Which Accelerometer is Best for Measuring G-Force in a Car-Bicycle Collision?

[ConnorM]

I am trying to decide on an accelerometer to use that will help me measure the g-force that a crash test dummy’s head experiences during a car on bicycle collision. The dummy will be mounted on a bicycle and launched at 20km/h, it will then be struck from behind by a car driving at 30km/h. I expect the main part of the collision to occur over approximately 15ms.

My major need for this accelerometer is that I need to be able to sample it at 10,000 samples/second. I am currently looking at the ADXL1001 (+/-100g), http://www.analog.com/en/products/mems/accelerometers/adxl1001.html#product-quality. Do I need a accelerometer with a high linear frequency response like this one in order to reliable sample it at 10,000 samples/second?

I am a mechanical engineering student working on this project, so there is a lot that I don’t know and I’m looking learn how I can pick the best accelerometer for my application.

 

[JRMichler]

Start by reviewing your earlier thread: https://www.physicsforums.com/threads/accelerometer-impact-sensing.926724/. Then you need to spend several hours (or more) studying the data sheet for your accelerometer. Pay attention to antialiasing filters, transverse sensitivity, mounting recommendations, frequency response, sample rate, etc.

Is there a grad student working with sensors who you can get advice from? Or a professor who teaches a class in dynamic measurements? An excellent book on the subject is: Applied Measurement Engineering, by Charles Wright.

 

[ConnorM]

Thank you. I'll do some reading then and post back with any questions.

 

[Nidum]

Try using back of envelope calculations , ball park figures , data from other peoples work and your own intuition to estimate at least roughly what the response motion is actually going to be before doing any more design work on this instrumentation system .

 

[ConnorM]

After doing some more reading I think that an ADXL377 accelerometer (1600Hz X and Y) should be enough for my case. I was looking at "Vehicle Crash Mechanics", by Matthew Huang, and the SAE J211 standard for electronic sensors in impact tests. These resources talk about "Classes" for accelerometers based on their placement around the body, for me since I am placing my accelerometers inside the head they would be Class 1000. Class 1000 accelerometers based on the SAE J211 standard require a frequency range from F low = 0.1 Hz, F high = 1000 Hz, and F cutoff = 1650 Hz.

This range fits almost perfectly with the ADXL377's range. Now I understand that the rate that I sample at has nothing to do with my sensors frequency range, so I can sample at 10,000 S/s and my limiting factors just have to do with my microcontroller/SD writetime/etc.. In "Vehicle Crash Mechanics" Huang states that sampling at a rate that is 5 times your frequency range should allow for a perfect recreation the acceleration change over time (1650 Hz * 5 = 8250 Hz --> 10,000 S/s exceeds this)

Does this seem to make sense now? I just want to make sure that I understand what I am trying to do.

 

[JRMichler]

You have the frequency part correct. Next is to check the peak acceleration against the full scale of the accelerometer. And making sure to mount the accelerometer correctly for the magnitude and direction of acceleration that you will be measuring. Also accelerometer output voltage against the A/D converter voltage range. Etc.

 

[ConnorM]

I predict that this accelerometer's range (+/-200g) will be quite a bit larger than the actual g-force range in my crash test. I think this because the car will be traveling at 30km/h and the bike at 20km/h, the collision will be a rear collision so deltaV=10km/h. This is a relatively low velocity for a crash and I think that a large amount of impact energy will be absorbed by the bicycle/helmet/etc.. However, there is a possibility if all goes well that we will have the car travel upwards of 50km/h.

Anyways,

If I could find a slightly lesser g-force range maybe +/-100g I could achieve a bit better resolution. The accelerometer is going to be mounted about the heads center of gravity, I will need to look into mounting it so that I do not cause any extra damping.

I am using a Teensy 3.6 with the ADC running 12 bit resolution (4095). So +200g=4095, -200g=0, 0g=2047.

Based on this do you think it would be worth looking into some accelerometers with similar frequency response but a lower g-force range? Or do you think that a +/-200g range is reasonable?

 

[Tom.G]

Or do you think that a +/-200g range is reasonable?

You get 1/10th g resolution with that combination, how much do you need? But have you considered the g-force when the head meets the pavement?... or the car windshield?... or the lamp post at the side of the road?
The impact strength of the skull is probably the limiting factor. Unless you just want to know if the neck breaks.

Perhaps some of the folks in the Biology and Medical forum would have some insights.

 

[ConnorM]

The dummy will have 3D printed skin, a helmet on, and I expect the bike will absorb some impact energy (The relative velocity between the bike and car will only be about 10km/h for the majority of testing), so I don't expect very large g-forces that may occur in say a metal on metal collision. However, like you said I hope to be able to relate this g-force measured to the probability of neck injury/concussion/etc. potentially using HIC (head impact criterion).

As for resolution being able to measure accurately a difference of 0.1g would make me happy. This is an engineering project through school so we have quite a limited budget and we aren't going for industry standard measurement requirements

 

[JRMichler]

With 0.1 G resolution, your largest errors will come from misalignment between the accelerometer sensitive direction and the acceleration vector. Keep in mind that the acceleration vector has six degrees of freedom (three linear and three rotation), while the accelerometer is measuring acceleration in one linear direction. Whatever you do, it will not be perfect. Just do the best you can, and put a paragraph in your report on recommendations for further research. You could headline it "What we would have done knowing what we know now and if we had an unlimited budget and time". :)