Page 1 of 3

A Hub-Axle Treatment Evaluation Protocol

Posted: Thu Aug 07, 2003 7:58 pm
by Stan Pope
I mentioned in another thread a setup to evaluate Hub-Axle treatments with high sensitivity and repeatability. Following is some stuff that I composed (mostly) in February, 2002. Additions, Alternatives, Concerns welcome.

Stan

Form a "wheel" by fabricating a "ring" from wood.
The OD is about 2".
The ID is about 1/32" larger than a typical PWD wheel, allowing a press fit when shimmed with three strips of rubber band.
The thickness of the ring should be about the same as a PWD wheel hub.
The total mass of the wheel assembly is about 2 ounces. The mass of the ring dominates. (increased from 1 ounce, 8/11)

The rim of the ring should allow attachment of a thread such that the thread may be secure tangentially but loose radially. A headless pin radially into the rim should work for this.


Experiment setup:

Attach a weight of about 1/4 ounces onto a 4 foot(+) thread. Loop the other end. (The weight is chosen to induce the approximate the angular velocity of an actual wheel on a car at the bottom of the slope.)

Attach the loop to the headless pin in the rim and wind the length of thread around the rim.

Mount an axle support about 50 inches above the floor so that a pwd axle can be inserted into the support horizontally, adjustable to level or inclined.

Mount a small eye screw on a support so that the eye lies in the plane of rotation and so that the thread makes a slight (5 degree) bend as it comes off the wheel, through the eye, and toward the floor. (The purpose of the eye screw is to assure that the thread comes off the wheel in its plane of rotation even if the axle is sloped off horizontal.)

The mass and moment of inertia of the wheel assembly and the angle of thread deflection must remain constant.


Experiment measures:

1. Measure and Record the time elapsed from (1) release of the weight until (2) the wheel stops turning. Repeat trials and average measured times.

Alternative Measure: Measure and record the number of rotations completed by the wheel. Measured by LED + Photodiode across the wheel. Drill 1 hole to measure total rotations, drill N equally spaced holes to measure in units of 1/N rotations. Use off-the-shelf pulse counter.


Question 1: What treatment produces the lowest turning resistance under pwd loads?

Configurations to evaluate:
1. aligned level polished (baseline)
2. reversed direction of rotation
a. control: rotation in polished direction
b. rotation in opposite direction
3. sloped axle
a. flat head
b. tapered head
4. notched axle
a. control: polished axle
b. single wide notch, polished as above
c. multiple, narrow notches, polished as above
5. reduced diameter axle (hub diameter unchanged, moly graphite lube)
a. -0.005"
b. -0.010"
c. -0.015"
6. threaded hub on polished shaft
7. Lubrication:
a. unlubricated
b. Lemon Pledge
b.1. Same, wiped off carefully
c. graphite, regular
d. graphite, moly
e. white lube
e.1. Same, wiped off carefully
f. petroleum jelly film
g. salad oil film
h. silicon spray lube
h.1. Same, wiped off carefully
i. 3in1 Oil
j. cumulative effects of graphite lubrication
k. new vs. spun in graphite
l. "Rain Shield"
m. white lube, washed
8. axle shaft polishing - (head treated and polished. shaft bumps removed):
a. untreated - control
b. polished - toothpaste (Pearl Drops)
c. polished - jeweler's rouge
d. polished - polishing compound
e. polished - polishing compound + jeweler's rouge
f. polished - rubbing compound + polishing compound
g. polished - crocus cloth + rubbing compound + polishing compound
h. polished - emery cloth + crocus cloth + rubbing compound + polishing compound
i. bleached, not polished
j. polished and bleached
k. polished and bleached and polished
l. polished opposite direction - rubbing compound + polishing compound
m. polished opposite direction - crocus cloth + rubbing compound + polishing compound
n. polished with Mother's Chrome Polish

Question 2:
What is an effective method for removing the various lubricants?
Chemical:
detergent
mineral spirits
paint thinner
brush cleaner
Mechanical:
plastic scrubber
steel wool


Things to think about:
Review the sequence to minimize the needed number of trials.
Carefully evaluate the sequence to minimize the effect of prior treatments on the materials.
Rerun "baseline" at strategic points to identify the effect of prior treatments.

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Fri Aug 08, 2003 8:03 am
by RACER X
That's quite an extensive testing procedure but I am sure it would net some great results.

I have always wondered what happens to axle/wheel performance when load or the weight of the car is applied.

Any thoughts?

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Fri Aug 08, 2003 8:51 am
by Stan Pope
RACER X wrote:That's quite an extensive testing procedure but I am sure it would net some great results.

I have always wondered what happens to axle/wheel performance when load or the weight of the car is applied.

Any thoughts?
Well, that is the point of this setup ... simulate real life. I think that the setup transforms the car body weight and inertia with sufficient accuracy that it would produce meaningful results. I'm open to concerns.

The benefit of additional angular inertia in the spinning object is that it amplifies the spin time and/or spin count to get extra significant digits in the measurement.

The key reason for most of the decisions in the setup was to isolate the one interaction and remove the vagueries caused by everything else going on when a car runs down the track.

I (or someone who is a better engineer than I) ought to compute the formula to translation from spin time (or spin count) to coefficient of friction. I don't think that it would be necessary in order to make A:B comparisons, but would be useful in moving data to an engr. sim. program.

I'm up in the air about the takeoff angle for the weighted thread. The 5 degrees was a first thought simply to keep the thread force tangential. I wonder now if 30 degrees would more accurately simulate actual physical forces on real life car and wheel? Thoughts?

Also, I don't see a way to simulate the increased G-force that the car experiences due to radius of curvature. I don't think that the omission affects coefficient of friction, but might be some subtle influences due, perhaps, to changes in relative axle location.

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Fri Aug 08, 2003 2:17 pm
by Jewel
This set up looks brilliant. I was considering a standard roll out track with a small drop and measuring how far the car rolls. However, the repeatability of a roll out type set up is not as good as using a fixed mass dropping a fixed distance and isolating the frictional force in the Hub-Axle. As a final reality check however a roll test might be useful to verify the results found using the wood wheel set up and seeing if the data translates into rolling out further. Or perhaps the best test would be to use a standard Pinewood track to validate the results with some racing times and see if there is a correlation. This would also make it more clear how much wheel axle friction is a factor. I generally see three factors in building a good car, good alignment, low total friction, and wheels(lowest legal mass, small axle hole ID, tread concentricity). This kind of data may make it clear where you get the most speed for your efforts.

In terms of the angle that the thread comes off the wooden rotating mass, I would think it would not matter to much as long as the same angle was used every time and results are repeatable. Are there any issues between front and back wheels, since the rear wheels have twice as much weight resting on them?

Overall this seems like a great way to get some useful data with a simple set up. I also like the idea that I could make the same diameter and mass wood ring so I could try and make a wheel that spins as good as yours. Could be fun.

:D

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Sun Aug 10, 2003 9:19 pm
by Stan Pope
An additional question that should be addressed.

Question 3. How do the various lubricants perform over sucessive runs?

Run 20 trials on the best performing cases.
If there is performance decay below other cases, then run additional trials for those other cases.


Trial count for each condition needs to be set according to the variance observed. The need might be estiamted by running a shortened version of the protocol with limited conditions and limited trials per condition.

I need some input from someone with experimental statistics background regarding how to set the trial counts!!! Randy? Anyone?

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Sun Aug 10, 2003 10:10 pm
by gpraceman
Stan Pope wrote:I need some input from someone with experimental statistics background regarding how to set the trial counts!!! Randy? Anyone?
My statistics background is getting very rusty since I am in software development now instead of Quality Engineering. However, assuming that all other conditions are kept the same, a minimum 20 sample test for each type of lubricant will give enough data to tell if there is a statistically significant difference between the average values. I'd recommend that one of those data sets be using no lubrication and would be the baseline.

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Sun Aug 10, 2003 10:26 pm
by Stan Pope
gpraceman wrote:... assuming that all other conditions are kept the same, a minimum 20 sample test for each type of lubricant will give enough data to tell if there is a statistically significant difference between the average values. I'd recommend that one of those data sets be using no lubrication and would be the baseline.
Therein is the rub (if you will pardon friction puns). Most graphite studies I've read suggest that there is a gradual decline in effectiveness beginning with the 2nd or 3rd run and continuing to decline through at least the 12th run.

To avoid: Renew the treatment before each trial.


What do I have to do to demonstrate that measurement setup gives "highly consistent" results, akin, for instance, to micrometer measurements of a rod diameter. How many times do you measure a rod diameter before you conclude that it is, in fact 0.250 inches? I need to understand this process a bit better!


Baseline ... The "no graphite" case historically shows greatest variance. What damage does that do to it?

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Sun Aug 10, 2003 11:53 pm
by gpraceman
Stan Pope wrote:Therein is the rub (if you will pardon friction puns). Most graphite studies I've read suggest that there is a gradual decline in effectiveness beginning with the 2nd or 3rd run and continuing to decline through at least the 12th run.

To avoid: Renew the treatment before each trial.
Bad pun :wink:

For it to work with a lubricant that deteriorates, then you would have to reapply after each run. That, of course, will induce extra variation since it is hard to be consistent with how much you reapply and the lubrication procedure itself. You may want to check with Michael Lastufka to find out the procedure he used for the DOE experiment that he has been working on.
Stan Pope wrote:What do I have to do to demonstrate that measurement setup gives "highly consistent" results, akin, for instance, to micrometer measurements of a rod diameter. How many times do you measure a rod diameter before you conclude that it is, in fact 0.250 inches? I need to understand this process a bit better!
To be sure about your results, you first need to be sure about your measuring system. Your measuring system error should be a factor of 10 or more better than your measurements. You can use the spreadsheet posted on my GrandPrix Race Central weigh scales page to do this analysis, just change the unit of measure. Once you are sure that your measuring system error is acceptable, then you can evaluate your data. For your rod example, you would know that it is 0.250 +/- the measurement error after one measurement.

However, when you are not measuring the exact same thing all of the time, you need a sufficient sample size to be able to determine a "statistically significant" change in the data distributions of the different data sets. There is a way to calculate the minimum sample size, but like I said, I am getting very rusty and would have to refer back to my stats books and notes. If it wasn't for me being behind schedule on getting version 3 of my race software out, I would. Hopefully, someone out there might know some stats and can help out.
Stan Pope wrote:Baseline ... The "no graphite" case historically shows greatest variance. What damage does that do to it?
A comparison of the amount of variation in each data set would be useful to know. For each data set you should end up with a distribution that is normal (a bell shaped curve) that has an avearage and a standard deviation. No two ditributions will be the same, but will the difference in the averages or the standard deviations be different enough to be deemed statistically significant? This brings us back to having a sample size big enough to tell the difference. The bigger the sample size the better.

Sorry, I can't nail it down better than that right now.

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Mon Aug 11, 2003 12:04 am
by Stan Pope
Very helpful.

Before the theory will fit in my head, I must have a functioning intuition. It is starting to form.

Thank you!

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Mon Aug 11, 2003 9:45 am
by MaxV
I completed a set of experiments over the summer which I will publishing in the Pinewood Derby Times newsletter over the course of the coming season.

Not everyone will agree with the methodology I used, but I believe it provides reasonably accurate results while still being practical to implement.

I found that graphite maintains its ability to lubricate for at least 15 runs, and then begins to very slowly deteriorate. I only ran 20 heats so I don't how long it will actually last.

As we know graphite takes a few runs to break in. So, in a race situation, lubricating between race heats is not a winning strategy.

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Mon Aug 11, 2003 10:15 am
by Stan Pope
MaxV wrote:I completed a set of experiments over the summer which I will publishing in the Pinewood Derby Times newsletter over the course of the coming season.

Not everyone will agree with the methodology I used, but I believe it provides reasonably accurate results while still being practical to implement.
Sounds great!

Is there anything in the process you used that could be used in this protocol to better isolate the hub-axle interface? Additional steps and replacement steps are welcome. This is the easiest time in the project to make major revisions!

There is value, too, in independent testing. Both repeating the same protocol and using an entirely different method. Those methods which are sound should reach consistent conclusions.

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Mon Aug 11, 2003 1:50 pm
by Stan Pope
The analogies that support the validity of results of this protocol.

1. Force between axle and wheel

Real Car: Each rear axle bears approximately 2 ounces of weight.

Test Setup: The wheel plus ring creates a controllable weight on the axle. For maximum analogy, the weight of the ring+wheel should be about 2 ounces.

2. Momentum causing rotation to continue

Real Car: The linear momentum of the car body and the angular momentum of the wheels cause the car wheels to continue spinning as the car goes down the track, slowed by wheel-axle friction and other frictions.

Test Setup: The angular momentum of the wheel+ring cause the car wheel to continue spinning, slowed by wheel-axle friction and other (hopefully lesser) frictions.

3. Friction causing rotation to slow.

Real Car: The wheel-axle friction is proportional to the car body weight suspended by that axle, approx 2 ounces for a rear wheel. Other frictions also in play.

Test Setup: The wheel-axle friction is proportional to the wheel+ring weight suspended by that axle, approx. 2 ounces. Some other frictions also in play.

4. Hub face frictions against nail head and car body.

Real Car: Axle tilt can cause the hub to rub on the car body or nail head with force determined by the axle tilt and axle alignment.

Test Setup: Axle tilt can cause the hub to rub on the car body or nail head with force determined by the axle tilt.


Some analogies are still soft.

The analogy between either "total time to stop rotation" or "total number of rotations completed" in the Test Setup should be analogous to "elapse heat time" for a Real Car. That is to say, if the angular inertia of the wheel+ring and the force+distance from the weighted string are held constant, one should be able to compute one from the other. Might have to get Michael L in here to do that, but it should be possible!


Anyone see relevant analogies that I missed?

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Mon Aug 11, 2003 5:08 pm
by Jewel
Stan,

Is stiction an issue? Is there any start up friction that may be a factor and something that could be reduced or measured?

Impressive working through of the problem above.

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Mon Aug 11, 2003 5:25 pm
by Stan Pope
Jewel wrote:Is stiction an issue? Is there any start up friction that may be a factor and something that could be reduced or measured?
"Stiction" is probably present and would present in the same manner as on a real car. The starting point for each trial is a stationary wheel with the weight suspended ready to "take off down the track."

Perhaps I should have identified this as another analogy between the two systems.

Re: A Hub-Axle Treatment Evaluation Protocol

Posted: Tue Dec 02, 2003 8:45 pm
by Stan Pope
I have posted the article "A Method to Compare Pinewood Derby Wheel-Axle Frictions" at http://members.aol.com/standcmr/pwafm.htm with sketch showing test setup.

Thanks all for inputs.

Stan