Pergamon J. Biomechrmics, Vol. 28, No. 9, pp.

1103-1109, 1995
Copyti&t 0 1995 Elseviex Scienct Ltd
Printed in Great Britain. All rights rwewed
mz-929ops 39.50 + .a0

0021-9290(94)00139-l

TECHNICAL NOTE

FORCE AND MOMENT DISTRIBUTIONS AMONG OSSEOINTEGRATED

DENTAL IMPLANTS

M. J. Morgan *t and D. F. Jarnest

*Department of Oral and Maxillofacial Surgery, Faculty of Dentistry, University of Toronto, Toronto,
Ontario, Canada M5G 1G6; and $Department of Mechanicaf Engineering, University of Toronto,
Toronto, Ontario, Canada M5S lA4

Ahatraet-Distributions of force, bending moment and torque are determined by structural analysis for an
osseointegrated dental implant system. The system is a dental prosthesis rigkiiy connectad to bone by
implants. Since the implants have the lowest flexural rigidity of the structural commts, they are
considered to be the only elastic components of the system. In the analysis, the number and position of the
implants are variable and the magnitude, direction and location of the applied load are arbitrary. The
distributions found by force and moment balances are in the form of simple algebraic equations, a form
which is useful for clinicians in determining the number and location of implants so that forces and
moments are shared equitably. One immediate result of the structural analysis is that the bending moment
due to the vertical component of the applied load-a moment which has previously been nef#e&d-can
produce stresses in the implant which are an order of magnitude larger than the direct axial stresses.

INTRODUffION Skalak (1983) analysed the distribution of forces in

a model structure, assuming that the connections between
An osseointegrated dental implant system (Fig. 1) consists of the implants and prosthesis are flexible. Since the
implants which connect a prosthesis to the mandible or connections are in fact rigid (Fig. 2) an improved analysis is
maxilla such that a force due to biting or chewing is needed, one that accounts for the bending moments and
distributed over the bone. The necessary tight connection to torque which develop as a result of the solid connections. In
bone is ma& possible by osseointegration, which allows the the analysis of Rangert et nl. (1989), a bending moment is
bone and implant to act as a unit in withstanding loads. This included for the two implants closest to the load. but the
implant system is used for reconstruction in the oral cavity. results cannot predict forces and moments on the remaining
At the first surgical procedure, the fixture component of the implants. Brooke-Smith (1988) presented three analyses for
implant is placed in the bone and left unloaded for four to six the distribution of a vertical force; because the prosthesis was
months to achieve integration. During the second surgical modelled as a simply supported beam, bending moments
procedure, the abutment component which is the trans- were not considered.
mucoaaJ element, is co~cctcd to the fixture by an abutment
screw (Fig. 2). The prosthesis is then made, which usually METHOD OF ANALYSIS
consists of a silver-palladium alloy framework with an
acrylic superstructure simulating teeth. Gold alloy cylinders
which are cast into framework fit precisely onto the The analysis of forces and moments is carried out for
abutments and abutment screws, and become the a model structure, one which is a simple yet realistic
intermediate elements which anchor the prosthesis to the representation of the actual structure. The model is like the
implants. one introduced by SkaJak (1983), consisting of a rigid
Mechanical failures have been reported (Adell et al. 1981) prosthesis attached to elastic implants which are anchomd in
for implant components made of commerciafly pure bone. But, in contrast to Skalak, the connections between the
titanium The fracture of a fixture is particularly important implants and the prosthesis in the present work are rigid.
because further surgical procedures are required to remove The meshed area (Fig. 3) represents the occlusaf plane of the
and replace the fixture. Morgan et al. (1993) demonstrated prosthesis which also defines the x-y plane. A force applied
that five fixtures which had fractured in uiuo had failed by to the prosthesis can be resolved into two components: one
fatigue. To investigate the mechanics of such failures, the vertical, or parallel with the long axes of the implants, and
distribution of forces and moments among the implants the other horizontal, or in the plane of the prosthesis. These
resulting from a concentrated load on the prosthesis is components will be referred to as Pv and PM, respectively,
required. Knowing these distributions can also provide and it is known that PH is up to about one-tenth of Py
valuable information in the treatment planning phase of (Brunski, 1988). The force is applied at the point (x,.y,). The
a clinical case by opitimixmg implant placement and by four implants shown are a typical arrangement, but the
minimixing adverse loading conditions. analysis is for an arbitrary number, n. The coordinates of the
ith implant are (xi, yi), where the origin of the coordinate
system coincides with the geometric centre (centroid) of the
TAuthor for correspondence. implants, i.e. Zy=, x, = 0 and I: ;=, y, = 0. The y-axis is
Received in final form 4 September 1994. parallel with the mid-sagittal plane of the patient.

1103
1104 Technical Note

Fig. 1. The prosthesis is rigidly attached to the mandible by implants.

The above description implies that the implants are modulus of elasticity, E, and the second moment of area, I, is
vertical or parallel to the z-axis. Clinically they may be a measure of resistance to bending. The modulus of elasticity
inclined up to 15” to improve bone contact, in which case Pv of titanium is 110 GPa (Hertzberg, 1989) and that of bone in
and P, have to be resolved into components parallel with the axial direction is about 20GPa (Ashman and Van
and perpendicular to the longitudinal axis of the implant. Buskirk, 1987). The second moment of the area of a typical
Since the angle is small, however, the parallel component of hollow elliptical cross-section of the mandible is estimated to
Pv is almost Pv itself, and the difference can be ignored. At be lo-sm4, while I for an implant with an effectively solid
the same time, since Pv is G(lO)Pn, the perpendicular circular cross-section is 4.5 x lo-i2 m4. Consequently, the
component of Pv may be significant and would have to be flexural rigidity EI of the mandible is about 400 times that of
added (vectorially) to PH in any numerical evaluation, By the implant, making the mandible the much more rigid
allowing for these modifications, the present work applies to component. It is assumed, then, that there are no deflections
inclined as well as to vertical implants, providing all the in the bone or prosthesis and that all of the deformation
inclinations are the same (with respect to the z-axis). If the occurs in the implants. The implants are taken to be identical
inclinations vary significantly, the present method may not and linearly elastic, and to have the same length exposed
be su5ciently accurate and a more complex analysis would above the marginal alveolar bone. They are also considered
be required. to be rigidly attached to the bone because of osseointegra-
The distributions of forces and moments throughout the tion.
structure depend on the relative flexural rigidities of the To fmd the forces taken by the implants, force and
components and on the rigidity of the connections. Some moment balances are made on the prosthesis. In carrying out
assumptions about rigidity must be made in order to simplify these balances, it is simplest to treat the components of the
the system but still retain the features necessary to give force separately, and the vertical component is considered
clinically useful results. First, the prosthesis is considered to first.
be a rigid plate because of its size and because the The free body diagram (FBD) for the vertical case is the
metal-acrylic superstructure is stiffer than the commercially projection of the prosthesis onto the z-x plane (Fig. 4). As
pure titanium fixtures (by about 2.5 times, according to our &l&ted the force applied by the ith imp&it is Fi, and for
calculations). Secondly, the bone in which the implants are static [Link], ZY-. F, = Pv. The load causes the plate
embedded is considered to be rigid, also because of size and to translate downward and to rotate about a horizontal axis
because its flexural rigidity is much larger than that of an through the centroid. When Pv is applied at the centroid
implant. Flexural rigidity, defined as the product of the there is no rotation and the load is shared equally amount
 Technical Note 1105

moment has two components, M, and M,, about the two

acrylic horizontal axes, and the M, components are shown (Fig. 4).
superstructure
I 1 i Obtaining a solution for Fi requires the relationship
between the bending moments and the slope of the plate. To
find this, consider the implants to be vertical beams (Fig 5).
The bending moment M at the end of a beam-in our case,
at the top of an implant--depends on beam lenffth and end
slope (with respect to the baam axis). The appropriate length
for the implant is the exposed portion above the alveolar
bone. The end slope is the slope a of the plate with respect to
the horizontaf. From Heam (1985),
Ml = Ela, (2)
where 1 is the length, and El is the flexural rigidity of the
mwxa implant. It is assumed that t is the same for all implants, and
abutment screw
this length is negligibly aRcted by extension (or compres-
sion) under load The end slopes of the implants are the same
because the plate is rigid and the implants are rigidly
attached to it. Hence, the moment M is the same for all
implants.
The vertical displacement z at any point (x, y) on the plate
is given by
2 = ax + by,
fixture where z is &fined as positive upwards and where a and b are
~ projections of the slope a onto the z-x and z-y planes,
respectively (see the appendix). Because Ri is proportional to
Zi at (Xi. Yi),
Ri = - k(axi + by,),
where k is the linear stiffness of the implant. The negative
Fig. 2. A cross-sectional view of an implant demonstrates sign is necessary because a positive value of z-as on the
how the prosthesis is connected to the implant. right side of the FBD (Fig. 4)--produces tension and R, (and
F,) is defined as positive in compression. The summation of
moments about the y-axis (the axis of rotation in Fig. 4)
yields
nM, = c [ - waxi + by,)xJ - Pvxp, (3)
the implants. When Py is applied at any other point, the
plate translates the same amount but also rotates, and then where the summation is over the n implants. Similarly,
taking moments about the x-axis yieIds
&:i=+&. (1) nM, = c [ - k(axi + by,)y,] - Pvy,. (4)
n
Equations (3) and (4) may be solved for a and b knowing that
R1is proportional to the extension or compression caused by
the rotation because the implants are modelled as linearly
elastic.
When the plate rotates, a bending moment is created at
each irnp~t~~~s joint, and the axis for this moment is as derived in the appendix. For later convenience, we instead
parallel with the axis of rotation of the plate. The bending present the solutions for a’ and b’. defined by a’ = - kalp,

Fig. 3. Definition sketch of the problem. A concentrated load, with vertical and horizontal components Pv
and PH, is applied to a rigid curved beam (the prosthesis, shown as a mesh) to which elastic implants are
rigidly attached. The origin of the coordinate system is the centroid of the implants.
1106 Technical Note

rigid plate (prosthesis)

plate, unloaded

Fig. 4. The free body diagram is used for the force and moment balance about the y-axis when a vertical
load Pv is applied.

r plate

X-Y plane

Fig. 5. The slope of the deflected beam (implant) at the end connection to the plate (prosthesis) is the angle
a between the plate and the x-y plane.

and b’ = - U/P,. The solutions for a’ and b’--after rule. The correctness of the signs can be illustrated by
substituting AE/l for k, where A is the implant cross- referring to the FBD (Fig. 4): xr, is negative and thus a is
sectional area-are positive; consequently, a’ is negative [as equation (6)
con6rms for yr = 01, and M, is positive (in the + j direction)
as @ven by equation (9) and as shown in the FBD.
Y,&Y~ - xACY: + nV4
Similarly, a free body diagram is used to find the force and
a’ = (ox& - (ni/A)c x: - (nl/A)&$ - cx;M - @I/A)” moment distributions for the horizontal component PH (Fig.
(6) 6) in the x-y plane. The direction of PH is detined by the unit
vector n,; e is the vector from the centroid to the line of
action of P, and e is its magnitude. The reactive shear force
applied by implant i to the prosthesis is
b’ = (&yJ2 - (d/A) xx: - (d/A)M - CxfB: - @I/A)”
(7) PH
Hi c-t-4) + Si(-ni),
n
Ri is now known, and thus
in which the first term is the reactive load due to the
translation of the plate, and S, is the force caused by the
F< = $ + Pv(dXi + b’J’J* (8) rotation of the plate around the z-axis when PH does not act
through the centroid. The direcbon of S, is defined by n,, the
unit vector perpendicular to r, in the plane of the plate, where
where Fi is positive in compression. r, is the vector from the centroid to implant i.
The bending moment components are then Because the implant/prosthesis joints are solid, each
implant is twisted about its longitudinal axis when the plate
rotates. Since the angle of twist of each implant is that of the
M,= -yb’, PVI
M, = - -a’, plate, and, since the implants are identical, the reactive
(9)
A torque from each implant is the same. Letting the magnitude
of this torque be T, and taking moments about the z-axis,
yields
where a’ and 6’ are given by equations (6) and (7) above. The
negative signs for M, and Mr are the result of the right-hand (exP,)-nnL+E(rixHi)=O.
 Technical Note 1107

Fig. 6. Forces and moments which develop as a result of the horizontal load component Pn are analysed,
with the aid of this free body diagram of the prosthesis.

Since (Hearn, 1985)

I’i X Hi = - ’ (4 X Op) - S&i X ai),

n
then where J is the polar moment of area, and G is the shear
modulus of elasticity. Substituting equations (12) and (13)
C(ri x Hi) = - kZSiri, into equation (11) it is found that
because ri = xii + yij and xxi = 0, Cyi = 0. It follows that Pi+
’ = (lZEZ/l~)~rf + nJG/l ’
nT = P,e - CSiri, (11) PHeri
(14)
where T is positive in the clockwise direction and PHe is si = CTi’ + nJGl’/12EI ’
positive in the counterclockwise direction. To eliminate S,, and the torque is
consider the implant to be a vertical beam with built-in ends
(Fig. 7). Let Si cause the beam end to be displaced an amount
Ai in the direction I&; because Ai is ri& then from Heam PHeJG
(198% ’ = (12EI/l’)crl’ + nJG (15)
Consequently, from equation (lo), the force applied to the
prothesis by the, ith implant is
where 4 is the angular displacement of the plate in the
counterclockwise direction, and where E, 1 and 1 have been PH PHeri
defined previously. Torque is related to angle of twist by Hi = y(-%) + Cr: + nJG[2/12EI (-ni)’
(16)

A bending moment Mi is induced in each implant about

a horizontal axis. M, depends on the magnitude of Ht and on
the length l, according to the formula (Heam, 1985)
prosthesis

where 1) indicates the magnitude of the vector.

AAAA.+,AAAAAAAA
I/
ilI
implant
RESULTS

The main results of this paper are the analytical expres-

sions apportioning the applied force among the dental
implants. For the vertical component of the force, equations
(6H9) give the axial forces and bending moments applied by
bone the implants to the prosthesis. For the horizon&I
[Link]
component, equations (15)-(17) give the shear forces,
moments and torques applied to the p~o&[Link]
Fig. 7. The built-in beam (implant) undergoes a deflection Ai and moments applied by the prosthesis to the implants are
in the x-y plane. equal in magnitude but opposite in direction.
1108 Technical Note

These equations are of practical importance for a clinician, which the unsupported length of one implant is greater than
especially during case planning. Since the equations are in the others because bone has resorbed around it [see, for
simple algebraic form, the clinician can easily carry out example, Fig. 11 in Morgan et al. (1993)]. For these cases,
calculations varying the number of implants, their spacing, the present results are approximate and we are currently
and the cantilever length(s) of the prosthesis in order to seeking more accurate results by experiment and by incor-
distribute loads and stress equitably throughout the implant porating variable stiffnessesin the analysis.
system. The possibility of mechanical failures is then Secondly the implants are assumed to be rigidly attached
reduced. to the bone, but these connections are not perfectly rigid
In formulating our model, it was noted that Skalak (1983) because of the elasticity of bone. However, bone stiffness
had analysed forces on a model which was identical except does not affect the force and moment distributions provided
that the joints were flexible. The present equations can be the stiffnessaround each implant is the same.
modified to create this situation and the resulting equations Finally, the prosthesis is assumed to be rigid. Every pros-
can be compared to Skalak’s. Suppose, then, that the connec- thesis deflects somewhat and the magnitude of the deflection
tions in our structure are not rigid but are ball-and-socket depends on the alloy of the framework, on the material of the
joints, in which case there are no bending moments and superstructure, on the distance between implants, and on the
M, = 0 = M,, or equivalently, I = 0. Our force distribution length of the cantilever section. Any deflection tends to
equations [equations (6)48)] then reduce identically to the concentrate the force on the implant nearest the load, which
result given by Skalak for a vertical load. That his equation modifies the force and moment distributions. However, the
(2) actually applies to flexible joints can be confirmed by prosthesis is relatively stiff, as described in the development
using his solution to determine the moment at any junction of the model, and thus the present results are sufficiently
and finding that the moment is indeed zero. In a recent accurate for many purposes. For greater accuracy, the stiff-
publication, Skalak noted that his equation does in fact ness properties of the prosthesis would have to be incorpor-
apply to freely jointed connections (Brunski and Skalak, ated.
1992). Similarly, if a horizontal load is applied and if our As to how much influence stiffnessof the prosthesis might
joints are made flexible, they cannot withstand a moment have, the only available information comes from Brunski
about the z-axis. Equivalently, the torsional rigidity JG can (1992). He built a laboratory model in which implants were
be set to zero; in this case the torque vanishes and our fastened to ‘prostheses’, curved beams with rectangular
equation (17) is the same as Skalak’s equation (1). Hence, our cross-sections. Two beams with different flexural rigidities,
equations reduce correctly to these special cases. El, were made, one of the dental acrylic and the other having
Now that joint rigidity is accounted for and bending a cast silver-palladium alloy framework with an acrylic
moments arise as a consequence, how much does joint rigid- superstructure, fashioned to duplicate an actual prosthesis
ity affect the forces (and stresses) in the implants? For the (apparently). A known vertical load was applied at three
vertical case, equations (6)-(S) show that the contribution to different locations on each prosthesis and the portion of the
Fi depends on the magnitude of d/A compared to the load taken by each implant was measured by strain gauges
magnitude of Xx: (or Zy:). To make the comparison simple, and compared with the loads predicted by the Skalak equa-
consider the problem in the z-x plane with yi = 0. Since nl/A tion (which is for a riaid nrosthesisj. While there is broad
is O(nrz), where r is the outer radius of the implant, and since agreement between theory-and exp&iment, it is difficult to
Xx: is O(nx’) where x is an average distance to an implant, draw conclusions from his work. First, the sums of forces on
d/A << Xx; because r2 Q x2. Consequently, when typical the implants differ significantly from the applied load, and
values are used for the parameters in equation (8), the results there is no discussion of experimental error to explain the
for Fi are only slightly different from those calculated by discrepancies. The large experimental variability unfortu-
Skalak. But the added bending moment creates stresses and nately eliminates any detailed comparisons between the data
these may not be small in comparison with the axial stresses and analytical results. Secondly, clinical relevance is not
created by Fi. The bending stress u, is MJI, which is established: the El values of the beams are not given and the
O(Pvy,/r3) (with yi = 0), whereas the axial stress a, due to F, beam rigidities are not related to the rigidity of an actual
is O(P,/A) or O&./r’). The ratio of these stresses,ax/u,, is prosthesis. Without this information, it is difficult to assess
O(y,/r). Since this ratio may be easily much larger than one, the influence of prosthesis rigidity and further work is
the dominant stress is likely the bending stress and not the needed.
axis stress. Hence, Skalak’s result is accurate for estimating There are two studies in addition to Skalak’s which deal
the vertical force distribution but it may lead to serious with force and moment distributions. Mailath et al. (1991)
errors if used for stress analysis. used a finite element technique to find distributions in
In the horizontal plane, the difference between the Skalak a structural model in which the number of implants, the
equation and the present result is the additional term cantilever length, and the load position were varied. Predic-
nJGl*/12EI, which can be as large as 20% of the denomin- tions are given for six models in all, and so comparisons to
ator. This amount alters the shear stresses,but, since shear values from our algebraic relations are in order. But the
stressesare typically much less than the normal stresses,the coordinate axes are not defined with respect to a skeletal
additional term has a minimal effect on any calculation of reference, the implant positions are not given, and the posi-
principal stresses. tion and direction of the applied force components are in-
These order-of-magnitude stress calculations clearly show completely specified. Unfortunately, then, it is not possible to
that the stress due to bending may be the dominant stress, make comparisons. Glantz et al. (1993) measured vertical
which reinforces the need for including bending moments in forces and moments in uiuo using strain gauges attached to
the structural analysis. five of six abutments. However, coordinate axes are not
defined and thus implant positions and gauge orientations
DISCUSSION are not known. Again, comparisons cannot be made.
Hopefully, new experimental work will properly identify
Although this investigation into force and moment distri- implant positions as well as measure individual loads and
bution was originally undertaken to understand why im- moments, for then it will be possible to assessthe importance
plants fail mechanically, the primary use will hopefully be for of bending moments and to make proper comparisons be-
treatment planning of clinical [Link] foregoing analyses tween theory and experiment.
are based on a structural model and assumptions in the
model deserve further discussion. First, the implants are Acknowledgements-The authors areatfully acknowledge the
assumed to be. identical in diameter, in cross-section and in support ofihis work by the Natural Scien&. and Engineering
length exposed above the marginal bone, conditions which Research Council (Canada). through a Research Grant to
hold for most clinical cases. However, there are cases in D. F. James. ~ ” I
 Technical Note 1109

REFERENCES Skalak, R. (1983) Biomechanical considerations in osseoin-

tegrated prostheses. J. Pros. Dent. 49, 843-848.
Adell, R., Lekholm, U., Rockier, B. and BrBnemark, P.-I.
(1981) A 15-year study of osseointegrated implants in the APPENDIX
treatment of the edentulous jaw. Int. J. Oral Surg. 10,
387-416. Relationship of moment components to the slope of the plate
Ashman, R. B. and Van Buskirk W. C. (1987) The elastic The relationship of the bending moment components, M,
properties of a human mandible. Adv. Dent. Res. 1,&l-67. and Mr, to a and b can be found by working in the ortho-
Brooke-Smith, M. (1988) The study of the fatigue life of small gonal coordinate system (t,u) which coincides with the axis of
gold locating screws used in the osseointegrated implant rotation of the plate. That is, let the axis of rotation coincide
technique. [Link]. thesis, Department of Mechanical and with the t-axis, so that the equation of the inclined plane is
Process Engineering, The University of Sheffield, U.K. z = au (Positive deflection is still in the positive z direction).
Brunski, J. B. (1988) Biomaterials and biomechanics in den- Let the t-axis be an angle fl from the x-axis (Fig. Al) with
tal implant design. Int. J. Oral Maxillofac. Imp. 3, 85-97. 6 defined as positive in the counterclockwise direction. Since
Brunski, J. B. (1992) Biomechanical factors alfecting the x=tcose-usin0andy=tsine+ucos&
bone-dental implant interface. Clin. Mater. 10, 153-201.
Brunski, J. B. and Skalak, R. (1992) Biomechanical consider- tan e = - s,
ations. In Advanced Osseointegration Surgery: Applications
in rhe Maxillofacial Region (Edited by Worthington, P. b a
and B&remark, P.-I.), p.23. Quintessence, Chicago. a
cos8 sin 0
Glantz, P.-O., Rangert. B., Svensson, A., Stafford, G.D.,
Arnvidarson, B., Randow, K., Linden, U. and HultCr, J. It is noted that when a and b are positive, 0 is negative
(1993) On clinical loading of osseointegrated implants. (Fig. Al). The magnitude of the bending moment about the
Clin. Oral Imp. Res. 4, 99-105. t-axis, M,, applied to the end of the implant, is proportional
Hearn. E. J. (1985) Mechanics of Materials (2nd Edn), Vol. 1, to the plate slope a. Hence, according to equation (2),
pp. 95, 146-147. Pergamon, London. ElcL
Hertzberg, R. W. (1989) Defonnotion and Fracture Mechanics M’=T
of Engineering Materials (3rd Edn), p. 8. Wiley, New York.
Mailath. G.. S&mid. M.. Lill W. and Miller. J. f1991) 3D- A positive slope a yields components M, and M, on the
finite-elemente-analyse’der biomechanik von rtm in&n- implant which are in the + i and - j directions, respec-
tatgetragenen extensionbriiken. 2 Zahntirztl Implantol. 7, tively; that is M, = M,(i) + MY( - j). Hence, it is found
205-211. that
Morgan. M. J.. James, D. F. and Pilliar, R. M. (1993) Frac- Elb Ela
tures of the fixture component of an osseointegrated im- Mx=? M,=I.
plant. Inc. J. Oral Maxillqjii. Imp. 8, 409-414.
Rangert, B., Jemt, T. and Jiimeus, L. (1989) Forces and These moments relate to the implant; the moments on the
moments on Brinemark implants. Int. J. Oral Maxillofac. prosthesis, as needed in the text, are equal in magnitude but
Imp. 4, 24 I .- 247. opposite in direction, i.e., in the - i and + j directions.

Fig. Al. The (x,y) and (t,u)coordinate systemsare related by the angle 8. The rigid plate (prosthesis) rotates
about the t-axis such that vertical displacements on the plate are given by s = au.