Actual data for the chain cleaning debate

Rick Onanian wrote:

On 6 Jan 2004 06:02:40 -0800, (Art Harris) wrote:
TBGibb wrote:
Date Miles Elongation Elongation
of washed of unwashed
01/05/04 2739.52 1/16" 1/32"


This data indicates that NOT washing the chain in solvent doubles
chain life. How do you explain that? Chain "stretch" is caused by wear
on the pins. Washing in solvent should remove internal grit better
than scrubbing with a toothbrush, and therefore result in longer life.

I imagine that washing in solvent also removes internal lubrication
better, and probably leaves some solvent in it's place. Then, when
you lube it, the lube either never displaces the solvent, or is
diluted by the solvent.

washing doesn't remove the grit from inside the chain, it just makes it
mobile again. Then you have to ride quite a while with a grinding chain
before the dirt particles again wear off from the bearing surfaces

--
Marten

On Tue, 06 Jan 2004 16:17:49 -0500, Rick Onanian
may have said:

At $10, it's so disposable that there's REALLY no reason whatsoever
to clean a chain, unless you put on a really huge amount of miles.
There's barely a reason to lube it -- mostly so it doesn't squeak.

Oiling it also lubes the cassette and chainring teeth, hopefully
helping to prolong their life.



--
My email address is antispammed; pull WEEDS if replying via e-mail.
Yes, I have a killfile. If I don't respond to something,
it's also possible that I'm busy.
Words processed in a facility that contains nuts.

TBGibb wrote:
To interject some real data into the chain lubrication debate I devised the
following method for comparing two methods.

Results:

Date Miles Elongation Elongation
of washed of unwashed
chain chain
07/11/03 0
07/14/03 137.1 0 0
07/23/03 328.6 1/64 1/64
08/01/03 607.24 1/64 1/64
08/11/03 819.71 1/64 1/64
09/02/03 1076.67 1/64 1/32
09/19/03 1430.39 1/64 1/32
10/23/03 1801.17 3/64 1/32
11/07/03 2257.55 3/64 1/32
01/05/04 2739.52 1/16 1/32

I find it very interesting that the rate of wear of each section of
chain is much different. In one month, the washed chain stretched 1/32".
No other data points show that much "chainge" in such a short period.

If I chart the data, I the washed chain showed elongation that was
approximately linear with elapsed time and mileage. The unwashed chain
would have a much poorer linear fit, since it stretched during the first
1100 miles and then stopped stretching for the last 1600 miles. For the
stat-minded, my spreadsheet shows RSQ=0.87 for washed and RSQ=0.66 for
unwashed data.

Two theories that have been proposed elsewhere in this thread might
explain the differences. First, if the chain were accidentally switched
in midstream (i.e. at the 10/23/03 measuring), the trends would make
much more sense. Second, if the unwashed chain has grit filling in the
gap, the gap might stabilize because a growing gap will continue to get
filled with more grit. I'm a little skeptical of this second theory.

I don't think the theory about the packing grease makes much sense in
light of the unusual trends in the data.

Other explanations might exist, but I can't think of any at the moment.
Well done, Tom.

Dave
dvt at psu dot edu

In article , "S. Anderson"
writes:

"TBGibb" wrote in message
...
To interject some real data into the chain lubrication debate I devised
the
following method for comparing two methods.

snip NASA test procedure..


Yes, please excuse me, I have an vehicle maintaince appointment on Mars.

Tom Gibb

In article .net, Jose Rizal
writes:

You might be able to get better results if you measure before putting
the oil on.

I did.



How about just shaking in solvent just once, instead of 5-6 as you did?
It will be interesting to see if washing in solvent just the once is as
effective as washing many times.

Good subject for another trial.



Tom Gibb

On Tue, 06 Jan 2004 21:37:35 +0000, Zog The Undeniable
wrote:
Did you get all the solvent out afterwards? I wash in strong detergent,
rinse and dry the chain in the oven or on a hot pipe before re-oiling it
- otherwise the solvent will mix with, and dilute, the oil. I get at

You do all that work for a $10 chain? That's like working for
$2/hour.
--
Rick Onanian

dvt wrote:
TBGibb wrote:
Date Miles Elongation Elongation
of washed of unwashed
chain chain
07/11/03 0
07/14/03 137.1 0 0
07/23/03 328.6 1/64 1/64
08/01/03 607.24 1/64 1/64
08/11/03 819.71 1/64 1/64
09/02/03 1076.67 1/64 1/32
09/19/03 1430.39 1/64 1/32
10/23/03 1801.17 3/64 1/32
11/07/03 2257.55 3/64 1/32
01/05/04 2739.52 1/16 1/32

I find it very interesting that the rate of wear of each section of
chain is much different. In one month, the washed chain stretched 1/32".
No other data points show that much "chainge" in such a short period.

If I chart the data, I the washed chain showed elongation that was
approximately linear with elapsed time and mileage. The unwashed chain
would have a much poorer linear fit, since it stretched during the first
1100 miles and then stopped stretching for the last 1600 miles. For the
stat-minded, my spreadsheet shows RSQ=0.87 for washed and RSQ=0.66 for
unwashed data.

The measurements could only be made to the nearest 1/64 inch,
and presumably have some error. That explains why the measurements
appear to jump. Presumably the unwashed chain is still stretching
during all the time it is sitting at 1/32" measured stretch, but
the measurement can't resolve it.

When you're measuring in very small numbers of what your instruments
can resolve, effects like this happen all the time. It's well
known that differentiating data to get a rate of change amplifies
noise. That's why I wouldn't worry about the month where the
washed chain appeared to stretch 1/32".

The reasonable way to make the measurement about 2x more accurate is
to measure the stretch in the entire half-chain, not just a foot
of it, but (from experience) doing this to 1/32" or 1/64" will be
a minor nuisance. One needs, at least, a bigger ruler. Anyway,
I know there's nothing more annoying than people who come along
afterward and tell the experimenter how he or she _should_ have
done things. Thanks for making the experiment, Tom.

On Tue, 6 Jan 2004 16:16:53 -0800, Rick Onanian wrote
(in message ):

On Tue, 06 Jan 2004 21:37:35 +0000, Zog The Undeniable
wrote:
Did you get all the solvent out afterwards? I wash in strong detergent,
rinse and dry the chain in the oven or on a hot pipe before re-oiling it
- otherwise the solvent will mix with, and dilute, the oil. I get at

You do all that work for a $10 chain? That's like working for
$2/hour.
--
Rick Onanian

Only if you stand there and watch it dry.

It only takes me around 10 minutes of actual work to clean and dry a chain:
remove, shake in plastic soda bottle with citris solvent, shake with water,
hang to dry, re-install and lube.

In article ,
says...
S. Anderson wrote:
"TBGibb" wrote in message
...

To interject some real data into the chain lubrication debate I devised

the

following method for comparing two methods.


snip NASA test procedure..

Tom Gibb


Yup, I concluded that years ago. My aggravation caused by cleaning a filthy
chain != cost of a new chain. In face, aggravation cost caused by cleaning
a filthy chain cost of new chain. I blow the $25 for a new chain every
year and toss the old one in the garbage. I just keep applying teflon "dry"
lube every other ride or so and that works for me. Scrape off the
accumulated crud (which is hard as a rock..) once a month or so..works for
me, but I ride in mostly good conditions. Your mileage may vary...

Cheers,

Scott..


Thanks TB. I used to clean my chain in solvent after every 200k. I've
stopped doing this out of laziness. Like Scott writes I just put on some
lube whenever my chain seems to want some. My only gripe is that after a
while you get all this crud build-up on the jockey wheels.

I just scrape that off with a screwdriver while spinning the cranks
backwards by hand.

--
Dave Kerber
Fight spam: remove the ns_ from the return address before replying!

REAL programmers write self-modifying code.

In article , Harris
writes:

So Tom, did you leave the packing grease on both halves of the chain at
the start?

Yes.

Tom Gibb