Suspension strain gauge loads vs. time

[riff_raff]

Apologies, Tim, but I think the second description is probably a better fit. Normally, 4 strain gauges are fitted, 2 "active" gauges are mounted with their sensitive axis aligned along the direction of the load, placed diametrically opposite (to cancel out beam bending) & 2 "passive" gauges are mounted with their sensitive axis across the direction of the load. The 4 gauges are wired to form a full bridge Type 3, with the active gauges diagonally opposite.

The arrangement works well if the strains in the host structure are constant over the gauged area (i.e. away from regions of stress concentration). That is probably something of an issue in the present application.

Good comment. When strain gauging something like a metal pushrod insert, it is important measure strains along multiple axes. A pushrod insert would likely have combined tension/compression, bending, and torsion loading. Thus you would need multiple strain gauges on opposing surfaces to discern the relative contributions of tension/compression, torsion or bending to the measured surface strains. A suspension pushrod normally requires some angular displacement at each end attachment joint during bump/rebound. And this angular motion at the end joints produces bending in the pushrod. The degree of applied bending moment on the pushrod is a function of the end joint stiffness/frictions and their radial loads.

[Tommy Cookers]

a force transducer made for the application would have a structural form such that the undesirable loads (eg bending moment) give negligible strain in the regions where the strain gauges are positioned (to respond to the desired loads)

that is it would have much better rejection of 'cross-axis' components of load (than would straingauging of a pushrod shape)

load cells are intended for 1 component of load (eg weighing a piece of steak), and are not certified for cross-axis use


anyway, thanks for the response, especially @speedsense

I shall re-read the thread with regard to the original purpose and topic

[gixxer_drew]

Because measuring acceleration is not going to give you push rod force

Accelerometers on the uprights and the chassis, work out the positional. It has a level of accuracy, but so does everything else we talked about here. Cheap enough to do to your family car since I dont think anyone is going to post up their team's strain gauge data.

[Greg Locock]

Nah, doesn't work well enough. Trust me if we could get away with accelerometers instead of strain gauges, load cells and wheel force transducers we would. Sometimes some manager says we can use accelerometers to scale measured forces from one car to those on a different car. That's one less guy we pay attention to.

[gixxer_drew]

Nah, doesn't work well enough. Trust me if we could get away with accelerometers instead of strain gauges, load cells and wheel force transducers we would. Sometimes some manager says we can use accelerometers to scale measured forces from one car to those on a different car. That's one less guy we pay attention to.

Forgive me that I am inexperienced in this regard, I have always used strain gauges. I presumed the accuracy was lower, I just wondered if it would be sufficient to get what he is looking for.

It was my understanding that the main issue with accelerometers is that the inaccuracy of each reading stacks up over time and the positional values degrade to be worthless. I was envisioning a short duration test to validate a mathematical model, which could be done in multiple runs (one turn at a time) recalibrating positional after each test?

[DaveW]

It was my understanding that the main issue with accelerometers is that the inaccuracy of each reading stacks up over time and the positional values degrade to be worthless....

You may be right, & possibly Greg is wrong (although I understand his comment).

It is possible to employ a model, validated by the totality of measurements that are available, to generate estimates of those parameters required but not measured. Available measurements might include airspeed, position transducers, accelerometers (sprung & unsprung), and the parameters required might include push rod loads. In effect, the model of the vehicle is acting as a "filter" on the measurements.

The quality of the results depends upon the accuracy of the model, of course, amongst (many) other things. I believe that many F1 teams use this strategy to extract the "measurements" they would like to see, but haven't actually measured.

[Greg Locock]

That's the main use of ADAMS for durability. Once you have a model that correctly predicts the laods where you can measure them, you can be reasonably confident that the model loads where you can't measure them in real life are representative.

There are many issues with trying to generate near 0 Hz forces from accelerometers. We always fit them because they are cheap and reliable, in the hope that if all else fails we'll have some data to think about, but if I take the measured loads from one vehicle, and the measured accelerations from two vehicles, i have yet to see good usable correlation between predicted and actual loads for the second vehicle on the basis of the results from the first vehicle. One of the main reasons is the shock absorbers.

[Brian.G]

Just spotted your name - great to see you on here Greg,

Brian,

[DaveW]

.......

Ouch! As I said, I do understand your comment. I would simply add that I wouldn't try to design a suspension based solely on measured PRL, & perhaps your last statement should read "One of the main reasons is the shock absorber models"

[Greg Locock]

Yes that it better, it is a modelling problem, and to be honest it is a laziness problem, there are better shock models around.

[riff_raff]

a force transducer made for the application would have a structural form such that the undesirable loads (eg bending moment) give negligible strain in the regions where the strain gauges are positioned (to respond to the desired loads)

that is it would have much better rejection of 'cross-axis' components of load (than would straingauging of a pushrod shape)

load cells are intended for 1 component of load (eg weighing a piece of steak), and are not certified for cross-axis use


anyway, thanks for the response, especially @speedsense

I shall re-read the thread with regard to the original purpose and topic

TC- You are correct about how most force transducers are intended to function. For the most part, these devices typically use strain gauges or piezo elements. However, the OP asked about the best method to collect data on the suspension movements in order to make a correlation with other data acquired from the car during a lap of the track. For sure, a single axis force transducer at the pushrod end would provide the tension/compression force components acting through the instrument's gauge point. But the full amount of deflection produced in the suspension members would require measuring the total forces/strains within every elastic component.

[speedsense]

Because measuring acceleration is not going to give you push rod force

Accelerometers on the uprights and the chassis, work out the positional. It has a level of accuracy, but so does everything else we talked about here. Cheap enough to do to your family car since I dont think anyone is going to post up their team's strain gauge data.

We do use Front/Rear Accelerometers over the axle plane of the suspension (inboard on the chassis center), mostly for US/OS navigation and calculation of Yaw (even with Yaw sensor). There are uses of accelerometers at the uprights, but mostly for accelerations and shock work. The presence of upright acceleration is used more frequently in off road racing but again for accelerations of the wheel. These are normally in conjunction with either load measurement through load cells (built into shock tops for instance) or strain gauging pushrods.
Also to add, the steering tie-rods are also strain gauged, for reasons of measuring under steer.

[Tim.Wright]

Because measuring acceleration is not going to give you push rod force

Accelerometers on the uprights and the chassis, work out the positional. It has a level of accuracy, but so does everything else we talked about here. Cheap enough to do to your family car since I dont think anyone is going to post up their team's strain gauge data.

Position won't give you a reliable measure of strain or force on the spring though. Consider the case where you have steady cornering on a smooth road (turn 8 at Turkey). The vertical hub accelerometer will be reading practically zero but you have massive loads from cornering and vertical aero force going through the spring. Its these constant "DC" loads which accelerometers miss.

Hub accelerometers are mainly useful when trying to recreate bump profiles on a circuit. Its then quite a convuluted path to get from accelerometer reading to spring load (or push rod strain). The only way I can think to do it via "calculation" would need good knowledge of the dampers dynamics and an iterative approach.

We do use Front/Rear Accelerometers over the axle plane of the suspension (inboard on the chassis center), mostly for US/OS navigation and calculation of Yaw (even with Yaw sensor). There are uses of accelerometers at the uprights, but mostly for accelerations and shock work. The presence of upright acceleration is used more frequently in off road racing but again for accelerations of the wheel. These are normally in conjunction with either load measurement through load cells (built into shock tops for instance) or strain gauging pushrods.
Also to add, the steering tie-rods are also strain gauged, for reasons of measuring under steer.

A couple of questions speedsense;
Why do you use front and rear accelerometers if you have a yawrate sensor?
What does the tierod strain gauge have to do with understeer?

Tim

[Jersey Tom]

What does the tierod strain gauge have to do with understeer?

I'd like to hear the answer to this one as well.

[gixxer_drew]

Because measuring acceleration is not going to give you push rod force

Accelerometers on the uprights and the chassis, work out the positional. It has a level of accuracy, but so does everything else we talked about here. Cheap enough to do to your family car since I dont think anyone is going to post up their team's strain gauge data.

Position won't give you a reliable measure of strain or force on the spring though. Consider the case where you have steady cornering on a smooth road (turn 8 at Turkey). The vertical hub accelerometer will be reading practically zero but you have massive loads from cornering and vertical aero force going through the spring. Its these constant "DC" loads which accelerometers miss.

I think my point on that may have been taken a bit differently than intended ... in that case you would have measured the positional change during the acceleration and if the car is steady you retain the value as calculated until it reverses and then you return to position. Also everyone keeps just saying its not "accurate" or "reliable" to do it that way, of which I full acknowledged and was aware originally. I feel like nobody is taking into account that accuracy is a relative term and the real question is how accurate do you need to be? Give up a ton of accuracy and you are learning something useful instead of just saying well, its all worthless if inaccuracy is worse than 10% (or whatever amount). It didn't sound like this is a race team or a business in the original post that can shell out $100k in sensors and strains are a huge pain in my ass its good thing i get paid to deal with them. I was just trying to think of any way to derive any useful data for cheap. I still think there is useful things to be learned on the original question via accelerometers on the uprights even without the rest of the standard array of gear.