Writing a paper that advocates a new way for:
NHTSA to rank vehicles' (both new-from-factory as well as used/owned) safety by taking into account the factors that CAUSE accidents, rather than just survivability in the event OF an accident.
The factors generally fall into two categories:

agility- the ability for a car to handle any given road loading within some performance envelope, avoiding vehicle-handling-failure accidents

distraction- the tendency of a vehicle to distract its driver from the act of driving, causing head-down/late-reaction accidents

Agility is largely based on a measure of a vehicles' handling, and I just thought of option A as an alternative to original safety/handling/agility measure I was going to use, option B.


So.... the options:

A=
laptime per unit ____
(where ____ is some sort of combined measure of speed increasers that don't help handling like hp, hp/lb or ton, tq, tq/lb 1/Average-Overall-Gear-Ratio-over-a-lap. e.g. tq/(ton*AOGR))

advantages:
-gives easily order-able numbers that consider entire performance with minimum of arguing as to how to calculate
-manufacturers and owners get a clear message: "If it makes your car faster around a track without merely making it faster in a straight line, we'll say it increases the agility of the vehicle."
-much easier to explain in a 30 page paper!

disadvantages:
-requires track testing, which means standardizing a track, which means severely limiting how often tests can be done or limiting the type of loadings that the test track can be selected to apply
-may encourage the fitment of things that degrade the ability of a vehicle to endure over time, perhaps encouraging the fitment of components that require more frequent replacement, increasing the cost to the consumer as well as increasing the odds that any given vehicle is riding on worn components.
-conservative anti-racing backlash will see the measure as encouraging careless driving and shopping for the raciest car rather than the "safest"
-improvements in performance on such tests may result in ruining the ride qualities of many vehicles, putting shopping priorities of safety and comfort at odds with one another, compromising the ability of many consumers to appreciate the new standard.
-more likely that consumers might use the instructions for conducting a test and the simplified calculations to "unofficially" award their own cars a score, creating a "black market" of safety values of low quality, taking advantage of consumers who aren't aware of the need to see the government certified score.

or

B=
fully computative measure based on position and dimensions of all suspension/tire/wheel components as well as measuring center of gravity, recursively "deflecting" a suspension and measuring the estimated tire loadings and spring loadings and roll center, then estimating how/ when grip "gives up", then ranking the vehicles according to how large the envelope is.

advantages:
-removes the act of measuring from organizing any sort of tightly-regulated event, allowing repair shops and dealers to conduct measures any time they have enough time to remove the wheels from a car and can look up tire data
-eliminates driver skill variation from the measure
-more equitably associates safety with design, encouraging better designed vehicles rather than merely better equipped vehicles that are subject to performance variation with component wear
-allows for rapid variation of how the data is valued, merely changing the software/ computation method according to change in policy rather than requiring a new track arrangement or
-keeps the actual awarding of a score in NHTSA's hands, since reporting values makes the actual measurement a government responsibility

disadvantages:
-unlikely to be able to include any aero/rolling resistance effects without going to some sort of finite element analysis that defeats the easy measurability of this option
-correlation between grip regularity and actual ranking will be more easily debated than if a vehicle in fact is slower but gets around a track quicker than another car
-correlation between the computed values and the buyer thinking of their true odds of having an accident requires more faith in the consumer to delve into the value of the computation

option c: do both and list as separate measures:
-laptime per unit "accelerability"
-computative grip regularity and grip failure mode

Advantages:
-at least one part will appeal to groups that prefer one over the other, and the point is to create a measure that induces a market
-debateability of one measure is partially compensated by the other measure

disadvantages:
-may defeat the decisiveness of either measure, nullifying the market incentive to buy a better handling car.
-having two measures may quash consumer curiosity to understand by exposure to too many numbers
-clutters the safety sticker, will look messy next to 5-star rankings, regardless of whether those are as valuable to safety.


[Edited on October 3, 2007 at 9:09 AM. Reason : .]

10/3/2007 9:06:54 AM

G meter.

pick a batch of 5 different vehicles.. they will all be able to do the same thing.

you dont need track testing for safety.

just get a big parking lot, set up some cones and test.


oh wait, thats an autoX.

10/3/2007 9:14:54 AM

^right, I kind of imagined a government-run autoX / scca, but including suvs and minivans too!

cones in a lot certainly magnify the availability of testing times and sites, but decrease the quality of the information circulating, owing to variations in grip/weather at varied locations, and encourage unofficial "equivalent" testing.

I should point out that I haven't fully justified the point of doing these tests. Right now, in impact testing, it always helps to be higher and heavier, and that's created a lot more fatal car-truck interactions and single-vehicle rollovers. Incorporating a handling standard penalizes the vehicles that are magnifying their own personal "safety" level at the expense of other vehicles.

These agility measures should really be set up to make a 540i look dramatically safer than an X5 and epitomize miatas and elises over vipers and evos over suvs and minivans, etc. lower, lighter per accelerability.

Also, these measures can be performed on both existing vehicles as well as brand-new, about-to-be-crash-tested vehicles, allowing govt regulation to encourage safety throughout the fleet rather than just among new-car purchases (and then waiting for new car purchases to displace the fleet over the following decade). So... buying a used car could be compared to a new car, and the vehicles that maintain their value and thus remain the fleet longer could be the ones with higher safety scores... which in this case would mean better handling vehicles.


[Edited on October 3, 2007 at 9:28 AM. Reason : .]

10/3/2007 9:17:04 AM

No manufacturer is going to be keen on testing used vehicles. They'll point out that handling is affected by wear and tear in shocks, bushings, etc. not to mention that most people don't put on the OEM tires when replacement time comes.

Really though, I see this as more of a common sense thing than something that the NHTSA can standardize. The best handling car in the world is worthless in the hands of a bad driver. Do you really think the minivan soccer mom is going to be doing many emergency lane changes if you put her in a 540i instead of her normal POS? Probably not.

10/3/2007 11:03:27 AM

^of course driver skill is more important than whether a particular car is good handling in avoiding an accident, but I disagree that it doesn't matter. If the car in front of you slams on its brakes to avoid squashing a bunny at 80 mph, do you want to be in an Acura NSX or a cadillac escalade EXT? These things absolutely do matter.

Also, I haven't mentioned it here, but these ratings are by no means permanent. A rating would be issued with an expiration date, depending upon how old or how many miles the car had at the time of measuring. There would be different "tiers" of quality, allowing manufacturers to perform their own tests or let the government test... but the used cars would merely be an equivalent test and could be
performed by dealers or repair shops or individuals, etc...

This is where the computative measure tends to win out... you could better estimate a vehicle's performance as it decays on the basis of an inspection since fewer people would opt to regularly go to something resembling a testing event whereas every time you rotate your tires the shop could just do the work for you...

but it definitely hit me like a thunderbolt yesterday that I might be wasting my time trying to come up with a vaguely accurate performance estimation when all the pertinent stuff goes toward making a vehicle quicker on a track! so I thought I'd ask.

[Edited on October 3, 2007 at 11:48 AM. Reason : .]

10/3/2007 11:26:25 AM

well appareltly the new 335 BMW with a 17 year old kid that doesnt have any driving experience can out perform good ole' M. Schumacher himself.

10/3/2007 11:27:25 AM

^I'm not sure if you really meant that.

In terms of preventing an accident, a 17 year old who fails to turn his car on and go anywhere is infinitely safer behind the wheel than michael schumacher running around a track in a ferrari (when he was racing). The point here is to rank vehicles on their ability to FACILITATE driving safely, which I correlate with handling and wear.

seriously though, I know there's a billion questions to ask here and ways to criticise, but given that what I'll be suggesting absolutely addresses handling and both new and used cars and conducting tests that would have to be repeated every few months on as many vehicles as possible... please pick between A, B, or C and I would love to hear your reasons.

10/3/2007 12:00:56 PM

it was nothing i was just ranting about some bmw salesman a while back to said that the 335 is the best handling car out there. he said it pretty much drives itself.

10/3/2007 12:09:58 PM

I propose you rank vehicles based on the likelihood of a woman purchasing it.

Or use the lane-change maneuver speed, like Car and Driver has been doing for years.

10/3/2007 12:26:21 PM