Bump detection using accelerometer inside car, estimating suspension damping

[zacharoni16]

Homework Statement

This is not a homework question, I am working on a project to get it data from how rough the road is, mostly road bumps/pot holes. My project is for a capstone project, I want to use an accelerometer to measure the upward G's on the car frame when goes over bumps, or into a pot hole. I imagine I will have to filter out the high frequency noise from the engine and drive train.

I'm calculating the upward G force that a car frame would move up/down to going over like 3 inch high bump at 40mph, I'm getting like 22G's but I believe the suspension will will dampen the the G's experienced because the final velocity won't be much over horizontal so it should be less than half this value. do you think I would be safe with a 6G accelerometer to order for testing?

Homework Equations

I basically used constant speed, and that the bump would only redirect a few degrees above horizontal. This will be a sudden large acceleration, a brief jolt but with great force

I'm getting like 5.5 G's now lol... more sensible

basically I used F = dP/dt

Calculated initial momentum let's say the car weighs liek 2000 pounds

The initial momentum is mass * velocity = 2000 lb / (32 ft/s2) * 200 MPH * (1.46 ft/s /1MPH) = 911.8 pounds

did some collision calculations

Time of collision is very brief

found the magnitude of the collision force

Found the horizontal / vertical forces

The force was like 11000 pounds

Then the G force was like 11k/2000 =5.5 G's


----------------------------------------


I questions are, can I estimate the total damping of the cars suspension? I was thinking of measuring the fender height with the car sitting there, and then adding in weight (barbell weights) in 50 pound increments in the backseat and calculate the change in displacement? I'm not really sure if this is a way to do it, anybody have any ideas? Also, how would the damping coefficient of the suspension affect my above calculation for the G force? Not sure how to apply it.

I'm planning on starting small scale with a remote control car with a simple spring suspension, I assume I would have to take the Accelerometer data and do an FFT on it? do to all the noise, would putting the accelerator on the frame of the car in the back be the best position? The real car in the large scale is a 1994 Nissan pathfinder SUV has a ridgid frame not unibody and shocks in the rear with no springs.


The ultimate goal of this project is to be mounted on the car, detect rough road conditions and use GPS positions to record and log their locations on the road. Then I will have to come up with a way to average the roughness of the road and have it draw on say Google Maps different colors where the rough and smooth parts of the road are.

I know there is the International Rough Road Index, would this be useful here? Anybody have any idea or suggestions of the best way to approach this?

 

[berkeman]

It would seem that you lose a tremendous amount of information by putting your accelerometers inside the car. It would be far better, it would seem to me, if you put them on the "unsprung truck" part of the car -- the suspension members that attach to the wheels themselves. You could attach them to the front independent suspension members to gather the data from both wheeltracks, and just run the wires into the cab of the car to your recording equipment.

Is that an option?

 

[zacharoni16]

That could be an option, if it was on one "wheel" component would it detect bumps if the other wheels would hit a bump and the wheel it is attached to doesn't?

 

[berkeman]

That could be an option, if it was on one "wheel" component would it detect bumps if the other wheels would hit a bump and the wheel it is attached to doesn't?

Not very much, because of two reasons. First, there are two sets of shock absorbers between one wheel and another, so the isolation effect is even bigger than just between one wheel nd the suspended cab of the car. Second, since you are tuned for sensing vibration at the wheel member, the signal levels will be much higher than what you would have been detecting in the cab, so it will be much less sensitive to other inputs coming from other areas of the car...

 

[berkeman]

Oh, and third, the accelerometer is attached to the metal member that attaches to the wheel, so vibration coming from above would have to accelerate that member somehow -- there is a little compression available in the tire but not much. So pushing down on the member from the spring and shock above will not cause much acceleration of the wheel member, if any at all.

 

[zacharoni16]

AH well my idea is to put the accelerometer on the frame itself between the two wheels past the back axle. However, I'm very curious on how to estimate the "spring constant" or damping coeffient of the suspension, I was thinking of adding weight and measuring displacement of how much the frame moves. I'm not sure if this is a correct estimation or not.

Also, I'm very curious after I estimate the damping coefficient, how it would affect the "measured" G force. the way I did it above is 5.5G's and I used change in momentum,

I'm very curious how to apply the damping to change that number to get a better model not sure what equations to use or how to apply it

This is for experienced bumps to the cab of the car because my project is to classify, record, rough roads while driving

 

[berkeman]

AH well my idea is to put the accelerometer on the frame itself between the two wheels past the back axle. However, I'm very curious on how to estimate the "spring constant" or damping coeffient of the suspension, I was thinking of adding weight and measuring displacement of how much the frame moves. I'm not sure if this is a correct estimation or not.

Also, I'm very curious after I estimate the damping coefficient, how it would affect the "measured" G force. the way I did it above is 5.5G's and I used change in momentum,

I'm very curious how to apply the damping to change that number to get a better model not sure what equations to use or how to apply it

This is for experienced bumps to the cab of the car because my project is to classify, record, rough roads while driving

How about recording 5 channels at once -- at the 4 wheels, and in the cab of the car. That way you can deduce your transfer function at the same time you gather all the data you need...

 

[zacharoni16]

I need to estimate the overall damping property of the suspension to tie all the theory together and have complete information, can you point me to methods or equations. It doesn't have to be exact just a ball park estimation to compare real world results to.

 

[zacharoni16]

How about recording 5 channels at once -- at the 4 wheels, and in the cab of the car. That way you can deduce your transfer function at the same time you gather all the data you need...

I like this idea, from the data how would I deduce the transfer function? One accelerometer on each wheel then the accelerometer in the cab of the car. The one in the cab would be the affects of all four, but I think the front and rear of the car have different suspension, how would it all tie together to the transfer function

 

[berkeman]

I need to estimate the overall damping property of the suspension to tie all the theory together and have complete information, can you point me to methods or equations. It doesn't have to be exact just a ball park estimation to compare real world results to.

What you want is the mechanical transfer function from the wheel member to the "suspended truck". There are two main ways that you can derive this in the case of simple linear damping by the shocks (real shocks are often much more complicated to improve the ride) -- you can use the impulse response of the suspension system, or you can use the frequency response of the system.

To test it physically, you would in the case of the impulse response drive over a sharp bump and measure the response of the cab to the bump. From the impulse response and information about the system, you can derive how the suspension will act for other types of bumpy rides.

For the frequency response alternative, you would need to either drive the wheels with large actuators (which car companies use in their test labs), or find a test track with bumps that are spaced at a decreasing distance and drive over them and measure the response.

You may be able to calculate the transfer function in a simple case if you estimate the spring constants of the suspension, and estimate a constant friction force that would approximate simple shock absorbers. Use the weight of the car in the calculation, and you should be able to calculate a transfer function from the road to the cab.

 

[berkeman]

I like this idea, from the data how would I deduce the transfer function? One accelerometer on each wheel then the accelerometer in the cab of the car. The one in the cab would be the affects of all four, but I think the front and rear of the car have different suspension, how would it all tie together to the transfer function

I believe you would end up with at least two transfer functions -- one from either front wheel to the cab, and one from either rear wheel to the cab.

You might have to also consider the transfer function from the frame to the seats, if you are talking about what passengers feel on the "seat of their pants"...

 

[zacharoni16]

You may be able to calculate the transfer function in a simple case if you estimate the spring constants of the suspension, and estimate a constant friction force that would approximate simple shock absorbers. Use the weight of the car in the calculation, and you should be able to calculate a transfer function from the road to the cab.

Yeah I want to estimate it, I'm not sure how though. I just need a simple approximation , I was just brainstorming fender ride height and loading increasingly heavier weights into the car and measuring the displacement, but I'm not sure if that's correct way to do it

 

[berkeman]

Yeah I want to estimate it, I'm not sure how though. I just need a simple approximation , I was just brainstorming fender ride height and loading increasingly heavier weights into the car and measuring the displacement, but I'm not sure if that's correct way to do it

Testing with various load weights would be one way to do it, just be sure that at each load weight the shock absorbers are neutral (there can be some stiction in shock absorbers).

Another way would be to look up the approximate spring constants used for car suspension. I just did a Google search on some keywords, and got lots of good hits...

 

[zacharoni16]

Thanks, if I run into trouble could you help me? First I want to do this on a remote control car that has a spring suspension, adding small weight to it, would it scale to a real car as an estimate using the same method just bigger weights?

 

[berkeman]

Thanks, if I run into trouble could you help me? First I want to do this on a remote control car that has a spring suspension, adding small weight to it, would it scale to a real car as an estimate using the same method just bigger weights?

I think it mostly scales, except for my comments about the additional transfer function between the car frame and the surface of the seats.

Have you worked problems with spring-mass-friction before? That's what this is in the simplest model. You can draw a FBD of the system, and using the values for spring k and mass m and friction μ, you can calculate the transfer function versus frequency for a given forcing function acting on the bottom of the spring... Have you written those equations yet? Now would be a good time to take a first cut at them...

 

[zacharoni16]

I've done transfer functions for electronics, mechanical is kinda new to me, trying to figure it out haha