NumericalExperimental Method

J. A. Simoes
Department of Mechanical Engineering,
for the Validation of a Controlled
University of Aveiro,
Aveiro, Portugal Stiffness Femoral Prosthesis
The aim of this paper is to describe a new numericalexperimental method to determine
J. Monteiro the stiffness of a conceptual proximal femoral prototype. The methodology consists of the
comparison of the numerical and experimental displacement distributions of the prosthe-
M. A. Vaz sis loaded as a cantilever beam to validate a design concept: controlled stiffness pros-
thesis. The manufactured prototype used to test the applicability of the numerical
Department of Mechanical Engineering and experimental procedure integrates a stiff metal core bonded to a composite material made
Industrial Management, of an epoxy resin reinforced with carbon-glass braided pre-forms. The prosthesis with an
University of Porto, embedded controlled stiffness concept was obtained by varying the geometry of the core
Porto, Portugal with the composite layer thickness. DOI: 10.1115/1.1375162

Keywords: Femoral Prosthesis, Stiffness, Holography, Displacement, Composite

Material

Introduction tion, have stimulated the development of compliant and tailored

The design of a hip prosthesis is a challenging problem that has
stimulated many investigators. The femoral component of a hip
prosthesis is essentially composed of a stem incorporating a ball
and is characterized by its geometry and materials, what is
known as stiffness elastic modulus multiplied by the second mo-
ment of area. Stiffness is an important design parameter and
many authors have addressed its influence on the prosthesis per-
formance. Stiffness is known to play a key role in the stress
shielding phenomena and on the relative interface bone-implant
micromotion, the predominant causes for long-term failure of total
hip replacements 1. Stress shielding depends on the design of
the intramedullary reconstruction and mainly on the material of
the femoral component, which influences the mechanism of load
transfer from implant to the surrounding bone and can have an
alarming effect on a revision procedure. Interface micromotion
can lead to a premature failure of the implant and is design and
material dependent. Therefore, two important design goals have to
be addressed in the design cycle of a hip prosthesis to balance the
advantages of flexible and stiff materials concerning stress shield-
ing and micromotion. These design objectives are strongly af-
fected by the prosthesis materials. Flexible stems provoke less
stress shielding in the surrounding bone, but higher proximal in-
terface stresses: cement stresses for cemented designs. Thus, ex-
cessive bone resorption is more likely to occur around stiff stems,
but proximal interface failure is more likely to occur around flex-
ible stems 1. Having this in mind, a suitable compromise has to
be achieved because the femoral component cannot be stiff and
flexible at the same time. The design problem is to know how to
minimize stress shielding keeping interface stresses and interface
micromotions at acceptable levels 1,2.
Conventional proximal femoral prostheses are manufactured
with a single constant modulus material. Such prostheses can ad-
dress either the stress shielding effect or the interface stresses, but
not both simultaneously. However, with optimized controlled
stiffness, it is possible to obtain a compromise between the two
conflicting design factors to attenuate the problems described
above. There has been substantial active research with composites
for total hip replacement applications 310, namely for his pros-
theses. The possibilities of stress shielding reduction, and there-
fore bone resorption minimization, leading to thigh pain allevia-
Contributed by the Bioengineering Division for publication in the JOURNAL OF
BIOMECHANICAL ENGINEERING. Manuscript received by the Bioengineering Divi-
sion June 26, 2000; revised manuscript received December 13, 2000. Associate Edi-
tor: V. K. Goel.
prosthesis designs made from composite materials. Due to the
anisotropic nature of composite materials, they offer to the de-
signer the possibility to tailor the desired optimized stiffness and
strength and are potentially excellent materials for hip prostheses
applications. However, it is necessary to prove the benefits of
these composites clearly, namely those properties related to bio-
compatibility and biofunctionality performance.
Published reports suggest that composite prostheses perform no
better than similarly shaped titanium alloy prostheses with regard
to maintaining bone stock. Further, unpublished reports from
clinical trials have shown early failure of femoral components
made from composite materials 9. One of the major problems in
the design procedure of femoral components when these materials
are used is the degree of design complexity. Laminated composite
technology has been widely developed by the aerospace and aero-
nautic industry. The design concepts developed do not seem to be
appropriate for femoral component design, and have been ignored
by many authors.
Previous studies have suggested that a controlled stiffness pros-
thesis is required to address the conflicting design requirements of
minimizing stress shielding and interface stresses 2,11. A pre-
liminary investigation of a novel controlled stiffness proximal
femoral prosthesis was performed by Simoes et al. 11,12. A
conceptual prototype of a novel prosthesis consisting of a cobalt
chrome core surrounded by a flexible composite material was de-
veloped and manufactured embedding a controlled stiffness con-
cept, based on optimized elastic modulus data of a cementless
prosthesis design 2,13. The aim of this work was to carry out an
appropriate experimental analytically validated procedure to de-
termine the variation of the prototype prosthesis stiffness and to
validate a concept.
Materials and Methods
To determine the variation of the prosthesis stiffness, the
method idealized and implemented consisted of the comparison of
the prosthesis lateral displacement field measured using a noncon-
tact experimental technique, namely ESPI Electronic Speckle
Pattern Interferometry with the one obtained with an analytical
model. The study was performed with a modified Freemans
femoral prosthesis, whose front orthographic projection is shown
in Fig. 1. The conceptual prosthesis was manufactured using a
composite technology, which allowed the implementation of the
concept 12. The prototype was studied and developed based on
findings of material stiffness optimization of a femoral component
13. These findings suggest that an optimized femoral component

234 Vol. 123, JUNE 2001 Copyright 2001 by ASME Transactions of the ASME

Fig. 1 Orthographic projection of the test model prosthesis
must have a relatively high proximalmedial modulus and a low
distal modulus matching cortical bone modulus at the prosthesis
tip 2,14. The model used in this study was manufactured to have
a modulus of 80 GPa at the proximal lateral part of the prosthesis
varying to 20 GPa distally. The modulus of the proximal medial
wedge was varied from 150 to 100 GPa at the intersection with
Fig. 2 Manufactured composite proximal femoral prosthesis
Journal of Biomechanical Engineering
Fig. 3 Prosthesis load as a cantilever beam
the straight-backed stem 12. Using a single modulus material
geometry cannot be deeply changed, it is not possible to physi-
cally materialize the optimized numerical results. Even using
composite materials, it seems difficult to achieve, within the same
structure, such highly differentiated modulus. To overcome this
manufacturing design obstacle, the prototype prosthesis was
manufactured combining the geometry of an inner cobaltchrome
core E200 GPa with the thickness of a relatively flexible com-
posite outer layer E12 GPa 11,12. The core was modeled and
machined with Powershape CAD/CAM software Delcam, plc.,
Birmingham. For the composite structure, hybrid commercial
braided pre-forms A&P Technology, KY of glasscarbon fiber
reinforcement were used. The composite layer modulus and Pois-
sons ratios were obtained experimentally testing tubular speci-
mens with internal pressure. Axial and circumferential strains and
therefore the axial circumferential stiffness properties were ob-
tained (E 8.9 GPa; E z 12.7 GPa; v z 0.461 and v z
0.323) 12. The prototype was manufactured localizing and
bonding sequentially carbonglass pre-forms around a stiff metal
core with different number of layers from distal to proximal. A
mean thickness of 0.35 mm was considered for each layer and
therefore, the total number of layers was calculated at different
locations on the prosthesis. The composite structure material was
impregnated with an epoxy resin, placed in a mold, and left to
cure at room temperature for 24 hours. The prototype was then
submitted to a curing cycle during 24 hours at 80C. Figure 2
shows the prototype manufactured used to test the controlled stiff-
ness concept.
The controlled stiffness concept has shown its adequacy for the
design of a proximal femoral prosthesis considering the compro-
mise between the strain energy distribution a parameter used to
determine the stress shielding effect and therefore the degree of
bone resorption and the cancellous bone stresses parameter to
determine the prosthesis migration 11,12,15,16. By combining
a stiff material with a more flexible one, it is possible to obtain a
prosthesis with controlled stiffness, which meets the requirements
of the compromise between the two parameters mentioned: stress
shielding and migration. Following Simoes et al. 12, the con-
trolled stiffness prosthesis showed a very similar numerical be-
havior to that observed with an identical cemented one.
To determine the prosthesis stiffness, an experimental
numerical procedure was idealized and its applicability studied.
JUNE 2001, Vol. 123 235
The method consisted of the comparison of the lateral displace- where R(z) is the radius of a circular section A as a function of
ment field measured using ESPI and the same data resulting from coordinate z as follows:
an analytical model. In both the numerical and experimental mod-
els the prosthesis was loaded as a cantilever beam Fig. 3. No 60 z
R z for 0z96 (3)
torsional force due to the anterior loading component of the hip 17 17
joint force was taken into account within the experimental setup.
The objective within the study was to validate a controlled stiff- For the proximal part of the model prosthesis, the geometry sec-
ness concept, by determining an optimal variation composite tion B was considered as a sum of four elementary geometries
thickness, assuming that the bending stiffness ElyyYoungs Fig. 1. Therefore, the second moment of area function for this
modulus x second moment area of the optimized prosthesis is the part of the prosthesis was obtained by
same as the bending stiffness of a prosthesis composed of a metal 4
core and a composite layer 12. For an equivalent torsional stiff-
ness, the variation of the optimal composite layer thickness was
I y y z
i1
y y i z I y y 1 z I y y 2 z I y y 3 z I y y 4 z

inferior to the one obtained for the equivalent bending stiffness,

therefore playing a second-order effect concerning the study for 96z162 (4)
hereby presented. where
Analytical Model. An easily understood analytical model
was developed to obtain the numerical solution for the lateral
displacements of the prosthesis loaded as a cantilever beam. The
I y y 1 z R 4 sin 2 4 sin2
4

8

9
R2
2R sin
3
G z 2

displacement distribution of a beam can be obtained by integrat- (5)

ing the equation:
d2y z

Mz
(1)
I y y 2 z
rR 3 cos3
18
rR cos
2R cos
3
G z 2
(6)

dz 2 E z I y y z
rB 3 L z
where y(z) is the displacement function, E(z) the Youngs modu- I y y 3 z 2rL z R cos G z 2 (7)
6 2

lus, M (z) the moment due to the loading, and I y y (z) the second
moment of area, respectively, all functions of the z coordinate. 8 r2 4r 2
I y y 4 z r 4 L z R cos G z
The second moment of area function was analytically obtained 8 9 2 3
having in consideration the prosthesis geometry and dimensions (8)
Fig. 1. Considering the prosthesis main sections section A and
B on the coronal plane, the function characterizing the second and

moment of area was obtained for each of them. For the middle-
r
distal part of the prosthesis, this function was determined using L z ztg30R cos 180arcsin r (9)
the equation: R
R z 4 The variable G(z), center of mass of section B, was determined
I y y z (2)
4 by the equation:

2 3 2 r2 2
R sin rR 2 cos2 rL z 2 2rL z R cos R cos L z r 2 r 3
3 3 2 2 3
G z (10)
r2
R 2 rR cos 2rL z
2

Finally, the analytical solution for the displacement distribution
was obtained replacing the function of the variability of the opti-
mized Youngs modulus E(z) and the function for the second
moment of area I y y (z) approximated to a second-degree function
in Eq. 1.
The applicability of the second moment of area function to be
used within Eq. 1 was tested using a finite element model. The
lateral displacements of a constant Youngs modulus prosthesis,
loaded as a cantilever beam, were obtained by integrating Eq. 1
and using the finite element method and data obtained by both
methods compared. For the finite element analysis, the prosthesis
was modeled with SolidWorks 99 computer aided design software
and simulated with Cosmos/Works 5.0 finite element analysis
software. The finite element model was composed of 2139 tetra-
hedral four-node elements and 702 nodes. The numerical integra-
tion of Eq. 1 was performed with Mathematica software 17.
The prosthesis was loaded with a load of P700 N.
To determine the numerical displacement field of the prototype,
a linear relationship between the optimal elastic modulus and the
axial prosthesis coordinate was used:
130
E z 20
z GPa (11)
162
for z 0,160 . Replacing Eq. 11 in Eq. 1, the following equa-
tion was numerically integrated to determine the displacements
along the lateral side of the controlled stiffness prosthesis:

d2y z Pz

(12)
dz 2 130
20 z 103 0.00092z 2 2.24105 z7.25108
162

where P was the load applied.

236 Vol. 123, JUNE 2001 Transactions of the ASME

 Fig. 4 ESPI setup to measure in-plane displacements

Experimental Model. To perform the experimental measure- second moment of area function in Eq. 1 to obtain the displace-
ments, the electronic speckle pattern interferometry ESPI tech- ments of the controlled stiffness prosthesis prototype analytically.
nique was used. With this technique, the assessment of the dis- Table 1 shows the displacement values measured experimen-
placement filed of the prototype was possible with no need of tally and those obtained analytically. Relative to the ESPI experi-
contact or surface preparation and with very high resolution, the ment, Fig. 5 shows phase map a, filtered phase map b, and
order of magnitude of the laser wavelength. The experimental unwrapped phase map c obtained for the prototype model pros-
displacements were obtained with a manufactured prosthesis pro- thesis. It is a typical phase map of cantilever-loaded beam. Figure
totype to determine the variation of its stiffness by comparing 6 shows the graphic comparison of the ESPI and numerical inte-
with identical results derived from the analytical model. The pros- gration displacement fields. A relatively good correlation between
thesis was fixed and loaded as assumed in the analytical model, as the displacement fields was obtained using the method described,
a cantilever beam. A stiff support was used to hold the prosthesis and so we can state with some certainty that the manufactured
30 mm from its distal end. The magnitude of the displacement prosthesis prototype had embedded equivalent optimized stiffness.
field measured was of the order of the micron, therefore a very This means that the prosthesis was manufactured with an equiva-
low load of P0.2 N was used, which allowed us to obtain a lent Youngs modulus as stated in the materials and methods sec-
reasonable number of fringes to calculate the displacements val-
ues easily. The experimental setup used Fig. 4 is a ESPI setup.
The interferometer arms were oriented in such a way that the Table 1 Displacement field determined analytical and mea-
sensitivity vector was coincident with the direction of the dis- sured experimental on a controlled stiffness femoral prosthe-
placements measured, that is, perpendicular to the lateral surface sis
of the prosthesis.
The displacement field obtained resulted from the phase map
calculation by using an image-processing algorithm based on
phase stepping techniques. The 2 phase difference corresponds
to a /2 displacement; therefore, the displacement of a point on
the prosthesis surface was obtained by:

y z fringe_order (13)
2
For the laser used, 632 nm, and replacing this value in Eq.
13, the displacements were obtained along the lateral side of
prosthesis.

Results and Discussion

The applicability of the second moment of area function of the
prosthesis model was verified by comparison of the displacements
calculated with the finite element method and the numerical inte-
gration of Eq. 1. The results obtained for both methods were
similar along the prosthesis length. The maximum displacement
observed was 1.704 mm for the numerical integration of the ana-
lytical model and 1.720 mm for the finite element method. This
observation gave the necessary confidence to use the analytical

Journal of Biomechanical Engineering JUNE 2001, Vol. 123 237


Fig. 5 Phase map of the measured displacement field
Fig. 6 Comparison between the experimental and numerical
displacements
tion. However, at the distal part of the prosthesis where it was
rigidly fixed, the highest displacement difference was observed,
essentially due to the natural experimental perturbing region. The
displacement difference near the prosthesis fixation device was
400 percent and gradually diminished up to a difference of 2.5
percent at the last observed fringe, which was localized at a dis-
tance of 15 mm from the applied load. The higher differences
observed at the fixation region of the prosthesis may have resulted
from a less efficient fixation of the prosthesis. In the analytical
numerical model, the fixation of the prosthesis was considered
ideally. We must say that the measuring displacements are of the
order of a micron and can easily be influenced by the experimen-
tal setup. The maximum displacement obtained with the ESPI
238 Vol. 123, JUNE 2001
technique was 4.3 m; for the analytical model, the maximum
displacement was 4.1 m, a relative difference of 5 percent.
Conclusions
The methodology idealized and described within this paper
proved to be suitable to validate the purpose of the study. Having
developed a conceptual novel prosthesis with an objective to ma-
terialize numerical optimized modulus data, there was a necessity
to validate experimentally the variability of the prototypes stiff-
ness and therefore the concept. The experimental method based on
ESPI was revealed to be suitable to determine the prosthesis dis-
placements and the comparison with the ones obtained with the
analytical model proved the adequacy of the procedure imple-
mented. Relative to the conceptual prosthesis, it is shown that it is
possible to materialize optimized Youngs modulus data by over-
coming possible composite manufacturing difficulties.
Acknowledgments
The authors would like to thank Carlos Relvas from the Depart-
ment of Mechanical Engineering of the University of Aveiro and
Jose Carlos from Norcam-Engenharia e Design Industrial for the
machining of the mold and core for the prototype.
References
1 Huiskes, R., 1993, Failed Innovation in Total Hip Replacement. Diagnosis
and Proposals for a Cure, Acta Orthop. Scand., 64, No. 6, pp. 699716.
2 Kuiper, J. H., and Huiskes, R., 1993, Numerical Optimization of Hip-
Prosthetic Stem Material, in: Computer Methods in Biomechanics and Bio-
medicine, Middleton et al., eds., Gordon and Breach, New York, pp. 7884.
3 Blake, T. A., Davy, D. T., Saravanos, D. A., and Hopkins, D. A., 1992,
Numerical Optimization of Composite Hip Endoprostheses, in: Proc. 4th
AIAA/USAF/NASA/OAI Symposium on Multidisciplinary Analysis and Optimi-
zation, pp. 119129.
4 Chang, F., Perez, J. L., and Davidson, J. A., 1990, Stiffness and Strength
Tailoring of a Hip Prosthesis Made of Advanced Composite Materials, J.
Biomed. Mater. Res., 24, pp. 873899.
5 Claes, I., Burri, C., Neugebauer, R., and Gruber, U., 1983, Experimental
Investigations on Hip Prostheses With Carbon Fibre Reinforced Carbon Shafts
and Ceramic Heads, in: Ceramics in Surgery, P. Vincenzini, ed., Amsterdam,
pp. 243250.
6 Christel, P., Meunier, A., and Leclerq, S., 1987, Development of a Carbon
Carbon Hip Prosthesis, J. Biomed. Mater. Res., Applied Biomaterials, 21,
No. A2, pp. 191218.
7 Davidson, R., Brabon, S., Lee, R. J., Nelson, K., Unwin, P., and Roughley, P.,
1996, The Development of CFRP Based Hip Stems, Proc. 7th European
Conference on Composite Materials, ECCM-7, London, pp. 513517.
8 Devanathan, D., 1991, Orthopadic Composites International Encyclopedia
of Composites, Stuart M. Lee, ed., New York, 4, pp. 7486.
9 Fu, W., 1998, Design Optimization of a Laminated Composite Femoral
Component for Total Hip Joint Arthroplasty, Ph.D. thesis, Graduate School
of Clemson University.
10 Widmer, M., Callenbach, T., Isler, J., Frohlich, M., Meier, D., Mayer, J.,
Wintermantel, E., Tschanz, P., Luthi, H., Matzenauer, K., and Klostermann,
L., 1995, Injection Moulding of Carbon Fibre Reinforced PEEK for Aniso-
tropic Hip-Prostheses: Preliminary Results, Proc. III Portuguese Congress in
Biomedical Engineering, Porto, Portugal.
11 Simoes, J. A. O., Taylor, M., Marques, A. T., and Jeronimidis, G., 1998,
Preliminary Investigation of a Novel Controlled Stiffness Proximal Femoral
Prosthesis Proc. Inst. Mech. Eng., 212, Part H, pp. 165175.
12 Simoes, J. A., Marques, A. T., and Jeronimidis, G., 2000, Design of a
Controlled-Stiffness Composite Proximal Femoral Prosthesis, Comput. Sci.
Eng., 60, pp. 559567.
13 Kuiper, J. H., 1993, Numerical Optimization of Artificial Hip Joint De-
signs, Ph.D. thesis, University of Nijmegen, The Netherlands.
14 Simoes, J. A. O., Antonio, C. A. C., and Marques, A. T., 1996, Material
Stiffness Optimization for a Composite Hip Prosthesis, in: Proc. 1st Inter-
national Conference on Composite Science and Technology, S. Adali and V.
E. Verijienko, eds., Durban, pp. 477482.
15 Taylor, N., Tanner, K. E., Freeman, M. A. R., and Yettram, A. L., 1995,
Cancellous Bone Stresses Surrounding the Femoral Component of a Hip
Prosthesis: an ElasticPlastic Finite Element Analysis, Med. Eng. Phys.,
79B, pp. 544550.
16 Taylor, M., and Tanner, K. E., 1997, Fatigue Failure of Cancellous Bone: a
Possible Cause of Implant Migration and Loosening, J. Bone Jt. Surg., Br.
Vol., 79B, pp. 181182.
17 Wolfram, S., 1991, MathematicaA System for Doing Mathematics by Com-
puter, Addison-Wesley Publishing Company, Inc., 2nd ed.
Transactions of the ASME