View Full Version : joint forces in pelvis CS: summary

Mark Thompson
09-18-2000, 08:29 PM
Dear Biomech-l readers

Thank you to those who responded to my posting (original included at the

Ben Stansfield has transformed joint forces calculated from kinematic and
force plate data into a pelvis CS, and he emailed me data from several
subjects. These showed monotonic anterior motion of the reaction force,
heel strike between 50 and 20 degrees posterior to the vertical in the
sagittal plane and toe off between 0 and 20 degrees anterior to the

I had a reply from Dr Roumen Dobrev who had also noted the discrepancies
between the papers reporting joint forces in a pelvis CS. He suggested an
alteration to the optimsation routines used for calculating muscle forces,
to include a step optimising the forces in the bone taking account of
antagonistic action.

Hartmut Witte acknowledged that the inaccuracies arising from using noisy
kinematic data from one subject and noisy joint force data from another
were considerable. He suggested that the Pedersen paper remains the most
reliable source.

I am going to use a loading scheme that reflects both Pedersen's paper and
the data from Ben Stansfield, with joint forces that sweep from posterior
to anterior.


================================================== =======
Mark S Thompson PhD Student
IRC in Biomedical Materials, QMW College, London E1 4NS
================================================== =======

original posting:

Dear Biomech-l readers

I've been working recently with the FE model of the human pelvis developed
by Michel Dalstra (1993), using the joint reaction forces he calculated.
Starting with hip joint force data from Bergmann (1990) he transformed
into the pelvis CS accounting for flexion-extension angles only (measured
on a different subject).

I've compared these with more recently published work: Witte H et al.
(1997) Acta Anatomica 160 (4) 269-280 used the Bergmann force data
(EBL277) but transformed it using kinematic data (Zebris) from 4 different
subjects, choosing one for detailed analysis.

Pedersen DR et al (1997) J Biomech 30 (9) 959-965 used instrumented hip
implant data and kinematic data from the same subject. They used inverse
dynamics and non-linear optimisation routines on force plate measurements
taken during collection of the kinematic data to calculate a hip joint
force. This technique was validated by comparison with the measured hip
joint forces. In this paper they transformed the calculated hip force into
a pelvis co-ordinate system using the same kinematic data.

There are some differences between these hip joint reaction forces, though
they all point medially and superiorly :).

The Dalstra forces all point anteriorly and move monotonically more
anterior during the stance phase.

At heel strike the Witte forces point posteriorly, but at the first
(largest magnitude) force maximum they point anteriorly, at the second
maximum posteriorly and then at toe off anteriorly.

All the forces calculated by Pedersen point posteriorly, moving
monotonically anteriorly from heel strike to toe off. Pedersen also finds
that the second peak force is larger than the first.

The monotonic anterior motion of the joint reaction force in a pelvis CS
seems intuitively correct - the joint reaction force stays relatively
constant in direction in a femoral CS and the femur-pelvis relative motion
seems to be smooth during stance phase.

Obviously the simplicity of the transformation Dalstra used will introduce
some errors, but it is hard to see how two groups starting with similar
data can produce such different results.

I know very little kinematics but am trying to decide upon a scheme of
loading (in a pelvis CS) for the above mentioned FE model of an implanted
pelvis. I understand that the variability of component position, hip
geometry and gait patterns will of course produce a wide variation in
joint reaction force, but would like to use loads that at least represent
the case in one individual with some idea of the worst cases.

I would appreciate any comments on the differences between the papers I've
mentioned and will post a summarised version of responses.

Thank you very much

Mark Thompson

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l