There are two behind-the-scenes clips on Discovery's website. One is about a machining mistake I made that wasted some time. The other shows Mark's perspective on why both designs failed.
There're actually still at least two competing theories here, but we can start by ruling a few things out. Static friction was not the problem. If it was, the package never would have started sliding at all, and just blown up in the starting position. The brakes were also not simply too tight. We set the brakes to slide at 800 pounds (see calibration section below), which is enough to decelerate the 210 pound package-plus-carriage at 3.8 g's. For the package to exceed the maximum 25 g's that it must have, it must have seen about 6-1/2 times this acceleration. Further, if the brakes had somehow gotten that tight, the package would only have slid half as far down the rails as it did.
Looking closely at the video, it appears that the package starts sliding correctly with a gradual deceleration, then comes to an abrupt halt. This indicates that the brakes were working about correctly, then something changed which drastically incrased the braking force.
We had aluminum brakes sliding on steel rails. This was a poor material selection. Aluminum can gall, causing it to effectively friction weld to the steel. A crude description of galling: as it slides, the outer layer of the aluminum flakes off, which causes bigger flakes, which causes bigger flakes, and so on. This can lead to an exponential increase in friction and an abrupt stop.
I'm inclined to believe this was the true failure mode.
In short: we didn't know any better. We knew enough to use dissimilar metals, and I vaguely remembered that you want one hard material and one soft material for bearing surfaces. I asked Corey what he thought we should use. He said aluminum, and I said that sounded good. I think we asked the rest of the team if they had any suggestions/objections with none forthcoming, but I can't truly remember. We will be forever kicking ourselves for not just using brakepads or brakepad material.
Deformation of the rails. During the collision, the rails bent. After the crash, the rails could be seen to be bowed downwards in the middle. This causes problems in a few ways:
We should have had one end of the rails attached via a non-rigid floating joint so that when the frame deformed, the rails could remain straight. We thought of this, but didn't think we had enough time to implement it. In actuality, we probably did have enough time, but didn't know we were ahead of schedule until the frame was already done and it was too late to change it. We did attempt a poor-man's version of this (which, I believe, helped significantly, but not enough) by reinforcing the rear support for the rails, but not the front support, to make the rear much more rigid than the front.
We also attempted to make the carriage and brake system as fault-tolerant as possible so it could slide well even with significant deformation of the rails. There was plenty of tolerance in the sliders so we'd have needed much more massive bending than we saw to actually jam the carriage. The brakes were held against the rails by leaf springs (admittedly made from carbon steel), which we tried to make as soft as possible, so small changes in rail thickness should have had minimal effect on braking force. I think we did a reasonable job of making the brakes fault-tolerant, so I'm inclined to think galling was the primary problem, though rail deformation may have also contributed.
We calibrated the brakes using a chain hoist (from the shop) and a dial scale (purchased as part of our budget). We used the hoist in line with the scale to tow the carriage down our rails. We could then read the braking force off of the scale and adust the clamping bolts until we had the desired 800 pounds. Once adjusted, we actually found the force to be consistent over the full length of the rails, even when sliding past our welds (confirming that we did not overly deform the rails with our welds).
We set the brakes to 800 pounds because we calculated that that would cause the package to slide 80 percent of the way down the rails, giving us quite a bit of sliding room but some margin to not hit the end.
We thought both teams would succeed at this challenge (we were wrong), so we went with the design that would have the minimum maximum acceleration (our definition of best or optimal). This is achieved by decelerating with a constant acceleration over the longest straight-line distance possible. We built rails the full length of our allowed envelope to maximize the length of travel. Friction brakes were ideal because they provide constant force, which means constant acceleration if you believe F=ma.
Realize that this was a 3 day build. That includes sourcing parts, and in this case, it also includes getting to know the shop, which we had never been in before. I'd rather not talk about how long it took us to get the plasma cutter working that first time. The result of this time pressure is that we can't design, then order parts, then build, then test, then iterate. Instead everything gets jumbled together with very little testing and no iteration.
Amy, as team captain, did a great job leading the initial brain-storming. We all drew our ideas from the blueprint challenge on the whiteboard then explained them. All the ideas involved constraining the box to one degree-of-freedom motion and some system for removing the energy. I point out that we have two orthogonal (unrelated) decisions to make:
Somewhere around here, Dan starts tearing apart the truck because we know whatever we go with, we're going to want to anchor to the frame of the truck, and it'll take some doing to get clear access to it. I think we all felt better with somebody making real, physical progress on the build.
We pretty quickly decide to go with rails over the 4-bar mechanism both for their simplicity and because the geometry allows for the longest straight-line motion. Corey and Amy then start work on the rail system and helping Dan dig down to the truck frame, leaving Andrew and myself to analyze energy-removal systems and report back.
Andrew and I look at a bunch of options and slowly start ruling them out or deciding we don't like them. Springs either result in the package bouncing back (unless combined with another method) or finishing the event with a lot of energy stored in the springs. We have trouble sourcing dampers in the size we need that we can get overnight within our budget. I don't really remember what other options we looked at, but we decided friction brakes were the simplest thing that would work and had the nice property of supplying constant force. Sometime on the second day, we presented several options to the rest of the team, going over the strengths and weaknesses, and they agreed with our plan to go with brakes.
Then we all worked on fabrication. I mostly worked with Corey on the brakes themselves. When we got to the end and we had some extra time, we built two backup systems.
We built two additional systems that would kick in if the brakes did not provide enough braking. Unfortunately, we had the opposite problem, which is harder to protect against, so they could not save us.
The first system was trivially simple. At the very far end of the rail, we added a simple pusher plate attached to two coil springs to cushion the collision if the box made it all the way to the end. The springs provided about 6 inches of compression, which wouldn't be enough on its own, but would help if we were mostly slowed down when we reached the end.
The second backup system was a pair of dampers. This is what we're arguing about at lunch. My concern was that attaching dampers to the package would increase the stopping force, which might actually make things worse rather than better. I broke the stalemate when I figured out we could attach them via a slack cable, so they would only kick in if the box slid too far.
I've always liked strange conservation laws. I'm just going to quote Andrew for this one since he was our team's victim (culprit?):
" So the funniest thing that happened on this episode is what we labeled the "Conservation of H's." You might recall the plasma cut sign I made that said "CRUSHER" on the front of our (Red Team) truck. This was not the first sign created. The first sign was beautiful! I took my time and made the cuts look really good. After finishing the sign, I showed it to Corey, Dan, and Amy and they loved it! Then Eric saw it and asked one question, "Why is there an extra H?" WTF?! Yep, I spelled "CHRUSHER!" It was pretty funny, only because everyone else saw it and didn't realize the misspelling and it took Eric to point out the obvious error. Unbeknownst to us, on the blue team, Joe created a shirt for the competition day that was suppose to say "Mr. Murphy;" however, in his haste, he left the "h" off, having to draw it in post completion of writing the word. Thus the label "Conservation of H's." " - Andrew
One of the behind-the-scenes clips on Discovery's website is about how I drilled some holes wrong and had to redo them. The more you rush, the more mistakes you make. In the real world, you work slowly enough so that you make few, if any, mistakes. In this environment it makes sense to try to find the place where the amount of time you're wasting by redoing stuff is less than the amount of time you're saving by rushing. So, yeah, I made a mistake that cost time, and it certainly won't be the last. However, I'd say that even with this, I was a productive member of the team, and it didn't really matter anyway because it was a relatively minor setback and we had extra time anyway. Corey can maybe take 10 percent of the blame for not noticing the problem when he was tapping the holes I'd drilled. Blaming Amy is just silly. The team leader needs to know what's going on with everybody, but expecting her to check details like that is a bit much.
When Amy selects me, she says I look like a controls person. She had seen my application video on youtube and knew my particular specialty is in control systems. We didn't end up using any controls on this build.
We actually went around saying who the strongest and weakest person on the team was (besides ourselves). Of course, they only showed the negative part. Right before she called me the weakest member of the team, Amy also called me the strongest member (not as contradictory as it sounds). Also, I can't speak for the rest of the team, but I didn't take this particular eliminations too seriously because I kind of figured we may have failed, but the blue team failed more, so it'd be a blue team member getting eliminated. I think I called Corey the strongest.
I was terribly disappointed that we didn't succeed. We did so many things right and (probably) just one thing wrong (bad material selection for the brakes). But without any ability to do a high-speed test, we had to do everything right on the first try. Despite the results, we were really happy with our team, and I at least came out of this hoping to have as many of the same teammates for the second challenge as possible.
Back (Eric's TBBT Home)