Cedar Point guest hit by Top Thrill Dragster part while in queue

I am not certain that those are bolts holding the plate in place as it doesn't look like there is room on the backside to get a nut in there. It looks more likely that they are screws , and I bet they are countersunk to keep the head from sticking out and potentially catching on something.

If it is a screw that would explain why part of the "bolt" was still on the train - the head snapped off and the threaded shank stayed screwed in.

Or maybe my definition of a bolt being that it requires a nut is not quite right.

I've never differentiated between screw or bolt really, but I'm certainly not the source of truth either.

Just my casual observation: A screw is secured by the material it is attached to. A bolt is secured by the nut on the other side of the material.

If I'm seeing the photo on page 2 correctly, it was attached with countersunk screws. That seems more prone to failure to me, but if it was making contact with something along the track, that might not matter.

I'm no engineer and I'm not all that mechanically inclined, so take that with a grain of salt.

I have a new theory here...

It's unlikely both fasteners would fail in the same way. My thought is that the failure of a single fastener occurs due to whatever reason, which allows the bracket to rotate around the axis of the other fastener. That bracket is long enough, it appears, and positioned in such a way close down low to the track (looking at it's position in the wheel bogey), that it could rotate around that axis and then put the bracket in the path of other obstacles along the track (sensors or the track itself).

If the bracket now hits the track cross-tie at a high rate of speed, that of course is a brutal shock event and likely snaps the remaining fastener off at the head as well, turning the bracket into the projectile at question.

It's likely that during morning inspections, they both would have visually been fine and in place, with nothing unusual to be seen. The more I think of it and we see how it's mounted, where and what occurred, that it just more and more sounds like a freak incident and there may be no "at fault" here, maintenance wise or anything.

Again, without NDT on every fastener used, you may have not seen the issue until it well, failed. Then starting the chain of events.

That's more or less the only scenario I can visualize with a 25-foot path of visible damage of some kind. One connection broke free, and the piece dangled/rotated enough to come into contact with the structure...but said contact did not last long (I'm not doing the math but that train is going to cover 25 feet pretty rapidly) and the piece broke free and ricocheted into the queue and struck the victim.

Then it was stolen.

Then guilt set in.

And then it was returned.

I forget whether I said it here or on Pointbuzz, but what SteveWOA said two posts above is what I was envisioning the other night after they had the press conference. One fastener failed and that caused the bracket to rotate and dangle down and catch the cross members of the track - several of them - until the other fastener failed and sent the bracket flying. I'm wondering what the sound was like when that thing was hitting the track over 25 feet. Might have been almost like a machine gun.

I made the statement that there are probably going to be some pretty simple and inexpensive things they can do to mitigate the risk of this ever happening again. They could add a third fastener so that the failure of one would not cause the positioning of the bracket to change. They could make these fasteners a point of emphasis in daily inspections if they weren't already. They could weld the thing on there instead of or in addition to the fasteners. It seems to my naked eye that it would be more ideal to have a bolt going through with a nut on the back. Also can't tell from the naked eye whether that would even be possible though.

Or is there a way to re-engineer the sensor system (at least this part of it) altogether without throwing the proverbial baby out with the bath water?

These sensors simply tell the computer where the train is at, yes? Isn't there a simpler way to accomplish this goal rather than trying to tweak the current system that eventually failed?

This is the family therapist in me thinking here, meaning I'm looking at this from a systems-theory perspective.

What I'm saying is this:

1) Per Dave, this was the current system when the ride was designed, but it wasn't working well. The solution was to keep the current method and lengthen (re-design) the plates and make the existing sensors faster. From a systems perspective, this is simply a first-order change, which is doing more or less of something when it initially doesn't work. Sometimes it does work; i.e., your kid doesn't respond to discipline "X", but when you make "X" harsher, he does respond in the desired way.

2) Now this happened, and one could do another first-order change, like adding a third bolt. But I would assume from an engineering and maintenance perspective maybe this is an opportunity to do something different that doesn't involve another plate or another bolt attached to the train that gives everyone one more thing to check and torque and inspect and potentially fail.

I'm not a a roller coaster engineer, but if the plate/sensor job was to simply tell the computer where the train is on the course, it seems like there could be a simpler less error-prone design that isn't doing more of the same.

Or maybe this is the only way to do it. (not true)

If one of the two bolts failed, its looks like it would just be friction from the second bolt holding the plate from rotating. I have no idea how tight they torque the bolts, but its certainly possible that its not tightened enough to keep it from rotating, especially if some grease were to make its way in. Once it has rotated, then it could easily hit the track.

Intamin's answer for everything seems to be more sensors for everything. I mean, look at all of those prox switches in the low speed areas of track. As a novice system designer, I can't imagine any world where that makes sense. In software, when we design for scale, we're always trying to figure out how to do less, not more.

And that approach isn't just limited to this I believe there are two switches on every brake fin on the launch. That's a lot of stuff to break. When the fins move in groups, why not have longer moving fins? It's like, B&M used one long bar to trigger the release restraints on the inverts, instead of one per car.

Do less, not more.

To accomplish the multimove of trains, the amount of sensors are needed to do that, do they need to be a foot apart? Probably not. But as I understand it, the sensors are installed in pairs for redundancy purposes. This allows two sets of I/O to confirm the positions of the trains. If you look at other coasters with proxes you'll see two sets of them around 6-8 inches apart, sometimes on opposite sides of the track.

From my observation TTD's prox sensors are in the places where trains need to move for load/unload, beginning of launch and each section of brakes. I'm assuming there are some on the downside of the tower to allow the second train to move forward to the launch area. There could be some on the launch, but I don't remember ever seeing them.

Update on the victim...

"In a statement released to WOIO on Monday, Hawes' family confirmed that she remains in the hospital's ICU in critical condition with a brain injury."

https://people.com/human-interest/woman-struck-by-roller-coaster-pi...q2fTrDYheA

Simple solution to the bolts shearing off would be to weld the flags in place.

OhioStater said:

Or is there a way to re-engineer the sensor system (at least this part of it) altogether without throwing the proverbial baby out with the bath water?

I'm a Controls Tech at Electrolux in TN -- so I work with these types of systems on a daily basis. Unfortunately, there's not a great way to redesign the system in question. But I have a thought or two.

Dave's statement regarding elongating the plate and increasing the sensor speed seemed to pertain to when a car is travelling at speed, outside of the "transfer" areas of the ride. In my earlier post I described how sensors could be found at three different locations along the width of the track. The sensors that needed to be higher speed were likely those closest to the center line of the track, and would not be flagged by these plates. These plates are used for the precise positioning within the move blocks of the ride.

It definitely seems like the amount of sensors in the sections are overkill, but if you want to know exactly where the train is at all times, there's not another great way. You could measure tire drive rotations and crosscheck where you expect the train to be with fewer sensors, but that's not a perfect solution.

Also, I agree with welding the plates on would be a better solution, but we don't know if there's other components underneath that maintenance needs to be able to access. Do we have a photo of what's underneath when the plates are stripped from the train?

I get why there are so many sensors, but isn't it because the intention was to move the trains in pairs, like on Millennium Force? They're not actually doing that though, right? If trains are moving from standby to unload to load to launch one at a time, you don't need to know precisely where the train is, just that it entered or exited a block, like every other blocked coaster ever made.

And the brake fin grouping design, that still makes no sense.

That would entirely depend upon how the PLC is programmed. It could easily be programmed for them to be required to move in pairs, independently, or both.

I agree that they only really need to know when the train enters/leaves a block. You may want a larger grouping for getting the train nailed down to a specific spot for load/unload. The park may have requested for this type of monitoring -- who knows.

cmwein said:
It definitely seems like the amount of sensors in the sections are overkill, but if you want to know exactly where the train is at all times, there's not another great way. You could measure tire drive rotations and crosscheck where you expect the train to be with fewer sensors, but that's not a perfect solution.

Also, I agree with welding the plates on would be a better solution, but we don't know if there's other components underneath that maintenance needs to be able to access. Do we have a photo of what's underneath when the plates are stripped from the train?

If Top Thrill was being designed today, I have to wonder if some of this "precise placement" could be handled using image processing, similar to how manufacturing lines detect items and the do Q/A on them. However, I have got to believe that the system employed is probably way cheaper than image processing would be, as a sensor is honestly rather cheap in the big picture of things (though modern computing is making image detection much cheaper as well).

In terms of the screws breaking loose, I honestly have to wonder if this simply was a failure mode that was never considered. If you look at a coaster car, the torque on the wheel assemblies, etc. seems to be routinely checked on coasters and is clearly specked somewhere. But, I know not every screw is checked every day on every coaster. These screws, while critical to holding this fin in place, probably were not viewed as "safety critical" like the wheel and bogie assemblies.

You're absolutely correct - image processing is cost prohibitive for an amusement park. You also have a greater control of the environment in manufacturing, and those processes are most typically used for quality measures with safety guards implemented separately. For a roller coaster, the hardware has to be more robust in the factors it can handle. These sensors can handle that, and also double as a part of the safety system of the ride.

My plant is currently undergoing a roughly half billion dollar upgrade and expansion, and there's practically zero image processing throughout any of it.

Might not be the best example since it's still Intamin and also doesn't have a great reliability track record, but Hagrid has multiple trains moving constantly around the load/unload areas. I've yet to ride it but I imagine it would have to have some sort of more advanced/modern positioning sensor system that would be a candidate as an alternate approach for Dragster multi-move if you were to design it today.

Hagrid trains never stop moving in the station, the platform has a conveyor. The trains are in constant contact with a bus bar, but I don't see an army prox switches on the track.

I'm not exactly certain what Hagrid's uses for it's positioning, but it most likely has some kind of encoder tape and scanner along the busbar. But again, cost. I cannot see how TTD would benefit being fitted with this type of system. Hagrid's on-ride effects made that type of system be the better choice for the ride.