Computer Methods in Biomechanics and Biomedical

Engineering

ISSN: 1025-5842 (Print) 1476-8259 (Online) Journal homepage: https://www.tandfonline.com/loi/gcmb20

Development of a method for quantifying the

midsole reaction model parameters

Roozbeh Naemi & Nachiappan Chockalingam

To cite this article: Roozbeh Naemi & Nachiappan Chockalingam (2013) Development of a
method for quantifying the midsole reaction model parameters, Computer Methods in Biomechanics
and Biomedical Engineering, 16:12, 1273-1277, DOI: 10.1080/10255842.2012.666795

To link to this article: https://doi.org/10.1080/10255842.2012.666795

Published online: 11 Apr 2012.

Submit your article to this journal

Article views: 107

View related articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=gcmb20
Computer Methods in Biomechanics and Biomedical Engineering, 2013
Vol. 16, No. 12, 1273–1277, http://dx.doi.org/10.1080/10255842.2012.666795

Development of a method for quantifying the midsole reaction model parameters

Roozbeh Naemi* and Nachiappan Chockalingam
Faculty of Health, Staffordshire University, Leek Road, Stoke on Trent ST4 2DF, UK
(Received 21 February 2011; final version received 12 February 2012)

Midsole force – deformation pattern has important implications in determining the kinematics and kinetics of foot during
locomotion. Furthermore, the midsole stiffness and viscosity determine the midsole’s force – deformation behaviour.
Despite the importance of stiffness and damping components of the midsole reaction, which determines the shoe-specific
ground reaction forces during locomotion, there is still a lack of methodology to quantify them separately. The purpose of
this study was to develop a method of extracting the shoe-specific midsole stiffness and damping components during
uniaxial compression testing. For this purpose, the force – deformation behaviour of the sole was modelled as a system
consisting of a nonlinear spring and a nonlinear damper. Based on the fact that the stiffness and damping component of the
midsole reaction force acts in favour during loading, and work against each other during unloading, the stiffness and
damping components were separated. Utilising a curve-fitting technique, a parametric curve represented by the stiffness and
damping components of the midsole reaction force model was fitted to each components of force-deformation data to extract
the parameters. Statistical tests indicated that the proposed method is reliable for extracting the midsole reaction model
parameters with the stiffness and damping components producing favorable results (R 2 0.998 ^ 0.000 and
0.984 ^ 0.018/root mean squared error of 5.550 ^ 0.954 and 3.286 ^ 2.504, respectively).
Keywords: midsole reaction model; parametric curve fitting; footwear; ground reaction force; midsole deformation

Introduction structure like the heel pad mechanical properties could

Whilst wearing a shoe, the sole complex determines the
mechanical properties of the interface between the ground
and human body. As a generic midsole consists of
viscoelastic material, the reaction behaviour is similar to a
mechanical system represented by a nonlinear spring and a
nonlinear damper. Hence, the reaction force can be
formulated according to the following equation (Cole et al.
1996):
F ¼ Aðax b þ cx d VÞ
Where F represents the vertical ground reaction force, A
represents the plantar area of each contact point of the foot,
x represents the deformation and V represents the
deformation velocity of the shod foot. The constants a,
b, c and d are the midsole reaction model parameters,
whereas the first term in the bracket (ax b) represents the
nonlinear stiffness term and the second term (cx dV)
represents the nonlinear damping term. Previously these
constants were determined using a trial and error
procedure in which the force –deformation curve predicted
from the simulated pendulum impact tests is compared
with the force deformation curve of the pendulum test data
provided by Aerts and de Clercq (1993). Furthermore, the
parameters within these studies were quantified when the
participant was wearing the shoe. The variations in the foot
have influenced the reported results.
The sole complex that is the interface between the body
and the ground can be considered as a system consisting of
multiple shock attenuators including the insert or orthoses
and the insole, midsole and the outsole. All the components
included in the interface exhibit viscoelastic behaviour
with a different viscosity (damping) and stiffness (rigidity).
Since the midsole forms a dominant part of most shoe sole
structures, the mechanical property of the total sole/sole
ð1Þ complex is mainly determined by the viscoelastic
behaviour of the midsole. Depending on these character-
istics, the energy return ability that is defined as the
percentage of impact energy that is not dissipated in the
material during the loading and unloading cycle is
different. Furthermore as the sole complex serves as a
foot bed, over which the foot is supported during the stance
phase, the midsole reaction model parameters determine
the foot orientation and the ground reaction force
relationship and consequently the forces and moments
applied to the whole body during movement.
Despite the importance of the viscoelastic character-
istics of the midsole, the mechanical properties of the
midsole have not been investigated in specific detail in
such a way that the stiffness and the damping properties
can be determined as two interacting components. Unlike
the stiffness term that is only dependent on the

*Corresponding author. Email: r.naemi@staffs.ac.uk

q 2013 Taylor & Francis
1274 R. Naemi and N. Chockalingam
deformation, the damping term is correlated with the
deformation as well as with the deformation rate (Equation
(1)). As a result, quantifying the stiffness and damping
components separately allows prediction of the midsoles
reaction force at different sets of deformation and
deformation rates. Furthermore, this would allow quanti-
fying the frequently used terminologies within footwear
science, such as viscous versus elastic as an indicator of
damping level, and soft versus hard as an indicator of
stiffness level, for which there is a clear paucity of
quantitative measurements in the published literature. The
existing knowledge and categorising criteria for midsole is
based on using shore hardness that is defined as a
material’s resistance to permanent indentation. This
measure, although useful to get a general understanding
of the material behaviour, does not provide adequate
information on how the midsole complex behaves during
loading. This makes it difficult to evaluate how the
reaction force and deformation characteristics of a midsole
complex are related. Although the compression loading of
the midsole complex provides the force –deformation
curve during loading and unloading, there is still a lack of
methodology to quantify the damping and stiffness
components separately. Separating these components can
allow quantitative comparison of viscous/elastic and
hard/soft concept between the two designs. This can
potentially lead to testing and development of a sole design
comprising materials with stiffness and damping proper-
ties that can fulfil the performance requirement for specific
purpose. These can range from the viscoelastic material
that can provide the required stiffness and damping
component requirements for a running shoe to the
performance criteria of a walking shoe for diabetic
patients. Therefore, the overall aim of this study was to
propose a method of extracting the midsole reaction model
parameters. Although one might argue that the sole
complex consists of various components with varying
material properties, within the scope of the study we have
used ‘midsole’ as a term to refer to the ‘footbed’ within
any shoes. The initial objective of this study was to
develop an innovative method to separate the stiffness and
damping components of the midsole reaction force and to
extract the midsole reaction model parameters correspond-
ing to each of these two components. A further objective
was to validate this proposed method and to discuss the
implications of these parameters in practice.
Methods
A 3 kN universal testing machine (Lloyd Instruments, West
Sussex, UK) was used to carry out uniaxial compression
testing with a 0.042 m diameter top head. A conventional
running shoe was loaded on the lateral heel region to a
maximum load of 2500 N. The deformation rate of
0.5 m min21 was used as the maximum deformation rate
that can be carried out by the machine. The test was
repeated at deformation rates of 0.3 and 0.4 m min21 to
compare the parameters at different deformation rates.
Force and deformation were sampled at 100 Hz. Within the
midsole, like any typical suspension system, the stiffness
element (spring) resists compression, whereas the damping
element (damper) resists changes in deformation, by
applying the force opposite the direction of deformation.
Based on this fact, during loading the stiffness and damping
elements act in favour of each other, i.e. both the spring and
damper resist further deformation. Hence, the midsole
reaction equation can be rewritten as Equation (2)
F l ¼ A½ax b þ cx d V
ð2Þ
Based on the fact that the damper resists any changes in
deformation and the spring resists compression, during
unloading the stiffness and damping act against each other.
As the damper resists rebounding the midsole complex to its
original shape, the spring helps the rebound.
The midsole reaction equation can be rewritten as
Equation (3)
F u ¼ A½ax b 2 cx d V
ð3Þ
Summing up Equations (2) and (3), the elasticity
component can be written as
Fl þ Fu
¼ Aax b ð4Þ
2
While subtracting loading (Equation (2)) and unloading
(Equation (3)), the viscosity term can be found as follows:
Fl 2 Fu
¼ AVcx d : ð5Þ
2
To find the loading – unloading forces that correspond to
the identical deformation, the force –deformation curve for
loading and unloading fitted with a polynomial curve.
Then, the forces were reconstructed for the entire
deformation range from 0 to maximum deformation
within 0.05 mm increment.
The loading and unloading forces corresponding to the
same deformation were used to calculate the left side of
Equations (4) and (5). These stiffness and viscosity
components along with the corresponding deformation
were imported into Matlabw curve-fitting toolbox. A power
function represented by Equation (4) was fitted to the
stiffness component –deformation data. a was calculated as
the ratio of the scaling factor of the power fit to the top head
area (A) and b was found as the exponent of the power fit.
Similarly, a power function represented by Equation
(5) was fitted to the damping component – deformation
data. c was calculated as the ratio of the power fit scaling
factor to the top head area (A) times deformation rate (V).
The scaling factor of the fit was found as d.
Since the damping component represents a non-
conservative force, the work done against this force will
 Computer Methods in Biomechanics and Biomedical Engineering 1275

be dissipated to the environment through heat during the Er

loading and unloading cycle. On the other hand, stiffness Energy efficiency ¼ : ð8Þ
Ei
represents a conservative force, hence the work against
the stiffness component is stored as strain energy during
loading and returned to the system during unloading. Results
Based on these, the shoe-specific midsole reaction model The stiffness reaction force component has a higher
parameters can be used to quantify the energy return contribution to the reaction force than the damping
efficiency of the midsole. The area under the loading component. The contribution ratio of the stiffness to the
curve in a force– deformation graph represents the work total reaction force increases as the midsole undergoes
input, and the area under the unloading curve represents further deformation (Figure 1).
the work output. Knowing the equation for the loading Once the data were applied to the model, statistical
and unloading curves, the work input and output can be analysis indicated an R 2 value of 0.998 ^ 0.000 and
calculated using integration techniques (Equations (6) 0.984 ^ 0.018 for the stiffness and viscosity components,
and (7)). respectively. Furthermore, the root mean squared error
ð Xmax (RMSE) was recorded as 5.550 ^ 0.954 N and
Ei ¼ A ðax b þ cx d VÞdx 3.286 ^ 2.504 N for the stiffness and damping parametric
X min curves.

X max Figure 1 shows the stiffness and damping component
a c of the midsole reaction force and the corresponding power
¼A x bþ1 þ x dþ1 V ; ð6Þ
bþ1 dþ1 X min fits.
Table 1 shows the midsole reaction model parameters.
Table 2 shows the midsole reaction force calculated for
ð X max three deformation rates at deformations of 4, 8 and 12 mm.
Er ¼ A ðax b 2 cx d VÞdx The effect size is calculated as the ratio of standard
X min
deviation to the mean for the reaction force.

X max For comparative analysis, the reaction force applied to
a c
¼A x bþ1 2 x dþ1 V ; ð7Þ a contact area of 32.2 cm2 as the surface area of the
bþ1 dþ1 X min
pendulum reported by Cole et al. (1995) was calculated for
the tested shoe. This was then compared against the forces

Figure 1. The stiffness and damping components of the midsole reaction force and the corresponding fits. The vertical axis denotes force
(N) and the horizontal axis represents deformation (m).
1276 R. Naemi and N. Chockalingam

Table 1. Midsole reaction model parameters.

Deformation
rate (m/min) a b c d
0.3 80,376,099 1.205 802,172,249 0.889
0.4 104,279,503 1.266 272,758,785 0.773
0.5 88,175,397 1.226 508,254,257 0.928

applied to the same area using the. typical hard and soft
sole model parameters reported in Zadpoor et al. (2007)
Figure 2 shows the reconstructed reaction force for the
prototype shoe and the categorical hard and soft shoe
(based on Zadpoor et al. 2007) at three deformations of 4,
8 and 12 mm.
For the values mentioned in Table 1, and the loaded
area of 13.8 cm2 (the top head area) work input of 3.5 J,
work output of 2.3 J was calculated. This will result in an
energy return efficiency of 65.4% at deformation rate of
0.5 m/min. Based on the midsole reaction model
parameters, energy efficiency values of 61.8% and
67.2% are calculated for deformation rates of 0.3 and
0.4 m/min, respectively. The difference between the work
input and the work output is the work done against the
resistance of damper during loading and unloading that is
dissipated to the environment in the form of heat.
Discussion
The technique reported in this study allows separating the
stiffness and damping components of the midsole reaction
force and to quantify the reaction model parameters using
a parametric curve-fitting approach. The stiffness com-
ponent plays a dominant role in the total midsole reaction
force as compared to the damping component at the
deformation rates under which the prototype was tested.
Validity
The fit goodness for the force – deformation curve to the
raw displacement data highlights the validity of the
reported technique. High R 2 value and low RMSE
Table 2. The midsole reaction force calculated at three
deformation rates for deformations of 4, 8 and 12 mm.
Deformation (m)
Reaction force (N) @ 0.004 0.008
Def. rate 0.3 m/min 185 407
Def. rate 0.4 m/min 168 380
Def. rate 0.5 m/min 175 394
STDEV 8.27 13.55
Average 175.9 393.7
Effect size (%) 4.70 3.44
Figure 2. The reconstructed reaction force for the prototype
shoe and the categorical hard and soft shoe (based on Zadpoor
et al. 2007) at three deformations of 4, 8 and 12 mm.
indicated that the underlying assumptions and principles,
which were used in the process of developing the method,
are valid. Comparing the midsole reaction forces at
different deformation rates revealed that the deformation
rate does not make considerable change in the predicted
force (maximum 4.7% effect size as the ratio of standard
deviation to the mean at deformation of 4 mm).
Furthermore, the fact that the predicted force from the
proposed model is close to what is predicted using the
categorical reaction model parameters previously pub-
lished within the literature (Zadpoor et al. 2007) indicates
that the assumptions made within this research for
developing the proposed method were realistic.
It is clear that the reaction forces for the shoe that was
tested within this study show higher values than the ground
reaction force of the categorical hard shoe. This may be
due to the fact that the data collection procedures were
different (Aerts and de Clercq 1993).
Energy efficiency
Although the area under the curve can be calculated
using numerical means by approximation of the area
below the curve as the sum of areas of trapezoids, the
mathematical integration technique introduced here
allows a further depth to be added to the energy efficiency
aspect of the midsole complex. For example using
optimisation methods, the stiffness and damping par-
ameters can be modified with a view of increasing the
energy efficiency.
0.012
648 Determining the rear-foot motion
616 As the midsole provides support for the foot during
636 locomotion, the midsole reaction model parameters can
16.07
633.6 have implications in determining the foot orientation.
2.54 The midsole reaction model parameters determine
the deformation pattern of the midsole when loaded.
 Computer Methods in Biomechanics and Biomedical Engineering
During a stance phase, as the different regions of the
midsole go under loading, the orientation of the foot
changes according to the relationship between the pressure
and deformation of each region underneath the plantar
surface.
The midsole reaction model parameters can also be
useful in determining the effect of footwear on the
movement kinematic/kinetic. In a sense, the type of
modification a midsole design imposes on the movement is
directly determined by the midsole reaction model
parameters.
This can have practical applications, i.e. to produce
a customised midsole component for footwear that
canf accommodate such a reaction force –deformation
requirement, in a way to modify gait that may
have important implications in determining the effect
before manufacturing the footwear. This allows the
use of midsole models, with certain midsole reaction
model parameters in simulation software in which
the footwear comprising such a midsole can be worn by
the virtual model and the effect can be determined on
the movement prior to having the physical prototype
made.
Application in mass –spring– damper modelling
The ground reaction model parameters that are
commonly used in simulation can only provide
categorical midsole reaction model parameters for the
hard and soft midsole. While these parameters proved to
be useful in generic model-based mass– spring – damper
models in a number of studies (Nigg and Liu 1999; Liu
and Nigg 2000; Zadpoor et al. 2007), for a simulation to
be realistic the ground reaction force models need to be
modified and defined based on the specific midsole
design. This allows the body mass models consisting of
spring and damper to be simulated with more realistic
ground reaction model parameters specific to the
midsole.
1277
Conclusions
The method proposed in this study can distinguish between
the stiffness and damping characteristics of the midsole.
The shoe-specific midsole reaction stiffness and damping
parameters then can be quantified for a shoe-specific
midsole. These parameters have important implications in
categorising the cushioning characteristic of a midsole and
in determining the energy efficiency. Both damping and
stiffness characteristics of the midsole play a major role in
the mechanical behaviour of the midsole and in
determining the force – deformation behaviour of the
midsole.
Acknowledgements
This work was funded in part through, the European
Community’s Seventh Framework Programme under grant
agreement no. 222468 (Heelless). The authors would also like
to thank Abbas Daragheh for his assistance with data collection.
References
Aerts P, de Clercq D. 1993. Deformation characteristics of the
heel region of the shod foot during a simulated heel strike:
the effect of varying midsole hardness. J Sport Sci. 11:
449– 461.
Cole GK, Nigg BM, van den Bogert AJ, Gerristen KGM. 1996.
Lower extremity joint loading during impact in running.
Clin Biomech. 11:181 – 193.
Liu W, Nigg BM. 2000. A mechanical model to determine the
influence of masses and mass distribution on the impact force
during running. J Biomech. 33:219 –224.
Nigg BM, Anton M. 1995. Energy aspects for elastic and viscous
shoe soles and playing surfaces. Med Sci Sports Exerc. 27:
92 – 97.
Nigg BM, de Boer RW, Fisher V. 1995. A kinematic comparison
of over ground and treadmill running. Med Sci Sport Exerc.
27:99 –105.
Nigg BM, Liu W. 1999. The effect of muscle stiffness and
damping on simulated impact forces during running.
J Biomech. 32:849– 856.
Zadpoor AA, Nikooyan AA, Arshi AR. 2007. A model-based
parametric study of impact force during running. J Biomech.
40(9):2012– 2021.