Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 72 (2014) 405 – 410

The 2014 conference of the International Sports Engineering Association

A method for sports shoe machinery endurance testing:

Modification of ISO 22675 prosthetic foot test machine for heel-to-
toe running movement.
Felix Starker*, Florian Blab, Florian Dennerlein, Urs Schneider
Fraunhofer IPA, Biomechatronic Systems, Nobelstr.12, 70569 Stuttgart, Germany

Abstract

This paper describes a method for testing in running shoes based on a test method for endurance testing of prosthetic feet (ISO
22675:2006).
A specific step from human running motion was selected and processed to be feed into the test machine software. A motion
capture system tracks the motion of the shoe and the machine. Forces and moments are acquired with a 6-DoF load cell.
The test shows the capability of this approach to reproduce dynamics of heel-to-toe running motion. For this specific test
scenario 60% of real time running speed was achieved while forces and moments were reproducible.

© 2014
© 2014 The
Elsevier Ltd. Published
Authors. This is an by
open accessLtd.
Elsevier article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Selection and peer-review under responsibility of the Centre for Sports Engineering Research, Sheffield Hallam University.
Selection and peer-review under responsibility of the Centre for Sports Engineering Research, Sheffield Hallam University
"Keywords: Shoe test; ISO 22675; Heel-to-toe running; test machine; prosthetic foot"

1. Introduction

1.1. Background of work

There are diverse methods for sport shoe testing. By Odenwald (2006), the methods can be classified in
subjective, biomechanical and mechanical test methods. To use the advantages of the different methods, a holistic
approach in testing of sports equipment is suggested and methods of all three categories should be used in the

* Corresponding author. Tel.: +49 711 970 3644;

E-mail address: felix.starker@ipa.fraunhofer.de

1877-7058 © 2014 Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Selection and peer-review under responsibility of the Centre for Sports Engineering Research, Sheffield Hallam University
doi:10.1016/j.proeng.2014.06.072
406 Felix Starker et al. / Procedia Engineering 72 (2014) 405 – 410

development process. Beside combinations of different test procedures, improvements can be achieved by
implementing test characteristics and advantages from different methods.
To improve mechanical test setups by focusing on biomechanical input data for movement control or a more
realistic loading of the shoe using special designed artificial test foot devices.
The aim of this study was to describe a machine setup that uses a standard test rig for prosthetic feet endurance
tests that was adapted with biomechanical human running data to simulate a heelstrike running movement.

1.2. Test machines and test methods

There are various methods and machines to test shoes, some of these are easy available some are custom build
according to requirements of customers. First test machines in shoe manufacturing were developed in the early
1930s. Miller (1938) mentioned the so-called Wheel of Torture by the U.S. National Bureau of Standards that
performed one of the first human-like wear test for shoes. This kind of testing was developed over the years and is
still in use for long-duration tests and quality tests in sports shoe industry today. Besides these machines, test rigs
were developed which mainly consist of a vertical movement axis. Standard test methods for impact testing (e.g.
ASTM F1976-06, DIN EN 23743) are reduced to this vertical loading and therefore only use simple stamps with a
defined size. Denoth et al. (1985) demonstrated that results from simple material measurements (DIN 55305) have
to be assessed critically if conclusions should be done to human movement. Heidenfelder (2010) described an
improved dynamic test method using a modified heel stamp in a vertical compression test.
The design of prosthetic feet requires robust, biomimetic designs that interact harmonically with the user. User
safety and to assure durability of these devices is of high interest as failure may cause injuries. Therefore a
prosthetics testing ISO group introduced ISO 22675 in 2006 that describes a test standard to perform dynamic heel
to toe walking for prosthetic feet. The related test machines have the following features in common: a rotating
table that follows the shank angle curve and a second linear axis that pushes the prosthetic foot on the table to
apply dynamic load during stance phase. The setup and dimensions of all components in the machine is described
in the standard to allow for intended anterior and posterior forces as well as moments.

2. Material & Method

2.1. Running data

Running data were captured in a motion lab. A subject (m, 27 years, 72 kg) was instructed to run on a straight
line passing two force plates (AMTI AccuGait, AMTI, Inc., Warren/MA, USA) at self-selected running speed with
its own used shoes. The motion was captured via an optical tracking system (Qualisys AB, Goteborg, Sweden)
with 9 cameras (Oqus 3+, Oqus 4). A Helen Hayes marker setup was applied on the subject with additional
markers focusing on the rear, middle and front section of the running shoe (ASICS Gel TN771, size UK 9). Four
rigid bodies were defined to analyze rotation angles in roll, pitch and yaw (Rearfoot, Midfoot, Forefoot and
Shank).

2.2. Data processing

Synchronized motion and force data were recorded in Qualisys Track Manager (V 2.8) and exported as matrix
file for further processing in MATLAB (R2013a, MathWorks, Inc., Natick/MA, USA). One single step with a
stance time of 0.243 s (cp. Fig. 1) was selected following the criteria of the subject to hit the force plate right in
center position, avoiding measurement errors as well. Shank in sagittal plane and vertical ground reaction force
(F z ) was exported to control the test machine (data rate 1000 Hz).
 Felix Starker et al. / Procedia Engineering 72 (2014) 405 – 410 407

(a) (b)

Fig. 1. Ground reaction forces (a) and moments (b) captured from human subject for data transmission

2.3. ISO 22675 test machine and modification

For the test, a hydraulically driven machine (Shore Western KS2-07) that tests prosthetic feet according to ISO
22675:2006 was utilized. The machine consists of two individually controllable pistons. First piston (LP) presses
the foot downward on a tilt table that is actuated by a second piston (RP) to allow for rotation in sagittal plane. The
LP has a fixed uniaxial load cell with acceleration compensation to control forces on the test specimen. The RP has
an angular sensor to control the second axis for synchronized motion. The prosthesis is attached to a prosthetic
pylon that is displaced posteriorly to the LP. A spring loaded cam joint is located above to provide compensation
of unwanted shear forces. The control system (RTAC) provides individually increased load on the prosthesis after
each step to find the specified load after a startup phase. As the standard test machine setup restricts higher
compensation of anterior- posterior forces due to the cam mechanism the setup was modified to allow for more
degrees of freedom by help of a ball joint. Hence the setup is able to articulate around a ball that is reset in swing
phase with rubber bands. To achieve damping behavior elastic foam was inserted. Furthermore as the setup of ISO
22675 was created for matching feet sizes, the increased size of the shoe covering the prosthetic foot has to be
compensated by a posterior shift of the shank setup. At a specified position (500 mm height, 65mm anterior) a 6-
DoF load cell was inserted working individually to the machine control algorithms.
The closed loop control of the machine uses the desired force curve as input parameter and modifies
accordingly the dynamic stroke curve of LP. Therefore deformation characteristics due to different shoe designs as
well as fatigue or wear of material can be detected as the LP curves alters during the cyclic test. As a result
dynamic stiffness (force vs. stroke in N/mm), wear (LP curve from first cycle vs. last cycle) as well as material
degeneration (characteristic points in the stroke or stiffness curve) through all load cycles can be observed. As a
second benefit of the closed loop control every test specimen (shoe) is tested according to the same dynamic
loading conditions. Therefore after assuring the same alignment and setup different samples can be tested and
compared. Furthermore load curves can be modified according to specific load scenarios (walking, running, etc.).

2.4. Machine Test setup

Seven motion capture cameras are placed around the prosthetic test machine. The running shoe is put on the
prosthetic foot (Soleus, size 26 cm, College Park Inc., Warren/MI, USA) with the same marker setup. The
prosthetic connector is used to align the foot with the shoe to be in an upright position (90° +/- 0.2° to floor) in the
unloaded state. Afterwards the specimen is mounted on the machine. The spring setup aligns the foot setup to hang
in straight downward oriented position. The overall length of the assembly is adjusted by help of a tube clamp
adaptor (Model 2W062, Wagner Polymertechnik, Silkerode, Germany) for an overall length of 700 mm.
408 Felix Starker et al. / Procedia Engineering 72 (2014) 405 – 410

(a) (b)

Fig. 2. a) Schematic test setup of modified ISO 22675 foot tester. b) Machine setup with reflective markers applied to the test specimen

3. Results

The input data was successfully transmitted to the machine and interpreted by the controls. The test specimen
performed the desired heel-to-toe motion at the desired force values (F z ). Creating efficient motion and decrease
unnecessary travel of the hydraulic pistons the shank angle was manually smoothed for swing phase.

(a) (b) (c) (d)

Fig. 3. Stance phases in machine test from a) initial contact to d) toe-off

3.1. Running Speed

Tests were performed with 20, 40, 50, 60 and 80% of real time speed. According to the test results all control
were in a stable range from 20% until 60% (stance time 1.215 - 0.405 s) of real time running speed of the subject.
At higher frequencies > 80% (stance time 0.304 s) the control algorithms were not fast enough to adapt for an
accurate force curve. Furthermore dynamics detected by the acceleration sensor on the LP were high enough to
engage the safety mechanisms of the machine.

3.2. Forces and Moments

 Felix Starker et al. / Procedia Engineering 72 (2014) 405 – 410 409

Forces and moments are acquired of the 6-DoF load cell. The machine is controlled by a single component load
cell that has acceleration compensation. Therefore force data of the machine is especially at higher speeds less
noisy compared to the 6-DoF load cell due to inertia effects. In Fig. 4, forces and moments at slow velocity test
(20%) are compared to high velocity test (80%). There are differences in the 80%-test between human input data
and machine force data. With reduced movement velocities, the accuracy increases tremendously with nearly
identical force output data in 20%-tests (correlation coefficient of r = 0.99). The accuracy of the loop control trying
to follow the preferred kinetics and kinematics is reduced dramatically after 60% of real time motion (r = 0.95 in
80%-test).

(a) (b)

(c) (d)

Fig. 4. Force data in human versus machine tests compared for slow test with 20% (a, c) and fast tests with 80% of movement velocity (b, d)

Motion

Simulation of human movement by the machine is high precisely. There is no difference in shank movement
simulation performed by RP during machine tests between the different movement velocities (cp. Fig. 4c-d). With
regards to the shoe sole angle to the ground, machine data differs especially at initial contact due to a dorsal flexion
movement of the prosthetic foot in this phase. The difference decreases in mid stance and late stance phase. There
seemed to be only little influence on this effect caused by higher velocities (cp. Fig. 5). This soft heel characteristic
is caused by the foot construction and could be influenced by a specific designed artificial foot for machinery shoe
test setups.
410 Felix Starker et al. / Procedia Engineering 72 (2014) 405 – 410

(a) (b)

Fig. 5. Movement angle of rearfoot rigid body in sagittal plane (pitch angle). Human running data versus machine data in a) slow (20%) and
b) fast machine velocities (80%). The machine angle data are referenced to the RP

4. Conclusion

The overall setup and data of the presented test approach is reproducible and time efficient. Shear forces were
found to follow not accurately but are related mostly to the used foot model. The test method is capable to perform
heel-to-toe running at reduced speed (60% of real-time running speed). ISO 22675 is designed for endurance test
and hence machines are suitable for endurance testing also for shoes. During analysis and post processing of the
resulting data significant difference was found in force data that was used for input data. The test machine used the
Fz force of the force plate. Comparing both coordinate systems reveals that a force measured at the shank (LP)
would have to be a converted and transformed force vector of the force plate. Therefore we suggest for more
realistic tests to use transformed forces as input data. Although maximum speed achieved within this study was
about 60% of real time speed there are ways for optimization. For example the current control frequency (1 GHz)
of the DAQ card that is down sampled by the operating system could be increase up to 2 GHz. Furthermore LP
motions, speeds and accelerations were already at the machines maximum (660 mm/s). As a potential solution a
curved surface on the tilt table might help to decrease these high piston travels when the prosthesis rolling over the
plate the height would be compensated by the convex surface of the table. For this test we used a durable, state of
the prosthetic foot model (College Park Soleus) but see high potential in more advanced foot models to provide
more biomimetic gait and load distribution in the shoe. Pitch angle differences at heel and forefoot show that there
is a difference between human foot and the prosthetic model. Regarding late stance, passive prosthetic feet will not
be able to create toe flexion as well as a net positive push off for distributing realistic loads on the foot.

References

American Society for Testing and Material, 2006. Standard Test Method for Impact Attenuation Properties of Athletic Shoes Using an Impact
Test. F1976-06, Beuth-Verlag, Berlin.
Denoth, J. , Kälin, X., Stacoff, A., 1985. Schockabsorption beim Jogging - Materialtests kontra Versuchspersonentests, Deutsche Zeitschrift für
Sportmedizin, 196–202.
Deutsches Institut für Normung e.V., 1999. Schuhe. Prüfverfahren für Laufsohlen. Kompressionsenergie. Deutsche Fassung EN 12743: 1999.
Beuth-Verlag, Berlin.
Heidenfelder, J., 2010. Entwicklung eines dynamischen Tests zur Prüfung der Rückfußdämpfung von Laufschuhen mittels biomechanischer
Messmethoden. Dissertation, Chemnitz.
International Organization of Standard, 2011. Prosthetics - Testing of ankle-foot devices and foot units. Requirements and test methods
11.040.40, ISO 22675:2006. Beuth-Verlag, Berlin.
Miller, J., 1938,.Torture test tell the truth, Mechanix Illustrated, 54–57.
Odenwald, S., Test methods in development of Sport Equipment, in: E.F. Moritz, S. Haake, The engineering of sport 6, Springer, New York,
2006, pp. 301–306.