ARTICLE IN PRESS
`
`Journal of Biomechanics 37 (2004) 1379–1386
`
`Heel–shoe interactions and the durability of EVA foam
`running-shoe midsoles
`
`R. Verdejo*, N.J. Mills
`
`Metallurgy and Materials, University of Birmingham, Birmingham B15 2TT, UK
`
`Accepted 16 December 2003
`
`Abstract
`
`A finite element analysis (FEA) was made of the stress distribution in the heelpad and a running shoe midsole, using heelpad
`properties deduced from published force-deflection data, and measured foam properties. The heelpad has a lower initial shear
`modulus than the foam (100 vs. 1050 kPa), but a higher bulk modulus. The heelpad is more non-linear, with a higher Ogden strain
`energy function exponent than the foam (30 vs. 4). Measurements of plantar pressure distribution in running shoes confirmed the
`FEA. The peak plantar pressure increased on average by 100% after 500 km run. Scanning electron microscopy shows that
`structural damage (wrinkling of faces and some holes) occurred in the foam after 750 km run. Fatigue of the foam reduces heelstrike
`cushioning, and is a possible cause of running injuries.
`r 2004 Elsevier Ltd. All rights reserved.
`
`Keywords: FEA; Foam; Heelpad; Running; Plantar pressure distribution
`
`1. Introduction
`
`Running involves a series of heel-strikes on the
`ground. The midsole foams of running shoes, by
`absorbing energy, limit the peak impact force in the
`heel-strike. Shorten (2000) showed that soft cushioning
`systems increased the duration of footstrike impacts and
`spread the load across a larger area of the plantar surface.
`Plantar surface pressure distributions were measured for
`running in various shoe types by Chen et al. (1994),
`Henning and Milani (1995, 2000), and Shiang (1997).
`However, there is no published information on how the
`pressure distribution changes with shoe use.
`The impact response of a heel on a flat rigid surface
`was shown to be non-linear, with energy absorption on
`unloading, by Cavanagh et al. (1984). In these experi-
`ments the heelpad deformation is overestimated, since
`the lower leg and knee deform to some extent. Aerts
`et al. (1995, 1996) measured the impact response of the
`isolated lower half of the foot, giving a better indication
`of the heelpad response. However, no one has simulta-
`
`*Corresponding author. Tel.: +44-121-414-5180; fax: +44-121-414-
`5232.
`E-mail addresses: raquel@verdejo.net (R. Verdejo),
`n.j.mills@bham.ac.uk (N.J. Mills).
`
`0021-9290/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
`doi:10.1016/j.jbiomech.2003.12.022
`
`neously measured the heelpad response and the pressure
`distribution at the foot/floor interface. Miller-Young
`et al. (2002) characterised the fat pad, taken from
`cadavers of the elderly, in compression, to generate data
`for finite element analysis (FEA). However the magni-
`tudes (0.01–0.1 Pa) of
`the moduli reported, appear
`unrealistically small. Aerts and De Clercq (1993)
`performed pendulum impact tests on shod heels, and
`showed that the heelpad compression was smaller with a
`harder shoe midsole; they reasoned that the heelpad
`response was rate dependent, and that the shoe heel
`counter constrained the heelpad. Gefen et al. (2001)
`measured the heelpad thickness of two 30-year old
`subjects as 11 and 13 mm. They found that the non-
`linearly elastic heelpad had an initial compressive
`modulus of 105711 kPa.
`The only detailed FEA of a foot–shoe interaction
`(Lemmon et al., 1996) was a two-dimensional analysis of
`the forefoot region for walking. Shiang (1997) per-
`formed FEA of polyurethane foam midsoles in the heel
`region, but gave no material parameters. Rather than
`model the heelpad, he used in-shoe pressure data as
`vertical loads for the upper surface of the midsole; this
`ignores shear stresses at the interface. He predicted a
`greater (mean) vertical strain in the centre of the heel
`contact area than elsewhere in the foam.
`
`MacNeil Exhibit 2179
`Yita v. MacNeil IP, IPR2020-01139, Page 1
`
`

`

`1380
`
`R. Verdejo, N.J. Mills / Journal of Biomechanics 37 (2004) 1379–1386
`
`ARTICLE IN PRESS
`
`Most running shoe midsoles are made from the
`foamed copolymer of
`ethylene and vinyl acetate
`(EVA), of density in the range 150–250 kg/m3. Foot-
`strikes, repeated at approximately 1.5 Hz, may cause
`fatigue damage to the foam, hence may lead to the foam
`bottoming out, causing injuries. Prior research has used
`laboratory fatigue tests, which only provides indirect
`evidence of performance deterioration in running.
`Misevich and Cavenagh (1984) used repeated, rapid,
`uniaxial compression tests on EVA foam, to show that
`the midsole force-deflection response changed with cycle
`number. No details of the EVA foam densities were
`given. Cook et al. (1985) used a prosthetic foot, tilted
`back by 15, to load the heel of the shoe from 0 to 1.5 kN
`at 2.5 Hz. After the equivalent of 500 miles running, the
`shoes had 55710% of their initial energy absorption.
`Bartlett (1999) discussed the cell geometry in sectioned
`EVA midsoles, claiming that cells next to the outsole
`became flattened after 3200 km of running. Mills and
`Rodriguez-Perez (2001) studied diffusion in EVA foam
`under creep loading, concluding that the air content of
`the foam cells decreased, reducing the cushioning.
`One objective was to study the mechanical interaction
`of the heelpad with running shoe midsoles, and to
`estimate the magnitude of internal heelpad stresses.
`Another was to clarify the mechanism of shoe midsole
`degradation, and to investigate the resulting changes in
`the peak plantar pressures.
`The approach taken was to use FEA of the heelpad
`and shoe to predict the pressure distribution at the heel/
`shoe interface,
`then use the pressure distribution
`experiments on runners to validate the analysis, and to
`follow changes in shoe cushioning.
`
`2. Methods
`
`2.1. Plantar pressure distribution
`
`Three healthy male long distance runners (Table 1) ran
`at 2.61 m/s for 10 min on a Quinton Instrument Co. 640
`treadmill; this short experiment avoided fatigue, hence
`possible changes in foot loading (Edington et al., 1990).
`The treadmill provides a standard running speed and
`running surface. They were all rearfoot strikers, did not
`use orthotics, and reported no lower extremity injury for
`the past year. The University ethical committee ap-
`proved the study and informed consent was obtained
`from the runners. They wore Reebok Aztrek DMX
`shoes that were new at the start of the experiment.
`Their plantar pressure distribution was recorded using
`the Tekscan ‘F-Scan’ system—a flexible, 0.18 mm thick,
`plastic sole-shape having 960 pressure sensors with
`spatial resolution of 5 mm. The resistance of pressure-
`sensitive ink, contained between two polymer-film
`substrates, decreases as the pressure, applied normal to
`
`Table 1
`Runner characteristics
`
`Run distance
`(km)
`
`Age
`
`Weight (kg)
`
`Shoe size
`(UK)
`
`Runner 1
`Runner 2
`Runner 3
`
`725
`700
`500
`
`34
`37
`49
`
`91.472.0
`82.171.4
`62.972.3
`
`11
`9
`8
`
`Fig. 1. Calibration of five new F-scan sensors, ‘non-averaged’.
`
`the substrate, increases. Ahroni et al. (1998) and Mueller
`and Strube (1996) reviewed studies on F-scan sensors;
`some researchers reported good reliability and reprodu-
`cibility, while others reported a decrease of sensor
`output with time at fixed pressure. Woodburn and
`Helliwell (1996) concluded that they were not suitable
`for accurate, repeatable measurements. However, peak
`pressure measurements in this paper were similar to
`those reported previously, allowing for differences in
`running speeds (Gross and Bunch, 1989).
`To check the sensor linearity, an area of 0.0079 m2 in
`the forefoot region was sandwiched between two 6 mm
`layers of soft, closed-cell ‘Airex 5230’ PVC foam, then
`loaded in uniaxial compression between metal plates
`using an Instron machine. Constant pressures of 31, 36,
`39, 155, 225, 329, 398 kPa were applied for 20 s, while
`data was recorded for 2 s. The loading cycle was
`repeated five times, on five new and three used sensors.
`Values were taken from 15 different frames of each 300
`frames ‘movie’. The manufacturer suggested using the
`‘weighted averaging’
`function to reduce cell-to-cell
`variation. There
`is a linear
`relationship between
`pressures measured by the F-scan sensor and those
`applied by the Instron (Fig. 1). Table 2 gives the values
`of the intercept a; the slope b and the correlation
`coefficient R2: Calibrations using ‘weighted averaging’
`and ‘non-weighted averaging’ are not
`significantly
`different. However when sensors were recalibrated, at
`
`MacNeil Exhibit 2179
`Yita v. MacNeil IP, IPR2020-01139, Page 2
`
`

`

`ARTICLE IN PRESS
`
`R. Verdejo, N.J. Mills / Journal of Biomechanics 37 (2004) 1379–1386
`
`1381
`
`tion was simplified as a hemisphere of radius 15 mm,
`attached to the end of a 20 mm long vertical cylinder of
`radius 15 mm (Fig. 8a). The heelpad outer geometry was
`taken to be a vertical cylinder of radius 30 mm; the lower
`surface was spherical with a radius of curvature of
`40 mm, typical of a foot; a smooth blend was made
`between this surface and the vertical cylindrical surface.
`The minimum heelpad thickness was taken as 12 mm,
`the mean of the values given by Gefen et al. (2001). The
`lower ends of the tibia and fibula are approximated as
`an annular projection on the upper part of
`the
`calcaneus. When a EVA midsole foam was present, it
`was taken as vertical cylinder of radius 35 mm and
`height 22 mm, with initially flat upper and lower faces.
`Its lower surface (or that of the heelpad when no foam
`was present) rested on a flat rigid support table.
`The support
`table was fixed, while the upper
`calcaneus boundary was
`ramped down by 20 mm
`(12 mm when no foam was present). The heelpad is
`allowed to expand freely at the sides; it is assumed that a
`shoe heel counter and upper would have no confining
`effect. The heel pad is assumed to be bonded to the
`calcaneus surface, which allows load transfer by shear
`stresses at
`the interface. The coefficient of
`friction
`between the heelpad and the foam or support table
`was taken as 1.0. Meshing was chosen to maximise the
`computation stability.
`
`3.2. Material models
`
`Table 2
`Tekscan sensor calibration
`
`Sensor
`age
`
`New
`New
`Used
`
`Data treatment
`
`Intercept
`a (kPa)
`
`Slope b
`
`Correlation,
`r2
`
`Non-average
`Average
`Non-average
`
`1.876
`1.894
`2.591
`
`1.023
`1.028
`1.136
`
`0.998
`0.998
`0.996
`
`Average values for five new and three used sensors.
`
`the end of their use, the slope was 11% larger than that
`for new sensors.
`The subjects were asked to run on hard surfaces (track
`or roads), and to keep a running distance diary. Every
`15 days the plantar pressure distribution was measured,
`after a standardised 24 h recovery time since the last run.
`The insole was trimmed to fit the subjects’ right shoe.
`The sensors were calibrated every session by the known
`weight of the test subject, standing on one foot. Data
`were recorded, at the beginning (once they had acquired
`their gait) middle and end of the 10 min run, at 150 Hz
`for 4 s, which gave an average of 5 strikes per movie.
`The peak pressures from each frame were corrected to
`an ‘Instron’ pressure value using the new ‘non-averaged’
`calibration of Fig. 1.
`
`2.2. Shoe midsole foam characterisation
`
`The midsole of Reebok Aztrek DMX shoes was
`20 mm thick and contained two foams: a section of area
`20 mm by 40 mm on the lateral side of the heel was
`coloured grey, and the rest white. The densities were
`measured using a hydrostatic balance. Differential
`scanning calorimetry (DSC) was carried out using a
`Mettler DSC 30. The melting point and crystallinity
`were taken from the second heating run; this eliminates
`the effect of thermal ageing. The melting point was
`taken as the peak of the heat flow vs. temperature curve;
`the degree of crystallinity was calculated by dividing the
`melting peak area by the 286.8 J/g enthalpy of fusion for
`polyethylene crystals (Brandrup and Immergut, 1975).
`Cross-sections of the midsoles, from new and used
`trainers, were fractured after
`immersion in liquid
`nitrogen, then vacuum coated with gold. They were
`examined using a JEOL JSM 5410 scanning electron
`microscopy (SEM).
`
`3. FEA
`
`3.1. Geometry and boundary conditions
`
`ABAQUS Standard FEA version 6.3 (HKS) was used
`with the large deformation option. The calcaneus
`geometry was simplified to have a vertical axis of
`rotational symmetry. The geometry of its lower projec-
`
`Ideally, energy losses in the heelpad and shoe should
`be considered. Linear viscoelasticity can be incorporated
`in Explicit FEA, but the heelpad and the foam are non-
`linear viscoelastic materials. Initially a hyperelastic
`model was used, allowing the use of the more-stable
`Implicit FEA. The hyperfoam model in ABAQUS uses
`the Ogden strain energy function:
`2 þ laiðlai1 þ lai 3 3Þ þ 1
`bi
`
`
`
`
`
`
`
`Jaibi 1
`
`;
`
`ð1Þ
`
`2mi
`a2
`i¼1
`i
`where lI are the principal extension ratios, J ¼ l1l2l3 is
`a measure of the relative volume, the mi are shear
`moduli, N is an integer, and ai and bi non-integral
`exponents. The latter are related to Poisson’s ratio ni by
`ni
`1 2ni
`X
`The initial shear modulus is given by
`
`
`
`
`
`
`
`
`
`XN
`
`U ¼
`
`bi ¼
`
`:
`
`ð2Þ
`
`m ¼
`
`mi:
`
`i¼1;N
`
`ð3Þ
`
`One of the hyperelastic models in ABAQUS is the
`Ogden strain energy function:
`ðlai1 þ lai2 þ lai3 3Þ þ 1
`Di
`
`
`
`
`
`
`
`
`
`ð
`
`J 1
`
`Þ2i
`
`;
`
`ð4Þ
`
`
`
`2mi
`a2
`i
`
`XN
`
`i¼1
`
`U ¼
`
`MacNeil Exhibit 2179
`Yita v. MacNeil IP, IPR2020-01139, Page 3
`
`

`

`1382
`
`R. Verdejo, N.J. Mills / Journal of Biomechanics 37 (2004) 1379–1386
`
`ARTICLE IN PRESS
`
`-20
`
`-10
`tensile strain %
`
`0
`
`10
`
`Fig. 3. Runner 3 pressure distribution after various distances run.
`
`400
`
`200
`
`0
`
`-200
`
`tensile stresskPa
`
`-400
`
`-30
`
`Fig. 2. Impact compression, and tensile stress–strain response for
`loading and unloading EVA foam of density 170 kg m3, compared
`with hyperelastic model prediction (dashed curve).
`
`:
`
`ð5Þ
`
`where the bulk modulus K is given by
`K ¼ 1
`Di
`Lemmon et al. (1996) used a N ¼ 3 hyperfoam model
`to fit compressive data for an EVA foam (‘Cloud’ from
`Soletech); the initial shear modulus was 479 kPa, while
`their parameters predict that the Young’s modulus in
`tension is much higher than that
`in compression.
`Soletech says the Cloud’ density is between 150 and
`200 kg m3 (personal communication).
`The heelpad was simulated using the Ogden hyper-
`elastic material, with initial compressive modulus close
`to that found by Gefen et al. (2001), and bulk modulus
`that of water (2 GPa). The non-linearity parameters
`were found by matching the force-deflection response of
`the isolated heel of a 24-year old male (‘Foot II’ of Aerts
`et al. (1996)). The response of the EVA foam from the
`Reebok shoes was measured in both uniaxial compres-
`sion and in tension. Since the response alters with cycle
`number, for the first few cycles of deformation, data was
`taken after 10 cycles. Fig. 2 shows the combined
`response for tension and compression, on loading and
`unloading for the 10th cycle. The EVA foam response
`was modelled using the Ogden hyperfoam material, with
`N ¼ 2 and shear modulus m1=1000 kPa, a1 ¼ 10;
`Poisson’s ratio n1 ¼ 0; m2 ¼ 50 kPa, a2 ¼ 4; n2 ¼ 0:4:
`Fig. 2 shows that this provides a reasonable match to
`both the tensile and compressive response.
`
`4. Results
`
`4.1. Plantar pressure distribution
`
`Fig. 4. Peak pressure in runner 3’s heel region as a function of run
`distance.
`
`pressure is a maximum in the heel region. This local
`peak almost has the axial symmetry assumed in the
`FEA, although the centre of this peak is offset to the
`medial side of the foot for runner 3. The peak pressures
`are lower than some reported values (Henning and
`Milani, 1995, 2000), due to the running speed being low,
`and the surface being more compliant than a running
`track.
`The peak pressures in the heel region increased with
`running time (in the 10 min experiment), with a greater
`increase in the second and subsequent sessions, when the
`trainers had been used for several 100 km. There is an
`average 100% increase in peak pressure over the total
`run distance for all three runners (Fig. 4).
`
`4.2. Foam characterisation
`
`The Tekscan measurements of pressure distributions
`(Fig. 3) are for the time in the footstrike when the peak
`
`The density and DSC results (Table 3) are approxi-
`mately the same for both white and grey foam samples,
`suggesting that they only differ in colour. The degree of
`
`MacNeil Exhibit 2179
`Yita v. MacNeil IP, IPR2020-01139, Page 4
`
`

`

`ARTICLE IN PRESS
`
`R. Verdejo, N.J. Mills / Journal of Biomechanics 37 (2004) 1379–1386
`
`1383
`
`in the
`
`crystallinity suggests a 18% VA content
`copolymer (Dupont, 1997).
`The elastic moduli of EVA foams increases with their
`density. Larger shoe sizes may have higher foam
`densities, to give extra support to the assumed heavier
`wearers. Samples were cut from the arch region of each
`used running shoe, a region that suffers little damage.
`The size 11 shoe has increased foam density (Table 4),
`but sizes 8 and 9 the same density, allowing for density
`variation from the manufacturing process.
`
`Micrographs of the midsole foam from the new
`trainers (Fig. 5) contain flattened cells close to the lower
`(outsole) and upper (insole) surfaces; they result from
`moulding the midsole into its final shape. The cells are
`slightly elongated along the shoe length direction.
`Hence, Bartlett (1999) may have incorrectly interpreted
`flattened, near-surface cells as evidence of foam fatigue.
`All the trainers at the end of the experiment (Fig. 6)
`contain some cell faces that are wrinkled. More severe
`damage, such as cell-face fractures, is present in the
`trainers of runners 1 and 2. Fractures in cut cell faces
`may be due to sample preparation; consequently only
`fractures in complete cell faces are considered to be due
`to trainer use.
`
`The best fit to Aerts et al. (1996) data for a heel pad
`impacted by a flat rigid surface was obtained (Fig. 7)
`using shear moduli m1 ¼ m2 ¼ 50 kPa, with exponents
`a1 ¼ 30 and a2 ¼ 4: Although the material model
`cannot simulate hysteresis on unloading, it predicts the
`shape of the loading curve.
`Figs. 8b and c show the vertical compressive stress s22
`contours, respectively, for the bare heel on flat anvil and
`the heel on shoe, at total deformations of 4 and 10 mm,
`respectively, when the force is close to 0.5 kN. The total
`force on the foot will be higher, since other parts of the
`foot also transit force to the ground (Fig. 3). The main
`
`Crystallinity
`(%)
`
`4.3. FEA results
`
`Table 3
`Characterisation of EVA foams in Reebok midsole
`
`Sample Density
`(kg/m3)
`
`Melting
`point (C)
`
`White
`Grey
`
`170
`173
`
`82.2
`82.0
`
`Melting
`enthalpy
`(J/g)
`
`56.7
`53.6
`
`19.8
`18.7
`
`Table 4
`EVA foam density in used trainers
`
`Runner
`
`Runner 1
`Runner 2
`Runner 3
`Unused
`
`Shoe size
`
`Density (kg/m3)
`
`11
`9
`8
`8
`
`19073
`16875
`16474
`17073
`
`Fig. 5. SEM micrographs of EVA foam from new trainers, with the midsole thickness direction horizontal, and the shoe length direction vertical: (a)
`near the lower surface, with thick cell faces on the right (next to outsole), and (b) centre of midsole.
`
`Fig. 6. The heel region of trainers after (a) 500 km and (b) 750 km, showing some wrinkled cell faces. In (b) there are holes in some internal cell faces.
`
`MacNeil Exhibit 2179
`Yita v. MacNeil IP, IPR2020-01139, Page 5
`
`

`

`1384
`
`R. Verdejo, N.J. Mills / Journal of Biomechanics 37 (2004) 1379–1386
`
`ARTICLE IN PRESS
`
`peak force during the footstrike of runner 3 was
`typically 1.0 kN, with an initial peak of 0.8 kN. These
`compares with values of 2.3 and 1.7 kN, respectively, for
`running at 3.6 m/s on a rigid surface (Gross and Bunch,
`1989). Their peak heel pressures increased from 300 to
`420 kPa, when the running speed increased from 3.0 to
`4.5 m/s.
`the predicted
`force close to 0.5 kN,
`For a heel
`maximum compressive stress on the heelpad/foam
`interface
`is approximately 0.7 MPa, while
`in the
`bare heel test (Fig. 8b), it is 2.0 MPa. The maximum
`stress at the bone–heelpad interface is 1.0 MPa for the
`shod foot.
`In the foot/shoe simulation, for forces less than 200 N,
`the majority of the deformation occurs by the flattening
`of the lower surface of the heelpad, and the increase of
`the contact area with the foam. However, at higher
`
`forces, the deformed heelpad does not decrease much in
`thickness, while the midsole upper surface becomes
`increasingly concave. Although the heelpad has spread
`laterally, the side of the EVA foam hardly bulges. The
`maximum foam stress occurs at the centre of the contact
`area on the foam upper surface (Fig. 8c). This confirms
`the Tekscan data for the pressure map on the upper
`surface of the shoe midsole—with a 300 kPa stress area
`of diameter approximately 20 mm.
`The energy of the footstrike was calculated as the
`integral of the force vs. deflection graph, for the heel
`plus midsole simulation. Fig. 9 shows the force is a
`nearly linear function of the footstrike energy; the result
`of the force increasing nearly exponentially with the
`deflection, but not as rapidly as for the bare heel in
`Fig. 7. The maximum vertical compressive stress in the
`foam, which occurs at the centre of the heel/foam
`
`force
`no shoe
`
`stress
`
`force
`with shoe
`
`2
`
`4
`
`6
`
`8
`
`10
`
`energy J
`
`2.5
`
`2
`
`1.5
`
`1
`
`0.5
`
`peak stress(MPa) or force (kN)
`
`0
`
`0
`
`Fig. 7. Force vs. displacement, for impacts on an isolated human heel,
`redrawn from Aerts et al. (1996). FEA prediction for heel model
`(dashed curve).
`
`Fig. 9. Predicted force (and peak compressive stress on heelpad
`surface) vs. energy input for heelpad on an EVA midsole, compared
`with force vs. energy graph without a shoe.
`
`Fig. 8. (a) Undeformed mesh. Contours of vertical compressive stress (kPa) for heel on flat surface; (b) no shoe; 424 N force, 4.0 mm deflection; and
`(c) EVA foam shoe, 502 N force, 10.2 mm deflection.
`
`MacNeil Exhibit 2179
`Yita v. MacNeil IP, IPR2020-01139, Page 6
`
`

`

`ARTICLE IN PRESS
`
`R. Verdejo, N.J. Mills / Journal of Biomechanics 37 (2004) 1379–1386
`
`1385
`
`interface, is a roughly linear function of the footstrike
`energy. Hence a footstrike of a given kinetic energy will
`produce a much lower peak force for the shod foot
`rather than the bare heel.
`
`5. Discussion
`
`Both the heelpad and the running shoe EVA foam act
`as shock-absorbing non-linear spring-dashpot struc-
`tures, reducing the peak force in a heelstrike. There is
`a synergy in their responses; the foam, by indenting on
`its upper surface, increases the load spreading to the
`plantar surface, which reduces the force on the heel area.
`The indentation also probably stabilises the angular
`position of the calcaneus, affecting foot pronation. The
`properties of EVA foam can be measured more easily
`than those of the human heelpad; the modelling of their
`interaction is also a method of comparing their relative
`mechanical properties.
`FEA has successfully analysed the non-linear, large
`deformation, problem of heel/shoe interaction. The
`predicted lack of bulging of the shoe foam sides is
`consistent with experimental observations; however
`such bulging is predicted for softer, low-density EVA
`foams. The EVA foam is more compliant than the
`heelpad, since it has a much lower bulk modulus, in spite
`of its shear modulus being higher. The initial foam shear
`modulus used here is double that used by Lemmon et al.
`(1996), for an EVA foam density inside the range
`estimated for their Soletech foam. Their Ogden strain
`energy function exponent a2 ¼ 3:9 is smaller than the
`a1 ¼ 10 used here, but both values provide significant
`hardening in compression. The heelpad initial shear
`modulus used here is 100 kPa,
`the same order of
`magnitude as their value used for forefoot soft tissue,
`and equal to the value given by Gefen et al. (2001).
`Hence the data of Miller-Young et al. (2002) must be in
`error by orders of magnitude. The uncertainty in the
`moduli is probably an order of two, since the exact
`dimensions of the heel tested by Aerts et al. (1996) is
`unknown, and since viscoelasticity was ignored in the
`modelling. It appears that heelpad is highly hyperelastic,
`with a Ogden strain energy function exponent close to
`30. This is higher than the value of 17 used to model the
`thigh tissue (Setybudhy et al., 1997). The predicted
`pressure distribution at the skin/midsole interface is
`confirmed qualitatively by the F-scan data. In an ideal
`experiment the interface pressure, and the deformation
`of the heelpad and foam, would be simultaneously
`measured for a runner; however this is impossible at
`present.
`FEA predicts a significantly higher peak heelpad
`pressure in a bare heel strike, compared with a shod heel
`strike with the same force; however barefoot and shod
`runners run differently. The pressure distribution on the
`
`midsole upper surface is non-uniform, with peak values
`near the centre of the heel contact area. The impacts
`cause fatigue damage to the EVA foam. Although air
`compression provides a major shock cushioning me-
`chanism in the foam, other experiments (Verdejo and
`Mills, 2004) suggest that air loss is negligible in running.
`Instead, weakening of the EVA structure causes soft-
`ening of the foam. The wrinkling of some cell faces is
`evidence of the foam fatigue. The consequent increase in
`peak plantar pressures is a real effect, since it is an order
`of magnitude larger than the sensitivity change of
`Tekscan F-scan sensors.
`One limitation of the plantar pressure measurements
`is that they were made on a treadmill. The biomechanics
`of treadmill running differ from those of overground
`running (Nigg et al., 1995), and the peak footstrike force
`will be lower on the more-compliant treadmill surface.
`This means that the peak plantar pressures will be
`higher for running on a road than those measured for
`running on a treadmill.
`Modelling of EVA foam midsoles, with a region
`weakened by the heel-strike stress field, has not yet been
`attempted; further FEA will consider a weakened region
`in upper centre of
`the midsole. Recent
`research
`(Taunton et al., 2003) suggests that running shoe age
`contributes to running injuries. The deterioration of
`running shoe cushioning may be an important explana-
`tory factor for such an effect.
`
`Acknowledgements
`
`R. Verdejo would like to thank EPSRC for support in
`the form of a Ph.D. studentship. We would like to thank
`the runners who took part in the experiments. We would
`like to thank one referee for suggesting a better source
`for the heelpad mechanical response.
`
`References
`
`Aerts, P., Ker, R.F., et al., 1995. The mechanical properties of the
`human heel pad. Journal of Biomechanics 28, 1299–1304.
`Aerts, P., Ker, R.F., De Clercq, D., et al., 1996. The effects of isolation
`on the mechanics of the human heel pad. Journal of Anatomy 188,
`417–423.
`Aerts, P., De Clercq, D., 1993. Deformation characteristics of the heel
`region of the shod foot during a simulated heelstrike. Journal of
`Sports Science 11, 449–461.
`Ahroni, J.H., Boyco, E.J., Forsberg, R., 1998. Reliability of F-scan in-
`shoe measurements of plantar pressure. Foot & Ankle Interna-
`tional 19, 668–673.
`Bartlett, R., 1999. Sports Biomechanics: reducing injury and improv-
`ing performance, E & FN Spon, London.
`Brandrup, J., Immergut, E.H., 1975. Polymer Handbook 2nd Edition.
`Wiley Interscience, New York.
`Cavanagh, P.R., Valiant, G.A., Misevich, K.W., 1984. Biological
`aspects of modelling shoe/foot interaction during running. In:
`
`MacNeil Exhibit 2179
`Yita v. MacNeil IP, IPR2020-01139, Page 7
`
`

`

`1386
`
`R. Verdejo, N.J. Mills / Journal of Biomechanics 37 (2004) 1379–1386
`
`ARTICLE IN PRESS
`
`Frederick, E.C. (Ed.), Sports Shoes and Playing Surfaces. Human
`Kinetics, Champaign, IL.
`Chen, H., Nigg, B.M., de Koning, J., 1994. Relationship between
`plantar pressure distribution under the foot and the insole comfort.
`Clinical Biomechanics 9, 335–341.
`Cook, S.D., Kester, M.A., Brunet, M.E., 1985. Shock absorption
`characteristics of running shoes. American Journal of Sports
`Medicine 13, 248–253.
`Dupont,
`1997.
`www.dupont.com/industrial-polymers/footwear/
`index.html. Thermal properties of Elvax measured by differential
`scanning calorimetry.
`Edington, C.J., Frederick, E.C., Cavanagh, P.R., 1990. Rearfoot
`motion in distance running. In: Cavanagh, P.R. (Ed.), Biomecha-
`nics of Distance Running. Human Kinetics, Champaign, IL,
`pp. 135–162.
`Gefen, A., Megido-Ravid, M., Itzchak, Y., 2001. In vivo behavior of
`the human heel pad during the stance phase of gait. Journal of
`Biomechanics 34, 1661–1665.
`Gross, T.S., Bunch, R.P., 1989. Discrete normal plantar stress variations
`with running speed. Journal of Biomechanics 22, 699–703.
`Henning, E.M., Milani, T.L., 1995. In-shoe pressure distribution for
`running in various
`types of
`footwear. Journal of Applied
`Biomechanics 11, 299–310.
`Henning, E.M., Milani, T.L., 2000. Pressure distribution measure-
`ments for evaluation of running shoe properties. Sportverletzung/
`Sportschaden 14, 90–97.
`Lemmon, D., Shiang, T.Y., Hashmi, A., et al., 1996. The effect of
`insoles in therapeutic footwear a finite element approach. Journal
`of Biomechanics 30, 615–620.
`Miller-Young, J.E., Duncan, N.A., Baroud, G., 2002. Material
`properties of
`the human calcaneal
`fat pad in compression:
`experiment and theory. Journal of Biomechanics 35, 1523–1531.
`
`Mills, N.J., Rodriguez-Perez, M.A., 2001. Modelling the gas-loss creep
`mechanism in EVA foam from running shoes. Cellular Polymers
`20, 79–100.
`Misevich, K.W., Cavenagh, K.W., 1984. Material aspects of modelling
`shoe/foot interaction. In: Frederick, E.C. (Ed.), Sports Shoes
`and Playing
`surfaces. Human Kinetics, Champaign,
`IL,
`pp. 47–73.
`Mueller, M.J., Strube, M.J., 1996. Generalizability of in-shoe peak
`pressure measures using F-scan system. Clinical Biomechanics 11,
`159–164.
`Nigg, B.M., De Boer, R.W., Fisher, V., 1995. A kinematic comparison
`of overground and treadmill running. Medicine and Scienc in Sport
`and Exercise 27, 99–105.
`Setybudhy, R.H., Ali, A., Hubbard, R.P., et al., 1997. Measuring and
`modelling of soft tissue and seat interaction. In: Progress with
`Human Factors in Automotive Design, Vol. SP-1242. Soc. Auto.
`Eng., Warrendale, PA, USA.
`Shorten, M.R., 2000. Running shoe design: protection and perfor-
`mance. In: Pedoe, T. (Ed.), Marathon Medicine. Royal Society of
`Medicine, London, pp. 159–169.
`Shiang, T.-Y., 1997. The non-linear finite element analysis and plantar
`pressure measurements for various shoe soles in heel region.
`Proceedings of the National Science Council, ROC, Part B: Life
`Sciences 21, 168–174.
`Taunton, J.E., Ryan, M.B., Clement, D.B., et al., 2003. A prospective
`study of running injuries. British Journal of Sports Medicine 37,
`239–244.
`Woodburn, J., Helliwell, P.S., 1996. Observations on F-scan in-shoe
`pressure measuring system. Clinical Biomechanics 11, 301–304.
`Verdejo, R., Mills, N.J., 2004. Simulating the effects of long distance
`running on shoe midsole foam. Polymer Testing, accepted for
`publication.
`
`MacNeil Exhibit 2179
`Yita v. MacNeil IP, IPR2020-01139, Page 8
`
`