EFBE Frame Test and Pedal Loads

There was some discussion awhile back about this frame fatigue test. One of the criticisms was about
the magnitude of pedal loads.

I have some SRM data from several people on my site (including myself) that might shed some light on
the subject of "real world" pedal load magnitudes.

Please keep in mind that the SRM unit averages power over one pedal revolution, so it is
pretty reasonable to say that the peak pedal loads are on the order of two-times that
indicated in the plots.

http://tinyurl.com/bc8h

--
=======================
Kraig Willett RBT Subscription Specials www.biketechreview.com\rbt.htm
=======================

 

>There was some discussion awhile back about this frame fatigue test. One of the criticisms was
>about the magnitude of pedal loads.
>
>I have some SRM data from several people on my site (including myself) that might shed some light
>on the subject of "real world" pedal load magnitudes.
>
>Please keep in mind that the SRM unit averages power over one pedal revolution, so it is
>pretty reasonable to say that the peak pedal loads are on the order of two-times that
>indicated in the plots.
>
>http://tinyurl.com/bc8h
>

Kraig:

This is certainly some interesting data and I hope people look at it and think about it. As Damon
has/had on his website, "One measurement is worth 50 expert opinions."

What I thought after looking at it and thinking it over for a bit:

1. I think in general the assumption that twice the average force is equal to the peak is probably
a reasonable one, maybe even conservative.

However it it may be in particular cases where the average load is high, this is not because the
peak load is high but rather because the load was better distributed about the pedalling cycle. For
what its worth, when I make peak effort, it comes from a concerted effort to pedal in full circles
rather than because I pushed as twice as hard as normal.

2. But assuming that this factor of 2 is the correct assumptions, there are certainly some peak
loads that would then be in the 1300 newton range.

You do not indicate how long these rides were and with several overlapping rides is difficult to do
an analysis. Your hilly training ride seems to be the clearest and I will make the assumption that
this ride lasted about an hour, probably longer but for this sort of rough analysis, an hour is
probably OK.

Now I count about 20 (21 points, I think) points that are above 600 Newtons average load, 1200
Newtons, assumed peak load.

Dividing that into the 200,000 cycles, it would take 10,000 hours to achieve this many fatigue
cycles at this load level, which would probably be somewhere over 150,000 miles of "hilly training
with sprints."

3. In previous threads, not the most recent one but one probably about a year ago, I suggested that
a more realistic test would be one which used a broad spectrum of loads that accurately
represented the loading history.

What I am seeing here, says that likely these peak loads are a significant (even major) contribution
to the fatigue failure in frames but that other lower level but higher frequency of occurance loads
should also taken into account.

One could do analysis, knowing the S-N curve for the material of a given frame, using the frame test
data to put set determine the location on S-N curve and then use a histogram such as Kraig has
provided to determine the fatigue lives of different frames of different materials under this "real
world" loading spectrum.

4. It is important to remember the assumptions made when working with data. In this case the
principle assumption is that factor of 2. It may be that these outlying points are the result of
unique situations either in pedalling or in the data acquisition.

I have also assumed that indeed these points represent all the points in the ride and are not some
average values or something.

5. I think that as usual Kraig has upped the level here and I would like see the some significant
input on this thread.

I certainly have my point of view but I am also certainly open to being wrong on this, Kraig has
provided some "preliminary" data that raises important questions and has certainly made me rethink
my position on this.

I am interested in what Kraig has to say, he just the data out there without much comment.

Jon Isaacs

 

"Jon Isaacs" <[email protected]> wrote
> I think in general the assumption that twice the average force is equal to the peak is probably a
> reasonable one, maybe even conservative.
>
> However it it may be in particular cases where the average load is high,
this
> is not because the peak load is high but rather because the load was
better
> distributed about the pedalling cycle. For what its worth, when I make
peak
> effort, it comes from a concerted effort to pedal in full circles rather
than
> because I pushed as twice as hard as normal.

The "peak is roughly twice the average" is supported by the instrumented pedal studies (which is
about the only way to really know).

> You do not indicate how long these rides were and with several overlapping rides is difficult to
> do an analysis. Your hilly training ride seems to
be the
> clearest and I will make the assumption that this ride lasted about an
hour,
> probably longer but for this sort of rough analysis, an hour is probably
OK.
>
> Now I count about 20 (21 points, I think) points that are above 600
Newtons
> average load, 1200 Newtons, assumed peak load.
>
> Dividing that into the 200,000 cycles, it would take 10,000 hours to
achieve
> this many fatigue cycles at this load level, which would probably be
somewhere
> over 150,000 miles of "hilly training with sprints."

All of those points were when Kraig was practicing sprints, so the proportion of time during the
entire ride that they represent or the number of miles is a red-herring. Those points represent six
different sprints. You can see another graph of those sprints here: http://tinyurl.com/azfr but note
that I was using average pedal torque rather than average pedal force--you should divide the torque
by his crank length (maybe 175?? mm) to get the pedal load (and then multiply by two to get the
peak, if that's what you're interested in).

 

"Jon Isaacs" <[email protected]> wrote in message
news:[email protected]...
> >The "peak is roughly twice the average" is supported by the instrumented pedal studies (which is
> >about the only way to really know).
>
> I am sure that this is true in almost all cases. However I am suggesting
that
> in extreme cases this may not hold true.
>
> One has to be careful in extending generalizations to the extreme points because those are the
> very points which may lie outside the boundries
where
> those assumptions hold true.

Right, but if I understood your point properly, you were suggesting that *your* peak loads would
be *lower* than twice the average. If so, then doubling the average is a "safe" estimate of
maximum loads.

> >All of those points were when Kraig was practicing sprints, so the proportion of time during the
> >entire ride that they represent or the
number
> >of miles is a red-herring.
>
> It is not a red-herring. Kraig is presenting this data to approximate
"real
> world pedal loads."
>
> I am merely using this to try to get a handle on the numbers and the
frequency
> that such loads may occur.
>
> My intuition tells me that hilly training with sprints probably has a
higher
> than normal incidence of peak loads, but for the purpose of getting a ball
park
> understanding of the frequency of these high loads it seems relevent to
me.
>
> So please realize, I am not trying to dispute this data or suggest that it
is
> bad data.
>
> Rather I am trying to understand it and understand the assumptions that
lie
> behind it so that it can be applied to the issue at hand in an appropriate manner.

I was under the impression that he was definitely *not* suggesting that this was a normal ride. The
data were being presented as a way to examine peak loads. I looked at his data, and the sprints were
definitely not made in the course of normal riding. I took his data to represent real world max
sprint loads for someone who can produce 1200 watts, not real world loads while JRA. He could have
done the entire hilly ride without sprinting at all.

 

"Robert Chung" <[email protected]> wrote in message news:[email protected]...
>
> "Jon Isaacs" <[email protected]> wrote in message
> news:[email protected]...
> > >The "peak is roughly twice the average" is supported by the
instrumented
> > >pedal studies (which is about the only way to really know).
> >
> > I am sure that this is true in almost all cases. However I am
suggesting
> that
> > in extreme cases this may not hold true.
> >
> > One has to be careful in extending generalizations to the extreme
points
> > because those are the very points which may lie outside the
boundries
> where
> > those assumptions hold true.
>
> Right, but if I understood your point properly, you were suggesting
that
> *your* peak loads would be *lower* than twice the average. If so,
then
> doubling the average is a "safe" estimate of maximum loads.

http://www.isbweb.org/data/kautz/highw.avg

Average power: 192.6 W (single leg) Average cadence: 90.3 RPM

Phil Holman

 

"Phil Holman" <[email protected]> wrote in message
news[email protected]...
> http://www.isbweb.org/data/kautz/highw.avg
>
> Average power: 192.6 W (single leg) Average cadence: 90.3 RPM

Phil, during what type of effort and on which type of apparatus was this data taken?

Because, if my math is any good, the peak pedal load to average pedal load ratio in this case was
around 2.7.

--
==================
Kraig Willett www.biketechreview.com
==================

 

"Kraig Willett" <[email protected]> wrote in message
news:[email protected]...
> "Phil Holman" <[email protected]> wrote in message
> news[email protected]...
> > http://www.isbweb.org/data/kautz/highw.avg
> >
> > Average power: 192.6 W (single leg) Average cadence: 90.3 RPM
>
> Phil, during what type of effort and on which type of apparatus was this data taken?
>
> Because, if my math is any good, the peak pedal load to average pedal load ratio in this case was
> around 2.7.

D'oh - one leg over 360 degrees - my bad.

--
==================
Kraig Willett www.biketechreview.com
==================

 

"Kraig Willett" <[email protected]> wrote in message
news:[email protected]...
> "Phil Holman" <[email protected]> wrote in message
> news[email protected]...
> > http://www.isbweb.org/data/kautz/highw.avg
> >
> > Average power: 192.6 W (single leg) Average cadence: 90.3 RPM
>
> Phil, during what type of effort and on which type of apparatus was this data taken?
>
> Because, if my math is any good, the peak pedal load to average pedal load ratio in this case was
> around 2.7.
>

These data appear to be the average of 14 "elite endurance" cyclists, and they must have come from
instrumented pedals. I'm guessing Ft is tangential force and Fn is normal force. How do you get 2.7?

 

"Robert Chung" <[email protected]> wrote in message news:[email protected]...
> How do you get 2.7?

2.7 is what I get as a result of not thinking it through and neglecting one of the components of
pedal force! ;-)

 

For my own sanity, here is what I come up with using the data that Phil provided a link to (after
making the assumption that the other leg is 180 degrees out of phase and is identical to the one leg
we have data on - if that makes sense):

Average Torque: 41.6 N-m (should be close to what SRM uses to calculate an average power for the
pedal stroke)

Assuming a 172.5 mm crank length this means the calculated average pedal load would be ~ 241.2 N
(this does not match the _actual_ average pedal load as reported in the data - due to crank angle
and the pedal force direction issues).

When one compares the reported maximum pedal load magnitude (both tangential and normal components)
to the calculated average pedal load using the average torque (as I did with the data collected with
the SRM) one gets:
473.3 N/241.2N = 1.96.

The measured peak load is approximately two times the pedal load calculated using the SRM-like data
reduction method.

I have probably confused the issue, but _I_ feel better about the assumption anyway.

I think it is clear that even mortals like myself (at about 75kg) are capable of creating the
1200-1300 N loads used in the EFBE test. To Jon's point, it now becomes a question of frequency of
these events and how that relates to the life of the product. I think it is also reasonable to
assume that forces in excess of this are applied by individuals that are larger than I.

How one uses this type of data to develop/design product is another issue entirely - in my limited
experience, one designs a part based on how confident one is in knowing the load spectra.

I will also say that in my experience with stems, the amount of "damage" done to the part in the low
load stages seemed inconsequential once the large loads were applied. After evaluating the part
failures, I don't think the results would have been much different had I started with a higher
initial load. The next time that I test stems, I will be starting at a higher initial load. There
are also practical reasons to consider when developing tests (time).

For what it was used for in the magazine article (a comparative evaluation), the EFBE test protocol
seems appropriate and even attempts to respect the flexible boundary condition at the rear dropout.

Phil probably has the most experience when it comes to designing and testing critical parts, so I
would be curious as to what he thinks about the EFBE frame test.

--
==================
Kraig Willett www.biketechreview.com
==================