Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Computer Program to Generate Dimensional

J. h. BARTZ
Principal Engineer.
and Inertia! Properties of the Human Body
A digital computer program has been developed to calculate dimensional and inertial
0. R. OIANOTTI
properties of the human body. The program has been designed so that the user may
Assistant Mathematician. either select a data set from a program library, or compute a data set from a geometric
Calspan Corp., Buffalo, man-model. From primary program inputs of sex, standing height, seated height,
N. Y. and weight, the routines compute body segment link lengths, contact surface dimensions,
masses, and moments of inertia from inputted sets of anthropometric data. Overall
validity of the formulation and techniques has been established by comparing computed
results with measurements on the human body reported by various investigators.

Introduction mon characteristics with the present model, such as geometrical


simplification in modeling the human body.
T.
I HIS research effort1 was performed to automate, Primary assumptions applied in formulating the present model
are as follows:
at least partially, the calculation of dimensional and inertial
properties of the human body. These properties are required 1 For purposes of computing segment dimensions and inertial
for the preparation of input data sets for digital computer properties, the body is approximated by 15 rigid body segments
simulations of the motor vehicle crash victim that have been connected by 14 joints, in the manner illustrated in Fig, 1.
developed recently at Calspan. The simulations, the planar 2 These body segments are approximated with simple
(two-dimensional) Revised Occupant Simulation model (ROS),
[l],2 and the ROS model as modified by the General Motors JOINTS
Corporation (MODROS), [2], and the Three-Dimensional HEAD PIVOT
NECK PIVOT
Crash Victim Simulation computer program (3-D), [3-6], each WAIST
require hundreds of inputs to specify the simulated crash vic- PELVIS
RIGHT HIP
tim and the crash environment. These numerous inputs are RIGHT KNEE
RIGHT ANKLE
required because of the detail of representation incorporated LEFT HIP
in the simulations. Automated generation of crash victim LEFT KNEE
LEFT ANKLE
data sets offers a means to reduce overall computation cost, RIGHT SHOULDER
rapidly perform parametric variations, and decrease the possi- RIGHT ELBOW
LEFT SHOULDER
bility of human error in input preparation. LEFT ELBOW

The basic approaches and assumptions applied are similar to


those reported in other efforts to compute the inertial properties
of the human body. These include models to calculate the inertial
properties of body segments developed by Lepley [7], and Patten
and Theiss at Calspan [8], and a model to calculate inertial
properties of the whole body in various orientations that was
developed by Hanavan [9]. Work space evaluation models, such
SEGMENTS
as reported by the Boeing Company [10], Kroemer [11], and
H HEAD
Chaffin and Snyder [12], and a model for a manned spacecraft N NECK
UT UPPER TORSO
application developed by Conway [13], also share certain com- CT CENTER TORSO
LT LOWER TORSO
RUL RIGHT UPPER LEG
RLL RIGHT LOWER LEG
RF RIGHT FOOT
LUL LEFT UPPER LEG
^ponsored by the Motor Vehicle Manufacturers Association. LLL LEFT LOWER LEG
LF LEFT FOOT
RUA RIGHT UPPER ARM
'Numbers in brackets designate References at end of paper. RLA RIGHT LOWER. ARM
LUA LEFT UPPER ARM
Contributed by'the Bioengineering Division and presented at the Winter LLA LEFT LOWER ARM
Annual Meeting, Detroit, Mich., November 11-15, 1973, of T H E AMERICAN
SOCIETY OF MECHANICAL ENGINEERS. Manuscript received at ASME Head-
quarters, July 26, 1973. Paper No. 73-WA/Bio-3. Fig. 1 Fiftaen-segment man-model

Journal of Engineering for Industry FEBRUARY 1975 / 49


Copyright © 1975 by ASME
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Table 1 Comparison of anthropometric data obtained by several Table 2 Required dimensional data
Investigators
MEASUREMENT SYMBOL

SUBJECT MEASUREMENT- STOUDT


REFERENCE
SAHLEY DAMON CLAUSER
"HLE' DAMON ET*. STOUOT

ET AL ET AL ET AL

5TH PERCENTILE HEIGHT 59.0 60.4+ 59.9 60.0


FEMALE SITTING HEIGHT, ERECT 30.9 32.0+ 32.4 31.7
SEAT BREADTH (SITTING! 12.3 13.3+ 13.5 13.3 SITTINCHE.H,
KNEE HEIGHT 17.9 Yi.2

SOTH PERCENTILE HEIGHT 62.9 64.1 63.9 63.6 HEADBfltADTH :;::;;


FEMALE SITTING HEIGHT, ERECT 33.4 33.6 34.1 33.7 •fa
SEAT BREADTH (SITTING) H.3 15.0 15.0 15.0
KNEE HEIGHT 19.6 18.8 , : ™ r .,«.„„••
50TH PERCENTILE HEIGHT 68.3 69.0 69.1
MALE SITTING HEIGHT, ERECT 35.7 36.2 36.0
SEAT BREADTH IS1TTING]
KNEE HEIGHT
14.0
21.4
14.0
21.7
13.9
21.7 ;
95TH PERCENTILE
MALE
HEIGHT
SITTING HEIGHT. ERECT
SEAT BREADTH (SITTING)
72.8
38.0
15.9
23A
73.2+
37.0+
1S.S+
23.<H
73.1
38.0
15.4
23.3 ; •
b WAI5T CIRCUMFERENCE

=
BUTTOCK CIRCUMFERENCE
E (S AMEI

KNEE HEIGHT
ZZZZ"ZT
,»„, *,.
•NOMENCLATURE OF STOUDT ET AL
+CALCULATED ASSUMING GAUSSIAN DISTRIBUTION ,»«, u
™ ™
; = = « r a
u»r™™. •*11

geometrical shapes; in the present model, ellipsoids are used to


represent each segment.
I
WRIST CIRCUMFERENCE

b KNEE HEIGHT
" " " ' " " " • • " ' " " »

UffEB THSGH CJBCUUFEBEHM ir "


» ^ E R LEG CIRCUMFERENCE L
™,Vcr" ci " cu "' - •*li

3 During body motion, the various body segments move !• KNEE CIRCUMFERENCE

Eiir !S,mwrF!"tWi
K
s miNDCUMFtl"r'CE'

about joint locations that are fixed in the adjacent rigid body It ANKL* CIRCUMFERENCE ISAMEI ISAMEI -*lt
segments. »
UNKieHeffiKT.OUrSIDt
FOOT BREADTH ISAMEI
LATERAL HftttEOUS HEIGHT
ISAMEI
-
ISAMEI
J"n

4 For any given body segment, variation of segment density »


31. FOOTIENGTH ISAMEI ISAMEI ISAMEI
*X
*l\

with location is sufficiently small that the average density, taken


as uniform, is used to compute segment inertial properties.
These assumptions are similar, if not identical, to those that
were applied in the models used to calculate inertial properties comparison that sets of dimensional measurements can be used
of the human body that have been previously discussed, refer- interchangeably for practical engineering calculations with
ences [7-9]. The present model, however, differs from those little error.
described in [7, 9] in that it also serves to define input data Stoudt [18], presents a rather detailed review of existing data
sets for models in recent Calspan computer simulations of for children, together with two sets of average dimensional data
the motor vehicle crash victim. for children. They are values that were derived by Stoudt from
a large number of anthropometric studies and, therefore, should
The 15-segment, 14-joint ellipsoidal man-model coincides
be quite representative.
identically with the body dynamics model and contact model of
the 3-D crash victim simulation, and largely defines the plotter The required inertial data on body segments were obtained
graphics model of that program as well. To compute the re- primarily from results presented by Drillis and Contini [19],
quired parameters for the eight-segment, seven-joint planar which includes a summary of the results obtained by other in-
ROS/MODROS computer program, additional mathematical vestigators as well as their own results. The inertial data re-
relationships are required. ported by Santschi, et al. [20], obtained on living humans in
various body positions, were most useful in testing the predic-
In development of this model, emphasis has been placed on
tions of the model.
obtaining results that are sufficiently accurate for use in practical
engineering calculations that require dimensional and inertial
properties of the human body. Consequently, all simplifications Development of The Model
and approximations that were made in the modeling effort were
As a starting point in the formulation of the model, a list of
aimed primarily at program utility rather than an extremely
anthropometric measurements required to completely define the
high degree of precision of the calculated results.
model was compiled. In developing this list, emphasis was
To aid in the development, a literature survey of applicable placed on using only measurements which are commonly ob-
anthropometric data was performed, with emphasis placed on tained in anthropometric surveys, as discussed in references
data obtained recently on large, representative samples of the [16, 18]. It was found, however, that in order to provide suf-
U. S. population. The required dimensional data for adults ficient detail to completely specify the 15-segment model, some
were taken primarily from Sahley [14], for two reasons. First, less common dimensions were required. The primary intent
a large number of measurements are presented in that reference, was to tabulate the minimum number of dimensions that would
which provided information for a detailed definition of the model. adequately define the model. This list, presented in Table 2,
Second, the measurements are based on a large sample (6283) uses the nomenclature of Sahley [14], but also includes the
of the TJ. S. civilian population, both male and female. Other nomenclature of Damon, et al. [16], Clauser, et al. [17], and Stoudt
sources served as useful supplements, however. For example, [18], whenever their terminology differs from reference [14]. In
Stoudt, et al. [15], also contains a tabulation of a rather large addition, a symbol corresponding to each measurement is listed.
number of measurements obtained on thousands of TJ. S. civil- In some cases, two measurements are given in Table 2, either of
ians. Damon, et al. [16], also presents a rather extensive tabula- which may be used to define a single model dimension. The pro-
tion of dimensional data obtained on large samples (thousands) vision for two measurements in the model was included because
of TJ. S. military personnel, both male and female. A recently the survey of available data showed some nonuniformity in the
completed study of the anthropometry of Air Force women choice of measurements, such as the measure of chest depth.
(nearly two thousand) by Clauser, et al. [17], also provided ad-
ditional useful information. To illustrate the availability of the required data, a tabulation
of these dimensions is presented in Table 3, as follows:
In Table 1, a comparison is made between some of the data
presented by Stoudt, et al., and the corresponding values pre- 1 Values of the mean and standard deviation of each available
sented by Sahley, Damon, et al., and Clauser, et al. It can be dimension for civilian males and females, from Sahley [14].
seen from Table 1 that the measurements for a given size do not 2 Values of the mean and standard deviation of each available
vary much over a wide range of body sizes. It is clear from this dimension for males and females in military service, obtained

50 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 3 Values of the dimensional data (in inches) from references given in Damon, et al. [16], and from Clauser,
SAfKIY « m l M I "kSflrAiSfi*1
et al. [17].
.«« MMMaiAlMLIMHVI 1T0UOT

3 Values of the mean of each dimension for 30- and 50-lb


UtA*
««. 515: H..N
SK'. «.* •IS: kUH
S* sa sa
MV.
children from Stoudt [18].
*3
In some cases, it was necessary to estimate the required di-
mensions for the 30- and 50-lb children from the dimensions given
-c,
by Stoudt. Stoudt considered only the seated position in his
l*-0
study, which was concerned primarily with child restraints for
*». E «•' motor vehicles. In any case, the availability of most of the re-
E 'oil
1.H'
i quired dimensions is obvious from Table 3. It should be noted
» ™ that the missing values could be easily and rather accurately
estimated from other data in the table, because there are only
"]•'
•*!•.

«™"
small variations in any given dimension among population
E samples.
• ^
« Following the identification of sufficient data to adequately
define the model, the mathematical model was formulated. To
«?„
aid in constructing the geometric model, a set of spheres, centered
on the principal joints in the extremities, was defined. These
J* spheres are illustrated in Fig. 2 for the right side of the body,
together with the dimensions that define their vertical position
from the floor for a standing position. The nomenclature is
given in Table 2, with the exception of the sphere radii, which
for the right side are defined as follows:
••*••«;-

Joint Symbol Sphere radius

right shoulder RS fRS =


2ir

Table 4 Average body segment specific gravity, p„ Lit


right elbow RE fRE =
N o m e n c l a t u r e of N o m e n c l a t u r e of iff
R e f e r e n c e 19 3-D P r o g r a m

head and n e e k (head, neck) 1. I l l Ln


right wrist RW «"RW =
(upper t o r s o , center
2w
trunk 1.031
torso, lower torso)
Lu
upper a r m (upper a r m ) 1. 081 right hip RH fRH
2TT
forearm (lower a r m ) 1. 1Z2
Ln
thigh (upper leg) 1.069 right knee RK »*RK =
2T
shank (lower leg) 1.095

1. 100
Lis
foot (foot) right ankle RA fRA =

Table 5 Average body segment mass (in percent of whole body mass)
The spheres, defined in this manner, have a circumference
N o m e n c l a t u r e of N o m e n c l a t u r e of
R e f e r e n c e 19 3-D Profiram equal to the measured circumference at or near the corresponding
h e a d , neck, a n d
joints.
(head, neck, u p p e r
trunk torso, center torso The remaining geometry, illustrated in Fig. 3, follows in a
lower torso)
straightforward manner from definition of the spheres. The
u p p e r a r m s (each} (upper a r m ) geometrical model is constructed entirely from ellipsoids, with
f o r e a r m s and (lower a r m ) an ellipsoid corresponding to each of the 15 body segments.
h a n d s (each) The ellipsoid semiaxes are computed from the breadth, depth,
t h i g h s (each) (upper leg) and height of the corresponding body segments. It should be
shanks (each) (lower leg)
noted that the lengths of the neck and torso segments are some-
what arbitrary. The ellipsoid semiaxes were selected with the
feet (each) (foot)
following considerations in mind, in the order listed:
1 Minimum overlap between segments is desirable to avoid
Table 6 Test of prediction of segment masses—values of R„ for large corrections in the computation of segment masses, which
various segments
are based on the total ellipsoid volume.
, Measured Segment Mass/
Segment n Pi edicted Segment Mass 2 The lengths of the neck and torso segments are based on
head adjacent body dimensions, not arbitrarily chosen, to insure that
neck a reasonable geometry will be produced with widely varying
upper t o r s o 1.22 sets of anthropometric dimensions.
center torso 3 The model illustrated in Fig. 3 defines both the contact
lower t o r s o model and much of the plotter graphics display model of the
upper a r m 1.08
Three-Dimensional Crash Victim Simulation computer program.
lower a r m 0. 83
upper leg 1.02
Consequently, dimensions were selected to provide realistic
lower leg 0.89
contact surfaces and also a realistic overall representation of the
foot 1.04 human form. For example, the ellipsoids for the limb segments

Journal of Engineering for Industry FEBRUARY 1975 / 51

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Tf\
w RUA

LT RH

7~? / / / / /~y / / / / ,Mu/ / /


HEIGHT

Fig. 2 Spheres used in constructing the model

RUL
were selected to be tangent to the corresponding pairs of spheres.
This assures a reasonable geometry for the limb contact surfaces
when relative motions of the limb segments occur, and also
provides a more realistic appearance in the graphics display.
Algebraic relationships for the ellipsoid semiaxes corresponding
to each segment and the locations of the joints with respect to
the adjacent segment centers of gravity are presented in detail
in reference [21]. For the sake of brevity, they are not repeated
in this paper.

Validation of the Model


To test the algorithms, some of the anthropometric data of
Table 3 were used in a series of calculations to compute the model RLL
geometries from four data sets:
50th percentile male
Sahley data, reference [14]
50th percentile female
50-lb child
Stoudt, reference [18]
30-lb child
Model geometries for these four data sets are shown in pictorial
form in Pig. 4. It can be seen that the algorithms yield realistic
results over a wide range of stature, weight, and relative propor- immnmmmw
tion. Fig. 3 Geometry of the modal
As a more critical test of the algorithms, additional model
geometries were computed in a series of calculations, again
using the data of Table 3, as follows:
were obtained from values of the mean and standard deviation
95th percentile male given in Table 3. For a Gaussian distribution, the value of a
Sahley data, reference [14] given dimensional measurement Y for a specified percentile P
5th percentile female
is related to the mean value of that measurement, m, and its
Values of all of the 31 dimensional measurements, Table 2, standard deviation, s, by the relation

52 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


mation of segment shapes by ellipsoids, and density variation
50th
PERCENTILE
, MALE
within a given segment. To test the assumptions applied in the
50th model, values of R„, the ratio of measured segment mass to pre-
PERCENTILE
FEMALE dicted mass of the entire ellipsoid of uniform density, were
computed for various body segments. Using the set of dimen-
sional measurements reported by Sahley for a 50th percentile
male, Table 3, and a total body weight of 164 lb for civilians from
Damon, et al. [16], the values of R„ were calculated from
u Ra
Ma
(1)
30 tb irptidaKCn
=•< CHILD

where M„ is the measured segment mass, p» the segment specific


V;/ gravity, and a„, bn, c„, the segment ellipsoid semiaxes.
P The values of R„ were obtained using the values of p„ from
ri Table 4 and M„ from Table 5. The results are given in Table 6,
where it can be seen that the predicted mass for the entire
ellipsoid is typically within about 10 percent of the experimental
value for the arm and leg segments. The maximum difference is
22 percent for the other segments. These results are regarded
as quite good, considering that available experimental data are
7777777777fr77777777Zr777777777777777777777777777777777777777777~
quite scarce and are, after all, only a first approximation to a
Fig. 4 Test of the algorithm used to compute model geometry 50th percentile male. An empirical correction factor that does
not differ greatly from the value of unity lends support to the
assumptions applied in the development of the model.
Segment moments of inertia predicted from the ellipsoidal
model were compared with experimental results obtained on
Y — m + sX, where eight cadavers by Dempster and on 20 living human subjects
by Drillis and Contini. Because of the scarcity of human data,
X relates the percentile and the Gaussian exponential function measurements of segment moments of inertia made by Calspan
through

95th
e-'<*£,
^ 2^ J
PERCENTILE
P = MALE

ft/ —O

where £ is a dummy variable.


Results for these two data sets are shown in Pig. 5. Again, it
can be seen that the algorithms are satisfactory over a rather
wide range of stature and height.
The validity of the model was assessed by comparing predic-
tions of segment masses and moments of inertia with experi-
mental results reported in the literature. Because of the scarcity
of such data on individual body segments, comparisons were
also made between moments of inertia for the whole body that
were computed from the algorithms and the corresponding ex-
w
perimental results obtained on living human subjects in various
positions.
For the comparison of predicted and measured segment
masses, mean body segment specific gravities (densities) p„ and
body segment masses M„ (in percent of the whole body mass)
were required. The specific gravity data, Table 4, were obtained
from Drillis and Contini [19], and represent the average values
of measurements on cadavers reported by Harless and Dempster •i)iinrin/)niiii/i)//n/i/)///ii/i/i)uiu/iinn
(references given in reference [19]), and measurements on living
Fig. 5 Test of the algorithm used to compute model geometry
subjects reported by Drillis and Contini [19].
The average values reported by Drillis and Contini were used Table 7 Test of prediction of segment moments of inertia in the
because there is only a small variation (several percent) between sagittal plane for various segments (all values in Ib-secMn.)
the results reported by these three groups of investigators.
Segment Prediction Measurement
The average body segment masses (in percent of the whole body
mass), Table 5, were also obtained from Drillis and Contini Calapan Measure-
Dempster Drillis & Contini ments (Sierra 292-
[19] and represent the average values of measurements obtained {cadavers) (living subjects) 1050 Dummy)
on cadavers reported by Harless, Dempster, and Braune and upper a r m 0. 150 0. 123 0. 122 0. 165, 0. 167
Fischer, and measurements obtained on living subjects by Meeh lower a r m 0. 198 0. 165 0.218 0.255, 0.259
and Bernstein (references given in reference [19]), and Drillis upper leg 1.242 0.955 0. 791 0. 727, 0. 703
and Contini [19]. Again, the average values reported by Drillis lower leg 0.415 0.369 0.438 0.435. 0.437
and Contini were used because the variation between the re- foot 0.0400 0. 0274 0.0177 0.0434
0. 930 1. 18
ported results is not large (typically several percent). entire a r m 1.333 -
6.62
In the formulation, an empirical correction factor R„ is used to
entire leg 8. 38 6. 16 -
0. 980 0. 920 0,991
account for overlap between adiacent body segments, approxi-
lower leg
and foot
-

Journal of Engineering for Industry FEBRUARY 1975 / 53

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 8 Test of prediction of head segment principal moments of TABLE 9 (continued)
Inertia (all values in Ib-sec'-in.)
Configurator
Santschi, et al ent Model Hanavan
Prediction Measurement (Becker)
50th r centile male
0. 163 0.226
0. 194 0.240 nter of gravity

67
0. 133 0. 128
7.9 8.6
y 0 0*
2 26.5 26.5

moment of inertia
Table 9 Test of prediction of whole body moments of Inertia (centers
of gravity in inches, moments of inertia in Ib-sec'-in.) X 61. 1 52. 8 50.0
y 66.6 66.3 63. 9
z 33. 5 24. 8 27.4
Configuration Measured
Santschi, et al Present Model Hanavan
95th percentile male sitti "fi
rcentile male - standing
center of gravity
X 8. 5 9.6
X* 3.5 4.4 3.4 y 0 0*
Y 0** 0** 0** z 28. 4 27. 9
31.0 30.6 31. 5
moment of inertia
moment of inertia
X 78.0 66. 1
x t 115.0 107. 7 108. 1 y 85. 7 86.5
Y* 103.0 104.3 103, 9 43. 0 34. 8
z * 11.3 6.8 8. 0

95th percentile male - standing


measured from back of torso
symmetry about the sagittal plane assumed
center of gravity measured from top of head
A-P direction-negative x
to the right of the sagittal plane - positive y
S-I direction-positive z

ent of inertia
146. 8 141, 8 151. 5 specific gravity p that were used are given in Tables 6 and 4,
132.4
14, 9
135. 8
11.0
143. 2 respectively, and using values of a, 6, and c, the ellipsoid semi-
axes of the head segment, that were calculated from Sahley data,
Note: The momenta of inertia tabulated are with respect to the corresponding
Table 3, using the algorithms of the present model.
centers of gravity of the whole body. The average value of the moment of inertia (about the principal
axis perpendicular to the sagittal plane) of these 13 heads, as
reported by Hodgson, et al., was 0.20 lb-sec2-in., which is within
on a 50th percentile Sierra 292-1050 anthropometric dummy 3 percent of the value of 0.194 lb-sec2-in. for the moment of
in related research, references [4 and 5], were included in the inertia of the head of a 50th percentile male calculated from the
comparison. The moment of inertia3 of a segment n in the sagittal algorithms of the present model. This calculation of moment of
plane (erect standing position), <p„u, is given by inertia <j>y for the head segment was performed using equation
M„ (2) and the calculated values of it, a, and c previously discussed.
0n„ = -~ (0„J + C„2) (2) Becker [23], presents measured principal moments of inertia
O
on several cadaver head segments of varying sizes. The experi-
where the segment masses are given by mental results for male subject number 2986, [23], were used
for this comparison, because the measured head weight of the
M„ = - wR„p„a„b„cn. (3) subject, 9.72 lb, was closest to the calculated head weight
o of 10.7 lb for a 50th percentile male. This calculation of prin-
cipal moments of inertia <j>„x, tj>„v, and (j>„x for the head seg-
Using the set of dimensional measurements reported by
ment was performed using the expressions
Sahley for a 50th percentile male, a total body weight of 164 lb,
the specific gravity data of Table 4, and R„ values of Table 6, M
moments of inertia of individual segments <t>ny were calculated <t>ni = —" (6nJ + C»)
5
from equations (2) and (3) and were compared to the correspond-
ing measured values. In addition, moments of inertia of selected M„
groups of connected segments were .calculated from the $„„ and 0n„ = — (On* + C„s)
O
M„ values and compared to their corresponding measured values.
These comparisons are presented in Table 7. The agreement is Mn
again considered quite satisfactory, in view of the limited «*. = W» + &„')
available experimental data.
Additional comparisons of model predictions with more recent and the calculated values of M, a, b, and c previously discussed.
experimental results have also been made. Hodgson, et al. [22], The corresponding measured and calculated principal moments
presents inertial data on 13 male cadaver heads, which are con- of inertia are presented in Table 8.
sidered to be a useful supplement to the inertial data previously The agreement between the measured values and the cor-
discussed. The average weight of these heads reported by Hodg- responding prediction is considered reasonable, because a single
son, et al., was 10.0 lb, which is within 7 percent of the 10.7-lb male subject is, at best, only a first approximation to a 50th
head weight of a 50th percentile male, calculated from the percentile male.
algorithms of the present model. This calculation for head mass As a further, and perhaps more critical, test of prediction of
M was performed using equation (3), where values of R and the model, results of measurements of whole body centers of
gravity and moments of inertia of living male subjects in various
positions, reported by Santschi. et al. [20], were utilized in this
•TJnleas otherwise stated, moments of inertia are referenced to the corres-
ponding center of gravity. study. Whole body centers of gravity and moments of inertia

54 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


were computed from the present model for two positions, stand- Table 10 Summary of four trial cases (legend-numbers inside
circles indicate inputs to program)
ing and sitting, and these predicted values were compared to
STANDING SEATED
Santschi's measured results. In addition, comparisons were HEIGHT HEIGHT WEIGHT
made with the predictions of a man-model developed by Hana- INCHES PERCENTILE INCHES PERCENTILE POUNDS PERCENTILE

van [9] for computing whole body inertial properties. Results


of these comparisons are presented in Table 9 for four subjects:
64.5

(66.4)
( 9

26
) 34.6

(M,6J
© 73 167
0 (nj

50th percentile male—standing


95th percentile male—standing
74.S

(75.2)
(We)

99.4
39.0

(^3)
©87
nia)

(264)
66

99.8

50th percentile male—sitting


95th percentile male—sitting. Table 11 Measurements on subject JB
Ratio,
The predictions of 50th percentile male properties from the Measured/
Di m e n a i o n a l D a t a Computed Measured Computed
present model were obtained from a series of computations using
the calculated segment masses Mn, segment moments of inertia L 1 Standing Height 66.40 66.4 1.00
<frn and link dimensions obtained in the manner previously L 2
L 3
S h o u l d e r Height
A r m p i t Height
5 7.69
53.55
55.8
50.0
0.97
0.93
outlined. The 95th percentile properties were obtained in a L 4 Waist Height 42.67 36.5 0.86
L 5 S e a t e d Height 36.60 36.6 1. 00
similar manner, using values of the mean and standard devia- L 6 Head Length 7.82 8.2 1.05
L 7 Head B r e a d t h 6. 13 6. 1 1.00
tion of the set of dimensional measurements of Sahley, Table 3, L 8 Head to C h i n H e i g h t 9 . 15 9.0 0.98
to compute a set of dimensional values Li for the 95th percentile L 9
L10
Neck C i r c u m f e r e n c e
Shoulder Breadth
14.94
18. 19
15.0
17.6
1.00
0.97
male, as previously discussed. Values of p„ and B„ for the 95th Lll Chest Depth 9. 18 9.3 1.01
L12 Cheat Breadth 12.95 13.2 1.02
percentile cases were obtained from Tables 4 and 6, respectively. L13 Waist Depth 8.47 8.2 0.97
L14 Waist Breadth 11.11 11.4 1.03
The results of Santschi, et al., are presented in terms of mean LIS Buttock Depth 9 . 13 9.8 1.07
L16 Hip B r e a d t h , S t a n d i n g 13.95 13.8 0.99
values and standard deviations for 66 subjects. They were L17 S h o u l d e r to E l b o w L e n g t h 14.78 13.8 0.93
F o r e a r m - H a n d Length 19. 17 18.3 0.96
converted to the form shown in Table 9 by assuming that the L18
L19 Bicep3 C i r c u m f e r e n c e 12.96 13.0 1.00
mean is approximately equal to the 50th percentile, and by LZ0
L2 1
Elbow C i r c u m f e r e n c e
Forearm Circumference
12.37
1 1.45
11.7
11.7
0.9S
1.02
assuming a Gaussian distribution to calculate the 95th percentile L22 Wrist Circumference 7.03 7.2 1.02
L2 3 Knee Height, Seated 20.92 19.9 0.95
values from the values of mean and standard deviation pre- L24 Thigh C i r c u m f e r e n c e 19.89 21.0 1.06
L2 5 Upper Leg C i r c u m f e r e n c e 14.66 15.0 1.02
sented by Santschi, et al. L2 6 Knee C i r c u m f e r e n c e 14.25 14.6 1.02
L2 7 Calf C i r c u m f e r e n c e 13. 12 14. 0 1.07
It can be seen from Table 9 that the primary parameters of L2 8 Ankle C i r c u m f e r e n c e 8.29 8.5 1.03
interest, the whole body moments of inertia about the x and y L29
L30
Ankle Height, Outside
Foot Breadth
2 . 72
3.68
2.5
3.9
0.92
1.06
axes (defined in Table 9), are predicted with good accuracy L31 Foot Length 10.00 10.3 1.03

(typically a few percent) by both the present model and the


Hanavan model for the four configurations considered. The
calculated moment of inertia about the z axis for both models is
consistently too small, with the present model differing most relatively small difference between erect and normal seated
from the experimental results. This difference is discussed in height, typically less than 2 in. A tabulation of these measure-
some detail in reference [21]. Except for the center of gravity ments was reviewed, and four subjects were selected, using the
in the x direction for the standing position, the center of gravity widest variation of parameters as the primary guide in the selec-
is, in general, predicted with good accuracy by the present model. tion process. Data (standing height, seated height, and weight)
measured for these four male subjects were used as program
A parametric study was performed to test the algorithms that
inputs.
define the ellipsoidal man-model, and at the same time to obtain
. an overall check of the coding of the computer program. To These cases are summarized in Table 10, together with the
check the program in an efficient manner, its tabular output selected form of inputting them to the computer program, as
was inputted to the plotter graphics program that was developed indicated by the values that have been circled. The set of values
for the Three-Dimensional Crash Victim Simulation program that are circled covers all available combinations of primary
[4]. The plotter program produces results in a form that is more program inputs for male subjects. It can be seen from this
readily comprehended than a data tabulation. Emphasis in tabulation that the primary variables range from about the 1st
this study was placed on testing each of the primary program to the 99th percentiles, with some combinations of the three
options, using inputs that represent wide variations in human primary parameters that should critically test the algorithms.
stature and weight. In the interest of brevity, only the results One of these cases was used to assess a rather basic assumption
from the four male trial cases considered are presented herein:4 applied in the calculation of segment properties. This assumption,
2 standing height/seated height input options X 2 weight input or more properly, rule is as follows:
options = 4 cases. If the standing height and seated height percentiles differ,
This matrix of four trial cases covers all primary program the algorithm computes head, neck, torso, and arm dimensions
options for male subjects, since both heights and weight may be corresponding to the seated height percentile, and leg and
inputted as either a dimension (inches or pounds, respectively), foot dimensions corresponding to the standing height per-
or as a percentile. Relaxed standing height, erect seated height, centile.
and weight data were measured for a number of males at Cals- This rule was adopted because it seemed reasonable, and be-
pan. All subjects were selected with emphasis placed on ob- cause no obvious alternate rule was suggested by the review of
taining wide variations on these three parameters. The erect published anthropometric data. To assess this rule, one of the
seated height, rather than normal seated height was measured, trial cases, subject JB, was studied in some detail. This case was
because erect height is probably better defined than normal selected because the difference between the measured standing
seated height and because it is more commonly measured in height percentile (26) and the measured seated height percentile
anthropometric surveys. It should be noted that there is a (73) was the greatest among the trial cases and, consequently,
should provide the most critical assessment of the rule.
*Results from an additional four trial cases for females are documented
The value of each of the dimensional parameters listed in
in reference [211. Table 2 was measured for subject JB, and the results are pre-

Journal of Engineering for Industry FEBRUARY 1975 / 55

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 6 Plotter graphics displays of four trial cases

sen ted in Table 11, together with the corresponding program (1) inputting his own computational tables
predictions. In this tabulation, LI, the standing height, and L5, (2) using the internal computational tables
the seated height were inputs to the computer program, so that (3) using the data library.
corresponding "computed" and measured values of those two (C) Parameters for inputting sex, standing height, seated
parameters are identical. It can be seen from Table 11 that height, weight.
the maximum difference between measured and computed re- The following output capabilities have been incorporated:
sults is 14 percent, and a simple calculation shows that the (A) The title of the case.
average difference for all 31 parameters is only about 4 percent. (B) Echo of input data (documented).
Although this comparison, with measurements obtained on a (C) A tabular listing of the dimensional data and weight
single subject, cannot be considered an absolute test of the rule tables which were used, along with titles.
previously discussed, it does indicate that* for the case con- (D) A tabular listing of all the 3-D output parameters (op-
sidered, the rule is reasonable, and there is no reason to expect tional).6
that the rule will result in errors that are excessive for practical
engineering calculations.
The plotter graphics displays, corresponding to the four trial Table 12 Primary input options
cases summarized in Table 10, are shown in Fig. 6. sex male or male or male or male or
In spite of the vast differences in program inputs for these female female female female
four cases, the algorithms are considered to generally behave standing height percentile percentile dimension, dimension
satisfactorily. It is also evident from these plots that extreme inches inches
differences between inputted standing and seated heights could seated height percentile percentile dimension, dimension
result in unrealistic geometrical configurations. Consider, for inches inches
example, the plot of subject JB (26th percentile standing height, weight percentile value, percentile value,
73rd percentile seated height). The shoulder height, calculated pounds pounds
for the seated (73rd) percentile, is somewhat inconsistent with
the standing height (26th percentile). As previously discussed,
a detailed comparison between computed and measured di- Table 13 Printout of input
mensions for subject JB is given in Table 11. sex m a l e or female

standing height p e r c e n t i l e and dimension, inches


Computer Program Summary seated height p e r c e n t i l e and dimension, inches

weight p e r c e n t i l e and value, pounds


To provide maximum flexibility for the program user and also
to simplify the input of anthropometric data published by many
different investigators in a variety of forms, the following input
options, Table 12, have been incorporated in the program. Table M Anthropometric data set options
Regardless of the option chosen, however, the printout of (male and female)
input includes the following, Table 13. For each data table:
In addition, the following options have been incorporated dimensional data, L,-: dimensional data as a tabular func-
for the anthropometric data sets or tables, Table 14. tion of percentile—15 entries per
For the data library, a collection of previously computed or table (see Table 2)
and
gathered output data sets, a total of 17 data sets are available, 9 mean and standard deviation of each
corresponding to inputs for the 3-D Program, and 8 correspond- dimensional measurement
ing to the ROS/MODROS Program. weight data: weight data as a tabular function of
The following input capabilities have been incorporated: percentile—15 entries
and
(A) Input title for each run. mean and standard deviation of
(fi) Input parameters to indicate whether user is: weight

56 / FEBRUARY 1975 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


(E) A tabular listing of all the BOS/MODROS output pa- 2 Danforth, J. P., and Randall, C. D., "Modified ROS
rameters (optional).5 Occupant Dynamics—Simulation User Manual," Research Pub-
lication GMR-1254, Research Laboratories, General Motors
(F) Punched deck of all the 3-D output parameters (op- Corporation, Oct. 1972.
tional).5 3 Bartz, J. A., "A Three-Dimensional Computer Simulation
(G) Punched deck of all the ROS/MODROS output pa- of a Motor Vehicle Crash Victim—Phase 1—Development of
rameters (optional).6 the Computer Program," CAL Report No. VJ-2978-V-1, PB
No. 204172, July 1971.
Certain output features have been included to aid the user 4 Bartz, J. A., and Butler, F. E., "A Three-Dimensional
with the various input options provided. To serve as a con- Computer Simulation of a Motor Vehicle Crash Victim—Phase
venient reference, at the beginning of a run containing one or 2—Validation Study of the Model," Calspan Report No. VJ-2978-
more cases, a list of all input cards and the various options on V-2, Dec. 1972.
5 Bartz, J. A., "Validation of a Three-Dimensional Math-
each card is output by the program. In addition, the options ematical Model of the Crash Victim," Proceedings of the Human
chosen for each particular run are listed along with card images Impact Response Symposium, General Motors Research Labora-
and explanations. All tables used are output, and description of tories, Warren, Mich., Oct. 1972.
the simulated crash victim and all output parameters are listed 6 Bartz, J. A., "Development and Validation of a Computer
Simulation of a Crash Victim in Three Dimensions," SAB Paper
in a tabular form that is designed for ease of comprehension. No. 720961, Proceedings of the Sixteenth Stapp Car Crash Con-
In addition, the computed set of dimensional data Li cor- ference, Detroit, Mich., Nov. 1972. Also 1972 SAE Transactions,
responding to the inputted heights and weight is tabulated. Society of Automotive Engineers, New York, 1972 (in press).
7 Lepley, D. A., "A Mathematical Model for Calculating
the Moments of Inertia of Individual Body Segments," GM
Report TR67-27, May 1967.
Conclusions 8 Patten, J. S., and Theiss, C. M., "Auxiliary Program for
The model predicts body segment link lengths, surface di- Generating Occupant Parameter and Profile Data, CAL Report
No. VJ-2759-V-1R, Jan. 1970.
mensions, masses, and moments of inertia with acceptable ac- 9 Hanavan, E. P., "A Mathematical Model of the Human
curacy for most practical engineering calculations. This con- Body," AMRL-TR-64-102, Oct. 1964.
clusion is based on the fairly detailed comparisons between 10 "Cockpit Geometry Evaluation," JANAIR Report
model predictions and measured whole body and body segment 690104, Boeing Report No. D162-10128-1, Jan. 1969.
11 Kroemer, K. H. E., "COMBIMAN: COMputerized
dimensional and inertial properties that have been presented. Blomechanical MAN-model," AMRL-TR-72-16, 1972.
To insure acceptable results, the model should not be applied 12 Chaffin, D. B., and Snyder, R. G., "New Advances in
far beyond its intended range, , which is nominally between the Volitional Human Mobility Simulation," Proceedings of the
1st and 99th percentile of anthropometric parameters. In Human Impact Response Symposium, General Motors Research
Laboratories, Warren, Mich., Oct. 1972.
addition, combinations of input parameters that are unlikely 13 Conway, B. A., "Development of Skylab Experiment
in a large sample of a population, such as extreme differences in T-013 Crew/Vehicle Disturbances," NASA TN D-6584, Jan.
standing height percentile and seated height percentile, should 1972.
not be used, to avoid results that may be ambiguous and un- 14 Sahley, L. W., Dimensions of the Human Figure, Cleveland
Designers and Consultants, Inc., Cleveland, Ohio, 1957.
realistic. 15 Stoudt, H. W., Damon, A., McParland, R. A., and
Computer program anthropometric tables and inputs should Roberts, J., "Weight, Height, and Selected Body Dimensions of
be updated as more extensive experimental data become avail- Adult Humans," Public Health Survey Publication No. 1000,
able. Provision has been made within the program for this up- Series 11, No. 8, June 1965.
16 Damon, A., Stoudt, H. W., and McFarland, R. A., The
dating process. Human Body in Equipment Design, Harvard University Press,
Cambridge, Mass., 1966.
17 Clauser, C. E„ Tucker, P. E., McConville, J. T., Churchill,
Acknowledgments E., Laubach, L. L., and Reardon, J. A., "Anthropometry of Air
Force Women," AMRL-TR-70-5, Apr. 1972.
Special acknowledgment is due to Messrs. Leonard Garr and 18 Stoudt, H. W., "Anthropometry for Child Restraints,"
Seymour Samet for their advice and supervision on the pro- Harvard School of Public Health Report, DOT No. HS-800 535,
gramming effort. In addition, the comments and suggestions PB No. 204186, July 1971.
19 Drillis, R., and Contini, R., "Body Segment Parameters,"
of the Analytical Model Simulation Subcommittee of the Motor NYU Technical Report No. 1166.03, Sept. 1966.
Vehicle Manufacturers Association have been most helpful. 20 Santschi, W. R., DuBois, J., and Ompto, G., "Moments
Thanks are also due to Mr. Norman DeLeys for editing the of Inertia and Centers of Gravity of the Living Human Body,"
manuscript. AMRL-TDR-63-36, May 1963.
21 Bartz, J, A., and Gianotti, C. R., "A Computer Program
to Generate Input Data Sets for Crash Victim Simulations,"
References Calspan Report No. ZQ-5167-V-1, Jan. 1973.
22 Hodgson, V. R., Mason, M. W., and Thomas, L. M.,
1 Segal, D. J., "Revised Computer Simulation of the Auto- "Head Model for Impact," SAE Paper No. 720969, Proceedings
mobile Crash Victim," CAL Report No. VJ-2759-V-2, Jan. 1971. of the Sixteenth Stapp Car Crash Conference, Detroit, Mich., Nov.
1972.
23 Becker, E. B., "Measurement of Mass Distribution Pa-
rameters of Anatomical Segments," SAE Paper No. 720964,
B
Program output consists of body segment weights, principal moments Proceedings of the Sixteenth Stapp Car Crash Conference, Detroit,
of inertia, link dimensions, and contact Burface dimensions. Mich., Nov. 1972.

Journal of Engineering for Industry FEBRUARY 1975 / 57

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 12/13/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like