Is an impact required for vehicle airbags to deploy?
February 13, 2008 8:32 AM Subscribe
Is an impact required for vehicle airbags to deploy? From what I understand, deployment of the airbags is based on change in acceleration. So, for example, your car had chains attached to the frame on one end and and secured to a post on the other end, and enough slack to allow you to build up speed, when you reached the end of the chain and suddenly stop, your airbags should go off. No impact required. There are no sensors in the vehicle that are affected simply by parts of the vehicle being crushed inward. Is this correct?
"One of the simplest designs employed for the crash sensor is a steel ball that slides inside a smooth bore. The ball is held in place by a permanent magnet or by a stiff spring, which inhibit the ball's motion when the car drives over bumps or potholes. However, when the car decelerates very quickly, as in a head-on crash, the ball suddenly moves forward and turns on an electrical circuit, initiating the process of inflating the airbag."
From: http://www.chemistry.wustl.edu/~edudev/LabTutorials/Airbags/airbags.html one of the first google hits for "how air bags work."
posted by Cricket at 8:38 AM on February 13, 2008 [1 favorite]
From: http://www.chemistry.wustl.edu/~edudev/LabTutorials/Airbags/airbags.html one of the first google hits for "how air bags work."
posted by Cricket at 8:38 AM on February 13, 2008 [1 favorite]
Yes you are correct, the airbag is triggered by an accelerometer. In practice more than one since an accelerometer usually detects motion in only one direction.
posted by McSly at 8:39 AM on February 13, 2008
posted by McSly at 8:39 AM on February 13, 2008
The crushable contacts in tbe bumper is the old way of detecting impact. It was phased out after insignificant damage would cause the airbags to deploy.
posted by OldReliable at 8:43 AM on February 13, 2008
posted by OldReliable at 8:43 AM on February 13, 2008
Yes, but the trigger is a sudden deceleration (i.e. a negative acceleration), not a change in acceleration.
posted by ssg at 8:43 AM on February 13, 2008
posted by ssg at 8:43 AM on February 13, 2008
Response by poster: Wow! Superfast answers! Thank you!
"Yes, but the trigger is a sudden deceleration (i.e. a negative acceleration), not a change in acceleration."
So, if you are stopped, and you get hit from behind by someone going 50+, the airbags wouldn't go off?
posted by boymilo at 8:48 AM on February 13, 2008
"Yes, but the trigger is a sudden deceleration (i.e. a negative acceleration), not a change in acceleration."
So, if you are stopped, and you get hit from behind by someone going 50+, the airbags wouldn't go off?
posted by boymilo at 8:48 AM on February 13, 2008
The sensor doesn't set off the airbag, as such, but a control module. This matches the pulse trigger shape from the sensor against a crash pulse table - if it detects behaviour that suggests a crash that requires airbags, the airbags will deploy.
It is much more complicated that a green light from a sensor. You'd have to model a crash pretty precisely to get an airbag to go off (assuming no malfunction), as the inadvertent deployment of airbags is pretty much 'a bad thing' from a safety and marketing point of view. So lots of crash testing has been done to ensure that the module understands what a crash pulse looks like, and doesn't set the airbags off if the accident isn't big enough to warrant it, or just because water got in a sensor. The best way to model a crash is to drive the car into something...
posted by Brockles at 8:56 AM on February 13, 2008
It is much more complicated that a green light from a sensor. You'd have to model a crash pretty precisely to get an airbag to go off (assuming no malfunction), as the inadvertent deployment of airbags is pretty much 'a bad thing' from a safety and marketing point of view. So lots of crash testing has been done to ensure that the module understands what a crash pulse looks like, and doesn't set the airbags off if the accident isn't big enough to warrant it, or just because water got in a sensor. The best way to model a crash is to drive the car into something...
posted by Brockles at 8:56 AM on February 13, 2008
It would be triggered by a high magnitude acceleration, positive or negative. So if you hit someone stopped or someone hits you while stopped, either way you're going to change velocities rapidly, and thus experience a large acceleration.
posted by JiBB at 8:57 AM on February 13, 2008
posted by JiBB at 8:57 AM on February 13, 2008
So, if you are stopped, and you get hit from behind by someone going 50+, the airbags wouldn't go off?
Correct. Airbags only go off in accidents that you will (as the driver) need stopping from slamming into the steering wheel, or against the side window. You'll be primarily stopped/controlled by the seat in your example.
posted by Brockles at 8:57 AM on February 13, 2008
Correct. Airbags only go off in accidents that you will (as the driver) need stopping from slamming into the steering wheel, or against the side window. You'll be primarily stopped/controlled by the seat in your example.
posted by Brockles at 8:57 AM on February 13, 2008
I don't know that the chain-stopping-you example would work. As I understand it, there's a network of accelerometers throughout the car. As you crash into something head-first, first the front of the car decelerates, then the middle, then the back, then there's all sorts of restitution as you rebound away from whatever you hit. The control module is going to look at the entire *system* to determine whether or not it should deploy the airbag.
posted by Netzapper at 9:02 AM on February 13, 2008
posted by Netzapper at 9:02 AM on February 13, 2008
"Click and Clack" were just talking about a very similar question this past Saturday on Car Talk (for those reading this later, airdate 2008-02-09), in regards to a woman wondering whether her husband using his truck to push a dumpster would set off the airbag. See Segment 8 here [link doesn't appear to be permanent] (Short answer: no, they use accelerometers, and slow changes in pressure on the bumpers will not set them off.) The deceleration has to be greater than anything which could be achieved by braking alone to deploy the airbags, they said, so that gives you an idea of the deceleration necessary to set them off.
posted by DevilsAdvocate at 9:03 AM on February 13, 2008
posted by DevilsAdvocate at 9:03 AM on February 13, 2008
Standard airbags don't deploy in rear impacts, because they would not protect you from the typical rear-impact injury, which is whiplash as you head is forced back against the headrest. In fact, they might aggravate whiplash. Special bags that are built into the headrest for deployment in rear-impact crashes have been developed, basically they cradle your head to reduce the whiplash effect.
posted by beagle at 9:03 AM on February 13, 2008
posted by beagle at 9:03 AM on February 13, 2008
Yes, but the trigger is a sudden deceleration (i.e. a negative acceleration), not a change in acceleration.
posted by ssg
I think it's already been pointed out, but this is wrong. A sudden change in acceleration is all that is needed (one that exceeds the limits of the cars own abilities to stop and go by a predetermined amount). Negative acceleration as a concept is only useful in the case that there is only one way for an object to move, which is not the case in this situation.
posted by Grither at 9:05 AM on February 13, 2008
posted by ssg
I think it's already been pointed out, but this is wrong. A sudden change in acceleration is all that is needed (one that exceeds the limits of the cars own abilities to stop and go by a predetermined amount). Negative acceleration as a concept is only useful in the case that there is only one way for an object to move, which is not the case in this situation.
posted by Grither at 9:05 AM on February 13, 2008
not sure this is still the case, but the early versions would not deploy under 12 mph or so.
these days, with sophisticated data processing, may be different.
posted by KenManiac at 10:07 AM on February 13, 2008
these days, with sophisticated data processing, may be different.
posted by KenManiac at 10:07 AM on February 13, 2008
Grither: What matters is that there is a high-magnitude negative acceleration. That's what I would call a sudden deceleration. In other words, what matters is a change in velocity over time (this is the definition of acceleration: dv/dt), not the change in acceleration itself.
Here is an example:
You drive your car into a brick wall at a constant 3m/s (11km/h). Your airbag doesn't go off because the deceleration wasn't great enough. Just as a simplified example, say it takes 1 second for your air bag sensor to go from 3m/s to 0m/s, so you measure an acceleration of -3m/s2. Your change in acceleration is from 0m/s2 to -3m/s2.
Now you drive your car into a brick wall at 3m/s while you are stomping on the accelerator and accelerating at 3m/s2. You still have an acceleration of -3m/s2, but you have twice the change in acceleration: from 3m/s2 to -3m/s2. Should your airbag go off in this case because you have a large change in acceleration? Of course not, because what matters is the change in velocity, not the change in acceleration.
When you crash your car, your body will tend to continue to move forward at the same velocity (this is Newton's first law). The airbag helps by slowing the motion of your body, i.e. decelerating it, so that it doesn't stop very rapidly when it hits something hard (like the steering wheel). The airbag sensor measures the acceleration because it needs to know when there is a sudden change in the velocity of the car, i.e. when your body is going to need its velocity suddenly changed as well. It doesn't measure the change in acceleration because the change in acceleration isn't the problem.
However, Brockles is right that in contemporary cars this is computer controlled and more complicated than a reaching a simple threshold acceleration value.
posted by ssg at 10:31 AM on February 13, 2008
Here is an example:
You drive your car into a brick wall at a constant 3m/s (11km/h). Your airbag doesn't go off because the deceleration wasn't great enough. Just as a simplified example, say it takes 1 second for your air bag sensor to go from 3m/s to 0m/s, so you measure an acceleration of -3m/s2. Your change in acceleration is from 0m/s2 to -3m/s2.
Now you drive your car into a brick wall at 3m/s while you are stomping on the accelerator and accelerating at 3m/s2. You still have an acceleration of -3m/s2, but you have twice the change in acceleration: from 3m/s2 to -3m/s2. Should your airbag go off in this case because you have a large change in acceleration? Of course not, because what matters is the change in velocity, not the change in acceleration.
When you crash your car, your body will tend to continue to move forward at the same velocity (this is Newton's first law). The airbag helps by slowing the motion of your body, i.e. decelerating it, so that it doesn't stop very rapidly when it hits something hard (like the steering wheel). The airbag sensor measures the acceleration because it needs to know when there is a sudden change in the velocity of the car, i.e. when your body is going to need its velocity suddenly changed as well. It doesn't measure the change in acceleration because the change in acceleration isn't the problem.
However, Brockles is right that in contemporary cars this is computer controlled and more complicated than a reaching a simple threshold acceleration value.
posted by ssg at 10:31 AM on February 13, 2008
Now you drive your car into a brick wall at 3m/s while you are stomping on the accelerator and accelerating at 3m/s2. You still have an acceleration of -3m/s2, but you have twice the change in acceleration: from 3m/s2 to -3m/s2.
Some of your assumptions are off, mind you. Simplifying it like that is also, oddly, confusing the issue. The airbag probably would have gone off in that instance (I've had a near exact experience, and it sure as shit hurt when the airbag hit me in the face, I can tell you). I don't think adding numbers is helping, really. Also, there's no way the crash impulse would be a second long in that instance, and an 11kmh impact with a wall is a significant impact, particularly if the car is accelerating.
The system does measure change in acceleration (it's all part of the crash impulse). It's not as basic as you suggest.
A crash impulse is a signature of accelerations (positive and/or negative) that have been calculated as making the deployment of an airbag beneficial to the occupant. If the signatures match the impulse (or one of a certain number of parameters of impulses) it will fire. It's that simple in overview, but much more complicated in detail.
posted by Brockles at 10:53 AM on February 13, 2008
Some of your assumptions are off, mind you. Simplifying it like that is also, oddly, confusing the issue. The airbag probably would have gone off in that instance (I've had a near exact experience, and it sure as shit hurt when the airbag hit me in the face, I can tell you). I don't think adding numbers is helping, really. Also, there's no way the crash impulse would be a second long in that instance, and an 11kmh impact with a wall is a significant impact, particularly if the car is accelerating.
The system does measure change in acceleration (it's all part of the crash impulse). It's not as basic as you suggest.
A crash impulse is a signature of accelerations (positive and/or negative) that have been calculated as making the deployment of an airbag beneficial to the occupant. If the signatures match the impulse (or one of a certain number of parameters of impulses) it will fire. It's that simple in overview, but much more complicated in detail.
posted by Brockles at 10:53 AM on February 13, 2008
What matters is that there is a high-magnitude negative acceleration. That's what I would call a sudden deceleration. In other words, what matters is a change in velocity over time (this is the definition of acceleration: dv/dt), not the change in acceleration itself.
The sensor directly measures dv/dt or acceleration. The microprocessor samples the sensor output over time to compute da/dt or change in acceleration in order to determine a crash signature as Brockles explains. So both acceleration and change in acceleration are important.
posted by JackFlash at 12:08 PM on February 13, 2008
The sensor directly measures dv/dt or acceleration. The microprocessor samples the sensor output over time to compute da/dt or change in acceleration in order to determine a crash signature as Brockles explains. So both acceleration and change in acceleration are important.
posted by JackFlash at 12:08 PM on February 13, 2008
Is acceleration in reverse not acceleration? Could you trigger the thing by suddenly moving a parked car backwards?
posted by Pollomacho at 1:38 PM on February 13, 2008
posted by Pollomacho at 1:38 PM on February 13, 2008
boymilo writes "So, if you are stopped, and you get hit from behind by someone going 50+, the airbags wouldn't go off?"
Most SRS systems don't engage until a speed threshold is reached.
posted by Mitheral at 2:06 PM on February 13, 2008
Most SRS systems don't engage until a speed threshold is reached.
posted by Mitheral at 2:06 PM on February 13, 2008
Regarding getting rear-ended, some of the newer active safety systems use detectors to shift the head-rests to an optimal position if a rear impact is unavoidable. Lexus have a little animation here. Airbags aren't involved.
posted by jonathanbell at 2:21 PM on February 13, 2008
posted by jonathanbell at 2:21 PM on February 13, 2008
Could you trigger the thing by suddenly moving a parked car backwards?
Theoretically. The ignition not being on would mean they wouldn't fire, though. You'd have to move it back bloody fast, in the style of an accident, too. As has been mentioned. So, you'd have to crash into a parked car very hard (that was sat there with the engine running). And assuming it didn't have a minimum speed limit on the system.
Theoretical enough?
posted by Brockles at 2:39 PM on February 13, 2008
Theoretically. The ignition not being on would mean they wouldn't fire, though. You'd have to move it back bloody fast, in the style of an accident, too. As has been mentioned. So, you'd have to crash into a parked car very hard (that was sat there with the engine running). And assuming it didn't have a minimum speed limit on the system.
Theoretical enough?
posted by Brockles at 2:39 PM on February 13, 2008
This thread is closed to new comments.
posted by cmiller at 8:35 AM on February 13, 2008