View Full Version : responses to treadmill power exchanges

James Dowling
07-20-1995, 05:10 AM
This is my summary of the responses that I received to a query about the
energy or power exchanges between treadmills and the subjects running on them.
Below, you will find my original posting and the actual responses that I
received. The responses from Kistler followed my posing of the question
directly to them. I wish to thank each respondent because the dialogue is
quite informative. I have learned that others (more capable than myself)
have attempted to quantify the changes in power drawn by the motor without
success (see Paul Guy's response) and that in spite of significant gains in
treadmill instrumentation (see Kistler responses), the inability to measure
horizontal ground reaction forces prevents a power exchange calculation.
It is clear that the exchange occurs whenever the belt velocity changes and
that this does indeed happen. A paper "Power Transfer from Treadmill Engine
to Athlete" by Henk Schamhardt, et al. that was published in the proceedings
of the Canadian Society for Biomechanics (1994) seems to be the only
published method of quantifying the phenomenon (see response by Ton van
den Bogert for E-mail addresses of the authors). This method combined
horizontal GRF obtained from over-ground running with treadmill velocity
changes via integrated accelerometry in equine trotting.
It seems to me that it is quite possible to either use the accelerometer
method of this paper or the Kistler "Gaitway" or some other method to
quantify the velocity fluctuations in the treadmill belt. It also seems
possible to use linked segment analysis to determine the horizontal
accelerations of the whole body center of mass relative to that of the belt.
The multiplication of this acceleration by the whole body mass and the
velocity deviation from average should yield the power exchange. If this
can be done accurately (see Ton's response for an anomaly), then one does
not need to account for power drawn by the motor and other losses such as
friction, belt slip or stretch, etc.
The bottom line seems to be that this is still an unresolved issue but we
may be closing in on a solution. Thanks again to those responsible for the
following responses.
Jim Dowling
Dept. of Kinesiology
McMaster University
Hamilton, Ontario, Canada

Dear Biomch-l member,
Unlike over-ground running or walking, there exists the potential for
an exchange of energy between the surface and the subject in treadmill
locomotion. The subject performs negative work when landing on the
treadmill if the belt speed is reduced and energy can be returned to the
subject when the belt speed increases again. My question involves the
calculation of this exchange. We have used a current shunt to monitor
the electrical power drawn by the treadmill motor assuming that the
voltage stays constant. The current increases under conditions of
mechanical load but to calibrate this in Watts and to relate this as an
instantaneous measure effecting locomotion energetics is another matter.
Are there other energetic losses to the system (i.e. belt slippage or
transmission losses)?
Suggestions or useful anecdotes by others who have addressed this
issue are grateful solicited. As per the custom of this list, I will
share the compiled information. Thanks in advance,

Jim Dowling
Dept. of Kinesiology
McMaster University
Hamilton, Ontario, Canada

>From MCANELLY@uthscsa.edu Tue Jun 13 09:31:29 1995
Date: Mon, 12 Jun 1995 15:50:01 -0600
From: Robert McAnelly
To: James Dowling
Subject: Re: treadmill power exchange

Reply to: RE>treadmill power exchanges
I have heard of belt vibration, and of belt elasticity (stretching like a
rubber band under the foot, including stretching in response to shear forces).
There is also the energy loss/gain from holding onto handrails. No doubt, the
belt is also meeting resistance as it slides through the mechanical parts, and
so don't treadmills have some feedback mechanism to keep their velocity

I would think that the more expensive the treadmill, the less the tread would
speed up or slow down from foot impact. If so, then most researchers would
rather buy a better treadmill, rather than take acceleration/deceleration into
consideration. Best of luck with this complicated problem.
Robert McAnelly

>From paul@gaitlab1.uwaterloo.ca Tue Jun 13 09:36:44 1995
Date: Mon, 12 Jun 1995 22:45:47 -0500 (EDT#Canada)
From: Paul Guy
Subject: Re: treadmill power exchanges

Hi Jim,
A long time ago, we tried to determine either power or the forces
under the foot by measuring AC power parameters on a treadmill. I used
both the AC current and AC real power as measurements, and tried to
correlate them with expected force and powers from normal walking.
For treadmills with induction type motors (most of them), the
experiments were total failures. Either taking power out of the system
by a workload, or putting power into the generator by vigorous exertion
caused current increases in the same direction. If I remember
correctly, the measured power went in the proper polarities, but
sensitivity was very much reduced in the generation mode. I attribute
this to fact that current on a motor of this kind is proportional to the
'slip' (difference from the 60Hz rotating field), and it doesn't care
whether the slip is positive or negative. With synchronous motors,
especially 3 phase ones, this experiment might work. DC motors might
also work, a lot depending on their configuration. Unfortunately, I
don't believe any of the traditional treadmills use these kinds of
If you had to do it using an existing treadmill, I believe the best
way would be to attach some sensor that would measure the torque applied
to the belt driver on the treadmill. Using that and the speed (easily
measured), you might have something more useful - once you have a way of
eliminating the contribution of the belt friction. That could be done by
measuring the weight applied (a constant from the mechanism, plus the
subject's weight plus the vertical inertial components) and determining
the friction. From the speed and the vertical force, you could roughly
isolate the belt's power loss to friction under the feet.
This was something we never tried... we went to using more
forceplates, it seemed like a lot less hassle.
One question I'd like to fire back, do you think that treadmill
walking gives you the same results in terms of power, moments, and all
of the finer 3D measurements? We had many heated debates about this


Paul J Guy work phone:519-885-1211 ext 6371
paul@gaitlab1.uwaterloo.ca home/FAX/:519-576-3090
pguy@healthy.uwaterloo.ca 64 Mt.Hope St.,Kitchener,Ontario,Canada

>From Christopher.Johnston@ah.slu.se Tue Jun 13 09:38:49 1995
Date: Tue, 13 Jun 1995 08:04:37 +0100

Dear Jim Dowling

There is an study on this question in Canadian Society for Biomechanics 1994
by Schamhardt, van Bogert and Lammertink pp. 306-307 entitled Power transfer
from treadmill engine to athlete. They used a large treadmill designed for
horses and measured the change in speed by measuring the surface speed and
found a change related to the stance phase.

We work with horses also and have a treadmill with a coir mat which visible
buckles when the horse first contacts the mat with it's hoof. The mat
instantaneously stops while the drums continue. The treadmill is build to
do this because of observations from films from horses moving on race tracks
where the hoof "slides" for some milliseconds.

Finally, I recently received a thermovideo from a colleague in Washington
State who had measured heat transfer between hoof and treadmill.
Interestingly, the rubber treadmill surface became locally quite warm during
the stance phase to then cool down after the horse lifted it's hoof.

These are just some observations with large creatures but I hope they can help!

Regards and Happy Treadmilling


Chris Johnston, DVM
Equine Biomechanics Lab.
School of Veterinary Medicine
Uppsala, Sweden

>From TURN_KNICK@hrz.dshs-koeln.de Tue Jun 13 09:41:38 1995
Date: Tue, 13 Jun 1995 09:15:37
From: Axel Knicker

Hello Jim!
Is this a serious request? If yes, You should consider the law of
relativity. A runner on a track does also perform negative work as he
makes contact to the ground. His CM is then decelerated app. until CM
is right above the supporting foot. It is then again accelerated in
running direction. He performs positive work. To maintain a constant
running speed the relation between negative and positive work must be
balanced. In treadmill running the horizontal velocity of CM is
neglectable and to compare to overground contant speed you must
assume it to be zero. Thus the moving belt reflects the horizontal
path of the CM. The energy transmitted by the belt should have the
same magnitude as the energy transmitted by the interaction between
runner and ground in overground running both positive and negative
If you like I can send you an ASCII file with velocity, energy, work
and power data of some long distance runners calculated from
kinematic data taken at the WAC in Tokio 1991.


>From russell@dingo.vut.edu.au Wed Jun 14 09:30:10 1995
Date: Wed, 14 Jun 1995 16:42:50 +1000 (EST)
From: Russell Best


I recommend you speak to Alfred Zommers . The same
type of problems exist in cycle ergometry. Fred has built his own
ergometer for use with the athletes own bike. He has taken a treadmill
type approach and has had to work out slippage, temperature effects,
alternator issues, etc. He may be a useful person to talk to about energy
losses in such systems.


>From skane@fhs.csu.McMaster.CA Thu Jun 15 09:50:03 1995
Date: Wed, 14 Jun 1995 12:57:58 -0400 (EDT)
From: Sheri-Lynn Kane


it seems to me that the belt itself heats up considerably during
operation, ie friction between the belt and supports may be a significant
source of energy loss. What do you think? This may be difficult to

Jim Dickey

>From bogert@acs.ucalgary.ca Sat Jun 24 20:03:10 1995
Date: Wed, 21 Jun 95 9:50:21 MDT
From: Ton van den Bogert
Cc: "Henk C. Schamhardt"

Hi Jim,

You may have seen Henk Schamhardt's poster at the CSB conference
last year. If not, look up the abstract. I visited him again
last week, and we discussed this topic for a couple of hours.
Before seeing your posting!

>locomotion. The subject performs negative work when landing on the
>treadmill if the belt speed in reduced and energy can be returned to the
>subject when the belt speed increases again. My question involves the

I have intuitively calculated power exchange as F*deltaV, where
deltaV is the deviation from the average belt velocity. A very
funny thing happens if you decide to use a different reference
frame (effectively adding a constant to deltaV). The power
exchange pattern changes dramatically, but power still equals
rate of change of kinetic energy (when calculated in the same
reference frame). Fortunately, if the movement is cyclic, the
*average* power transfer is not affected since the integral of F
is zero. But instantaneous power does not appear to be
independent of choice of reference frame.

There must be something wrong here, but I can't pinpoint it. I
still think velocity should be expressed with respect to a
reference frame which moves at average belt speed. But why?

>calculation of this exchange. We have used a current shunt to monitor
>the electrical power drawn by the treadmill motor assuming that the
>voltage stays constant. The current increases under conditions of
>mechanical load but to calibrate this in Watts and to relate this as an
>instantaneous measure effecting locomotion energetics is another matter.
>Are there other energetic losses to the system (i.e. belt slippage or
>transmission losses)?

I don't think this works. Frictional losses (between belt and
underlying surface) are probably responsible for the major part
of engine power. On top of that, there is a small amount of
power which is transferred to/from the athlete. It would be hard
to separate the two.

Coming back to the issue of 'instantaneous power'. Even if the belt
speed is exactly constant, the engine will have to produce
alternating positive and negative power on the athlete. Still,
the athlete can't feel different from running on the ground
because he is moving in a perfect inertial reference frame. So,
is this instantaneous power measurement relevant? Or is only the
time integral meaningful?

Because of the large frictional losses, we came to the conclusion
that power exchange can only be measured using force between foot
and belt (should be possible in horses with an instrumented horse
shoe), in combination with velocity of the foot. But this
instantaneous power problem keeps bothering me.

I would be interested in your further thoughts. Once you have
collected some responses, a summary to Biomch-L would be good.

-- Ton van den Bogert

cc: Henk Schamhardt

>From biomech.kistler.com@ Wed Jul 12 21:55:07 1995
Date: Mon, 10 Jul 1995 10:49:39 -0400

The Gaitway instrumented treadmill is capable of measuring the
vertical ground reaction forces, and not the shears. Therefore, it is unable
to perform the necessary measurements that you require (namely, the horizontal
ground force).

> Quantification of this exchange is difficult but the power
>exchange could be calculated by multiplying the horizontal ground
>reaction force by the difference between the average velocity and the
>instantaneous velocity of the belt.

We are capable of measuring the instantaneous velocity of the belt, which is a
necessary feature for us to be able to accurately calculate the center of
pressure. Ideally, we would like to be able to calculate the shears, but to
this point we have not been successful due to the belt movement across the

So, a quick summary would be to say that at present we are unable to offer the
required measurements, but we will keep you informed if things change.
Namely, we will send you our
Kistler Biomechanics Newsletter (if you already don't receive it) three times
a year to keep you updated on our latest and greatest products.

Also, please feel free to stop down (we are 15 minutes from the Peace Bridge)
and see our treadmill firsthand. Just let us know you'll be coming and we can
give you directions.

Thank you for your question, if I have not answered to your satisfaction
please write back!


Ken Wagener
Biomechanics Systems Engineer

>From klavoon@kistler.com Wed Jul 12 22:16:19 1995
Date: Wed, 12 Jul 1995 17:06:44 -0400
From: klavoon@kistler.com

Dear James,

As you know, Kistler has recently gone into production of a force plate
instrumented treadmill system called "Gaitway". We may be able to provide you
with some limited information which can help you in your research. I know that
Ken Wagener has e-mailed you previously with some information and hopefully I
can add more.

I agree that energy is exchanged due to belt speed variations, but I am not
sure how I would calculate this exchange. In theory, if the belt speed did not
change, then walking on a treadmill would mechanically be identical to
overground walking, with the exception of no wind frictional forces. Thus, we
agree any differences must be due to (or at least result in) the speed
variations. Our production system includes an instantaneous speed measurement
channel (used for center of pressure calculations.) Generally, speed is
maintained within +/- 1% of nominal. Using a simple tachometer you can
measure this. One simple solution is to use a small DC motor driven by the
belt, calibrate its voltage out vs. speed. Another solution is to use a pulse
counting tach.

Our system is a vertical force only system, but in our development we did
implemented a prototype system with 3-component force plates, just as an R&D
exercise. The problem is that shear force measurements are severely effected
by the frictional forces between the belt and the top plate. The magnitude of
these force (AP direction) are roughly 1/3 body weight (coefficient of
friction approx. 0.3). Unfortunately, it seems the Coeff. of friction changes
+/- 15% at different parts of the belt or different locations on the top plate
(seam in the belt and uneven plate lubrication are the assumed culprits).

If you would like me to elaborate more on any of the above topics, please let
me know. I am looking forward to your conclusions.

Bill Klavoon
Kistler Instrument Corp.