New bumper rules (even more new changes) and allowed materials means that now we can think, at least a little bit, of how to optimize bumper construction for our robots. A handful of folks from all over have been working together to understand things as best as we can. Aaron from 1768 and Chris from 3128 have sent me a pile of materials to test, covering legal and some not-yet-legal options. Assembled bumper test sections from @mrnoble and crew should arrive any day now.
Some ground rules for this effort:
Consider this much like 95’s build threads: interactive. I am happy to accommodate reasonable requests for additional testing, different approaches to analyses, and answer as many questions as I can. This interactivity helps to drive the thread in the direction you, the audience, are most interested in.
I am going to break up testing and results sets across many posts. To do otherwise would make us all insane and further delay the distribution of results.
We’re only going to test core examples of materials. I cannot test every permutation of of every material/vendor/whatever at every strain rate and load situation. My time and patience are finite.
Only first-order effects are being considered. There will be a million and one reasons why these data and conclusions are not 100% applicable to your own situation. See 1. The aim here is to get an 80% answer for almost everyone and provide skills so everyone can get where they need to go.
The best is the enemy of good enough. I am using a traceably-calibrated load tester with currently available tooling for these tests. This lets me walk into the lab, test over my lunch break, and have everything back in order quickly. I am not going to rework tools/machines/whatever to perfectly fit them for robot impact analysis. We would all still be waiting for results as competition started. See 3. and 4.
Answers provided here are to be considered generalized. There will be no ‘this is the best’ conclusion or ‘the right answer for you is…’ This is simply the nature of fundamental level testing and analysis like this. It will help steer us all in the right direction(s) but is not a replacement for full-scale robot-level testing. What it will provide is a level of understanding to limit the number of full-scale robot tests any one team will have to do and help steer adjustments of bumper design.
[take a peek at this post from 95’s 2024 build thread to see what the test setup looks like]
Sample Size:
~5in “tall” (as tall as a conventional bumper)
~1.5in wide (about as wide as a truss pipe, the most dangerous thing to hit)
~2.5in deep (materials were all tested at different thicknesses, but compensated to simulate 2.5in thickness)
Sometimes samples were a single noodle, in which case they were compensated for stiffness/footprint rather than thickness.
Loading Method:
Start compression
Record data when load cell reaches 1N
Compress until 1332N reached (~2g applied to 150lb robot, and 50% of the machine’s capacity), and enough load to make most foams appear solid given the sample geometry
Yeah, yeah, there’s a little c’bore in the squisher tool to bolt it to the load cell, and you’ll see indentations in the test parts from that. It’s a compromise I’m willing to make to not build all new tooling or mess with anything else.
The first thing we tested was solid vs hollow pool noodles, e.g. stuff you can buy from AM, amazon, whatever, that is originally intended to be a literal pool noodle. The following data is from a sample bumper we constructed for this purpose. The ‘fresh’ tests from solid and hollow noodles are illuminating if unsurprising.
They way to interpret these plots is that ‘distance’ is how far the bumper as squished and ‘force’ is how much force a truss pole applies to deform the bumper that distance. Integrating under the curve shows how much energy is absorbed by the bumper (both as potential energy and dissipated through losses).
We see pretty clearly that solid noodles can absorb more energy and are ‘stiffer’ than hollow noodles. Both noodles start to get ‘solid’ towards the upper end of the displacement distance. As the Force/Distance slope steepens the bumpers have a stiffer spring rate.
Testing the same segment of solid noodle bumper over and over shows a steady degradation in the material.
The shifting start point of each curve shows how far the bumper has been crushed. The damaged material has less ‘area under the curve’ so to speak, and cannot absorb as much energy as fresh material can.
The biggest change is after the first step, which is also the only time we really heard cells popping in the noodles.
I wonder if the degradation happens only at the most crushed end of the scale (80%+?) or if even repeated crush to say 50% depth results in degradation?
Thank you for doing much more professional tests!!!
The impression in the foam is due to closed cells breaking. Think popping baloons
In my target to hammer compression tests I saw a LOT of hysteresis. And found out that the backer rod I chose to start with on the target is vastly stiffer than hollowpool noodles. Almost all of the deflection was in the pool noodle.
yeah, the weird part to me is that the impression has a bump in the middle, like the edges are more squished than the middle.
like, the anvil pushes on the middle, but the “free” part is still springy, so it bends the sample around the corner of the anvil, producing more stress than there would be if the anvil covered the whole sample evenly.
Comparing first to last in the repeat solid noodle test, and synchronizing the start position (rather than the maximum load position) we see that the degraded noodle is both softer at lower displacements and stiffer at high displacements. E.g. worse across the board.
After writing that out I think I understand your actual question. To answer that: it doesn’t take much load to start popping cells. Most people can grab a noodle and pop cells under very low loads.
It is not weird. The tested material is damaged, the cells certainly pop. The un-tested material right next to it is much less damaged (or not at all) and recovers more completely.
That’s because of the space left from from the counterbored bolt (that is facing down). The very center where the c’bore is does not get squished quite as much as the rest of the material.
I love this thread! Thank you so much to 95 for getting a setup like this to help the larger frc community. I look forward to seeing the results!
I’ve seen some teams experimenting with using either TPU printed corners or a corner backer made from PVC pipe cut into 3/4" curves. I am wondering if there are plans for testing such a setup, if that’s even doable on the device you’ve calibrated.
For the backer rods, you mentioned more permanent deformation in the hollow pool noodles. Which of the two seems to have more overall impact absorption? I hope I am reading these graphs correctly, I am an amateur relative to the MatSci field. Personally I would prefer to damage a noodle via proximity to 2056 and replace between tournaments than have a less damaged bumper but it absorbs less of the impact.
Based on both the data the bumper task force collected and looking at the data James posted, solid noodles are more protective in damaging collisions. Neither is likely to be as durable as EVA or crosslinked polyethylene. I haven’t seen any testing I see as conclusive on the durability of EVA or crosslinked yet, but I’m hoping to start testing these myself soon.
Thanks. That’s exactly what I was wondering. It appears that it is not just the (presumably less frequent) big collisions doing damage, but it accumulates even with “ordinary use”.
