Now that rolls are hibernating until spring, it’s time to bring back some of the mid-week posts starting with a fun new buggy research project. Apex driver Rachael Schmitt (the same driver that built the wooden buggy B1R with CIA) will be analyzing buggy crashes from previous years and plans to conduct a few tests of her own.
Below is an except from an email that I got from her where she explains a bit more about what she’ll be doing and where she could use some help from the buggy community (i.e. you guys).
Buggy Alumni and Current Teams,
There’s been a lot of discussion on BAA recently about safety, crashes, the forces harness systems can take, and how to make buggy safer for drivers without losing the spirit that makes us love it.
I’ve excited to say that I’ve received a SURG grant to crash test a few of these rare but very dangerous crashes.
I would love, input, ideas, critique, and any willing help from the buggy community. We won’t be able to do statistically relevant testing for each buggy type and make, but we can get more information about how a general buggy and driver reacts in a specific bad crash.
Here is the general scope of the project, and to answer a few of your preliminary questions: The Transportation Research Center has generously donated the use of a crash test dummy and six accelerometers, to measure the metrics of the crash.
I’ve been given access to Emergency Medical Services records on all buggy crashes, to chart the combination of frequency, types of crash, and injuries, with input from lovely EMS personnel.
I have an old buggy (Bethany) that other mechanics and I will use part of my SURG grant to fix up to be strong and rollable. I hope to acquire other old buggies as well. Please let me know if you know of any gathering dust that want to go to a good cause!
We have the option, due to kind Roboclub and ECE members, to build a remote control steering, similar to robobuggy. This would allow us to gain speed from actually rolling the course. This option has a lot of uncontrollable variables, but would be fun.
The question in all that text is: What would you like to see from this testing? Please discuss in the comments.
Feel free to contact me with questions, concerns, input, offers of help, anything!
Rachael Schmitt
reschmit@andrew.cmu.edu
With everything that has been going on in the last few years, this seems like a great opportunity for the buggy community to grow as a whole and perpetuate the importance of safety in the sport.
Ryan says:
“been given access to Emergency Medical Services records”
As a clarification, all the information that could identify the driver/patient has been removed from these records.
Rachael says:
Yes! I should have clarified that. All proper protocols are being followed, for both the Institutional Review Board, and EMS privacy rules.
Jerry says:
Having worked on Bethany/Quicksilver with you during the previous buggy SURG, I feel the need to add the cautionary note that there is likely no amount of carbon fiber and epoxy that will make that shell an accurate analogy for a well built modern buggy. The core material is very damaged in lots of places, and weakly adhered in others, so make sure you interpret your results with a grain of salt based on the shell’s poor condition.
How will you be getting it up to a realistic top.speed? A 25mph crash is very different from a 40mph crash. Will the crash test buggy have a speedometer?
I’m certainly very interested in the results of this experiment! Is the goal to provide information to create standards of construction for buggies? This might prove impossible to enforce, but I appreciate the need for something more than the current qualitative, subjective “strength tests” used today. The data may be much more immediately applicable for specifying minimum strength ratings of harness components. If you find that the crash test dummy experiences 10g peak acceleration, you’d need harnesses rated for ~1000lb to not fail.
Rachael says:
I will have a speedometer and accelerometers. The speed will be a controlled variable, no matter how we choose to crash it.
chayes says:
If you are just trying to figure out some accurate force numbers on the drivers and possibly mount points of the harnesses then the use of an actual buggy shouldn’t be necessary. Just build a board (death sled) with steering and some sort of “nose cone” (metal “roll cage” like structure?) so the dummy doesn’t hit the hay bales and launch it down hill. Also, driving it down the entire course shouldn’t be necessary, just borrow some hay bales, find a big parking lot, set the bales up against a curb ands push your sled up to speed with a car, then brake and let the dummy roll into the bales. If you design your sled with a push structure of some sort that can be pushed by a car, this would allow you to get more data points in a shorter amount of time and you could test it at almost any time of the day or week. You can also adjust for crash angles by driving the buggy at the bales head on, off right, off left etc.
Like jerry said, bethany won’t really give you any accurate readings on shell construction, for that you would have to build and destroy newly constructed structures that you know the exact layup schedules and construction methods, and can verify the quality of the layup. Would love to see the data whenever you get there, good luck
Rachael says:
We will most likely be testing crashes ‘worse’ than hitting the haybales, as there is a lot of data and generally minimal injury and worry there.
Eric says:
I only speak from fuzzy hungover memories, but I seem to recall some pretty nasty crashes where buggies slipped right under the bales and still caught a lot of curb. One that sticks out was with a Fringe buggy several years ago that led to a broke/sprained/fucked up wrist.
I think it’s worth looking at buddy design for this, since I’m willing to bet that some buggies are a lot more prone than others.
Glenn Philen says:
I think, since the impact performance of the shell depends so much on design, the construction skill, and the particular materials chosen, (as well as any damage that it has sustained) that we may not learn much about the composites…
I would be very interested in testing or 2 harness systems that teams currently use, measuring their performance in the kind of situations that have historically led to injury.
My $0.02, feel free to take it as such.
Elmo Zoneball says:
If all they need to do is test harnesses, they could use SAE’s “Limo” as the test bed — you could drive it into the monument at 40 mph and it would hardly notice. The hard part would be getting it moving that fast.
Ben says:
Mythbusters has done some similar stuff with car crashes that we might be able to repurpose. I believe this link might help explain the pulley system that they used to pull the cars into eachother.
http://kwc.org/mythbusters/2005/11/episode_41_compact_compact_and_1.html
Rachael says:
Thanks for the responses, guys!
We’re definitely going to shoot for some controlled variable testing of a standard buggy harness and a couple different kinds of mounting points. It should be easy enough to test each at several different forces.
I also think it is important to conduct a crash test without controlling variables other than speed and angle of crash. I want a ‘real’ crash, to see how the buggy reacts. Will the carbon fiber hold or crumple, how much? Will the front wheel smash into the driver’s face? How much will the driver shift in the buggy? How will the momentum of the buggy affect it’s trajectory at the angle we crash? The accelerometers will give us interesting and accurate crash data.
I have no intentions of simulating ‘every buggy ever’. I think the fact that buggies are varied makes it even more important to get another data point (hopefully a couple) of how one will react in a bad crash. However, John Dieser, the current safety chair, will attest to the fact that there are currently rolling buggies that have a very similar construction to Bethany.
Rachael says:
I meant to post this as a general comment.
Shafeeq says:
I think a lot of the interesting unknowns are in the human side and not the buggy side.
1) The medical reports tell you what the driver’s injury was, but what was the mechanism of injury?
I.e: The harness restrained the driver ideally, but applied more force than what the driver’s body could take. Or the driver’s unrestrained head banged off the inside of the buggy. Or the driver was decelerated by a body part that isn’t ideal for the task.
2) Most of the mass is in the driver, and at the forces involved in a crash, the buggy, driver, and harness are elastic. The buggy can be analyzed well, but what happens to the driver is unknown. High-speed video that shows how much motion even a properly fitted harness permits in a crash would give buggy designers an idea of how much clear space to leave in front of the driver’s face.
3) Other studies e.g: ejection seat injuries, show that both acceleration and derivative of acceleration (jerk) are significant – humans can take a larger acceleration if it builds up smoothly. Looking at the size of the haybale, one can reason about what the average acceleration must have been in a crash, but there’s no data on how that varies during a crash. If the dummy’s sensors can provide that, it would give a target for a potential haybale replacement. to improve upon.
Ryan says:
The medical reports tell you what the driver’s injury was, but what was the mechanism of injury?
The medical reports contain this information when possible, either in the form of the driver’s account or the EMS members account of the accident.
Rachael says:
1) and 2) This is a great point, and I would LOVE to put a high speed camera in there. Any donors?
3) Yes, the accelerometers will give us ‘jerk’ forces.
chayes says:
Honestly, coming from somebody who worked on it heavily, bethany isn’t going to give you a good idea of how a shell reacts in a crash. Its been wrecked many times and rebuilt leaving a bunch of weak spots that won’t be present on a new shell. You’ve got to build multiple shells built different ways or find donor buggies that are more structurally sound (highly unlikely) and do destructive testing to get legitimate data on how shells react in crashes. A better bet to round up information on how shells react in a crash is to talk to current students and alumni who are willing to divulge layup schedules on buggies that have been involved in big wrecks and discuss with them how they failed, what the impact was that caused it and how the driver reacted.
From a death sled with a high speed camera, and some distance markers you could figure out how far forward a driver moves in different harnesses and come up with a suggested minimum distance a forward trike wheel has to be from the drivers face.
Also shafeeq makes a really interesting suggestion. This could be used to find a better solution to hay bales as well. You can try out a bunch of different “barriers” and check the accelerometer results on which is least taking on the drivers
Rachael says:
You can consider the crash ‘absolutely worst case’, but there are, and will be in the future, buggies that have been rebuilt, and/or are structurally weak, on the course.
Yes, Bethany won’t react the same way a new shell would, and that should be noted, but we shouldn’t act like all buggies that will ever roll are in pristine condition.
This was never a, “This will prove how all buggies will react” experiment. We’re simply adding another data point, one from an older buggy (I don’t see that as negating testing at all–imperfect buggies can and do pass capes), on a bad crash. And who knows, maybe the results will surprise us.
That said, I like the harness testing idea.
hold your house says:
Keep in mind that the only thing worse than no information is false information. The buggy community will, and should pay attention to whatever results you obtain but if those results are inaccurate, you will mislead people and actually do more harm than good.
If you want to crash a buggy into a wall at speed because it’s fun, great, but realize that in that scenario, and with only one buggy (regardless of condition) it is impossible to get meaningful results that won’t be misleading. Besides the inherent variation in every shell, the chances are MUCH higher that you will get only a glancing blow that both destroys the buggy and fails to provide useful information.
Connor is absolutely right about the death sled. With that setup you can control the max speed precisely with a car, control the direction of impact precisely, test multiple harnesses multiple times each, and maybe actually reach some useful conclusions.
Again, if you’re doing it for fun, great, but if you want to get actual results, you need to eliminate more variables.
Matt says:
This will certainly be valuable work. But, as far as safety goes, the biggest problem I think is actually what goes on in the buggy rooms. I personally almost blew up Spirit House (although Bordick continues to think it was Carl Nott).
bordick says:
Look, I just think Carl was trying to blow me up.
I have the same concerns as “Hold your house” Bad data could be damaging if it becomes the foundation for new rulemaking.
And Shafeeq, you ejection seat comparison is slightly flawed. The human body can handle rapid accelerations (or decelerations), but they can’t be instantaneous. There are basically 3 types of acceleration that the ejection seat guys reviewed: Instantaneous, rapid onset and gradual onset. The human body breaks with instantaneous accel/decel, which is why cars have crumple zones. However, gradual onset limits the human body to about 9 g’s (with a G suit) of acceleration because the gradual onset allows blood to leave the brain (in the piloting case) thus leading to G induced loss of consciousness. The ejection seats have essentially staged rockets that will accelerate quicker that the body can sustain (15-20g’s) but applies it properly in rapid method.
As applied to buggy design and crashes, the method to ensure rapid onset would be most like a car’s crumple zones, only those zones are occupied, typically, by hands and arms. So while the data from the pending Bethany carnage may show deceleration numbers that indicate survivability with minimal cranial/core injury, it may sacrifice hands and elbows in the crumple zone.
In fact, the more I think about this, the more I think it’s a bad idea from a data collection standpoint. This is too complex to learn much from one crash. I vote for death sled.
Bordick
Ryan says:
Why not both?
Death sled would show what happens in the absence of any crumpling (Which may be the case for some of the newer buggies), and Bethany would show how that translates to a real buggy.
Sure Bethany is pretty weak, but It could be made to pass safeties, and I’d be willing to bet there is at least one buggy that is similar in structural integrity on the course now. If the data is presented as “This is what happens when a very rigid buggy analogue crashes into bales/curb and this is what happens when a single older, weaker buggy crashes into bales/curb” Then it is entirely on the teams and sweepstakes to determine if the data is A)Meaningful to buggy and B)Worthy of making rules changes and design changes for.
The pope says:
I am late to this party but my 0.02 is: do some fn math. Cmu is an engineering school. Before you test anything, create a model the describes what you think is happening. Not easy but not impossible. Then design a test that allows you to confirm/ refine the model. That is how we learn. Test before model = pkaying.
Josh Ayers says:
One thing that hasn’t been mentioned is converting the raw data from the crash test to a prediction of injury – you measure an acceleration of X, how do you determinine if the driver would have been injured. It’s not a trivial problem.
One simple possiblity is the Brinkley dynamic response model. It was originally developed for aircraft ejection seats, and is currently being used for NASA’s commercial crew program. There’s a description in the paper below, starting on page 13.
http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20080018587.pdf
It’s basically models compression of the spine using a single degree of freedom mass/spring/damper. You apply your measured acceleration as a base shake input. The level of injury is determined by the peak deflection of the mass. It’s not terribly accurate, but it’s very simple and easy to apply.
If you want something higher fidelity, there are much more detailed injury criteria used in automotive crash testing. There is some basic information here, under the “Guidelines for rating injury measures” links.
http://www.iihs.org/iihs/ratings/technical-information/technical-protocols
In particular, the “neck axial tension” limit seems relevant for a head-on crash in a buggy. That would require getting a neck load cell added to the test dummy. See this quote, in the side impact document: Neck injury risk is evaluated on the basis of upper neck axial force, which has been shown to be the best indicator of serious (AIS = 3) neck injury.
Good luck. Be sure to post some summary of your results when you have them.