Surg Radiol Anat (2006) 28:581587

DOI 10.1007/s00276-006-0140-x

O R I G I N A L A RT I C L E

Finite element analysis of the strain distribution in the humeral

head tubercles during abduction: comparison of young
and osteoporotic bone
Ph. Clavert M. Zerah J. Krier P. Mille
J. F. Kempf J. L. Kahn

Received: 4 April 2006 / Accepted: 3 July 2006 / Published online: 26 August 2006
Springer-Verlag 2006

Abstract
Aim The aim of this work was to design an accurate
3D digital model of the humerus and rotator cuff
muscles. This model was then used to study strain
distribution in humeral tubercles according to bone
density.
Materials and methods The geometry of bone and
muscle structures was reproduced using SURFDRIVER software, based on anatomical sections, CT scans
and MRI images from the Visible Human Project
image library. The contours were transferred to
PATRAN software to rebuild volumes and mesh
them. Calculations of strains and their distribution
were performed using NASTRAN software. All the
elements were considered to be isotropes.
Results The study of the distribution of stress
magnitude according to the type of bone modeled,
shows that some stresses in cortical bone are greater
than those in cancellous bone and are also greater in
old bone, implying more deformation in old bone at
constant force. This study also shows that stresses do
not penetrate deeply into cancellous tissue.
Conclusion Observing the simulation results led
understanding of the pathology of certain fractures of

Ph. Clavert (&) J. F. Kempf J. L. Kahn

Biomechanical Laboratory of the GEBOAS,
Institute of Normal Anatomy, Faculty of Medicine,
4 rue Kirschleger, 67085 Strasbourg Cedex, France
e-mail: philippe.clavert@chru-strasbourg.fr
M. Zerah J. Krier P. Mille
Laboratoire dIngenierie des Surfaces de Strasbourg Groupe
LISS du LGECO, INSA Strasbourg, 24 boulevard de la
victoire, 67084 Strasbourg Cedex, France

the proximal end of the humerus. This study also

helped explain why certain types of osteosynthesis
fail due to tubercles reconstruction failures.
Keywords Shoulder Humerus Finite element
analysis Stress distribution Biomechanics
Introduction
Many physical or computer models of the scapulohumeral joint have been developed to improve understanding of its function. Some were developed to
understand and analyze muscle action developed
during certain movements of the upper limb (inverse
dynamic theories) [4, 13, 15, 23, 24, 26]. Others, based
on the concept of deformable bodies, were proposed
and used to calculate the distribution of bone stresses
[3, 10, 16, 20, 22].
Anatomically, four rotator cuff muscles (supraspinatus, infraspinatus, subscapularis, teres minor) insert
into the humeral tubercles, where stress is concentrated during abduction movements. The purpose of
this work was to develop a 3D digital model of the
proximal end of the humerus and the rotator cuff
muscles, in order to analyze the effect of bone tissue
quality on strain distribution to the proximal end of the
humerus owing to the physiological action of rotator
cuff tendons during an abduction movement.
Materials and methods
Obtaining anatomical shapes in 3D
SURFDRIVER (SURFDRIVER, Kailua, HI, USA)
software was used for acquisition of the external

123

582

Surg Radiol Anat (2006) 28:581587

contours of the different elements of the bone (cortical

bone, cancellous bone, subchondral bone) and articular
cartilage found in the proximal third of the humerus
(Fig. 1). This was done using 153 anatomical and CT
scan sections from the male specimen of the Visible
Human Project image library (National Library of
Medicine, Rockville Pike, USA). They were millimetric coronal sections in 1 mm steps. Since the automatic
outlining abilities of currently available software are
not suitable for the complexity of the anatomical
structures studied, this outlining was then accurately
defined on each CT slice (digitization error lower than
0.7 mm, i.e. 2 pixels).
Using the same process, the muscle bodies of the
subscapular, supraspinatus (Fig. 2), infraspinatus and
teres minor were outlined. Finally, the insertion areas
of rotator cuff muscles and their tendons were defined
by comparison with anatomical sections and MRI
images.

Fig. 2 The muscle bodies of the subscapular, supraspinatus are

outlined and reconstructed

Meshing
The contours obtained were transferred to MSC
PATRAN v7.5 (MCS Software Corporation, Santa
Ana, CA, USA) to be assembled individually and thus
reconstruct the volume of each element before meshing them. Meshing or discretizing all the elements
constituting the reconstructed humerus, was carried
out with 8 node brick type isoparametric threedimensional elements or HEX8 [7, 8], creating 221,626
elements (Fig. 3). This type of element produced good
results convergence, with acceptable computing times.
It was during the meshing stage that the x, y, z baseline
was set up; it will be retained throughout the experiments so that results can be read clearly.
Mechanical characteristics of different elements

Fig. 1 External contours of the proximal third of the humerus

obtain with SURFDRIVER

123

All the rebuilt structures were considered individually

as isotropes, thats why Youngs modulus and Poissons
coefficient of each structure were taken into account
and incorporated in the model. The values applied
to the different elements in the model are given in
Table 1.
Buchler et al. [6] showed that mechanical characteristics applied to muscle bodies and tendons had no
effect on the model. The most important factor is still
the surface and site of tendon insertion. Therefore, in

Surg Radiol Anat (2006) 28:581587

583

Fig. 3 Meshing of the

reconstructed humerus with
HEX8 elements

Table 1 Youngs modulus and Poissons coefficient applied to the different elements in the model

Youngs modulus
Poissons coefficient

Cortical bone [2]

Cancellous bone
[10, 12, 19]

Sub-chondral bone
[5, 14, 21]

Articular cartilage [2]

Young
bone

Old
bone

Young
bone

Old
bone

Young
bone

Old
bone

Young
bone

Old
bone

16 GPa
0.3

12 GPa
0.3

550 MPa
0.3

250 MPa
0.3

3.5 GPa
0.3

3.5 GPa
0.3

2 MPa
0.3

2 MPa
0.3

Youngs modulus are obtained from specific studies of cortical and cancellous bone of the upper end of the humerus

this model we did not apply any specific Young

modulus and Poissons coefficient to the structures.
Once the model was discretized and its mechanical
characteristics defined, the solver (MSC NASTRAN
v7.5 MCS Software Corporation, Santa Ana, CA,
USA) performed three levels of calculation to display
the results:
1. To determine the matrices of rigidity for each
element, so as to assemble them to obtain the global
matrix of rigidity for the model and determine the vector forces resulting from analysis of the load cases used.
2. To calculate the system solution in terms of displacement.
3. To calculate the other data such as stresses and
deformation of the model.
Loading conditions
We loaded the humeral tubercles by simulating static
abduction movement (considering only a gleno-hum-

eral movement) in the plane of the scapula on our

model. Two positions were studied: abduction at 10
(anatomical reference position) and abduction at 90
(maximum amplitude involving only the scapulohumeral joint) (Fig. 4).
Action was applied to each muscle in one direction
only [25], according to the humeral position selected
(Fig. 4). Points at the proximal and distal insertions for
each muscle were defined using contours obtained via
SURFDRIVER. From these points, lines of action
were created using PATRAN. The muscular forces
were applied along these lines. The load was applied at
a single node (Fig. 4), then at multiple nodes (Fig. 5),
as it is anatomically, to avoid concentrating stresses at
the nodes in the model [18]. However, the number of
nodes was chosen to retain adequate amplitude for
observing the results.
Because this is a static model, a glue type interaction
between the glenoid and the humerus was chosen for
these two positions. Muscular forces applied to each

123

584

Surg Radiol Anat (2006) 28:581587

Fig. 4 The humeral tubercles

are loaded by simulating
static abduction movement in
the plane of the scapula:
abduction at 10 (anatomical
reference position) and
abduction at 90 (maximum
amplitude involving only the
scapulohumeral joint)

deformation force (J mm3) in cortical and cancellous

bone in the models tested, were taken into account
each in turn.
We displayed the total densities of deformation
forces so as to compare them. They were then
extrapolated to deduce the resulting deformations.
Indeed, for an element of volume V, subject to a range
of stresses (rx, ry, rz, sxy, syz, szx) and a deformation
field (
x, R
y,
z, cxy, cyz, czx), the deformation force
is: U 12 rx ex ry ey rz ez sxy cxy syz cyz szx czx
dV:The Vtotal density of deformation forces is then
obtained by dividing this force U by the volume of the
element.

Results
Abduction positions
Fig. 5 The muscular forces were applied at multiple nodes of the
tubercles

muscle for an abduction position of 10 and 90 are

given in Table 2 [26]. As our model refers to static arm
positions, only the forces developed in these positions
are taken into account [26].
Displaying the results given by the simulations
To display the results, the maximum values of von
Mises equivalent stresses (MPa) and total maximum

The type of force applied in the multinode model did

not give a usable value for stress in cortical bone, which
is why only the single node model was used. In this
configuration, the stress noted for 90 abduction in
young cortical bone was 0.72 MPa, whereas it was
negligible at 10 abduction (Fig. 6). A small part of the
transferred stress was recorded in the cancellous bone.
The comparison of values according to the positions
studied demonstrates that the reference position (arm
the length of the body at 10 abduction in the plane of
the scapula) exerts almost no force on the humeral
tubercles.

Table 2 The muscular forces applied to each muscle for an abduction position of 10 and 90 [26]

10 of abduction
90 of abduction

123

Deltoideus

Subscapularis

Supraspinatus

Infraspinatus/teres minor

4.5 N
116 N

2.5 N
70 N

1N
25 N

2N
60 N

Surg Radiol Anat (2006) 28:581587

585

Fig. 6 The type of force

applied in the multinode
model (left) did not give a
usable value for stress in
cortical bone, which is why
only the single node model
was used (right)

Comparison of the distribution of stresses according

to type of bone
The study of distribution of the total densities of
deformation forces according to the type of bone modeled, shows that these values are higher for old bone
than for young bone, which implies that there is more
deformation in an osteoporotic bone under constant
force (Fig. 7). The maximum stress values were recorded in cortical bone. The stress values recorded
varied considerably between the two positions studied,
with a maximum load at 90 abduction (Fig. 8).
Looking now at the depth of penetration of stresses
in cancellous bone (sections at the subscapularis muscle insertion level), two observations can be made.
Firstly, it can be seen that in the multinode model, the
depth of penetration is fairly shallow (4 mm maximum), and distributed over a wide area (Fig. 9). Secondly, the depth of stress penetration is similar for both
types of bone.
Discussion
The finite elements method involves evaluating displacements for all points of the structure being studied,

Fig. 7 Comparison of the total densities of deformation forces

(104 J mm3) according to the type of bone modeled (old
bone and young bone)

Fig. 8 The maximum stress values were recorded in cortical

bone at 90 abduction

by resolution of linear systems. From these displacements, deformations are obtained by derivation. The
stresses are then evaluated from the deformations
using a matrix characteristic of the structural material.

Fig. 9 Depth of penetration of stresses in cancellous bone; the

depth of penetration is fairly shallow, and distributed over a wide
area

123

586

The aim of this study was to develop a digital

model of the humerus, taking into account all the
elements constituting the bone (cortical and cancellous bone, subchondral bone and articular cartilage),
so as to perform a large number of non-destructive
tests on anatomical specimens and prepare more
accurate and fewer in-vitro tests. The acquisition of
accurate geometry for the tendons and the bones is a
fundamental requirement for the construction of
three-dimensional FE models of joints. CT-scan yield
superior spatial resolution than MRI and by this way
provides excellent images of bones. Matching CT-scan
views and anatomic preparations is of interest to
differentiate between each tendon and surrounding
bone and soft-tissue structures.
In this study, a model of the proximal end of the
humerus was developed to study the distribution of
stresses in the humeral tubercles for two arm positions,
using the forces representing muscle action in the
rotator cuff muscles. We also studied the effect of bone
quality on these data. This knowledge is important in
so far as, by recreating an accurate anatomy of muscle
bodies and their humeral insertions, we were able to
describe exactly the mode and areas of diffusion of
stresses in cortical and cancellous bone. This structural
distinction is important, because the laws of stress
distribution depend on regional variations in the
moduli of elasticity.
The von Mises stress distribution observed in this
model was correlated with clinical observations. The
regions in which a stress peak is observed correspond
clinically to areas where tubercles are torn and fractured. The stresses noted in bone material did not exceed 6 MPa, which matches results already published
based on in-vivo studies [11, 17], or evaluated using
digital models [2, 8]. In osteoporotic bone, the load on
cortical bone is increased, as is the area of distribution
of stress, an idea already put forward by Maldonado
et al. [18].
Knowledge of the load on the bone is important in
understanding the physiopathology of fractures and
treating them. One of the aims of this study was to
demonstrate the effect of bone quality on bone load,
using a reference bone relative to an osteoporotic
bone. The increase in load and its distribution for an
osteoporotic bone must be taken clearly into consideration in designing osteosynthetic implants for use in
treating fractures of the proximal end of the humerus,
in which cancellous screws increase the stress on deficient cancellous tissue, by reducing the stresses normally absorbed by the cortical bone. Similarly, it is
important to consider bone quality when planning
humeral reconstruction surgery.

123

Surg Radiol Anat (2006) 28:581587

Criticisms of the model

As for all numerical models, the boundary conditions
are a difficult task. In the case of the shoulder, it is
virtually impossible to reproduce exactly the real
boundary conditions with all the muscles acting
together. For these reason, the boundary condition was
chosen to reproduce pure abduction. This is also why
we only considered the rotator cuff muscles. However,
the muscles modeled are the most important ones
involved in abduction, which makes the investigation
possible, particularly when using a multinode application, because of the complexity of calculations.
Because of this, modeling mechanical actions using this
digital model is fairly simplistic: indeed, modeling the
line of action of muscles using a rope stretched
between two points is still approximate. However, it
helps validate the procedure used, producing the first
results so as to envisage gradual improvement.
The von Mises criterion could also be used in
this study, because we considered our materials to be
isotropic, although in reality they are not [1, 9, 12].

Conclusion
The simulations performed in this study led to the
development of an anatomically accurate digital model
of the humerus, used to understand, display and map the
distribution of stresses in the major and minor humeral
tubercles generated by muscle tendons in the rotator
cuff muscles during abduction. The results obtained
were used to validate the model, by comparing them
with published data, and to envisage more complex
simulations. We can now incorporate prosthetic parts
into the model to determine the maximum forces the
material can withstand without risk of rupture, and
complete the model with the scapula to evaluate the
distribution and amplitude of stresses in the joint contact
zone. Finally, this study demonstrates that, under
physiological load on the rotator cuff, maximum stresses
are observed at 90 abduction. The amplitude and distribution of stresses is influenced mainly by bone quality.

References
1. Anglin C, Tolhurst P, Wyss U, Pichora D (1999) Glenoid
cancellous bone strength and modulus. J Biomech 32:1091
1097
2. Barea C (1998) Modelisation 3D par la methode des elements finis dune articulation scapulo-humerale. Application
a` letude des contraintes sur une epaule saine et avec differents types de gle`nes prothetiques. Thesis. Universite Paul
Sabatier, Toulouse

Surg Radiol Anat (2006) 28:581587

3. Barea C, Hobatho M, Darmana R, Mansat P, Mansat M
(1998) Three dimensional finite element modelling of normal
and implanted glenoid: validations and simulations. In: 11th
Conference of the ESB. Toulouse, p 27
4. Biluart F, Gagey O, Skalli W, Mitton D (2006) Biomechanics
of the deltoideus. Surg Radiol Anat 28:7681
5. Braidotti P, Bemporade E, DAlessiot T, Sciuto S, Stagni I
(2000) Tensile experiments and SEM fractography on bovine
subchondral bone. J Biomech 9:11531157
6. Buchler P, Ramaniraka N, Rakotomanana L, Iannotti J,
Farron A (2002) A finite element model of the shoulder:
application to the comparison of normal and osteoarthritic
joints. Clin Biomech 17:630639
7. Buchler P, Ramaniraka N, Rakotomanana L, Iannotti J,
Farron A (2002) A finite model of the shoulder: application
to the comparison of normal and osteoarthritic joints. Clin
Biomech 17:630639
8. Couteau B, Mansat P, Estivalezes E, Darmana R, Mansat M,
Egan J (2001) Finite element analysis of the mechanical
behavior of a scapula implanted with a glenoid prosthesis.
Clin Biomech (Bristol, Avon) 16:566575
9. Frich L (1997) Bone strength and materials properties of the
glenoid. J Shoulder Elbow Surg 6:97104
10. Friedman R (1992) Finite element of the glenoid component:
effect of the design parameters on stress distribution.
J Shoulder Elbow Surg 1:261270
11. Frost H (1987) The mechanostat: a proposed pathogenic
mechanism of osteoporoses and the bone mass effects of
mechanical and non-mechanical agents. Bone Minerology
2:7385
12. Hobatho M, Rho J, Ashman R (1992) Atlas of mechanical
properties of human cortical and cancellous bone. In vivo
assessment of bone quality by vibration and wave propagation techniques. Part II. Van der Perre G, Lowet G, Borgwardt A (eds) ACCO, Leuven
13. Hughes R, An K (1996) Force analysis of rotator cuff muscles. Clin Orthop 330:7583
14. Huiskes R (1987) Finite element analysis of acetabular
reconstruction. Non cemented threated cups. Acta Orthop
Scand 58:620625

587
15. Karlsson D, Peterson B (1992) Towards a model for force
predictions in the human shoulder. J Biomech 25:189199
16. Lacroix D, Prendergast P (1997) Stress analysis of glenoid
component designs for shoulder arthroplasty. Proc Inst Mech
Eng 211:467474
17. Lanyon L, Hampson W, Goodship A, Shan J (1975) Bone
deformation recorded in vivo from strain gauges attached to
the human tibial shaft. Acta Orthop Scand 46:256268
18. Maldonado Z, Seebeck J, Heller M, Brandt D, Hepp P, Lill
H, Duda G (2003) Straining of the intact and fractured
proximal humerus under physiological-like loading. J Biomech 36:18651873
19. Mansat P (1996) Proprietes mecaniques osseuse de la gle`ne.
Application a` la modelisation tridimensionnelle par la
methode des elements finis. Thesis. Universite Paul Sabatier,
Toulouse
20. Orr T, Carter D, Schurman D (1988) Stress analyses of glenoid component designs. Clin Orthop 232:217224
21. Pedersen D, Crowninshield R, Brand R, Johnston R (1982)
An axisymetric model of acetabular components in total hip
arthroplasty. J Biomech 15:305315
22. Stone K, Grabowski J, Cofield R, Morrey B, An K (1999)
Stress analyses of glenoid components in total shoulder
arthroplasty. J Shoulder Elbow Surg 8:151158
23. Van der Helm F (1994) Analysis of the kinematic and
dynamic behavior of the shoulder mechanism. J Biomech
27:527550
24. Van der Helm F (1994) A finite element musculoskeletal
model of the shoulder mechanism. J Biomech 27:551569
25. Van der Helm F, Veenbaas R (1991) Modelling the
mechanical effect of muscles with large attachment sites:
application to the shoulder mechanism. J Biomech 21:1151
1163
26. Wuelker N, Wirth CJ, Plitz W, Roetman B (1995) A dynamic
shoulder model: reliability testing and muscle force study.
J Biomech 28:489499

123