Pergamon .I Rmwchanrcs, Vol. 28. No. 6. pp 733 75%.

{i/9$
Copyright Q 1995 Elscwer Saencc Lta
Printed in Great Britain. All rights rrwved
0021~ 9290’95 s9 50 4 !b.W!

0021-9290(94)00124-3

TECHNICAL NOTE

KINETIC ANALYSIS OF THE LOWER LIMBS DURING WALKING: WHAT

INFORMATION CAN BE GAINED FROM A THREE-DIMENSIONAL MODEL?

Janice J. Eng and David A. Winter

Dept of Kinesiology, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada

Abstract-Kinetic analyses (joint moments, powers and work) of the lower limbs were performed during
normal walking to determine what further information can be gained from a three-dimensional model over
planar models. it was to be determined whether characteristic moment and power profiles exist in the frontal
and transverse planes across subjects and how much work was performed in these planes. Kinetic profiles
from nine subjects were derived using a threedimensional inverse dynamics model of the lower limbs and
power profiles were then calculated by a dot product of the angular velocities and joint moments resolved in
a global reference system. Characteristic joint moment profiles across subjects were found for the hip, knee
and ankle joints in all planes except for the ankle frontal moment. As expected, the major portion of work
was performed in the plane of progression since the goal of locomotion is to support the body against gravity
while generating movements which propel the body forward. However, the results also showed that
substantial work was done in the frontal plane by the hip during walking (23% of the total work at that
joint). The characteristic joint profiles suggest defined motor patterns and functional roles in the frontal and
transverse planes. Kinetic analysis in three dimensions is necessary particularly if the hip joint is being
examined as a substantial amount of work was done in the frontal plane of the hip to control the pelvis and
trunk against gravitational forces.

INTRODUCTION 1798cm, range 162-186) participated in this study. ‘The

Sagittal plane kinetic analysis has been a useful tool in
understanding the mechanisms of normal and pathological
gait (Olney et al., 1991; Winter et af., 199Oa). However, the
sagittal view provides only part of the information, parti-
cularly at the hip joint where hip abductors are critical for the
balance control of the trunk in the frontal plane (MacKinnon
and Winter, 1993). Furthermore, the accuracy of motion
description especially of the hip joint, deteriorates as the
number of degrees of freedom considered is reduced due to
simplifying assumptions (Cappozzo and Gazzani, 1990). Re-
cently, a few studies have presented individual subject pro-
files of joint moments in three dimensions during walking
(Apkarian et al., 1989; Kadaba et al., 1989) but only one study
has examined the inter-subject variability found in the lower
limb joint moments (Andriacchi and Strickland, 1985). No
studies have extended the power and work analysis using a
threedimensional model. This study addressed the following
questions: (a) Does a characteristic profile exist for three-
dimensional joint moments and powers across subjects dur-
ing walking? (b) How much work is performed by the lower
limbs in the transverse and frontal planes?(c) Are the sagittal
profiles from three-dimensional analysis similar to those
reported from planar analysis?
MEmODS AND PROCEDURES
Nine male subjects (mean age of 22.2 yr, range 19-26;
mean mass of 77.2 kg, range 63.0-84.3: mean height of
Received in jinulform 27 July 1994.
procedures employed were approved by the University of
Waterloo Offloe of Human Research. Subjects walked along
an 11 m walkway at a self-selected pace wearing their run-
ning shoes. The principal optical axis of a frontal video
camera was set perpendicular to two sag&al video cameras
and parallel to the ground (Miller and Petak, 1973). Marker
coordinates and forceplate data were collected for one stride
at 60 and 500 Hz, respectively, and filtered using a fourth-
order Butterworth, zero-lag, low-pass cut-off at 6 and 30 Hz.
respectively (Winter, 1990). Multiple sets of two-dimensional
data were used to reconstruct the three-dimensional co-
ordinates. The RMS error of determining the three-dimen-
sional location of a marker in a 3 x 2 x 2 m space was less
than 0.5 cm (Jian et al., 1993).
A standing posture of the subject was used to define the
transformation matrix between the external marker reference
system and the principal axes of each body segment. The
principal axes of the pelvis and feet were parallel to the global
reference system where X = anterior/posterior (API. 1
= vertical, and 2 = medial/lateral (ML). The principal !‘
axes of the thigh and shank extended from the hip to knee
joint, and knee to ankle joint, respectively. The principal I
axes of the thigh and shank were parallel to the global 2 axis
while the principal x axes were the cross product of the y and
z axes. The hip joint was located 30% of the inter-ASPS
distance distal to the ASIS (verticalk 36% of the inter-ASIS
distance lateral to the pelvis centre of mass (ML) (Bell er al.,
1990) and aligned with the suwrior point of the iliac crest
(AP): The knee joint was midway *between the femoral
condyles, 2.5 cm distal to the lateral femoral condyle. while
the ankle joint was midway between the malleoli. 1 cm distal
to the lateral malleolus.
Three non-collinear markers were used to track each
segment (bilateral lower limbs and pelvis). Cardan angles (an

753
154 Technical Note

x-y-z rotation sequence) were then calculated. A three- RESULTS

dimensional inverse dynamic solution (Bresler and Frankel,
1950) was performed commencing with the most distal joint
(ankle). Mechanical joint powers were calculated from the
dot product of the joint angular velocities and joint moments
transformed in the global reference system (Robertson and
Winter, 1980). The time integral of the power curves (i.e.
work) was calculated (a) over one stride and (b) for each
power phase (two consecutive zero crossings). The work done
under each power phase was labelled by the joint (H = hip, K
= knee and A = ankle) followed by the plane@ = sagittal, F
= frontal and T = transverse) (eg. HI-S is the work done by
the hip joint for one of the power phases in the sagittal plane).
The total work over the stride was the sum of the positive
work and the absolute value of the negative work. Joint
moment and power profiles were normalized on a time base
of 100% stride and were ensemble-averaged across trials and
subjects.
Subjects walked at a mean velocity of 1.6 ms- ’ (range
1.4-1.8) and at a cadence of 108 stepsmin-’ (range 98-115).
The major proportion of the lower limb work was done at
the hip joint with a large amount of work being done in the
sagittal plane (74% of the total hip work) and a moderate
amount (23%) in the frontal plane. In the sagittal plane, a
large HI-S generation (mean + S.D.; 0.3 + 0.1 J kg-‘) con-
trolled the trunk and collapse of the stance limb, an H2-S
absorption ( - 0.12 + 0.053 J kg- *) decelerated the thigh
extension and an H3-S ‘flexor pull-of?’ generation (0.14
+ 0.06 J kg-‘) added energy to the swinging limb [Fig. l(f)].
The large stance hip abductor moment countered the upper
body which was medial to the stance hip [Fig. l(a)]. The hip
power phases were a result of the hip abductor muscles
controlling the pelvis; an Hl-F absorption ( - 0.081
f 0.05 J kg-‘) was an eccentric control of the dropping

0.76

f
Frontal a

0.00

-0.40 la. -1.50

1
0 20 40 50 50 100 0 20 40 50 50 100

f -0.550 Hi-T
I
0 20 40 50 DO 100

2.1

5
u

0.0

t 1
0 20 40- 5; DO
' 100
0 20 40 50 50 100

X of Strld, X of Stride

Fig. 1. Hip joint moments and joint powers normalized to body mass in the frontal, transverse and sagittal
planes. The dashed line indicates one standard deviation above and below the mean. Consistent hip moment
and power profiles were found in all planes except for the transverse joint power.
 Technical Note

Table 1. Means and standard deviations of the work (J kg- ‘) over one stride

Sagittal Frontal Transverse

Total Total
gent abs gen abs en abs f3en abs Total

Ankle 0.42 0.10 0.015 0.014 0.005 0.006 0.44 0.12 0.56
(0.09)* (0.05) (0.008) (0.007) (0.003) (0.002)
Knee 0.11 0.35 0.035 0.026 0.008 0.015 0.15 0.39 0.54
(0.052) (0.094) (0.013) (0.012) (0.004) (0.01)
Hip 0.48 0.14 0.092 0.10 0.009 0.022 0.58 0.26 0.84
(0.12) (0.053) (0.03) (0.05) (0.005) (0.013)

*Standard deviations in parentheses, N = 9 subjects.

tgen = energy generation, abs = energy absorption.
Note: The hip joint performs substantial work in the frontal plane

Frontal

‘;fj(2-F 2d.
-030
I
0 20 40 00 90 100

Transverse

1.2 ‘,U,O-s
,,’‘: s
k Sagittrl
: : u
8 0 ‘8
0.0 -

-0.75 2c. -2.1 K&S 2f.

I 1
0 20 A0 00 80 100 0 20 40 90 90 100

X of Slrldr X of Strldr

Fig. 2. Knee joint moments and joint powers normalized to body mass in the frontal, transverse and sagittal
planes. The dashed line indicates one standard deviation above and below the mean. The dominant power
profile was in the sagittal plane.
156 Technical Note

pelvis during weight acceptance while the HZF generation
(0.031 k 0.02 J kg-‘) and H3-F generation (0.057
+ 0.02 J kg-‘) raised the pelvis [Fig. l(d)]. In the transverse
plane, the hip moment profiles were consistent across sub-
jects with an external rotator moment in the first half of
stance and an internal rotator moment in the second half of
stance [Fig. l(b)]. A small absorption burst (Hl-T) during
weight acceptance was a result of the external rotators which
decelerated the forward rotation of the pelvis [Fig. l(e)].
At the knee most of the work was done in the sagittal plane
(85% of the total knee work); however, a small amount (11%)
was done in the frontal plane (Table 1).The KO-S generation
(0.039 +_ 0.03 J kg- ‘) resulted from a small flexor action at
heel contact to ensure that the knee was in a flexed position
to absorb the impact force. The knee extensors controlled
the knee flexion (Kl-S absorption: - 0.04 f 0.3 Jkg-‘)
and then extended the knee (K2-S knee generation: 0.31
+ 0.3 Jkg-I). A K3-S absorption phase ( - 0.13
+_0.06 J kg-‘) resulted from the knee extensors to control
the collapsing knee while the knee flexors absorbed energy
(K4-S - 0.14 +_0.3 J kg-‘) to decelerate the swinging leg
[Fig. 2(f)]. During the stance phase, a stabilizing knee abduc-
tor moment in the frontal plane (via the iliotibial tract, tensor
fascia latae, and passive forces from the ligaments) countered
the adductor stress from the upper body weight passing
medial to the knee [Fig. 2(a)]. The knee angular velocity in
the frontal plane was very low which resulted in power
phases which were less than 0.030 J kg-‘. The transverse
plane knee profile was characterized by an external rotator
moment during the first half of stance and an internal rotator
moment during the second half [Fig. 2(b)]. These moments
resulted from the knee ligaments which reacted to the
transverse rotation generated by the active hip moments. A
small but consistent power absorption (Kl-T) occurred
during weight acceptance as a result of the passive resistance
of the knee ligaments to the internal rotation motion which
accompanies knee flexion [Fig. 2(e)].
At the ankle, the work was done mainly in the sagittal plane
(93%). The plantarflexors eccentrically controlled the for-
ward rotation of the leg over the foot (Al-S absorption:

ODO

3d
-0.15 1 , I , I , -0.2s -I
I f
0 20 40 80 80 100 0 20 40 80 80 100

0.010 ,

0.000

-0.050 ,
-3

0 20 40 10 80 COO 0 20 40 00 80 100

8.0 , AZ-S
2.00 -9
,",

-0.50 - 3C.
I 1
20 40 60 10 100
;, 2b 40' 80' io 160 0

% of Strld,
X of Strldr

Fig. 3. Ankle joint moments and joint powers normalized to body mass in the frontal, transverse and
sagittal planes. The dashed line indicates one standard deviation above and below the mean. Large
variability was found in the frontal moment and frontal and transverse powers.
 Technical Note
- 0.096 & 0.04 J kg- ‘) and then concentrically generated a
rapid push-off (A2-S generation: 0.39 + 0.082 J kg-‘) which
was the single largest generation phase of all the joints
[Fig. 3(f)]. In the frontal plane, all subjects had an evertor
moment at heel contact. Eight of the subjects then switched
to an invertor moment during midstance. During the propul-
sive phase, seven of the subjects exhibited an evertor moment
while two exhibited an invertor moment. The power phases
were small and highly variable for both the transverse and
frontal planes [Fig. 3(d) and (e)].
DISCUSSION
There are a number of variables (e.g. temporal, kinematic,
kinetic) that may be used to identify a characteristic pattern
during gait, but the mechanical power profile is the only
single variable which reveals the functional role of the
anatomical structures as they shorten or lengthen under
tension. Furthermore, the time integral of each power phase
(i.e. work) is a unique variable in that it quantifies, with a
single value, specific functions during the gait cycle.Although
work and power are scalar terms, the work was partitioned
into three arbitrary planes to facilitate the functional inter-
pretation of specific events. A limitation of this technique is
that the work calculations will be underestimated if there is
simultaneous activity of the agonist and antagonist muscle
groups (i.e. co-contraction).
One would expect some variability of the results between
different investigators because of the location and orientation
of the principal axes of each segment and the location of the
joint centres. Nevertheless, the joint moment profiles were
consistent among the nine subjects and in general agreement
with other investigators (Andriacchi and Strickland, 1985;
Apkarian et a[., 1989). The exception was the ankle frontal
moment. The maximum ankle frontal moment in this study
was 1.5Nm compared to the 35-45 Nm magnitudes re-
ported by other investigators (Andriacchi and Strickland,
1985; Apkarian et al., 1989; Kadaba et al., 1989). Since the
vertical ground reaction force virtually dominates the ankle
frontal moment during stance and the centre of pressure is
limited to about + 3 cm within either side of the ankle joint,
the maximum frontal moment cannot exceed 24 N m. in a
subject who generates a vertical ground reaction force of
800 N. Larger values may be a result oferrors in the location
of the ML joint centre or the centre of pressure. A 1 cm error
in the ankle joint location can reverse the polarity of the
moment arm of the centre of pressure to the ankle joint
centre. Some of the variability of the ankle moments may also
be a result of anatomical differences in foot structure, the
amount of toeing in or out, or the initial foot placement.
MacKinnon and Winter (1993) found that the ankle frontal
moments reflect a control strategy to fine tune the medial
acceleration of the body mass during single support.
The sagittal work values were similar to the results of
planar studies which used similar cadences (Winter, 1983,
1991; Winter et al., 1990b) except the HI-S and H2-S work
values were larger than those previously reported. These
discrepancies are probably due to the fact that this study
examined the thigh segment relative to the pelvis segment
while previous planar studies examined the thigh segment
relative to the trunk segment. Minor differences may also be
a result of projection errors which can occur in a planar
analysis or from the fact that the planar studies used joint
moments calculated about the greater trochanter. The hip
joint centre reconstructed from the anatomical landmarks
selected in this study was a maximum of 1.9 cm posterior to
the greater trochanter in the first half of stance, and 1.8 cm
anterior to the greater trochanter in the second half. Hence,
the estimated hip joint coordinate would result in larger
moment, power and work values. The relationship we found
between the trajectory of the hip joint centre and a greater
‘?i’
trochanter marker during walking concur with the results by
Cappozzo and Gaxxani (1998) who compared a greater
trochanter marker to a hip joint centre identified by X-ray
techniques.
Characteristic joint moment profiles were evident in the
frontal and transverse planes during normal walking which
indicate defined motor patterns and functional roles in these
planes. As expected, the major portion of work was pcr-
formed in the sagittal plane since the goal of locomotion is to
support the body against gravity while generating move-
ments which propel the body forward in the plane of
progression. However, kinetic analysis in three dimensions is
necessary particularly if the hip joint is being examined as a
substantial amount of work was done in the frontal plane of
the hip to control the pelvis and trunk. For example, the H l-
F and H2-F phases will be important factors to quantify in
patients with osteoarthritis of the hip or hip arthrophtsty
where weak hip abductors are currently visually diagnosed as
a Trendelenburg. Although the power in the transverse plane
was low in normal subjects, such measures have potential for
elucidating some of the features in pathological gait patterns
Acknowledgements-This research was supported by MRC
# MT4343 to D. Winter. The technical assistance of M. Ishac
and P. Guy was greatly appreciated.
REFERENCES
Andriacchi, T. P. and Strickland, A. B. (1985) Gait analysis as
a tool to assessjoint kinetics. In Biomechanics oj~ormal
and Pathoiogicaal Human Articulating Join& NATO SI
Series (Edited by Berme, N., Engin, A. E. and Correia Da
Silva, K. M.), pp. 83-102. Martinus Nijho8, Dordrecht.
Apkarian, J., Naumann, S. and Cairns, B. (1989) A three-
dimensional kinematic and dynamic model of the lower
limb. J. Eiomechanics 22, 143-155.
Bell, A. L., Pedersen, D. R. and Brand, R. A. (1990) A
comparison of the accuracy of several hip centre location
prediction methods. J. Biomechanics 23,617-621.
Bresler, B. and Frankel, J. (1950) The forces and moments in
the leg during level walking. Tmns. ASME 72, 27-36.
Cappoxxo, A. and Gaxxani, F. (1990) Joint kinematic assess-
ment during physical exercise. In Biomechanics of Human
Movement: Applications in Rekubilitation, Sports and Er-
gonomics (Edited by Bernie, N. and Cappoxzo, A.).
pp. 263-274. Bertec Corp., Worthington, Ohio.
Jian. Y.. Winter. D. A.. Ishac. M. G. and Gilchrist. L. (1993)
Trajectory of the body C&I and COP during’initiatton
and termination of gait. Gait Posture 1, 9-22.
Kadaba, M. P., Ramakrishnan, H. K., Wootten, M. E.,
Gainey, J., Gorton, G. and Cochran, G. V. B. (1989)
Repeatability of kinematic, kinetic, and electromyography
data in normal adult gait. J. orthop. Res. 7, 849-860.
MacKinnon, C. D. and Winter, D. A. (1993)Control of whole
body balance in the frontal plane during human walking,
J. Biomechanics 26, 6333644.
Miller, D. I. and Petak, K. L. (1973) Three dimensional
cinematography. Kinesiology Ill. pp. 14.-19. AAHPER.
Washington, DC.
Olney, S. J., Griffin, M. P., Monga, T. N. and McBride, I. D.
(1991) Work and power in gait of stroke patients. Arch.
Phys.. Med. Rehabk 72, 3091314.
Robertson. D. G. E. and Winter. D. A. (19801 Mechanical
energy generation, absorption and transfer amongst seg-
ments during walking. J. Biomechanics 13,8&i-854,
Winter, D. A. (1983) Energy generation and absorption at the
ankle and knee during fast, natural and slow cadences.
Clin. Orthop. Rel. Res. 175, 147-154.
Winter, D. A. (1990) Biomechanics and Motor Control of
Human Movement (2nd Edn). Wiley, New York.
758 Technical Note

Winter, D. A. (1991) Biomechanics and Motor Control of its and Movement Organization (Edited by Winters, J. M.
Human Gait: Normal, Elderly and Pathological. University and Woo, S. L. -Y.), pp. 680-693. Springer, New York.
of Waterloo Press, Waterloo. Winter, D. A., Patla, A. E., Frank, J. S. and Walt, S. E. (1990b)
Winter, D. A., Olney, S. J., Conrad, J., White, S. C., Ounpuu, Biomechanical walking pattern changes in the fit and
S. and Gage, J. R. (1990a) Adaptability of motor patterns in healthy elderly. Phys. Ther. 70, 340-347.
pathological gait. In Multiple Muscle Systems: Biomechan-