Technology and Health Care 13 (2005) 5766 57

IOS Press
Simulation of manual materials handling:
Biomechanial assessment under different
lifting conditions

Omer G undogdu
a,
, Kurt S. Anderson
b
and Mohamad Parnianpour
c
Atat urk University, Department of Mechanical Engineering, 25240 Erzurum, Turkey
E-mail: [email protected]
b
Rensselaer Polytechnic Institute, Mechanical, Aerospace, & Nuclear Engineering, Troy, NY 12180,
USA
E-mail: [email protected]
c
Sharif University of Technology, Department of Mechanical Engineering, Tehran, Iran
E-mail: [email protected]
Received 25 May 2004
Accepted 20 September 2004
Abstract. Manual material handling (MMH) tasks were evaluated and compared under different lifting conditions. For the
theoretical evaluations, a two-dimensional sagittally symmetric human-body model was established to compute the moment
and joint load time histories for MMH tasks for a variety of different lift specications and constraints such as lifting durations,
loads, and modes. Nonlinear control techniques and genetic algorithms were utilized in the optimizations to explore optimal
lifting patterns. Since the kinetic measures such as joint moments are vital metrics in the assessment of the likelihood of injury,
the simulation results obtained may be compared using these metrics for each lift type, so that the superiority of a lifting method
or protocol relative to another may be determined.
Keywords: Lifting, manual materials handling, back pain, squat, stoop, joint strength
1. Introduction
For the last several decades, manual materials handling has attracted great interest from researchers
in many disciplines, primarily because of the huge amount of work and nancial losses, and human
sufferings caused by low back pain and injuries. Consequently, it is a major concern to researchers and
organizations to develop means to predict, control, and prevent such injuries. Thus, much research has
focused on the establishment of ergonomic workspace designs and employee training.
Biomechanics modeling plays an important role in estimating individuals lifting capacities, comparing
different lifting modes, and designingworkspace conditions. Using such models, the potential for injuries

Corresponding author.
0928-7329/05/$17.00 2005 IOS Press and the authors. All rights reserved
58

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment
Fig. 1. A two-dimensional sagittally symmetric model for human lifting.
can be estimated in advance and greatly reduce the need for often difcult, expensive and potentially
invasive laboratory measurements [7,18,19].
Optimal control techniques are being used to form lifting trajectories and predict associated joint
moments during lifting. Optimal control is used in the biodynamics modeling is primarily due to two
reasons. First, locomotion is believed to obey a certain principle of optimality [2,15]. Since optimal
control theory aims to determine the control laws that will minimize (or maximize) an objective function
subject to some constraints [12], such techniques, when applied to an adequate dynamic model for
the system, provide a practical means for determining muscle forces and joint torques. Secondly, the
dynamic model of acceptable accuracy that must be developed for use with the above described optimal
control scheme is itself a valuable analysis and assessment tool for the prediction of the muscle forces
and joint moments that produce the desired movement. Unfortunately, this dynamic model construction
and analysis is not a simple task, given that the musculoskeletal system considered is highly redundant,
i.e., the number of independent muscles acting on a particular joint exceeds the number of degrees of
freedom of that joint. Moreover, many muscles can affect more than one joint at a time, which brings
complex coupling to the system. Therefore, there is no direct or unique solution to the problem of
performing a specic task. However, the above-mentioned difculties can be overcome by using optimal
control techniques to estimate muscle forces produced during lifting [2,15,18].
In this paper, moment time histories obtained for different lifting conditions are compared to deduce
information on safe lifting patterns. The results obtained to date clearly indicate that one should include
kinetic measures in the objective functions to be optimized and in the assessment of lift simulation
accuracy and safety [1,6]. The primary reason behind computing moments is that resultant forces at
joints can be determined from this information. Since the forces produce stresses in bones, muscles and
connecting tissues, and thus contribute directly to injuries. Therefore, while comparing lift performances,
the kinetic measures should be considered, because looking at results at kinematics level (e.g. matching

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment 59

0 5 10
-20
0
20
40
A
n
k
l
e
/
M
o
m
e
n
t

[
N
m
]
0 5 10
-50
0
50
100
K
n
e
e
/
M
o
m
e
n
t

[
N
m
]
0 5 10
-500
0
500
H
i
p
/
M
o
m
e
n
t


[
N
m
]
0 5 10
-500
0
500
Time steps
S
h
o
u
l
d
e
r
/
M
o
m
e
n
t

[
N
m
]
0 5 10
-500
0
500
1000
Time steps
E
l
b
o
w
/
M
o
m
e
n
t


[
N
m
]
1/2 sec
1 sec
2 sec
Fig. 2. Comparison of moment-time histories for all joints under different lift durations.
predicted trajectories to those which are experimentally obtained) will not consider these important
quantities. With this in mind, kinetic measures, such as moment time histories were compared in this
paper to evaluate manners to perform more safe lifts for different lifting conditions such as lift load,
mode, and duration.
2. Theory
2.1. Physical model
A two-dimensional sagittally symmetric human body model, Fig. 1, was constructed as a ve rigid
link mechanism for the biomechanical simulation of manual lifting tasks. These links possessed the
same length, mass, and inertia properties as estimated for their human counterparts. Therefore, any
movement or conguration could be described with the ve generalized coordinates describing the
relative orientation of each of these ve links with respect to the parent body.
Joints at the ankle, knee, hip, shoulder, and elbow were all treated as one-degree of freedom revolute
joints. Spinal column was considered as one rigid link that includes mass of the head and neck
appropriately located. The hands were also modeled as parts of the forearms, and their relative motion
with respect to forearms were neglected. It was further assumed that subject did not walk with the load
during the lift, i.e., foot was xed on the ground [4,11].
The joint reaction forces and joint moments for a typical rigid link i in an n-link open chain mechanism
can simply be obtained by utilizing a Newton-Euler formulation recursively [20].
60

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment
1 2 3 4 5 6 7 8 9 10
100
200
300
400
500
600
700
800
Time st eps
E
n
e
r
g
y

[
N
m
]
1/ 2 sec
1 sec
2 sec
Fig. 3. Energy change of the total lift for different lifting durations.
2.2. Optimization
One of the most signicant problems in optimization of biomechanical systems is the choice of a
proper cost function reecting the most important aspects of locomotion and loading. In this paper, it
was chosen to minimize integration over the time of sum of the square of the ratio of the predicted joint
moments to the corresponding joint dynamic strength [5,6].
J =

t
f
0
5

i=1

M
i
(,

,

)
S
i
(,

)

2
dt (1)
where t
f
is the lifting duration, M
i
are moments and S
i
are joint dynamic strengths for the ith joint.
In Eq. (1), the moments and the strengths are given in terms of ,

, and

, which represent joint
angular displacements, angular velocities and angular accelerations for each joint, respectively. The
joint strengths were considered as the measures of joint capacities under different postures and joint
angular velocities [9,10].
The dynamic strength values were used in the objective function as opposed to static ones because
dynamic strengths better replicate the joint behavior and improve the simulation [9,10]. They were
dened to be functions of joint angular positions and velocities for each joint i [8] in the following form
S
i
(,

) =
i0
+
i1

i
+
i2

i
+
i3

2
i
+
i4

2
i
+
i5

i
(2)
The coefcients
1
through
5
were determined based on experimental results and they were directly
taken from [8]. The ratio between the moment and joint strength in the objective function above (Eq. (1))
is called the muscular utilization ratio (MUR).

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment 61

0 0.5 1
-50
0
50
100
A
n
k
l
e
/
M
o
m
e
n
t

[
N
m
]
0 0.5 1
-50
0
50
100
K
n
e
e
/
M
o
m
e
n
t

[
N
m
]
0 0.5 1
-200
0
200
400
H
i
p
/
M
o
m
e
n
t


[
N
m
]
0 0.5 1
-200
-100
0
100
lifting time [sec]
S
h
o
u
l
d
e
r
/
M
o
m
e
n
t

[
N
m
]
0 0.5 1
-200
0
200
400
lifting time [sec]
E
l
b
o
w
/
M
o
m
e
n
t


[
N
m
]
5 kg
10 kg
15 kg
Fig. 4. Comparison of moment-time histories of all joints under different loads.
The constraints on the objective function were of four types: kinematic, kinetic, stability, and penetra-
tion. Kinematic constraints were the ones that each joint operate within a certain range. For example, an
elbow cannot be extended over 180

. Consequently, every joint shared this type of geometric constraint.

The second type of constraint was related to some kinetic measures, in which the maximum moments
generated by joints during a lift were restricted not to exceed a certain limit (i.e., a strength capacity).
Thirdly, the stability of the body had to be maintained. For this purpose, the center of mass of the
subjects body and the load were forced to remain directly over the subjects foot. Lastly, load lifted was
forced not to penetrate into the body during the simulations. All these constraints were implemented in
the genetic algorithm as penalty functions of cubic order of the error (e.g. constraint violation).
2.3. Numerical formulation of the problem
The problem is difcult because it is highly nonlinear and is of innite dimension. One possible
approach is to try problemas a two-point boundary value problemthat is parameterized by approximating
the states and/or control variables by a polynomial and/or a Fourier series [4,17]. For this study, joint
angles were approximated as seventh order polynomial in the form

i
=
7

j=0
a
i,j
t
j
(3)
for the ith joint. Since the boundary conditions (initial and nal angular positions, angular velocities,
and angular accelerations) were known for a lifting experiment, six of the coefcients can be determined.
The other two coefcients were added to the polynomials to introduce extra degree of freedom for
62

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
100
200
300
400
500
600
700
lifting time [sec]
E
n
e
r
g
y

[
N
m
]
5 kg
10 kg
15 kg
Fig. 5. Energy change of the lift for different loads.
optimization. By substituting these polynomials and their derivatives into Eq. (1), the problem becomes
a nite dimensional parameter optimization of the form
J =

t
f
0
f
1
(a
i,j
, t)dt (4)
where i is the joint number, and j coefcient index of the polynomial. Since the lifting duration is
known, the problem can further be simplied by discretization in integration time steps of t as
t =
t
k
(5)
where t is time, k is the number of integration steps. Then, the problem becomes minimizing another
function including only the polynomial coefcients, a
i,j
, and the integration step size, t as follows
J = f
2
(a
i,j
)t (6)
Once the coefcients in the polynomial are estimated, the optimized path and moments for a lifting task
can easily be determined.
A genetic algorithm implementing Goldbergs [3] algorithm in MATLAB was developed for opti-
mizations. It used xed population size with string length of 30, a crossover probability, P
c
, of 0.001,
and a mutation probability rate, P
m
, of 0.002.
3. Results and discussion
The simulation and prediction program was coded for designing sagittally symmetric lifting tasks
and for utilization as a computer experimentation tool that provides an extensive amount of information

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment 63

0 0.5 1
-10
0
10
20
A
n
k
l
e
/
M
o
m
e
n
t

[
N
m
]
0 0.5 1
-20
0
20
40
K
n
e
e
/
M
o
m
e
n
t

[
N
m
]
0 0.5 1
-200
0
200
400
H
i
p
/
M
o
m
e
n
t


[
N
m
]
0 0.5 1
-200
0
200
400
Lifting time [sec]
S
h
o
u
l
d
e
r
/
M
o
m
e
n
t

[
N
m
]
0 0.5 1
-200
0
200
400
Lifting time [sec]
E
l
b
o
w
/
M
o
m
e
n
t


[
N
m
]
free
squat
stoop
Fig. 6. Moment time history for each joint for different modes of lifting.
about movements, and inter-joint loads under different lifting scenarios. For example, if the specied
lifting duration is changed, the predicted effect of this change on joint moment levels may be easily
investigated. In this manner, the analyst may similarly investigate the individual effects of various
system parameters on lifting (task execution) performance, and assess the likelihood of incurring injury.
Examples include, but are not limited to, consideration of individuals specic body structure (height,
weight, body geometry, etc), individuals specic constraints (e.g. presence of prior injury, or disability,
etc.), or modication of task objective or task execution protocol.
3.1. Effect of lifting duration
As an example, given below are the results for three different lifting durations, specically 0.5 sec
(fast), 1 sec (moderate), and 2 sec (slow), for a subject with mass of 94.7 kg, height of 1.85 m. For
this purpose, all other parameters are held constant. The load lifted was 5 kg. Sample results for joint
moment changes were given in Fig. 2 for subjects all joints, and the total energy change for the lift in
Fig. 3.
These results are presented with respect to the ten equally spaced time steps during the three lifting
durations in the simulation and measured data, i.e. the 1st and 10th steps represent the beginning and the
end of the lift.
It is obvious from Figs 2 and 3 that higher levels of joint moments, as well as signicantly more
energy are required in fast lifting. Although the change in energy of the payload from a pure physics
(conservative system) perspective is the same for each case, energy expended in lifting or accelerating
a body cannot be reclaimed because the human form is not a conservative system. Thus, if one too
greatly increases the rate at which a task is accomplished, the task may require over exertion, which can
be injurious to the individual involved. It is also interesting to observe that the plots collapse upon one
another as the task duration increases so that the system becomes effectively quasi-static.
64

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
150
200
250
300
350
400
450
500
550
600
650
Lifting time [sec]
E
n
e
r
g
y

[
N
m
]
free
squat
stoop
Fig. 7. Total energy change of the lift for different modes of lift.
3.2. Effect of load mass
Keeping the time and other parameters unchanged in this example, the mass of the load to be lifted is
increased from 5 kg to 10 kg and 15 kg for a subject with a mass of 94.7 kg and a height of 1.85 m, with
the associated results provided in Figs 45.
The results predicted are consistent with what one would physically expect. Specically, an increase
in load lifted requires more energy consumption (Fig. 5). Also, higher levels of moment applications
are necessary to accomplish the task, especially at the start of the lift (see Fig. 4). Therefore, it can
be advised here that one should not lift a load that is sufciently heavy to require a joint moment that
exceeds the associated joint strength for the individual, which can be injurious.
3.3. Effect of lifting mode
In this case, all parameters were kept constant, except the initial angular positions of body segments.
They were changed so as to have a lifting mode of a squat, a stoop, or a free lift. In the stoop, the ankles
and knees were kept straight, whereas in the squat, the back was kept to be vertical during the lift. The
results for the three aforementioned cases for a subject with a mass of 94.7 kg and a height of 1.85 m
were given in Figs 67.
As seen from moment diagrams (Fig. 6), stooping requires more exertions in the hip, shoulder and
elbow joints, whereas squatting requires higher moments at ankles and knees, as expected. However,
free lifting requires the most moderate moment exertion levels in almost all joints. As also obvious
from the total energy of the lift (see Fig. 7), the most demanding job is the stooping. The lowest energy
requirement is in the free lifting. Therefore, it is advised that one should choose a free style when lifting.
This was also concluded by other researchers in the literature based on other criteria [13,14].

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment 65

4. Discussion
A prediction program was developed to simulate the manual materials handling tasks for investigating
the effects of different parameters such as lifting duration, load, and mode. The program was coded in
MATLAB [16], which provides users a very rich collection of functions in mathematics, plotting and
animation of the results.
The prediction program rst approximates the states of the system in the equation of motion, which
are derived via a Newton-Euler formulation. Then, this program optimizes the system performance with
respect to the objective function, which is postulated as integration over the time of sum of the square
of the ratio of the predicted joint moments to the corresponding joint dynamic strength [6]. Genetic
algorithms are then used in the optimization portion of the program which returns estimates for the
coefcient values for the seventh order polynomials in time, which are associated with each predicted
joint angle time history. Lastly, this optimized coefcient information is fed back to the equations of
motion to simulate the lifting motion, and to obtain joint loads and moments.
In making a quantitative comparison of the safety of different lift conditions, it is imperative that
predictions be made on kinetic measures such as moment time histories [1,6]. As stated previously, these
moments are keys to assessing likelihood of injury. Moreover, these quantities directly affect the system
accelerations and angular accelerations, which in turn will inuence displacements, rates and thus joint
strengths. With this in mind, it is proposed that the objective functions used in evaluating the safety of
lifts should involve appropriate kinetics measures. In this paper, moment time histories for different lift
durations (slow, moderate, and fast), lift modes (squat, free, and stoop), and lift loads (light, medium,
and heavy) were computed and compared.
Acknowledgement
We greatly appreciate the invaluable assistance of Professor Kinda Khalaf and Professor Patrick Sparto
and the generosity with which they provided their experimental data for validation of our theoretical
model.
References
[1] K.S. Anderson and

O. G undogdu, Need for Kinetic Measures in Predicting Musculoskeletal Response with Optimal
Control, NSF/NIOSH Workshop on Next Generation Human-Assist Devices and Automation, September 14, 2000,
Baltimore, MD.
[2] C.K. Chow and D.H. Jacobson, Studies of human locomotion via optimal programming, Mathematical Biosciences 10
(1971), 239306.
[3] D.E. Goldberg, Genetic Algorithms in Search, Optimization, and Machine Learning, Addison Wesley, Reading, Mas-
sachusetts, 1989.
[4] W.A. Gruver and E. Sachs, Algorithmic Methods in Optimal Control, Pitman Advanced Publishing Program, Boston,
1980.
[5]

O. G undogdu, Quantication and assessment of objective function performance in manual materials handling, Ph.D.
Dissertation, Rensselaer Polytechnic Institute, Troy, New York, 2000.
[6]

O. G undogdu, Importance of kinetic measures in trajectory prediction with optimal control, J. Eng. Sci. 7 (2001),
151155.
[7] S.H. Hsiang and M.M. Ayoub, Development of methodology in biomechanical simulation of manual lifting, Int. J. Ind.
Ergonomics 13 (1994), 271288.
[8] K. Khalaf, Development of experimental and analytical models for biomechanical simulation and ergonomic assessment
of manual material handling tasks, Ph.D. Dissertation, The Ohio State University, Columbus, OH, 1998.
66

O. G undogdu et al. / Simulation of manual materials handling: Biomechanial assessment
[9] K.A. Khalaf, M. Parnianpour, P.J. Sparto and K. Barin, Determination of the effect of lift characteristics on dynamic
performance proles during manual materials handling tasks, Ergonomics 42 (1999), 126145.
[10] K.A. Khalaf, M. Parnianpour, P.J. Sparto and K. Barin, Feature extraction and quantication of the variability of dynamic
performance proles due to the different sagittal lift characteristics, IEEE Trans. Rehabilitation Eng. 7 (1999), 278288.
[11] K.A. Khalaf, L. Wade, M. Parnianpour and S.R. Simon, Biomechanical simulation of manual multi-link coordinated
lifting, Fifteenth Southern Biomechanical Eng. Conference, Dayton, Ohio, 1996, pp. 197198.
[12] D.E. Kirk, Optimal Control Theory, Prentice Hall, New Jersey, 1970.
[13] K.P. Kothiyal, J. Mazumdar and G. Noone, A biomechanical model for optimal postures in manual lifting tasks, Int. J.
Ind. Ergonomics 10 (1992), 241255.
[14] S. Kumar, The physiological cost of three different methods of lifting in sagittal and lateral planes, Ergonomics 27 (1984),
425433.
[15] B. Ma, An improved algorithm for solving constrained optimal control problems, Ph.D. Dissertation, University of
Maryland, 1994.
[16] MATLAB, The Student Edition of MATLAB, The Language of Technical Computing, Version 5, Users Guide, The
Mathworks, Inc., New Jersey, 1997.
[17] M.L. Nagurka and V. Yen, Fourier-based optimal control of nonlinear dynamic systems, J. Dynamic Systems, Measure-
ment, and Control 112 (1990), 1726.
[18] M.G. Pandy, F.C. Anderson and D.G. Hull, A parameter optimization approach for optimal control of large-scale
musculoskeletal systems, Trans. ASME J. Biomechanical Eng. 114 (1992), 450460.
[19] M.G. Pandy, B.A. Garner and F.C. Anderson, Optimal control of non-ballistic muscular movements: A constraint-based
performance criterion for rising from a chair, Trans. ASME J. Biomechanical Eng. 117 (1995), 1526.
[20] F.E. Zajac and M.E. Gordon, Determining muscles force and action in multi-articular movement, Exer. Sport Sci. Rev.
17 (1989), 187230.