Sunteți pe pagina 1din 15

Vinayak M Rao

20036966
Contents

Introduction
Validation methods for Segment models
Experimental methods and measurement
techniques
Equations used to estimate the moment
Sensitivity analysis
Results and discussion
Introduction and need for the current study
3-D Biomechanical models have been developed to estimate
stresses from Manual material handling tasks which are
recognized as a source of low back problems.
Existing biomechanical macromodels by Delleman, et al
consist of a segment model used to calculate net reaction
forces and moments at joints of interest and a distribution
model which distributes these forces and moments in the
surrounding tissues, ligaments and muscles.
These segment models suffer from inherent problems such
as errors in calculating the center of rotation of the joints,
skin movement, problems in estimation of segment mass,
center of mass and moment of inertia and external forces
and torques.
This necessitates the validation of segment models before
using a distribution model.
Validation methods for Segment models
1. Comparison between different models- De Looze et al
compared the L5 /S1 joint moment between two models, an
upper body model starting from hands and proceeding to the
trunk and a lower body model starting from the feet and
proceeding towards the trunk. The L5 /S1 joint moments
calculated gave a correlation of 0.99.
2. Comparison of the reaction forces of the model with those
predicted by a force plate- De Looze et al reported a
correlation between predicted and measured forces of 0.88
and 0.32 for the vertical and horizontal forces respectively.
3. Sensitivity analysis- Gagnon assessed the sensitivity of the
L5 /S1 joint moment by inducing variations of 3-10mm for joint
center positions and 3-22N for the external forces. Maximum
senstivity values varied between 3-12Nm.
Experimental methods and measurement
techniques
The current study examines the validity and sensitivity of two 3D segment
models, an upper body and lower body model to estimate the moments at
the L5 /S1 joint.
Three male subjects with a mean age of 28 years, mean mass of 69 kg and
mean height of 1.74 m performed the task of lifting a 9.6 kg load from a low
shelf, tuning 180 and placing it on a high shelf. Two speeds were used.
A peak motion measurement system was used to collect the 3-D positions
of 50 markers placed on the joint centres, the location of which was
provided by anthropometric data by Webb. The external forces and torques
on the foot were measured with AMTI force platforms, while those on the
hand were measured with load cells.
The lower body model included the feet, shanks, thighs and pelvis and
upper body model included the hands, upper arms, arms, head and trunk.
The position of the markers were filtered using a Buterworth filter.
Linear velocities and accelerations were calculated by hand using central
difference method and angular velocities were calculated by the product of
the direction cosines.
Fig.1 Illustration of the lifting task
Equations used to estimate the moment

The equations provided by Hof were used to estimate the moment with respect to a
moving point, P

[( ) ] + = + [( ) ] +

(Eqn 1)

, , are the positions of the external forces, moving point and the center of

mass of the segment respectively.
and are the external forces and torques.

is the segment mass, is the acceleration of the C.O.M, and are the
moment of inertia and angular velocity.
The equation was used to verify that the sum of external moments about the CG
was equal to the sum of inertial moments.
The net reaction moment at the L5 /S1 joint is derived from Equation 1 as follows,


L5/S1 = -=1[( L5/S1 ) ]
=1[ - =1[ L5 /S1 ]

+ =1[( L5 /S1 ) ] + =1 (Eqn 2)

Sensitivity analysis
Sensitivity analysis was performed to estimate the errors in
calculations of reaction moments. Gaussian noise was
added to each parameter of eqn(2) and the reaction
moments at L5 /S1 were recalculated.
The mass and center of gravity were calculated using
data provided by Zatsiorsky instead of Winter and results
from the data with and without noise were compared for
the L5 /S1 joint reaction moments.
The magnitude of noise added based on error analysis was
9-26mm for joint center, 1-4% for ground reaction force, 7-
19mm for center of pressure location, 4-10N for hand
transducers force, and 7-25mm for hand transducer
position.
Results
The correlation for upper and lower body models for joint reaction
forces and moments were above 0.95, with an exception of 0.76 for
longitudinal axis for one subject
The correlation between external forces and inertial forces were
found to be above 0.75 and the correlation between external and
inertial moments was above 0.57.
In cases where hand transducers were not available, acceleration
components of the load are used to substitute for the transducers
and the differences between the moments calculation at the L5 /S1
joint between the two methods varied between 10-53 Nm. The
differences were larger at the beginning and end of the task due to
peak accelerations being produced along with difficulty in
determining the exact time of the actions.
The sensitivity analysis showed that the rms data for the lower and
upper body models with and without noise varied from 4-6Nm and
5-12Nm respectively.
Fig.2 Percentage contribution of each term in Equation 2
Fig.3 External and inertial moment and force curves at
normal speed
Discussion
The correlation between the two body models were greater
than 0.97 for the forces and greater than 0.76 for the
moments. The RMS differences for the reaction forces
were below 15N and increased with high speeds, indicating
the errors were dependent on accelerations.
Small moment values correspond to smaller correlations
and therefore care must be taken while measuring
moments about longitudinal and sagittal axes where the
moments are less than 10Nm
The advantage of using equations 1 and 2 for validating
the models is that they include all the body segments and
measure forces in the global coordinate system which are
more accurate.
Equation 2 is used to compare the effects of a segmental weights
analysis(term 2), quasi static analysis(terms 1 and 2), fully dynamic
analysis(all four terms) on the reaction moment on the L5 /S1 joint.
For the lower body model, the major contributors were external forces
and torques with 95% of the moment being generated due to the quasi
static terms.
The inertial forces contribution was small as the task did not involve large
accelerations from the lower back.
For the upper body model, the external forces and segmental weights
were the major contributors to the L5 /S1 reaction moment. Only a static
or quasi static approach may not be recommended as the contribution of
the terms are not constant and depend on motion.
Therefore, the most important elements in estimation of the moments
are the estimation of L5 /S1 joint center positon and center of gravity.
Inertial moments were also important but the inertial acceleration
contribution was negligible in both models.
Conclusions
The trunk is an important factor in a lifting model but its
anthropometric parameters lead to errors as it is not a rigid
body and the L5 /S1 center of rotation is not fixed. Changing
the CG of the trunk by 0.05m leads to an error of 6Nm
about the transverse axis.
The use of a lower body model has many advantages, the
lower body model is less sensitive to anthropometric errors
It is also less sensitive to errors in the estimation of inertial
forces as the task did not involve large accelerations from
the lower part of the body.
Thus, it was concluded that a lower body model is to be
used if the position of the force plate can be estimated in
the global coordinate system.

S-ar putea să vă placă și