| |
| Finding fault: After a wreck, alumnus discerns what happened |
Eric Rossetter (MS 2000, PHD 2003 ME)
is a wreck detective. At the San Francisco firm, Principia, he and colleagues
use mechanical engineering and physics principles to determine from sometimes
scanty evidence, how a car accident really happened. Think of his job
as CSI for cars.
|
|
|
What does accident
reconstruction involve? |
|
|
Accident reconstruction is the
science of using all the post-accident information we have to reconstruct
the sequence of events that caused the accident. If cars are still available
we’ll go and look at crush damage. From the police report we’ll
get things like skid marks and where the cars ended up. We can use information
about the roadway surface, for example if it was raining or snowing, to
get an estimate of the coefficient of friction. We try to piece all of this
information together and reconstruct backwards what happened with the accident,
for example to get the speeds the vehicles were going or when someone started
applying the brakes. The bottom line is that our clients want to know who
caused the accident. |
 |
Is that something you
can usually tell? |
| |
Most of the time we can tell when
we have enough information. But a lack of information can certainly make
things difficult. Lots of times the accident may have happened three years
ago. While we can go to the scene and take measurements of roadway dimensions,
or see whether there was a stop sign or a stop light at the intersection,
all evidence is going to be gone from the roadway surface. We may be left
perhaps with only photographs of the vehicles involved in the accident because
they have all been scrapped or fixed. We have the police reports, but police
reports vary widely in their quality. |
|
How much does the testimony of the drivers matter? |
|
|
We try to stick with the physical evidence. Sometimes there is so little
evidence that you do want to hear what the drivers have to say. Frequently
you can look at what a driver says and, based on your scientific analysis,
you could say it definitely couldn’t have happened that way. This
can be very important in a legal case. In general, we find in this line
of work that people are very poor judges of things like speed and distance. |
|
What are the tools of the trade? How can
you discern the dimensions of skid marks, for example, from a photo? |
|
|
What we really do in accident reconstruction is work backwards. We have
a rest position for the vehicles, and a lot of times we can estimate from
evidence that was left at the scene where they came into contact. Based
on the distance between the rest and contact positions we can figure out
how fast they were going when they separated [from the collision]. We also
know that a certain amount of energy went into the crush damage of these
vehicles so we can estimate from there how fast they were going when they
first collided. And if there are skid marks we can work backwards from the
contact location to figure out how fast the vehicles were going when they
started to leave the skid marks. Skid marks can be directly measured or
estimated from photographs by comparing them to objects of a known dimension.
|
|
To what extent do you use computer simulations
as part of this? |
|
|
Quite often. You give certain inputs to the computer model and then let
it run and see if the output matches the physical evidence. You have control
over braking, throttle, initial velocity, and steering. Then you just let
it run and it will simulate the pre-impact motion of the vehicles, the collision
of the vehicles, and then their post-impact motion and final rest positions.
If you’ve done this correctly, you can put feasible values into the
program and the vehicle crush damage will look like it did on the vehicles
that were involved in the accident. Skid marks will be left where there
were skid marks measured and the vehicles will end up in the correct resting
positions. |
|
What has been one of your most interesting cases? |
|
|
I actually have a case right now that might go to trial in the next couple
of months. This is a single-vehicle accident and it’s a rollover.
In this case there was a vehicle traveling at a high speed late at night
with three occupants. It left the roadway, rolled over multiple times and
all the occupants were unbelted — well we know for sure that two of
them were. So the car ends up on its roof. One of the occupants died, one
of them was thrown 60 feet further down the road from where the vehicle
ended up. All of the occupants were over the legal limit for alcohol. This
is a case where they wanted us to come in and figure out who the driver
was. |
|
This seems like a whole different problem than
a multi-car accident. It’s not about the cars, it’s about the
driver. |
|
|
Well, it’s very similar in some respects because we still have to
figure out the exact motion of the vehicle, where it left the roadway, where
it initiated its rollover, how it rolled over. Did it rollover one time,
two times, two and a half times? How did that sequence happen? Then based
on the vehicle motion we can actually figure out where people would have
been positioned to end up where they did.
Rollovers are very chaotic events. This is one of those cases where our
initial intuition, mine and another coworker’s, was inconsistent with
the physical evidence. When you come up with a scenario, you’d like
to see all the physical evidence fit into that. There was one particular
area on the vehicle that had clear indication of contact with asphalt. Our
initial results and analysis had no way of explaining that contact. We had
to use that information and say, well maybe we’re just wrong. So we
spent a few days with a small toy model of the vehicle going over different
scenarios of how it could have rolled until we had a scenario that was physically
plausible and would leave all the physical evidence on the vehicle in the
appropriate locations. |
|
I bet the two survivors would just say the dead
person was the driver. |
|
|
Interestingly enough they did not. That was the interesting thing in this
case. They blamed each other. |
|
I notice that you also fly a small plane. Isn’t
there any irony here? |
|
|
There is a little bit of irony and some people in my field talk to me
and say, “Man you know how dangerous flying small planes is.”
I’ve read all the statistics and on an hourly basis it actually is
more dangerous than driving vehicles and I of course see tons of vehicle
accidents. But in my opinion, you only live life once and I’m careful
with what I do and I get enjoyment out of it. When you do this for a living
you realize that most accidents are caused simply by error in a driver’s
judgment. It is not usually caused by the vehicle having a mechanical failure.
I believe the same thing is true with small aircraft accidents. They are
almost entirely avoidable with sound pilot judgment. |
|
You studied here under Professor Gerdes. Were
you involved with his lane-keeping car project? |
|
|
I was the first student to really work on that system. It is two different
sides of the coin [accident prevention vs. reconstruction] but they are
both related. When I was a student working here at Stanford, I studied the
way vehicles moved, the way that throttle, braking, and steering affected
the motion of a high-speed vehicle. I used this information and these mathematical
models to design a safety system that could aid the driver in staying in
the lane. Now what I’m doing is using the same physical principles
to determine the motion of vehicles prior to and after an accident. |
|
How did you end up with your job? |
|
|
After finishing at Stanford I wanted a job that could utilize the analytical
and critical thinking skills I developed during my PhD. I also had an interest
in working in a vehicle related field and Professor Gerdes recommended I
look at Exponent, a large technical consulting company in Menlo Park that
does accident reconstruction and failure analysis. I took a position in
their accident reconstruction group. After just over a year at Exponent,
three of my colleagues and I decided to start our own technical consulting
company, Principia Engineering. |
| |
|
About Us