Future vehicle noise, vibration and harshness requirements for elastomeric isolators.

Noise, vibration and harshness (NVH) are important factors affecting the customer satisfaction of automobile owners, and elastomeric isolators play a key role in determining NVH. Noise and vibration sources in an automobile are many and various, and a multiplicity of isolators are needed, including:

* Engine mounts and exhaust hangers to provide isolation from the powertrain vibration;

* suspension bushings to provide isolation from road irregularities;

* frame and subframe mounts to provide general body isolation.

Whatever the type of isolator, the key noise and vibration concerns involve the dynamic stiffness of the isolator, that is its stiffness relative to vibration. The dynamic displacements are small, seldom more than a small fraction of a millimeter.

Issues and trends

Of course, we can't ignore other functional requirements for isolators, requirements which often involve relatively large static displacements. These include:

* Vehicle dynamics' characteristics such as steering and braking;

* package (can we fit the isolators in in the space allowed which is usually too small?); and

* durability.

Unfortunately the desired characteristics often conflict. For example, vehicle dynamics may require stiff suspension bushings while NVH requires soft bushings.

In the past, vehicle development engineers tried to resolve these problems by trial and error studies on prototypes. They took a big box of isolators of varying durometer and tried out various combinations, hoping to find some magic combination that would make everyone happy.

Of course, all this was done on one or two prototypes. In a production environment, build variation would cause problems if the elastomer stiffness was not robust against pre-load variation.

We have entered the computer age, so now the objective must be to design the vehicle by computer aided engineering (CAE). This will require very large amounts of data. Figure 1 shows some typical "static" stiffness data for a suspension bushing. Note that NVH is largely concerned with displacements of less than .01 mm, and forces less than 10 N. So this data covers a far greater range of displacement and force that are required for NVH.

[GRAPH OMITTED]

NVH and dynamic data

Most NVH issues are analyzed in the frequency domain, which means that noise and vibration problems are analyzed in terms of sine waves of specific frequency (hz or cycles per second). In the case of isolators the behavior of stiffness relative to frequency is important up to 500 hz. It is very important that the correct test method be used. Currently the swept sine method is commonly used, and it could be misleading.

As an example of the issues involved, consider road noise. The inputs to the vehicle's suspension tend to be random irregularities, so if we plot the sound pressure in the vehicle interior against time we get a rather messy looking plot shown in figure 2.

[GRAPH OMITTED]

To improve the picture, NVH engineers split the sound pressure into sine waves with different frequencies as shown in figure 3.

[GRAPH OMITTED]

Road noise sound pressure spectrum

The noise is spread over a wide range of frequencies, and the amplitude of the sine waves varies enormously with frequency. Two steps are frequently taken to make the situation easier to understand:

* The frequency range is split into bands whose width is proportional to frequency, and the energy in each band is represented as a single point. Standard set of bands are used, for example, 1/3rd octave bands which have each band centered at a frequency 2 1/3 times the previous band center. For each band, the sound pressure is represented by the sound pressure level which is proportional to the logarithm average squared pressure.

The resulting graph is called the sound spectrum, and it usually gives a clearer picture of the situation. Figure 4 shows a typical spectrum for road noise.

[GRAPH OMITTED]

For road noise the spectrum deviates relatively little from a straight line fit to the data. It is clear that the physical sound pressure is concentrated at low frequency, and we can assume that the forces through the bushings are largest at low frequencies.

There is, however, a complication. The human ear is relatively insensitive to low frequencies. If we correct the sound pressure for the sensitivity of the ear, we get subjective loudness, as shown in figure 5 (the straight line fit to the data becomes a smooth curve).

[GRAPH OMITTED]

This plot makes clear that when we take into account subjective loudness we must consider at least the frequency range 40 to 400 hz. This creates a severe technical problem, for it implies that we must understand the behavior of the isolator from low frequencies and relatively large deflections, to high frequencies and relatively low deflections. To be strictly accurate, we should consider both types of deflection simultaneously. A similar situation exists for other types of noise, for example powertrain or wind noise. However, above about 800 hz airborne noise dominates, and the behavior of isolators is reduced in importance.

Role of isolator

When NVH is simulated in computer models, elastomeric isolators are typically considered to be non-linear spring damper systems as shown in figure 6, where the symbols have the following meaning:

D = displacement of hot end of isolator;

F = force applied to vehicle body;

K = stiffness;

C = viscous damping coefficient for isolator.

[ILLUSTRATION OMITTED]

The force (F) at a frequency (f) is related to these quantities by:

F = [square root of [[K.sup.2] + [(2[Pi]fC).sup.2]]D] [approximately equals] KD]

While the spring characteristic (K) is usually the most important factor, the damping (C) may also be important. It is important to note that the parameters of this simple model will be frequency dependent; and this is one reason we consider the isolator to be non-linear.

NVH is the result of the isolator applying a force to the vehicle body. In most cases the displacement input to the elastomer is much larger than any body motion, so we will ignore the body motion in what follows.

The following approximations are reasonable for road NVH:

F ~ K x D (but note that the isolator is non-linear and K varies with frequency);

K ~ 1,000 N/mm (suspension bushing);

F ~ 1N at a frequency of 100 hz;

D ~.001 mm (or less) at a frequency of 100 hz;

D is independent of K above roughly 100 hz.

Based on these approximations we see that we would like K to be small above 100 hz.

Stiffness vs. frequency

The previous discussion should make it clear that the behavior of isolator stiffness with frequency is highly important. If we apply a sinusoidal displacement to the isolator we get a sinusoidal force at the same frequency as shown in figure 7. The dynamic stiffness is the ratio of force amplitude to displacement amplitude.

[ILLUSTRATION OMITTED]

Currently the standard test procedure is to measure the stiffness for sine wave input at a regular series of frequencies, for example every 10 hz from 0 hz to 200 hz. This method is known as the swept sine method. Typical results are shown in figure 8.

[GRAPH OMITTED]

There is evidence the results may be somewhat different when sine waves at several frequencies are applied simultaneously, as happens in the real world. So to be more realistic we should apply the actual operating input and measure the actual operating force. The stiffness would than be calculated by first expressing both force and displacement in terms of sine waves. Another problem with current methods is that it requires very expensive equipment to make measurements to high frequencies, and the test procedure is time consuming.

Design implications

The ideal isolator would:

* Be stiff at low frequencies and large displacements for durability and vehicle dynamics;

* be compliant at high frequencies and low displacement for NVH.

This is the opposite of the apparent situation, but better tests are needed.

Some attempts to achieve these ends include hydromounts and dual path or voided bushings. These solutions tend to result in cost, package, weight and durability issues. It is evident that a better theoretical understanding of isolators is vital for improved design.

Robustness against pre-loads

In some situations, isolators may be subject to pre-loads arising from manufacturing tolerances. For example, if a subframe is mounted to a body at four locations, and the body and sub-frame mounting holes are not aligned, then the mounts will be pre-loaded. If the pre-load is large enough it may change the static and dynamic stiffness of the isolator. Figure 9 shows the effect of lateral pre-load on vertical stiffness. It is important that isolators for this type of application are robust against pre-loads (ref. 1).

[GRAPH OMITTED]

Testing

Improved test capability is required. Tests should involve inputs which are as close as possible to actual operating inputs. The frequency capability should be up to 500 hz, and temperature effects also need to be examined.

Ford is sponsoring a university research project with Michigan Technological University to develop quicker, cheaper and more accurate ways to measure isolator rates (ref. 2). Figure 10 shows a schematic of a set-up for measuring exhaust hanger rates using an electromagnetic shaker.

[ILLUSTRATION OMITTED]

In this schematic, the shaker applies a force to the sample isolator via a wire which is kept under tension by an appropriate static pre-load. A force transducer and an accelerometer monitor the input to the isolator. The isolator then applies a force to the large mass. This output force is determined from the acceleration of the mass. The advantage of this system is that it is relatively cheap and makes it easy to apply relatively realistic loads to the exhaust hanger. Also, measuring the output force via the motion of an inertial mass avoids resonance problems which plague rigid fixture tests. Various cables support the mass and carry the signals from the accelerometers and the force transducer.

Isolator specifications

In the past isolator specifications were usually restricted to static stiffness and stiffness at one or two low frequencies, the displacement being relatively large. In the future it can be anticipated that isolator specifications will be extended to include dynamic stiffness requirements to 500 hz, or possibly even higher. Also, the stiffness will need to be determined for appropriate inputs such as random noise. In general tolerances will be tightened so that the standard deviation of the stiffness will at most be 10% of the mean.

All of this implies that much more data will be required in the fixture, and it will only be feasible to handle the data in electronic format. The use of standard file formats would simplify the task. For example, each electronic data file should identify:

* The component;

* the environment, particularly temperature;

* static pre-loads;

* type of dynamic load applied (swept sine, random noise, etc.);

* frequency range and frequency step;

* dynamic stiffness (K) and damping (C) at each frequency; and

* tolerance range

Summary

Vehicle interior noise and vibration is dominated by elastomeric isolator forces for frequencies at least as high as 500 hz. The isolator force is a product of input and isolator stiffness. Above 100 hz the displacement is roughly independent of the isolator stiffness. Isolator stiffness tends to increase with frequency thus degrading the interior noise. Consequently, the high frequency behavior of isolators will be increasingly important in the future, particularly the need to minimize the increase of stiffness with frequency.

Assembly tolerances can result in static pre-loads, particularly for frame and sub-frame mounts. These static pre-loads may result in large changes in the stiffness of the isolators, and this may adversely affect vehicle NVH. As a result it is important that the isolator design be robust against pre-loads. It can also be expected that there will be a need for reduced part to part stiffness variation.

Current testing methods using swept sine methods are insufficiently representative of real world operating conditions, and are slow and expensive. There is a need for test methods which make it easy and cheap to apply realistic inputs to isolators, for example random noise inputs over the range 20 to 500 hz. Work is in progress to develop procedures using electromagnetic shakers.

The use of computers in the design process will result in the need for large amounts of electronic data, for example, stiffness values will be needed over a wide range of frequencies and input displacements. It will be necessary to develop standard data formats to handle this data efficiently.

References

(1.) "Effect of body decking processes on NVH quality," James L. Swayze and Wei-Zen Shih, Sixth ISSAT International Conference on Reliability and Quality in Design, August 9-11, 2000.

(2.) "Measurement of dynamic parameters of automotive exhaust hangers," Peter Maikkula, Mohan D. Rao, Scott Gruenberg and Dave Griffiths, presented at the SAE Noise & Vibration Conference & Exposition, Traverse City, MI, April 30 - May 3, 2001.

Acknowledgements

"Future vehicle noise, vibration and harshness requirements for elastomeric isolators" is based on a paper given at the Detroit Rubber Group meeting in September, 2000. "Magnetic coating for primary vehicle door seals with improved sealing performance" is based on a paper given at the September, 1999 meeting of the Rubber Division.

"Fluoroelastomers and modern engine fluids" is based on a paper given at the April, 2000 meeting of the Rubber Division.