress (3) the maximum value should
be set in zero. While positive values of 1 refer to the maximum tensile stress,
negative values indicate tension-free region. While negative values of 3 refer to
the maximum compressive stress, positive values indicate compression-free
region. The simple adjustment on the scale already helps to visualize these
different areas. In addition, the distribution map of the figure can be complemented
with a figure of the orientation of 1 or 3 [54].

Fig. 8. Map distribution of maximum principal stress.


In many cases, the presentation in maps may not be enlightening to the
phenomenon under study. The researcher can then use other reports provided by
the software to build, in an intelligent but laborious way, other modes of most
enlightening presentation. For instance, results on interfaces are better represented
using graphs instead of the habitual map-distributed figure. The response is usually
plotted as a function of cumulative selected nodes distance [55]. It is important to
Characterization of dental biomaterials by Finite element analysis 17

define which side of the interface will be explored. For example, in a cement-post
interface, the user can chose to plot the stress in the cement or the stress in the post.
The two curves should also be plotted, but the presentation of average stress (of
cement and post) is not recommended because it lacks a physical sense.
Another important issue is that the statistical analysis is usually dispensable in FEA
studies for comparing results among different models. It is not well acknowledged
among researchers habituated with clinical or experimental study design, in which
the variability is inherent. Therefore, it is a common (but unsuitable) criticism from
some journals reviewers. The absence of statistic is explained due to absence of
experimental error. It is useless to run a certain FEA model more than once, waiting
random variation in the response. The response will be exactly the same in
different runs (if the model is the same regarding input data and mesh). The
judgment as to whether the differences are significant or not will be aided only by
the good sense of the researcher in recognizing that the discrepancy significantly
interferes in practice. It is reasonable to consider irrelevant, e.g., a stress difference
about only 1% for studies on specimen shape subjected to resistance test if the real
tests vary by more than 10%. In some specific FEA study design, the statistical
analysis is used, but it is the exception, not the rule.

7 Final remarks
• Finite element analysis presents a wide range of application in Dentistry. Some
stress studies can only be performed using this methodology, since stresses are
not visible experimentally.
• Simplifications are inherent to any model (even experimental models) and
impose specific limitations that do not necessarily invalidate the study. FEA is
no different. The researcher challenge is to distinguish the necessary
simplifications from the misrepresentative ones.
• Meticulousness and thoroughness in organizing and displaying the FEA results
facilitate the interpretation of the authors and the understanding of the reader.

8 Acknowledgments
We would like to thank CNPq for Alice Jikihara’s PhD scholarship of Capes for
Pavel Capetillo’s PhD scholarship, which were essential for enabling their help in
this chapter. We would also want to thank FAPESP for the constant financial
support in our FEA studies.
18 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

9 References
1. Ichim, I., M.V. Swain, and J.A. Kieser, Mandibular stiffness in humans: numerical
predictions. J Biomech, 2006. 39(10): p. 1903-13.
2. Liao, Z., et al., Computational modeling of dynamic behaviors of human teeth. J Biomech,
2015. 48(16): p. 4214-20.
3. Meira, J.B., et al., Understanding contradictory data in contraction stress tests. Journal of
Dental Research, 2011. 90: p. 365-70.
4. Barani, A., et al., Mechanics analysis of molar tooth splitting. Acta Biomater, 2015. 15: p.
237-43.
5. Li, H., et al., Fracture simulation of restored teeth using a continuum damage mechanics
failure model. Dent Mater, 2011. 27(7): p. e125-33.
6. Meira, J.B., et al., The suitability of different FEA models for studying root fractures
caused by wedge effect. J Biomed Mater Res A, 2008. 84(2): p. 442-6.
7. Ash, N., Wheeler's Dental Anatomy, Physiology, and Occlusion, ed. 9º. 2010. 401.
8. Gonzalez de Villaumbrosia, P., et al., In vitro comparison of the accuracy (trueness and
precision) of six extraoral dental scanners with different scanning technologies. J Prosthet Dent,
2016.
9. Dejak, B. and A. Mlotkowski, The influence of ferrule effect and length of cast and FRC
posts on the stresses in anterior teeth. Dent Mater, 2013. 29(9): p. e227-37.
10. Ammar, H.H., et al., Three-dimensional modeling and finite element analysis in treatment
planning for orthodontic tooth movement. Am J Orthod Dentofacial Orthop, 2011. 139(1): p. e59-
71.
11. Kim, K.Y., et al., Displacement and stress distribution of the maxillofacial complex during
maxillary protraction with buccal versus palatal plates: finite element analysis. Eur J Orthod, 2015.
37(3): p. 275-83.
12. Liu, C., X. Zhu, and X. Zhang, Three-dimensional finite element analysis of maxillary
protraction with labiolingual arches and implants. Am J Orthod Dentofacial Orthop, 2015. 148(3): p.
466-78.
13. Ausiello, P., et al., Numerical fatigue 3D-FE modeling of indirect composite-restored
posterior teeth. Dent Mater, 2011. 27(5): p. 423-30.
14. Meriwether, L.A., et al., Shrinkage stress compensation in composite-restored teeth:
relaxation or hygroscopic expansion? Dent Mater, 2013. 29(5): p. 573-9.
15. Benazzi, S., et al., Unravelling the functional biomechanics of dental features and tooth
wear. PLoS One, 2013. 8(7): p. e69990.
16. Junior, M.M., et al., Etch and rinse versus self-etching adhesives systems: Tridimensional
micromechanical analysis of dentin/adhesive interface. International Journal of Adhesion and
Adhesives, 2012. 35: p. 114-119.
17. Ludwig, B., et al., Application of a new viscoelastic finite element method model and
analysis of miniscrew-supported hybrid hyrax treatment. Am J Orthod Dentofacial Orthop, 2013.
143(3): p. 426-35.
18. Lee, N.K. and S.H. Baek, Stress and displacement between maxillary protraction with
miniplates placed at the infrazygomatic crest and the lateral nasal wall: a 3-dimensional finite element
analysis. Am J Orthod Dentofacial Orthop, 2012. 141(3): p. 345-51.
19. Tominaga, J.Y., et al., Effect of bracket slot and archwire dimensions on anterior tooth
movement during space closure in sliding mechanics: a 3-dimensional finite element study. Am J
Orthod Dentofacial Orthop, 2014. 146(2): p. 166-74.
20. Yu, I.J., et al., Comparison of tooth displacement between buccal mini-implants and palatal
plate anchorage for molar distalization: a finite element study. Eur J Orthod, 2014. 36(4): p. 394-402.
Characterization of dental biomaterials by Finite element analysis 19

21. Dejak, B. and A. Mlotkowski, 3D-Finite element analysis of molars restored with
endocrowns and posts during masticatory simulation. Dent Mater, 2013. 29(12): p. e309-17.
22. Ichim, I.P., et al., Restoration of non-carious cervical lesions Part II. Restorative material
selection to minimise fracture. Dent Mater, 2007. 23(12): p. 1562-9.
23. Rodrigues, F.P., et al., Finite element analysis of bonded model Class I 'restorations' after
shrinkage. Dent Mater, 2012. 28(2): p. 123-32.
24. Cuddihy, M., et al., Endodontic access cavity simulation in ceramic dental crowns. Dent
Mater, 2013. 29(6): p. 626-34.
25. Ausiello, P., et al., Mechanical behavior of post-restored upper canine teeth: a 3D FE
analysis. Dent Mater, 2011. 27(12): p. 1285-94.
26. Mattos, C.M., et al., Numerical analysis of the biomechanical behaviour of a weakened
root after adhesive reconstruction and post-core rehabilitation. J Dent, 2012. 40(5): p. 423-32.
27. Anchieta, R.B., et al., Effect of long-term storage on nanomechanical and morphological
properties of dentin-adhesive interfaces. Dent Mater, 2015. 31(2): p. 141-53.
28. An, B., R. Wang, and D. Zhang, Role of crystal arrangement on the mechanical
performance of enamel. Acta Biomater, 2012. 8(10): p. 3784-93.
29. Scheider, I., et al., Damage modeling of small-scale experiments on dental enamel with
hierarchical microstructure. Acta Biomater, 2015. 15: p. 244-53.
30. Ma, S., I. Scheider, and S. Bargmann, Anisotropic constitutive model incorporating
multiple damage mechanisms for multiscale simulation of dental enamel. J Mech Behav Biomed
Mater, 2016. 62: p. 515-533.
31. An, B., et al., Damage mechanisms in uniaxial compression of single enamel rods. J Mech
Behav Biomed Mater, 2015. 42: p. 1-9.
32. Podshivalov, L., A. Fischer, and P.Z. Bar-Yoseph, Multiscale FE method for analysis of
bone micro-structures. J Mech Behav Biomed Mater, 2011. 4(6): p. 888-99.
33. Miura, J., et al., Multiscale analysis of stress distribution in teeth under applied forces. Dent
Mater, 2009. 25(1): p. 67-73.
34. van Driel, W.D., et al., Time-dependent mechanical behaviour of the periodontal ligament.
Proc Inst Mech Eng H, 2000. 214(5): p. 497-504.
35. Huo, B., An inhomogeneous and anisotropic constitutive model of human dentin. J
Biomech, 2005. 38(3): p. 587-94.
36. Fill, T.S., et al., Analytically determined mechanical properties of, and models for the
periodontal ligament: critical review of literature. J Biomech, 2012. 45(1): p. 9-16.
37. Toms, S.R. and A.W. Eberhardt, A nonlinear finite element analysis of the periodontal
ligament under orthodontic tooth loading. Am J Orthod Dentofacial Orthop, 2003. 123(6): p. 657-
65.
38. Bergomi, M., et al., Hydro-mechanical coupling in the periodontal ligament: a
porohyperelastic finite element model. J Biomech, 2011. 44(1): p. 34-8.
39. Chen, J., et al., A periodontal ligament driven remodeling algorithm for orthodontic tooth
movement. J Biomech, 2014. 47(7): p. 1689-95.
40. Eser, A., et al., Predicting bone remodeling around tissue- and bone-level dental implants
used in reduced bone width. J Biomech, 2013. 46(13): p. 2250-7.
41. Oyama, K., et al., Effects of root morphology on stress distribution at the root apex. Eur J
Orthod, 2007. 29(2): p. 113-7.
42. Rudolph, D.J., P.M.G. Willes, and G.T. Sameshima, A finite element model of apical force
distribution from orthodontic tooth movement. Angle Orthod, 2001. 71(2): p. 127-31.
43. Kamble, R.H., et al., Stress distribution pattern in a root of maxillary central incisor having
various root morphologies: a finite element study. Angle Orthod, 2012. 82(5): p. 799-805.
20 J.B.C.Meira; A.N.Jikihara; P.Capetillo; M.G. Roscoe; P.M.Cattaneo; R.Y.Ballester

44. Roscoe, M.G., J.B. Meira, and P.M. Cattaneo, Association of orthodontic force system and
root resorption: A systematic review. Am J Orthod Dentofacial Orthop, 2015. 147(5): p. 610-26.
45. Schwarz, A.M., Tissue changes incidental to orthodontic tooth movement. Int J Orhodontia
1932. 18: p. 331-52.
46. Chan, E. and M.A. Darendeliler, Physical properties of root cementum: Part 5. Volumetric
analysis of root resorption craters after application of light and heavy orthodontic forces. Am J Orthod
Dentofacial Orthop, 2005. 127(2): p. 186-95.
47. Chan, E. and M.A. Darendeliler, Physical properties of root cementum: part 7. Extent of
root resorption under areas of compression and tension. Am J Orthod Dentofacial Orthop, 2006.
129(4): p. 504-10.
48. Barbagallo, L.J., et al., Physical properties of root cementum: Part 10. Comparison of the
effects of invisible removable thermoplastic appliances with light and heavy orthodontic forces on
premolar cementum. A microcomputed-tomography study. Am J Orthod Dentofacial Orthop, 2008.
133(2): p. 218-27.
49. Paetyangkul, A., et al., Physical properties of root cementum: part 14. The amount of root
resorption after force application for 12 weeks on maxillary and mandibular premolars: a
microcomputed-tomography study. Am J Orthod Dentofacial Orthop, 2009. 136(4): p. 492 e1-9;
discussion 492-3.
50. Hohmann, A., et al., Periodontal ligament hydrostatic pressure with areas of root resorption
after application of a continuous torque moment. Angle Orthod, 2007. 77(4): p. 653-9.
51. Hohmann, A., et al., Correspondences of hydrostatic pressure in periodontal ligament with
regions of root resorption: a clinical and a finite element study of the same human teeth. Comput
Methods Programs Biomed, 2009. 93(2): p. 155-61.
52. Field, C., et al., Mechanical responses to orthodontic loading: a 3-dimensional finite
element multi-tooth model. Am J Orthod Dentofacial Orthop, 2009. 135(2): p. 174-81.
53. Viecilli, R.F., et al., Three-dimensional mechanical environment of orthodontic tooth
movement and root resorption. Am J Orthod Dentofacial Orthop, 2008. 133(6): p. 791 e11-26.
54. Meira, J.B., et al., Residual stresses in Y-TZP crowns due to changes in the thermal
contraction coefficient of veneers. Dent Mater, 2013. 29(5): p. 594-601.
55. Santos, A.F.V., et al., Can Fiber Posts Increase Root Stresses and Reduce Fracture? Journal
of Dental Research, 2010. 89(6): p. 587-591.

View publication stats

You might also like