New 2006 Rules for Frontal Impact attenuator

[tohwj]

Does anyone has any idea about how to simulate a 300kg object crashing into the impact attenuator at 7m/sec. According to the new rule, the average deceleration cannot be more than 20g.

I am not sure whether most school has the capability to carry out such a huge scale testing? Does anyone have a better alternative?

[Storbeck]

My thoughts:

Find something that weighs 661 lbs

Hang it from something very sturdy, near a strong wall.

Attatch an accelerometer

Attatch impact attenuator

Swing heavy thing into wall at 7 m/s

measure results

[Cement Legs]

Originally posted by Storbeck:
My thoughts:

Hang it from something very sturdy, near a strong wall.

Swing heavy thing into wall at 7 m/s



Isnt that called a wrecking ball.....

[Storbeck]

That's why it has to be a strong wall.

Hmm, at 20G's that would be roughly-ish 13,000 lbs of force, better make that a VERY strong wall.

[Denny Trimble]

A quick stab at it in Excel:

http://students.washington.edu/dennyt/fsae/crushzone_calcs.jpg

That's gonna take a lot of beer cans! http://fsae.com/groupee_common/emoticons/icon_smile.gif

Or at least 5" of aluminum honeycomb, pre-crushed to avoid spiking, and with enough area to provide just less than 13,200lbf to crush it.

[Z]

I am not sure whether most school has the capability to carry out such a huge scale testing?
Tsk, Tsk...

Don't you guys watch Mythbusters? I'm thinking of the episode where they dropped the Cadillac on its nose...

So for uniformly accelerated motion;

A = G = 9.8 m/s.s.

V = A x t. So for V = 7m/s, t = 7/9.8 = 0.714s.

D = 1/2 x A x t^2 = 0.5 x 9.8 x 0.714 x 0.714 = 2.5 m (ie. 20x Denny's figure of 0.125m).

So take last year's car (or steel drum filled with gravel to 300kg) and push it off first floor balcony, nose down, onto concrete slab. Use accelerometers and load cells to measure G's/forces, or use high speed video to estimate G's, then send video and job application to Mythbusters... http://fsae.com/groupee_common/emoticons/icon_smile.gif.

Z

[Chris Clarke]

D = 1/2 x A x t^2 = 0.5 x 9.8 x 0.714 x 0.714 = 2.5 m (ie. 20x Denny's figure of 0.125m).

I think Denny's distance was the minimum crush zone depth possible for a max loading of 20g's, not the height to drop the test piece from.

[James Waltman]

When I run the numbers I come up with the same results as Denny.
http://dot.etec.wwu.edu/fsae/HostedPics/CrushStructure/Sample_Crush_Structure_Calcs.JPG

If anyone wants to check my math - here is the spreadsheet (http://dot.etec.wwu.edu/fsae/HostedPics/CrushStructure/Calculating_the_Crush_Structure.xls).

It is worth pointing out that the minimum distance is the minimum crush distance and not the minimum length of the crush structure. They usually have some thickness when they are done crushing.

I rambled on about this topic once before. See here (http://fsae.com/eve/forums/a/tpc/f/125607348/m/85510167411/r/85510167411#85510167411) .

[Denny Trimble]

Damn, your spreadsheet is way prettier than mine. I'm getting lazy http://fsae.com/groupee_common/emoticons/icon_smile.gif

And yes, Z, our numbers are different because you're coming up with the distance to accelerate from 0 to 7 m/s at 1G, and I'm coming up with the distance to decelerate from 7m/s to 0 at 20g's. Factor of twenty should do it.

[James Waltman]

My main motivation in life is not to be outdone by Denny Trimble.
Actually, this isn't the first time I've run these numbers.

[Z]

Originally posted by Denny Trimble:
And yes, Z, our numbers are different because you're coming up with the distance to accelerate from 0 to 7 m/s at 1G, and I'm coming up with the distance to decelerate from 7m/s to 0 at 20g's.
Yes, that was my point. Namely, you don't even need the equations or a calculator. Just multiply your (Denny's) distance of 0.125m by 20 - 'cos with same final velocity, distance and acceleration are inversely proportional.

My other point is that at 2.5 metres drop height (shade over 8 ft) you don't even need a crane, like they did in Mythbusters.

This is an easy test to do, and it should be a lot of fun! http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

Z

[rjwoods77]

I cant remeber the team who did it but they had a video of a hoisting up a bunch of freeweights(from a gym) to a distance of about ten feet with the crash att. attached to the bottom. They then dropped it on a hard floor. Think they had bar guides. Easiest way i can think to do it is using gravity like they did.

[James Waltman]

I think that you are talking about the University of Washington video.
The one called Impact Testing here (http://students.washington.edu/~auto/media.htm).

I think that the only way to make the weight drop test work for the new rules is to have a nice accelerometer on your weights that you can sample at a very high rate. I can't think of another way to measure the ride down for that test method. You are missing a critical part of the info if you don't know the ride down characteristics.

I happen to know what the ride down graph looks like for the crush structure in that video because I crushed a replica for them on our press at WWU last spring. I'm sure that UW won't be using that design again this year.

[Z]

Originally posted by James Waltman:
I think that the only way to make the weight drop test work for the new rules is to have a nice accelerometer on your weights that you can sample at a very high rate. I can't think of another way to measure the ride down for that test method. You are missing a critical part of the info if you don't know the ride down characteristics.

James, I would suggest a high speed video camera is just as good, possibly less deceptive. I don't know about current availability/cost/frame-rates of cameras (although they are getting cheaper/faster every year), but if you can manage about 5 frames during the actual crushing of the structure, then you should get pretty accurate estimates of forces/G's.

I think that the "spiking" that you mentioned in the other thread, which might be missed by a slow camera, is not really a problem for the driver. The front bulkhead of the chassis might experience a momentary spike of 200G over a few millimetres, and this might be registered by fast accelerometers or load cells, but it is extremely unlikely that any part of the driver (eg. his brain) would feel this. Any small compliance in the seatbelt or the driver's body will smooth-out these spikes (the driver's skull would have to be pressed hard up against the front bulkhead to feel the spike...). Apparently, most brain damage in severe accidents is caused by rapid and large rotations of the skull, the brain remaining relatively stationary.

Furthermore http://fsae.com/groupee_common/emoticons/icon_smile.gif, the crash velocity of 7m/s, as shown above, is "equivalent to falling from a first storey window" (actually quite a bit less). So who has never jumped from a first storey window??? Get rid of the seatbelt, steering-wheel, and front bodywork (use "side-stick" steering), and let the driver use his legs to break the "fall"! (Ok, just kidding, but it is not such a severe impact).

Also, testing safety equipment in typical laboratory conditions is very unrealistic. It is much more likely that an impact will be at some funny angle after a spin, than from precisely "head-on". That is why I don't like carbon composite crush structures. These can give very good results in carefully controlled conditions - a precise axial crush - but hit them sideways and they snap off with little absorbed energy. I'd feel safer with metals around me...

Z

[rjwoods77]

Originally posted by James Waltman:
I think that you are talking about the University of Washington video.
The one called Impact Testing here (http://students.washington.edu/~auto/media.htm).

I think that the only way to make the weight drop test work for the new rules is to have a nice accelerometer on your weights that you can sample at a very high rate. I can't think of another way to measure the ride down for that test method. You are missing a critical part of the info if you don't know the ride down characteristics.

I happen to know what the ride down graph looks like for the crush structure in that video because I crushed a replica for them on our press at WWU last spring. I'm sure that UW won't be using that design again this year.


At least play better music to get better results. http://fsae.com/groupee_common/emoticons/icon_smile.gif

[James Waltman]

Z,
The high speed camera is a good idea.

I agree with your reasoning for how other things would diminish the actual spike felt by the driver. The short G spike is not really that big for aluminum honeycomb. Pre-crushing it is such a simple fix that it seems silly not to do it. My comments and concerns were really for other structures that would behave differently. There are certainly structures that could have bigger and longer spikes than the honeycomb does. Example:
http://dot.etec.wwu.edu/fsae/HostedPics/CrushStructure/All_spike.JPG
I admit that this graph doesn't show much ride down (or energy absorbed). I can think of other structures that would have the high(er) and long(er) peak of the above graph and also a reasonable ride down after that. Certainly this could be a problem.

I agree with your off axis comments. As I said in the other topic:
"One area of concern for honeycomb is off axis impact. It simply doesn't behave in the same manner. I don't think that the off axis loads can be as high but they could be significant. Maybe someone else has looked into this and is willing to share some information."�

[Denny Trimble]

Our team has come a long way since 2002 http://fsae.com/groupee_common/emoticons/icon_smile.gif

[Timmay!]

Wow, ODU may actually have a heads up on some schools??? (j/k)

Our advisor is one of the top guys who investigates car impacts on the east coast. Anyways, we use a software program that displays energy disapated, speed, deceleration force, time, and on and on and on. Its a pretty pricy program but I dont remember off the top of my head what its called... I believe its a *nix program though.

[Travis Garrison]

Pretty nasty spike eh? I have a feeling that most steel structures seen in fsae look something like that, anyone else want to share? anyone have an example of one of those boxed structures?

[Greg H]

We have done some testing in the past with high-speed cameras provided by the department for drop tests. I can't find any of the data, but we tested from multiple angles. I think the biggest problem they had was shearing mounts in a non-ideal head on crash. Also, has anyone looked at aluminum foam? It seems like it would act the same from any direction.

[Peter Gibbons]

I have done extensive research and cannot figure out a few things about testing the attenuator. What exactly is the ride down? I was looking at the excel sheet posted above, and did not understand what the ride down was. Also, how did you get your graph? Was there a specific thing used? Any informations is greatly appreciated.



-"It's not that I'm lazy, it's that I just don't care"-

[James Waltman]

Ride down is just deceleration.

The spreadsheet I posted has ride down units in pounds per square inch. This is because our goal was always to match a material to the G requirements. We would test different crush structure samples to find the psi they crushed at. Then we could plug that psi in and figure out what size structure we needed. So we always (perhaps incorrectly) used ride down to describe both the deceleration (in G) and the crushing pressure required to match that.

For this competition you need to keep it under 20G. That is a little abstract but easy to work out. You take the weight and multiply it by the Gs. So a 661 pound car/driver combination at 20G works out to about 13,200 pounds. So to stay under 20G you need a structure that will crush (ride down) without needing more than 13,000 pounds of force. Any more force than that will mean higher than 20G. So you need a structure that will ride down/decelerate/crush with a constant 13,000 pounds (or less). Variations in the force needed to crush the structure are variations in the deceleration (G).

Read through the other topic I linked to for some better graphs (here) (http://fsae.com/eve/forums/a/tpc/f/125607348/m/85510167411/r/85510167411#85510167411) .
The ride down should be clearer in the aluminum honeycomb or carbon fiber honeycomb graphs.

I got the graph from our instrumented press. We had a nice 75 ton Dake hydraulic press (http://dot.etec.wwu.edu/fsae/HostedPics/CrushStructure/The%20Dake%20Press%20instrumented.JPG) that was hooked up to a data acquisition computer. It would record force and displacement. By entering the sample size we could also record the psi a sample crushed at (psi of the sample not the hydraulic system). I exported the data to Excel and graphed it. Pretty simple really. If you send me an email I'll see if I can dig up one of the files.

Does that answer your questions?

[Schumi_Jr]

Is anyone considering just doing a quasi-static crush on their designs? If your material isn't terribly strain-rate sensitive then this seems like a much easier way to test. All universities have access to an Instron (or equivalent) tester, whereas drop towers are much less common... Am I overlooking something?

[Peter Gibbons]

Thanks Waltman,

That does help! I believe I am going to do a lot of testing using IDEAS first to get an idea of what material to use and what geometry, then take those results and hope they will match a drop test result.

Thanks again

[B Hise]

a lot of testing using IDEAS


ah another IDEAS team... How do you like it? Also, what method are you using to test it in ideas? I havent done much dynamic work in that program.

Bryan

[O]

For those teams that financially challenged enough to not be able to afford the money for fast data aq systems, trick sensors or high speed cameras:

http://www.stretchpak.com/old/Products___Services/Shipi...shock_detectors.html (http://www.stretchpak.com/old/Products___Services/Shiping_Products/Shock_Detectors/shock_detectors.html)

The accelerometer equivelant of temperature strip stickers.

O

[chavez]

I was wondering how everybody is doing on impact attenuator testing. The data is due in less than 2 months and we still have a ways to go on having a attenuator that meets all the requirements.

[CMURacing - Prometheus]

OK, I've been doing some number "crushing" (i couldn't resist), and I've got some ideas to bounce off of people:

1) is anyone doing this solely on numerical calculations? Getting anything to do a physical test is rather difficult, as we're financially challenged this year. We do plan to use aluminum honeycomb, which facilitates the numerical solution, though.

2) I've seen teams with some reasonably complex shapes in past years (undoubtedly determined by the body guy, who says "the nose has to look like this"). Since I've been doing slightly more than basic calculations in MATLAB, I've come up with some pretty fancy shapes that can really change how the car decelerates as the crush structure rides down. How have you guys achieved these shapes? I found a description of a team from another sort of project that said they dipped the whole piece in parrafin wax and then just machined it like a solid block, anyone have experience with this?

3) This is probably a stupid question, but since the rules talk about an average deceleration, my instantaneous deceleration is essentially irrelevant, right? I've been working to keep the instantaneous under 20, which has been troublesome.

[kozak]

i talked to one of my profs about the attenuator. i tried to see if just numbers calcs were good enough. i think for what the rules say if you make an effort at calculating something they will let you slide. but anyway he said that w/o doing dynamic testing of a piece to get the stress strain curve you can't be sure of it's properties and by stress strain i mean drop testing beacuse of potential strain hardening of the piece. this is when the instantanious force applied to the piece actually makes it seem stronger than it is. the only thing i can think of is if the manufacturer had correction curves for the material.

[Ché]

i haven't been paying too much attention to this topic but have been doing quite a bit of thinking about it for WWU this past week. Here are some of my conclusions:

First understand that high g loading is harmful (to the driver) and that staying at a lower decelleration may be better. The average acceleration is determined by the distance you slow down in. Double the distance and your acceleration is cut in half. So how does this apply to the impact attenuator? Well the energy absorbed stays constant (because the initial and final velocity are the same) but the shape of the impact attenuator has to change if the ride down distance is doubled. If you run the numbers you'll see that volume of the impact attenuator stays constant for a given ride down (pressure per area)...but the force that the structure supporting the impact attenuator sees does not stay constant. 20g x 661lbs = 13,220lbs (this is simply f = ma (sorry for using lbs for mass...) but if instead you decellerate at 5g, then 5g x 661 lbs = 3305lbs...the load the rest of the car has to support is less and the driver didn't go through such a violent acceleration. However the length of the crush structure would increase 4 times (but the volume (and mass) stays constant, so the cross section is decreased)...of course there are minimum size requirements of the rules and limited materials availible, and maybe most important to some fitting a longer impact attenuator inside an existing nose may not be possible.

also be careful of the "spiking" effect some of the materials may produce (as ways already discussed) the ride down distance also has to be the distance that the matieral can "crush" a 3inch thick material may only crush 2.25 inches before it is solid...and all my above example assumed an average acceleration, in the real world it is probably not going to be as nice

[CMURacing - Prometheus]

kozak-

One of the advantages of aluminum honeycomb (illustrated by James Waltman in Impact Attenuators for Formula SAE found on this forum) is that, after pre-crushing and before exceeding the recommended crush depth, the stress-strain is dead-on constant. So a reasonably complex problem is simplified significantly, just by material choice. This is also one of the reasons you've seen me ask about MIL-STD-401B (now SAE-AMSSTD-401B), which is the testing procedure manufacturers use to determine these properties for a given material. (Another reason is our advisor, being a former army engineer, loves mil-specs...).

Che, while you're right, and instantaneous high-G loadings are bad, my maximum deceleration is still only in the neighborhood of 22-23G, so I think I'm safe to the spirit and letter of the law.

My second question still needs help though!

[Ché]

for an irregular shape divide it into subdivisions and calculate the cross sectional area and the length....i used 3 subdivisions for ours two of one type of material and a third of a different material....it will look somewhat like a tied wedding cake.

[kozak]

i thought that was what the precrush was for but i couldn;t remember. so can i just put a block (arbitrary size) in an instron and cruch it to get a curve then match the energy the moving 660lb car has and see if it will max out the material?

[CMURacing - Prometheus]

kozak-

Yes. Even better, get a letter from the manufacturer (or a white paper, or even, if it lists the standard, some marketing material) with the crush strength stated on it, and you don't even have to crush any.

[kozak]

exactly

[Peter Gibbons]

B Hise, I wouldn't consider us an IDEAS team. I know very little about it, used it a little bit for a class, but that was all. I have done a few analysis on our attenuator and I'm not sure what to think. IDEAS doesn't actually "crush" the material, it shows the material as if always elastic and not plastic, so no crushing. However, it has given me a pretty good idea of what materials are better than others and how making minor changes to the geometry can have some affects. So for my knowledge of IDEAS, I wouldn't say it is the way to go, but it can guide you if you have no idea where to start (ME). But there may be things IDEAS can do and adjustments that can be made that I don't know exist.

As far as testing our attenuator, I think we are going to start by placing it into a hydraulic press and see how this works. Another of mine, which I'm not sure how the team thinks of it is to:
Tie a rope to a mass of 661 lbs. Swing the rope over the vertical part of a basketball pole behind the backboard and use the pulley system to lift the weight to a specific height. attach the attenuator and accelerometer and see what heppens. I actually have a basketball court outside my room which gave me the idea. I'm not sure if it work, but make sure you video tape it for weekend laughs!

[uacampbell]

There's a lot of good info on here. Where are yall getting the Al honeycomb.

Chad

[Willy G.]

Hey,

I've been designing the impact attenuator for Brown this year using Cymat Al foam. I'm projecting a weight in the range of 12-15 lbs which seems high to me. Does anyone else have preliminary weight numbers? Does this sound resonable?

[James Waltman]

That looks like it's about 10 times heavier than the honeycomb the University of Illinois is selling in the For Sale section: http://fsae.com/eve/forums/a/tpc/f/412600868/m/52810192031

I think that the honeycomb they have listed would be less than 1.5lbs to meet the rules requirements. Anyone care to check my numbers?

What is the density and compressive strength of the Cymat foam you are looking at?

[Nihal]

We were also planning to use Cymat foam as well. I think the people doing the calcs and so forth were able to achieve the decel rate with the 7% density and minimum dimensions. I'll see if I can get those guys to get ahold of you.

[CMURacing - Prometheus]

james --

yes, i actually calculated ours to be about 1.4 lbs using the UIUC honeycomb. i did this after reading the post above and wondering just how heavy our crash structure would be.

on a side note, did you guys test any pieces with changing cross-section in your press way back when? I'm looking for something to correlate my (simple) computer simulations...

[Matt Gignac]

We're using the Cymat foam ourselves, and recently learned from the company that they are no longer using Al-MMC for the foam, but something more conventional. We asked them how much the properties would change, but we could not get a definite answer.

Therefore, we're pretty much stuck doing compression tests to caracterize the foam. I seem to remember the 7% density from my calcs somewhere too (it was the 200kg/m^3), so I think it would be safe to assume that it would be at least close to what we got in the technical manual.

Matt Gignac
McGill Racing Team

[Ché]

Mike...we will be testing some "tiered" stuff on the press in the next week or so...i'd be happy to share results. email me at che.silkisshero at gmail dot com

[James Waltman]

Mike,
We only tested rectangular pieces. If I were to do it again I would use a tiered setup with a lower and higher ride down honeycomb. That's what the WWU guys are planning this year and it looks like Ché is willing to share.

What kind of simulation are you doing? You mentioned MATLAB before and Peter mentioned IDEAS (I'm assuming FEA). If you choose an appropriate material then the simulations are basically cell phone math. Excel or MATLAB make it pretty easy to do simple iterations though. I wouldn't trust FEA for this sort of thing.

[CMURacing - Prometheus]

James (and Che) --

Yes, I'm using MATLAB for my simulations. I've been thinking that my math is too straightforward, actually, its been so simple. But my question, basically is this:

My simulation "crushes" the honeycomb in the classic computer derivative method: take a very small thickness slice, find the max force it can sustain, lather, rinse, repeat until v=0. One of my teammates, though, raised an interesting quandary. How does the actual material crush, does it deform uniformly over the whole thickness until there's nothing left, or does it deform as I've simulated? Will this change my results? I could redo the sim, I suppose, crushing it the other way (energizing the front area, then as it crushes, adding thin-wall sections like I was adding thicknesses), but then its not "cell-phone" math anymore.

If we had infinite funding and/or access to materials (I've actually just been informed we have an unnamed alum at hexcel), I'd look into your tiered approach, but as is, we're buying the stuff from UIUC, and I'm trying to fit it under our body and also use as little material as possible.

[Ché]

Mike...as far as the "cell phone" math method goes if you have a given material (with a constant ride down) then the volume (mass) is constant...so if you want to "use as little material as possible" then you should be looking for something with a higher ride down rate and a light material....if you are buying the UIUC stuff then the geometry will change based on your desired decel rate, but the volume (mass) stays constant.

For least amount of material you are limited by the rules. According to the minimum size requirements, and the 20g maximum decel rate, and an inital velocity of 23ft/s...the ride down would have to be in the 360-435psi range (depends on how much the material can crush before it goes essentialy solid)