My brain is going nuts this late at night...I need to get to bed...lol. I like the rig. I like the release. I like the trucks being identical. I like swapping wheelsets to eliminate possible false readings. BUT If the metal wheelsets do in fact add weight...as was tested by Rick Brodzinsky earlier. And you have them attached to each other. With the truck with metal wheels on top of the truck with plastic wheels...The forces on the bottom truck would be different then when you flip it over and the plastic wheels are on top. I hope that made sense. My mind says the trucks should be independent of each other. Equal weights glued on top of each truck would eleminate any discrepancies when releasing them. I am liking your approach to testing metal wheels and plastic wheels and roll resistance.
Playing devil's advocate here but with all the hallabaloo of recent times about the "slinky effect" of MT couplers, wouldn't increased rolling resistance be a benefit in preventing the "two steps forward, one step back" oscillation induced by the MT couplers or do we have to change out the couplers along with the wheelsets? Alternatively, we could use the old remedy of adding a spring to one of the wheelsets on the last car to induce more drag to stop the oscillation but wouldn't that make metal wheelsets superfluous? Also, there is the question of economics. How much money are you willing to spend trying to fix something that isn't really broke?
I have seen that posted before elswhere. It is a pretty nice idea, but, as you said, it doesn't really tell us the rolling resistance in something like engineering units. Still if everybody who wants to do these types of comparisons could make their test rigs exactly like yours, it would provide an objective comparison for all to use. But, "exactly" is the difficult part to make the measurements by different people actually be comparable. I noticed in the videos that you had a 1 or 2 cycle range in the results for the same truck on different trials, so there probably needs to be some testing protocol that specifies a number of trials over which an average is taken, plus an acceptable range of trial results that, if exceeded, indicates that there is some uncontrolled variable that is probably messing-up the results, so they are probably not comparable to others' results. There would also have to be some pretty detailed specifications for the track and its radius of vertical curvature. And, there should be some sort of leveling indication for the track to make sure that it isn't putting any more side force on the flanges than whatever is minimum for the wheel set on track that is level across the rails. Actually, it would be pretty important to cut the foam "roadbed" very perpendicular to the foam sheet's flat surface, or there would be some unaccounted for rolls to the rails that could affect the results from test rig to test rig. The test rig I am making uses a straight ramp with the track mounted on some square tubular stock to make sure that the rails are level across the track and the grade doesn't change along its entire length. The is a leveling indicator to allow me to assure that the rig is leveled no matter what table I set the rig on for testing. There is also a leveling indicator to tell me when the ramp is at 0° from the horizontal along the ramp. The ramp angle can be adjusted with a fine-thread screw to 0.1° grade increments per turn, so I could calibrate it to 0.01% increments for 1/10th turns. Because I have not completed my rig construction, I have not yet developed a test protocol. But, what I am thinking is to test both static (starting) and rolling resistance. Static resistance will be determined by slowly increasing the ramp slope with the car on the rails, and note the angle at which the car starts rolling. Of course, I will need to make sure the the slope-changing activity does not materially affect the "start" by inducing vibrations. Once I know what slope will start the car, I intend to repeatedly start the car from the top of the ramp with a gentle nudge with decreasing ramp slope until I find the angle at which the car will stop rolling while on the slope. Accurately measuring the weight of the car will allow me to use the slope and the gravitational constant to compute the forces that will start the car rolling and the minimum that will keep the car rolling. That should be useful to anybody who can make other types of measurements with other apparatus designs, so long as they can put their results into actual force values. Of course, this is still a plan that I have not yet proven will work well enough. But, I do intend to give it a try when life lets me get to it.
Actually there will be some difference whether the metal truck is on top or on bottom though I don't know how much of a significance. Potential Energy = mass x gravity x height. So actually with the metal truck on top of the plastic truck, it gives a slight advantage to the plastic truck during the test since now the over mass is higher. The height difference is very slight, I'm guessing 5mm or so (don't have trucks in front of me). The difference may be small and insignificant but there is some difference. Separate trucks, equal weight will eliminate the doubt as mtntrainman suggested.
A couple of points: Rolling resistance and slinky effect: the best approach is to make the rolling resistance come mostly from the last car, which minimizes the slinky effect for the entire train while still allowing low rolling resistance for a long train. The string-lining effect is caused (mainly) by the amount of rolling resistance that is on the opposite end of the curve from the locomotive. Effect of having metal wheels on top vs bottom: Proposed effect is not really true. The change in energy per inch of drop is the same no matter whether the metal wheels are on the top or the bottom, and the wheels drop the same distance either way. Also, The force of gravity doesn't change significantly due to the wheels being a fraction of an inch farther from the center of the earth.
I guess I should make a slight correction to my reply about the distance of the drop of the metal wheels between the top and bottom position on the joined trucks: There is a tiny effect, but opposite to what Mtntrainman was thinking. Because the track is curved, vertically, the wheels on the top will move on a slightly smaller arc as they roll from the top to the bottom. So, the top wheels will drop a slightly smaller distance than the bottom wheels. So, wheels on the top, when dropping to the center of the arc, will lose a little less potential energy, and thus contribute a little less speed (kinetic energy), than when they are on the bottom. But, it is a small effect. Working from the previously posted numbers of 19.5" of radius and 20.5" of track for the arc, I get 30° of angle at the tops of the track arc. Without going through a lot of trigonometry here, I'll just say that it comes out as the difference in the energy provided between the top and the bottom position of the wheels is proportional between the separation of the wheels on the paired trucks and the radius of the track arc, which is only 1%. And, that would probably decrease to something like 0.75% when the weight of the plastic wheels is considered, because they will be dropping more when the metal wheels drop less. So, because the differences that Tracktoo measured (between metal wheels on the top to metal wheels on the bottom) was on the order of 20-40%, this roughly 1% difference is not an issue to the relative difference in the results.
For those interested, I did individual truck tests, not mounted back to back, with basically the same comparative results (Newton, maybe? ). Because weight is often sighted as important and the individual truck tests didn't account for the weight difference I decided to mount them back to back so that the assembly weight didn't change at all, therefore the measurement from plastic to metal was assured to be rolling resistance only. The only other variable was the truck frames and that was canceled by swapping the axles and retesting again with the same results. As far as repeatable test rig designs, including arc uniformity, angle, etc, for this purpose and what it is measuring, comparative roll resistance, it doesn't really matter what arc you use, just that it's sufficient to cause enough action that simple counts of cycles are available, affording a percentage change number which can be useful in various ways, including determining how many more (or fewer) cars a locomotive might pull. The only important thing is to assure all test conditions are identical therefore assuring that the comparative benefits of any single change are accurately documented. If you made a rig with a different angle, different arc, different height, the cycle count might change but the comparative percentage change from tests completed on any one rig should repeat, one rig to another. Use that as you will.
For what it is worth sometime ago in the distant past I decided to make my own disconnected logging trucks. I chose to use trucks with metal wheels to get a lower center of gravity and the tab bit extra weight that those trucks would offer with metal wheels. Now weight for those trucks is important if you are going to model a string of empties behind a Shay. I was successful in my endeavor mainly because I used metal wheels and I do not think it would have been possible to run a string of twenty disconnects behind a slow moving Shay unless I had used the metal wheels instead of plastic.
