Why are Modern road cars set up so stiff

[DaveW]

That is the general idea, yes, .... plus if it needs fixing, then engine mounts are a good place to start (in this case).

[Greg Locock]

Yes, even in large cars the engine mounts have a profound effect on secondary ride, never mind small roofless cars. Lotus did know better, for example we spent a few days looking at the Z1 and identified the engine mounts as the fundamental problem.

[Yawpower]

Objectively, the Elise has a stiff set-up, so the short answer to your question is they don't. Here is the evidence. The plot contains estimates of "Comfort Rating" (CR) plotted against Heave mode natural frequency (FoH) for a range of road vehicles all taking during a rig test. CR is an accelerometer based measure of transmissibility (lower is better), whilst FoH is a measure of stiffness/mass. The trend line shows a very rough correlation between the two parameters. Over-plotted is the result obtained from the Elise, compared with a few other examples.

So why your perception? I don't know, but the car has a good structure, and it could be that you sampled it open-topped (i.e. without the sound box, and with other distracting sound sources).

Is body displacement an effective indicator of ride quality? Or would the derivative of that displacement be a better descriptor? Or maybe the second derivative?

For example, if car A had greater displacement than car B, but the minimal displacement of car B (Compared to car A) was rather abrupt, it would score a lower transmissibility number, but would the ride quality be better?

Or possibly substantially worse, depending on the abruptness (2nd derivative) of the displacement?

[Tim.Wright]

Typically acceleration (so second derivative of displacement) is used to measure ride comfort.

So in your case, car B could measure a higher acceleration than car A if its movement was abrubt enough.

[Blanchimont]

...CR is an accelerometer based measure of transmissibility (lower is better)...

[Yawpower]

...CR is an accelerometer based measure of transmissibility (lower is better)...

Hmmm...I clearly missed that and posted based on my assumption that transmissibility was a measure of displacement.

I'm going to go with the excuse that it was after midnight...

[DaveW]

Typically acceleration (so second derivative of displacement) is used to measure ride comfort.

You are correct, Tim, but there is a little more to it than that.

Take Greg's SiL car as an example, and concentrate on the rear axle. The transfer function of the rear axle body acceleration (mounted on the top of the suspension towers) for a unit input road acceleration (both averaged left & right) would represent the transmissibility of the rear suspension if the PSD of the input to the wheel was "white" (independent of frequency).

In reality, a typical road input would have a PSD whose velocity is (close to) "white". For that reason the transfer function is differentiated, so that the Magnitude goes from acceleration (output) per unit acceleration (input) to acceleration (output) per unit velocity (input) or, as the industry seems to prefer, jerk (output) per unit acceleration (input) - no matter, the two are the same.

Now it is required to scale the differentiated transfer function, and it is helpful (arguably) to filter the transfer function. The reason for the last is to weight responses to take account of the human body's selective response to different excitation frequencies. Fortunately, perhaps, the the International Organization for Standardization (ISO) has sponsored standards for the effect (we may disagree in detail, but it is an easy solution).

Here is the same plot (shown as a continuous line) after it has been differentiated and scaled (by 0.010), and also after it has been filtered by the (an) ISO standard representing a seated human response to vertical acceleration. The scale factor is fairly arbitrary, but it happens to deliver an integral value of unity for a "high end" Mercedes road vehicle. The legend contains some statistics for the filtered results. "Int" is the magic number. The observant will note that its value is higher than the number I quoted earlier. This is because the Comfort Rating is computed as the average of the front & rear axle response, see here.

The processed results show clearly why Greg was so dismissive of the vehicle. The large peak in the transfer function centered on 17 Hz. is indicative of a large dynamic response of the vehicle, which would be disliked by most "expert" drivers.

More can be discovered about the dynamic response by exploring the test results in an alternative way. A frequency response plot of estimated suspension load per unit body acceleration is shown here. Again the solid line represents the measurements, which have the units N/gn (i.e. effective mass*go). Here, however, I have identified a second order filter from the data, and over-plotted the results as the crosses. The legend contains model parameters.

Arguably, the measurements represent a second order system fairly closely, and experience would suggest that the power unit of the vehicle is around (look at the numbers) (1340.27*2/9.81) = 270 kg mounted in a chassis of (1542.79*2/9.81) = 314.5 kg. The natural frequency of the mounted engine is 12.16 Hz, with a damping ratio of 0.116. Interestingly, perhaps, the engine acts as a "tuned mass damper" at lower frequencies, but turns around at frequencies higher the natural frequency to reduce the effective sprung mass. It is the latter that grabbed the attention of Greg.

Possible solutions are:

1. Reduce the power train mass.
2. Increase the chassis mass (there is no substitute for mass).
3. Decrease the stiffness of the engine mounts (if packaging allows).
4. Use damped mounts (the preferred solution).

It is a fairly trivial task to re-plot the identified model in Excel, and then to play with parameters. This suggests that increasing the damping ratio from 0.116 to 0.300 (not unrealistic, I think) would increase the minimum effective mass by a factor of 2...

Edit: This is a case of a subjectively poor ride that has little to do with suspension set-up. I would suggest that no (sensible) changes to suspension springs &/or dampers would solve the issue, although suspension changes might be required once it has been "fixed". It is unfortunate that, in production, items like engine mounts are often "signed off" before subjective data is available, so that issues like the one discussed above cannot be addressed properly.

[ChrisM40]

Wheel diameter certainly makes a huge difference. A friend recently replaced the 15 inch alloys on his Rover 75 for 18 inch units from an MG ZT. The Rover is a very smooth riding car, but became rather fidgety on 18 inch wheels. An actual ZT has stiffer springs and increased rebound damping and is rather stiff and hard, but putting 15 inch alloys on do make it smoother, if a bit vague. They do however look stupid an such a big car.

[DaveW]

Wheel diameter certainly makes a huge difference...

Probably caused by differences in tyre stiffness. Generally tyres with a smaller side wall will be stiffer, and will therefore be less forgiving. But you don't need to change wheel diameter - there can be large differences between two makes/types of tyre having the same rim diameter. For example, the same car, with the same wheels & inflation pressures gave a vertical stiffness of 385 N/mm on Potenza's which was reduced to 271 N/mm on NCT's, both "run in". Changes in both handling & ride were as expected.

[ChrisM40]

I certainly agree, but the higher the sidewall the more undampened total suspension travel you have. On the example above if you fit 15 inch wheels to a ZT you tend to get a bouncy ride because the dampers are so much stiffer, you end up with just the tyres moving and the steering goes wooly, which it doesnt on the 75.

I recently fitted Goodyear Eagle F1s to my car, they are certainly softer than my previous Uniroyals, both are 225/45 R18.

[DaveW]

... but the higher the sidewall the more undampened total suspension travel you have. On the example above if you fit 15 inch wheels to a ZT you tend to get a bouncy ride because the dampers are so much stiffer, you end up with just the tyres moving and the steering goes wooly, which it doesnt on the 75.

I don't disagree with your conclusions (they are pretty much as expected), but I do question your explanation....

[bill shoe]

Edit: This is a case of a subjectively poor ride that has little to do with suspension set-up. I would suggest that no (sensible) changes to suspension springs &/or dampers would solve the issue, although suspension changes might be required once it has been "fixed". It is unfortunate that, in production, items like engine mounts are often "signed off" before subjective data is available, so that issues like the one discussed above cannot be addressed properly.

Fantastic post. It answered many questions I had, and explained many things that I could not have put into questions.

If the PSD velocity of real roads is close to white, then is this the reason most basic frequency sweeps on vehicle shaker rigs are done in terms of constant-peak velocity?

Basic ride research from Olley and later people suggested that perception of motion depended on frequency. At low freqs (maybe 1-6 hz) perception was proportional to constant peak jerk, at medium freqs (maybe 6-18 hz) perception was proportional to constant peak acceleration, and at higher freqs (18 hz and higher) perception was proportional to constant peak velocity. It seems likely this research is literally correct as far as the correlations, but are people actually feeling and objecting to different things at different freqs?

A colleague of mine once studied ride feeling over a long bridge that strongly excited the primary ride frequencies of typical cars for 4 or 5 large oscillations. These motions were of course in the frequency range where correlation to jerk would be expected based on research mentioned above. Passengers in the study car were given a button to push at the moments of most extreme/uncomfortable motion, so each bridge pass would result in 4 or 5 button pushes. A wide range of passengers agreed perfectly on the timing of the "bad" moments, and when these bad moments were compared to vehicle motion they aligned with jerk peaks, so initially all seemed as expected. Then the colleague considered the perceived direction of the bad motion and realized the bad feeling lined up better with velocity peaks (opposite phase to jerk). Careful thinking, repetition, and review seemed to confirm this.

Any perspective on that experiment, or on the general issue of ride perception at different freqs merely correlating to jerk/accel/velocity vs ride perception at different freqs actually coming from different kinds of motion (in some sense)?

[ChrisM40]

I wonder if weight has anything to do with it. Modern cars are heavy, maybe 25% or more heavier than they were 20 years ago, and many are taller. In order to make the handling and 'safe' feeling acceptable they need to be stiffer.

[DaveW]

If the PSD velocity of real roads is close to white, then is this the reason most basic frequency sweeps on vehicle shaker rigs are done in terms of constant-peak velocity?

I think so.

Basic ride research from Olley and later people suggested that perception of motion depended on frequency. At low freqs (maybe 1-6 hz) perception was proportional to constant peak jerk, at medium freqs (maybe 6-18 hz) perception was proportional to constant peak acceleration, and at higher freqs (18 hz and higher) perception was proportional to constant peak velocity. It seems likely this research is literally correct as far as the correlations, but are people actually feeling and objecting to different things at different freqs?

You have crossed the boundary between objective & subjective. I am happy about the former, but not about the latter, although I did some "research" into that in the past. This might be a good read, although I disagree with her answer to pre-test question 4, she clearly has not flown in high atmospheric turbulence!

A colleague of mine once studied ride feeling over a long bridge that strongly excited the primary ride frequencies of typical cars for 4 or 5 large oscillations. These motions were of course in the frequency range where correlation to jerk would be expected based on research mentioned above. Passengers in the study car were given a button to push at the moments of most extreme/uncomfortable motion, so each bridge pass would result in 4 or 5 button pushes. A wide range of passengers agreed perfectly on the timing of the "bad" moments, and when these bad moments were compared to vehicle motion they aligned with jerk peaks, so initially all seemed as expected. Then the colleague considered the perceived direction of the bad motion and realized the bad feeling lined up better with velocity peaks (opposite phase to jerk). Careful thinking, repetition, and review seemed to confirm this.

Any perspective on that experiment, or on the general issue of ride perception at different freqs merely correlating to jerk/accel/velocity vs ride perception at different freqs actually coming from different kinds of motion (in some sense)?

There is no doubt that the natural frequencies of internal body parts are important, the primary one being the heart/lung "mode" at around 5.5 Hz. (I once tested a senior test pilot's (John Cunningham) response to vertical seat inputs. A "comfort break" in the middle of the test transformed his responses, both objective & subjective). For me, the other important frequency is in the range 20-22 Hz. which is the control loop frequency of the eye stabilization system. The beam bending mode of an open LMP vehicle is often close, can couple with aero, and hence can cause problems (I can't see the braking boards...).

Edit: My "research" taught me two things: The human being is remarkably adaptive, and when processing test results, it was important to separate out "pre-lunch" from "post-lunch" tests - actually I repeated tests so that I had a set of each for all subjects.

[Greg Locock]

I've seen several people recommend that you should review or evaluate a car after a couple of hours, no more, otherwise you just get used to it.