Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02255523 1998-12-07
.~,
C A N A D A
TITLE: A Method and an Apparatus for
the ~Am;n~tion of Installed
Shock Absorbers (I)
v~NlORS: KARL EGLE
FRAh-K BEAUJEAN
CA 022~23 l998-l2-07
FIELD OF THE lNV~NlION
The present invention relates to a method and an
apparatus for the examination of shock absorbers installed on
vehicles.
BACKGROUND OF THE INVENTION
Apparatus for the measurement of the efficiency of
running gear of a motor vehicle without having to remove the shock
absorber from the vehicle, have already been proposed. There are
various different methods for this purpose, as for example using
a vibratory platform, on which one wheel of the motor vehicle is
stood and which is reciprocated with a suitable amplitude and a
variable frequency in a vertical direction in order to subject the
wheel to oscillations, the reading for the force exerted by the
running gear on the vibratory platform being evaluated.
The Method of the European Shock Absorber Manufacturers'
Association (EUSAMA), the largest organization of its kind,
involves a uniform testing method for shock absorbers in the
installed state. The examination is in this case performed using
a vibratory platform, which with the aid of an eccentric drive is
caused to oscillate with a stroke of exactly 6 mm. During
examination the measuring setup is excited to run at approximately
23 Hz and freely vibrates down to O Hz. The respective sinusoidal
dynamic wheel load is measured by sensors and stored. prior to
producing the vibrations the static wheel load Fs is measured. The
stored, dynamic wheel loads are investigated to find their minimum
Fmin. From the data the so-called relative road adherence A is
calculated as a percentage: A = Fs / Fmin/Fs.
CA 022~23 1998-12-07
In a real application the road adherence value is
corrected, since particularly light vehicles have only poor road
adherence figures. For evaluation the following table is employed:
100% - 60% good
59% - 40% fair
39% - 20% poor
19% - 0% insufficient.
The EUSAMA Method is widely employed but the data
provided by it depend not only on the condition of the shock
absorber but also on design features, which are characteristics of
the wheel suspension, as for example the ratio between the spring-
suspended mass and the non-spring-suspended mass, the type of
tires, the tire pressure and the type of suspension and
furthermore on the conditions of measurement, as for example the
loading of the vehicle or the temperature of the surroundings or
the operating temperature of the shock absorber. The dependence on
a large number of parameters means that in practice the value
defined by the EUSAMA Method is prevented from being
representative for the condition of the shock absorber itself.
Accordingly the EUSAMA method does not lead to any reproducible
and reliable results.
Similar problems have occurred with other known
measuring systems for shock absorbers, wherein the dependency of
the measurement on a plurality of physical parameters of the
running gear means that the measured data must be interpreted
using table or the like. This is something leading to high degree
of approximation and large ranges of error and for different types
of vehicle different tables are necessary, dependent on which
shock absorbers are employed.
CA 022~23 1998-12-07
The European patent publication 0476746 A1 discloses an
apparatus for testing running gear of a vehicle, which like the
EUSAMA Method operates with a vibratory platform means. A computer
processes the data and from such data produces a transfer function
for the wheel suspension, which is looked upon as a system with a
plurality of system elements. On the basis of a model of the
system the computer derives the values of the system elements from
the transfer function and with the calculated values of the system
elements calculates the values of at least one measurement
variable and compares such value with a stored value of the
dimensional variables. The computer assesses the wheel suspension
from such comparison and displays the result of the comparison.
Here as well suitable tables are necessary in a data base, which
must hold the values for all current types of vehicles and tires.
Furthermore the evaluation of the running gear by way of the
individual system variables is inaccurate.
The patent publication WO 96/07882 discloses an
apparatus for the examination of shock absorbers of motor
vehicles, in the case of which the wheel on a vibratory platform
is caused to vibrate and on the basis of the amplitude of
vibrations at different frequencies the results of measurement are
derived. In the case of this apparatus the resonance of the
suspension is sensed and then the frequency of the stroke and
accordingly the energy supplied is varied. The damping force of
the shock absorber is plotted against speed. In the case of this
model some important effects, as for example internal friction, as
will be described below, are not taken into account.
The European patent publication 0647843 A2 discloses a
system for measuring the damping coefficient of a shock absorber
mounted on a motor vehicle, as is recited in the preamble of claim
CA 022~23 1998-12-07
1 herein. The damping coefficient is calculated as a function of
the derivation of the frequency of the relative phase function
between the force transmitted to the vibratory platform and the
movement of the vibratory platform. In this apparatus an imperfect
vibration model is taken as a basis, the value termed the "degrees
axle damping path" only applying for the particular model employed
so that in practice there are substantial departures between the
results of this method and the actual degrees of damping of the
shock absorber being examined. In particular the effect of
external friction is not taken into account.
Methods in existence so far, in which evaluation takes
place on a basis including a vibration model, suffer from deficits
in theory. This is for instance an inappropriate dissection of the
simple vibration model into different model parts for different
characteristic frequency ranges; the inappropriate mathematical
- description of the resonant frequencies. In part methods on the
basis of free undamped elastic vibration systems are utilized. The
basis is however forced and heavily damped elastic vibration
systems; the invalid simplification or omission of parts of terms
of the mathematically described model; leaving out external
friction within the simple one quarter vehicle model and
infringing against the linear force law by progressive suspension
springs. Finally, the distortion of the appropriate sinusoidal
signal form by the different compression and tension stage
behavior is not taken into account in the known methods, or, in
other words, the measurement signals are not brought back or
compared to the sine-shaped signals upon evaluation.
Basically the different behavior of the shock absorber
in the tension stage and compression stage direction constitutes
a problem for all examination methods. Modeling the piston
CA 022~23 1998-12-07
kinematics is only possible with computer assistance even in the
case of known shock absorber characteristics. In this case at each
point of inversion a plurality of compensation effects begin,
which are not yet terminated by the time the next point of
inversion is reached. The association of a single parameter, which
describes the quality of the shock absorber, is thus even as such
problematical.
One object of the invention is to provide a method and
an apparatus for the examination of shock absorbers mounted on
vehicles, which so processes the measured characteristic curves of
the force response to the vibratory excitation by the vibratory
platform that a satisfactory basis is possible for the evaluation
of the results of measurement in accordance with a given vibration
model for the running gear.
In order to achieve this object the method of the
invention for the examination of shock absorbers installed on a
vehicle using a vibratory platform, on which one wheel of the
vehicle is stood and which is able to be reciprocated in a
vertical direction at a suitable amplitude and with a variable
frequency, an oscillating action is exerted of the wheel, the
damping of the shock absorber being derived from the force
response of the running gear to the oscillations of the vibratory
platform and the relative phase function between the force exerted
by the vibratory platform and the movement of the vibratory
platform is characterized in that the data signal, which
represents the time changes in the force response at a target
frequency, is split up by tension-compression stage split means
into a compression stage signal, which is characteristic for the
compression stage of the shock absorber, and a tension stage
signal, which is characteristic for the compression stage of the
CA 022~23 1998-12-07
shock absorber, and in that the compression stage signal and the
tension stage signal are separately further processed. In the
method of the invention the vibrations caused by the vibratory
platform are true sine-wave vibrations as desired for being able
to obtain also a sine-shaped response of the system.
SUMMARY OF THE lNvhNllON
In accordance with an advantageous development of the
method of the invention the signals obtained by frequency
splitting are subjected to digital filtering, and preferably
Fourier transformation and more particularly fast Fourier
transformation. This means that the corruption of information as
in previously conventional filtering of the overall signal is
avoided and the characteristic features of the respective stage
(tension and, respectively, compression stage) are preserved.
In an advantageous form of the method of the invention
after such digital filtering a quality assessment of the processed
signal is carried out as to the sine-conformity, wherein for each
measured value the relative departure from the sine-signal, which
represents the ideal filtered value, is calculated, the transverse
sum of all departures is found and those measurement points are
rejected, whose mean departure from the ideally filtered value
exceeds a predetermined value, such predetermined value preferably
being 5%. Thereby it can be achieved that two sine-shaped signals
are derived which are then feed to the further processing. Only by
means of the sine-shaped course of the signals the correctitude of
the calculating method with respect to the parameter calculation
and to the parameter estimation is guaranteed. Furthermore, the
measurement curve is smoothed such that further processing is
facilitated.
CA 022~23 1998-12-07
According to one of the advantageous developments of the
method of the invention the effect of the vibratory platform on
the result of measurement is compensated for, because in a dynamic
calibration run the frequency response of the vibratory platform
is recorded and the amplitude spectrum is interpolated by a
parabolic curve. This means that a significant interfering factor
is cut out.
In accordance with an advantageous development of the
method in accordance with the invention prior to the actual
measuring run a warm-up phase is provided for the vibration
absorber in order to heat the shock absorber during measurement up
to a predetermined temperature and accordingly to cut out the
effect of temperature variations on the result of measurement.
In accordance with an advantageous development of the
method in accordance with the invention for recording the support
points for estimation of parameters a quality test is performed,
the departure of the actual signal from the required sinusoidal
signal is detected and, if the departure is too large, the reading
is rejected and the next frequency aimed at will then lie below
the actual frequency by a predetermined value, as for example 1
Hz, while, if the quality of the signal is sufficient, scanning is
resumed in small steps of, for instance, 0.5 Hz. Accordingly the
pitch of the steps is automatically adapted to the circumstances,
something which leads to substantially improved results of
measurement.
In accordance with a further advantageous development of
the method of the invention as a basis of assessment for the
quality of the installed shock absorber the degree of axle damping
CA 022~23 1998-12-07
in relation to the quotient of the spring-suspended mass and the
non-spring-suspended mass is calculated, the readings of the
degree of axle damping being related to a characteristic curve of
a critical value degree of damping, which delimits the range of
unacceptable degrees of axle damping. The assessment of the shock
absorber is performed by the classification in the quality ranges
~very good, "medium" and "failed", with which a recommendation of
replacement of the installed shock absorber is coupled. The
assessment takes place in a manner dependent on the position of
the degree of axle damping found in relation to the characteristic
curve of the critical value degree of damping and not in a manner
dependent on the individual parameters of the one quarter running
gear, i. e. of the suspension of one wheel. Since the measured
value signals have been processed in the above mentioned manner,
it is also possible to obtain reproducible, i. e. satisfactory
results of measurement from the evaluation of these signals.
In accordance with a further advantageous development of
the method of the invention calculation of parameters is
performed, as same is recited in claim 26, and from the parameters
the degree of axle damping is ascertained, the parameters employed
here being a result of the model iteration. In tests it has been
shown that with this one quarter vehicle model satisfactory
results can be obtained in the examination and classification of
installed shock absorbers. In particular the problematical effect
of so-called external friction is taken into account by the model
iteration.
The advantageous development of the method of the
invention as defined in claim 27 means that the amount of
computation for the evaluation of the measurements is
CA 022~23 1998-12-07
substantially reduced, since the number of variables to be
processed can be reduced to six.
The advantageous development of the method of the
invention as defined in claim 29 leads to a still further
reduction in computation work without any reduction in the
measurements processed.
A further advantageous development of the method of the
invention is characterized in that after calculation of the degree
of axle damping and the plotting thereof in a graph of degree of
axle damping to spring-suspended/non-suspend mass a recommendation
can be made as regards the need for replacement of the shock
absorber.
A further advantageous development of the method of the
invention is characterized in that on determining the frequency
response the following steps are performed in the order given:
- Measurement of the static weight;
- Running up to starting frequency of for example 10 Hz;
- Warming up phase of shock absorber;
- Run up of exciting vibration of the vibratory platform to the
maximum frequency of for example 35 Hz;
- Recording of support points for the estimation of parameters;
- Recording of resonant frequency f3;
- Recording ~/2 frequency f2 and
- Transmission of data to evaluation system.
A further advantageous form of the method of the
invention is characterized in that for recording the frequency
points the following steps are performed in the order indicated:
- Run up to target frequency;
CA 022~23 1998-12-07
- Check to see if frequency becomes stable;
- Recording readings;
- Separation of tension from compression stage signal;
- Digital filtering, preferably Fourier transformation, of the
separated signals;
- Quality assessment of the signals;
- Check of road adherence;
- Check to see whether measurements have been recorded for a
given number of revolutions;
10 - Check to see whether measurements conform to the quality
standard for at least a fraction of the revolutions;
- Extraction of mean value for the measurements classified as
being good;
- Compensation of the effect of the vibratory platform.
A further advantageous form of the method of the
invention is characterized in that for evaluation of the results
of measurement the following steps are performed in the order
indicated:
- Check to see whether the road adherence is within
predetermined tolerance limits;
- Extraction of the characteristic frequencies;
- Parameter estimation as regards the calculation of kR, the
spring constant of the tire;
- Check of air pressure in tire;
- Check of signal form of the amplitude spectrum;
- Check on whether the ~/2 phase state is reached;
Check on whether the equation f2 = f3 is fulfilled;
- Parameter calculation
- Check to see whether parameters can be calculated;
- Calculation of degree of axle damping;
- 10 -
. .
CA 022~23 1998-12-07
- Assessment of degree of axle damping and issue of
recommendation concerning the need for replacement of shock
absorber.
The various different embodiments of the method of the
invention possess the following advantages: The preparation of the
measured values assesses the quality of a point of measurement,
makes it available for extraction of the mean value and separates
the upper and the lower half wave into two sinusoidal signals.
After digital filtering the corrupting effect of the vibratory
platform on the shear force sensors is compensated for.
The determination of parameters is based on mathematical
modeling of the running gear, which in its complexity can only be
solved by so-called computer algebra. The equation systems in this
case involve several hundred individual summands with complex 18th
order polynomials. This method, which is also known as parameter
calculation, is in addition combined with parameter estimation.
The assessment of the running gear is performed not by
using the isolated assessment of the model parameters, since for
each vehicle type, including its different forms of suspension,
the transmission ratios and the characteristics of the installed
shock absorbers and suspension springs must be held in the data
base. In view of the variety of possible forms such a data base
would not be practicable for real use on the customer's premises.
Here recourse is had to the EUSAMA Method, which provides a value
for road adherence which is unsatisfactory, but is not limited to
vehicle types. The so-called degree of axle damping, which is
evaluated on the basis of the ratio between spring-suspended and
non-spring-suspended mass, is to overcome the known disadvantages
of the road adherence value. On the basis of this assessment item
.
CA 022~23 1998-12-07
a recommendation for replacement is produced for the installed
shock absorber.
The advantages of this method are more particularly
valuable in the case of modern axle systems, as for example
employing pneumatic suspension, freely floating axles etc., in the
case of which the classical methods of assessment fail more or
less completely and as a rule classify such suspensions
unfavorably.
An apparatus for the performance of the method for the
examination of shock absorbers installed on a vehicle comprises a
vibratory platform, on which one wheel of the vehicle is placed,
means for causing the vibratory platform to vibrate in a vertical
direction with a suitable amplitude and variable frequency and to
cause the wheel to vibrate, the damping of the shock absorber
being derived from the force response of the running gear to the
vibrations of the vibratory platform and the relative phase
function between the force exerted by the vibratory platform and
the movement of the vibratory platform, is characterized by shear
force sensors on the vibratory platform and a pulse source for the
angularly equidistant scanning, by a means, by which the data
signal is split into a compression stage signal and a tension
stage signal and by a means in order to separately supply the
compression stage signal and the tension stage signal to the means
for the further processing of the signals.
An advantageous design of the apparatus of the invention
is characterized in that the assessment means assesses the quality
of the installed shock absorber using the degree of axle damping
as related to the quotient of the spring-suspended mass and the
non-spring-suspended, the readings for the degree of axle damping
CA 022~23 1998-12-07
being related to a characteristic curve of a critical value of the
degree of axle damping, which delimits the range of non-acceptable
degrees of axle damping.
The apparatus of the invention has the advantage that
the vibrations caused by the vibratory platform are true sine-wave
vibrations as desired for being able to obtain also a sine-shaped
response of the system.
Further advantageous developments of the invention will
be seen from the remaining dependent claims.
Other advantages, features and characteristics of the
present invention, as well as methods of operation and functions
of the related elements of the structure, and the combination of
parts and economies of manufacture, will become more apparent upon
consideration of the following detailed description and the
appended claims with reference to the accompanying drawings, the
latter of which is briefly described hereinbelow.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 of the drawings appended hereto is a flow chart
of the recording of readings.
Figure 2 of the drawings is a flow chart of the
preparation of readings.
Figure 3a and 3b of the drawings are graphs indicating
the principle of frequency splitting.
- 13 -
CA 022~23 1998-12-07
Figures 4A and 4B of the drawings are graphs of the
principle of amplitude splitting.
Figures 5A and 5B of the drawings are flow charts of
evaluation using the method of the invention.
5Figure 6 of the drawings is a graph to indicate the
action of disruption of road adherence on the frequency curve.
Figure 7 of the drawings is a graph to indicate the
action of good shock absorber performance on the frequency curve.
Figure 8 of the drawings is a graph to indicate the
10action of poor shock absorber performance on the frequency curve.
Figure 9 of the drawings is a graph indicating the
target line kR for the pre-check of tire pressure.
Figure 10 of the drawings is a graph, which represents
the kR line of a tire (UniroyalTM 155/70 R13).
15Figure 11 of the drawings is a graph, which shows the
signal form of a satisfactory frequency curve.
Figure 12 of the drawings is a graph to show the signal
form applying for a defective shock absorber.
Figure 13 of the drawings is a graph indicating the
signal form applying for a suspension with track jitter.
Figure 14 of the drawings shows the theoretical curve
form in the mu-k range.
- 14 -
CA 022~23 1998-12-07
Figure 15 of the drawings is a graph of the practical
curve form in the mu-k range.
Figure 16 of the drawings is a graph of the scan range
of model iteration in the mu-k range.
Figure 17 of the drawings is a graph of the pole
position and of the asymptote of the characteristic curve of the
degree of critical value damping.
Figure 18 of the drawings is a graph of the degrees of
axle damping of different compression stages.
Figure 19 of the drawings is a graph of the degrees of
axle damping of different tension stages.
Figure 20 of the drawings is a graph of dependency of
the degree of axle damping on tire pressure.
Figure 21 of the drawings is a graph of dependency of
the degree of axle damping on the loading of the vehicle.
Figure 22 of the drawings is a graph of dependency of
the degree of axle damping on the temperature of the shock
absorber.
Figure 23 of the drawings is a graph of dependency of
the degree of axle damping on the tires of the vehicle.
Figure 24 of the drawings is a graph of dependency of
the degree of axle damping on changes in the non-spring-suspended
mass mU.
CA 022~23 1998-12-07
Figure 25 of the drawings is a graph of the direction of
displacement of the parameters on the point of assessment.
DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT
The method of examination is divided up into two
principal steps, namely the recording of readings, in which
readings of data are made and processed, and the evaluation, in
which the processed data are further processed for making an
assessment as regards the quality of the running gear. The
examination means includes a vibratory platform, on which in
addition to the platform's shear force sensors a pulse source for
angularly equidistant scanning is provided.
During taking readings the static weight and the
frequency curve of the one quarter vehicle over an exciting
frequency of 35 Hz - 10 Hz range are recorded. The amplitude curve
then corresponds to the shear force sensor signal on the vibratory
platform in Newtons. The phase function corresponds to the phase
shift between the exciting oscillation of the vibratory platform
and the response oscillation of the one quarter vehicle in
radians. From the frequency curve the characteristic frequencies
or, respectively, frequency ranges are filtered out and further
processed.
In the following evaluation the three characteristic
frequency ranges are of special interest, namely the resonant
frequency of the non-spring-suspended mass f3 (here the amplitude
curve reaches the absolute maximum), the frequency f2 (here the
phase function is equal to exactly 90~ or, respectively, ~/2 and
the effective fraction of the amplitude curve is equal to zero at
- 16 -
CA 022~23 1998-12-07
this frequency) and the higher frequencies up to the maximum
frequency of 35 Hz. The measurement points are support points for
the parameter estimation, with which the spring constant of the
tire kR is ascertained.
The frequency-equidistant scanning of the frequency
range 35 Hz to 10 Hz is only a rough approximation. For high
quality and reproducible measurements a special measurement
algorithm is provided for each characteristic frequency and,
respectively, each characteristic frequency range, the chart of
such algorithm being illustrated in Figure 1.
The amplitude curve corresponds to the changing part of
the overall signal. The unchanged part is taken into account by
digital filtering. The static weight of the one quarter vehicle is
included as mS in the calculation of parameters.
At low environment temperature the viscosity of the
damping liquid increases and consequently also the damping
dissipation of power. The increases in power may here amount to
several 100~. Consequently a warming up phase is provided for, in
which in the environment of the upper resonant frequency f3 the
damping liquid is warmed up. The initially low force response to
the vibratory platform increases on warming up and converges
toward a maximum value, whose absolute value is irrelevant. After
reaching a maximum value the measurement of the frequency curve as
such is proceeded with.
In tests made so far at 0~C the duration of warming up
amounted partially to substantially less than 15 seconds. At even
lower temperatures the warming up time may be assumed to be under
30 seconds.
.... .. .
CA 022~23 1998-12-07
Starting with the maximum frequency of 35 Hz the support
points for the parameter estimation are measured. At the high
frequencies so-called bounce effects are to be seen in the reading
or data signal, which substantially falsify the signal in the
tension and compression stage direction. These support points
generally have the greatest weight for parameter estimation and
must be correct. At each measurement point a quality test is hence
performed. The quality test examines the departure of the actual
signal for the required sinusoidal signal. If the departure is too
large, the reading is rejected. The next target frequency is then
1 Hz below the actual frequency. If the quality of a signal is
sufficient, scanning is further performed in 0.5 Hz steps.
For determination of the resonant frequency the
following procedure is employed. If the amplitude curve in the
compression and tension stage direction exceeds its maximum, the
resonant point f3 of the non-spring-suspended mass has been passed.
This resonant range is then scanned again in 0.25 Hz steps. The
resonant frequency is thus ascertained with a resolution of 0.25
Hz. The algorithm also provides that in the case of suspensions
with a great change in track or shock absorbers with defective
valves several maxima may be swept successively. In all
problematical cases ultimately the actual resonant frequency is
detected with the necessary resolution.
As an alternative to the above mentioned manner of
proceeding the resonant frequency may be ascertained in such a
manner that the resonant frequency is measured in a single
frequency sweep with a resolution of 0.25 Hz. For this purpose it
is possible, if the amplitude curve exceeds its point of
inflection, for the assumed resonant range to be scanned in
smaller frequency steps, for instance in 0.25 Hz steps in the to
- 18 -
CA 022~23 1998-12-07
find the actual resonant frequency with the required resolution in
one sweep.
In the case of extremely strong damping effects the
phase function will not be able to reach the ~/2 level or, in
another extreme case with very weak damping, will run through the
~/2 level very steeply. For both cases the recording algorithm
must function. The ~/2 frequency must always be below the resonant
frequency f3. If the two frequencies were of equal size then the
vibrating system would be undamped. The phase function would rise
with a falling exciting frequency. If the phase value of 1.2
radians is reached, the step pitch if reduced to 0.1 Hz. The
resonant frequency f3 will then be set from these support points.
During the taking of readings and preparing readings of
a single excitation frequency, in addition to the evaluation of
the spectral variations the measured force value is further
processed. Such evaluation is based of the complex frequency
curves of forced, damped, elastic oscillating systems. This
implies strictly sinusoidal oscillations. The variations with time
of the data signal is however in practice not sinusoidal. On the
one hand the elongations in the compression and tension stage
direction are quite different and on the other hand the duration
in time of the compression stage stroke is slightly shorter than
the duration of the tension stage stroke.
25Figure 2 shows the taking of readings and the
preparation of data therefrom in order to achieve optimum accuracy
and to provide physically correct information for calculation.
In order to approach the next target frequency and to
stabilize same the frequency converter is provided with the
CA 022~23 1998-12-07
frequency value of the next scan point of the frequency spectrum.
The taking of a reading is only enabled when the target point in
time is reached and the system has become stabilized thereat. A
frequency point is deemed to be stabilized when its frequency
fluctuations no longer exceed a predetermined tolerance level. In
this respect the different frequency ranges are differently
evaluated. During run-up to the maximum frequency, attack or
transient phenomena have the most marked effects and have the
worst effects as regards the quality of the support points and
consequently for the estimation of parameters. Here longer
stabilizing times are to be preferred. In the case of low
frequencies the vehicle may come in the range of disruption of
road adherence. Here the road adherence of the suspension parts
resists acceleration due to the vibratory platform and abruptly
comes to an end as from a certain acceleration. The vehicle sags
into its suspension system. The speed of rotation controller is
not capable of dealing with this interfering factor when there are
substantial disruption effects. Frequency points, which are not
able to be stabilized, are omitted in measuring operations. The
analog/digital conversion operations are in this case caused by an
interrupt during measurement.
In order to be able to supply the physically correct
information for the following calculations the overall signal is
split up into two independent signal parts for the compression and
tension stage and two different split algorithms are employed.
The basic notion of such splitting is the association of
respectively half of the readings with the compression and,
respectively, tension stage signal. In such splitting there is an
exact maintenance of the exciting base frequency and the splitting
is relatively simple and rapid to implement. In this type of split
- 20 -
. .
CA 022~23 1998-12-07
the compression stage signal is exaggerated and the tension stage
signal is minimized to an even greater extent (see Figures 3A and
3B). The various different damping power dissipations in the
~ stages are spread out even furthe~. On the other hand the phase
shifts produced are not plausible. At high power dissipation
simulations on a simple one quarter vehicle model show a small
phase shift and low power dissipations show a large phase shift.
In the case of this type of stage split the phase shifts are
almost identical.
The basic notion of the amplitude split is the
association of the positive half wave as a compression stage
signal and the association of the negative half wave as the
tension stage signal. This is generally the same as the
association in shock absorber tests in the non-installed state. A
spread of the compression and tension stage amplitude does not
take place here (see Figures 4A and 4B).
At the frequency of oscillation which becomes
established of the two stages the compression stage is distinctly
above the excitation frequency, whereas the tension stage is
distinctly below the excitation frequency. This is however
something which can be compensated for by phase shift, which is in
agreement with the theoretical simulation. The narrow positive
compression stage half wave must be shifted by ~ to the right
(~it must have a lead") in order to reach the basic frequency of
the exciting signal. This is the same as a phase shift of -~. The
wide negative compression tension stage half wave must be shifted
by ~ to the right ("it must have a lag") in order to reach the
basic frequency of the exciting signal. This is the same as a
phase shift of +~.
CA 022~23 1998-12-07
The digital filtering means that the signal of the
exciting frequency is filtered out. The digital filtering must in
this respect be adapted to the corresponding compression-tension
stage split method, the classical form of Fourier transformation
being able to be employed for frequency split, since the
compression and tension portions are exactly equal to the exciting
frequency taken as a basis. For amplitude split the higher basic
frequency of the compression stage and the lower frequency of the
tension stage must be taken in account for the calculation of the
amplitude. For this purpose the Fourier transformation of the
entire signal may be carried out and the above mentioned phase
shifts are added.
The signal curves may be considerably distorted for the
most various different reasons. Even the digital filtering for
example by the Fourier transformation will then not provide any
usable data, since the original signal differs very substantially
from the sinusoidal form. These measurement points may then not be
used for the further processing.
For assessment of quality for each measured value the
relative departure from the ideal filtered value (sine-curve) is
calculated. The transverse sum of all departures is then employed
for an evaluation. Investigations have indicated that as from a
critical value of 60% mean departure the interference, in contrast
to 4~ or 5~ clearly increase on a scale which can not be
tolerated.
Alternatively for assessment of quality it is possible
for the differential area of the filtered curve (sine-curve) and
the actual value curve to be added on and related to the target
area, only one division being necessary.
- 22 -
.
CA 022 j j j23 1998 -12 - 07
In the case of extremely poor running gear or,
respectivelyt shock absorber performance it is possible in the
frequency environment of the resonant point f3 for the vehicle to
. lose contact with the .vibratory pla~form and even to come cle~ar of
the test bed.
In the case of road adherence of under 10% the taking of
measurements is discontinued for safety reasons. An evaluation of
the readings made so far is then no longer necessary, since the
running gear may then be classified ad hoc as being defective.
For each stored measurement point three complete cycles
are measured. If at least two cycles fulfil the quality criterion,
the qualitatively valuable measurement curve is employed to form
the mean value for the amplitude and the phase. If only one or no
cy le fulfills the quality criterion, then this frequency point is
not included in the spectrum.
The so-cal]ed vibratory platform is mounted on the shear
force sensors. If the system is excited, the oscillating mass of
the vibratory platform also has an effect on the shear force
sensors. This influence will increase in accordance with a square
law with an increase in frequency:
F = mp/ttlte SHU~ ~)
In the course of the measurement of the frequency
response of the vibratory platform however linear and constant
fractions were included in the amplitude. In order to mlnimize
this potential source of errors in a so-called 'dynamic'
calibration run the frequency response of tne vibratory platform
- 23 -
CA 022~23 1998-12-07
is recorded and the amplitude spectrum is interpolated with a
parabolic curve. The three coefficients of the interpolation
polynomial must here be permanently stored. The phase shift is,
but for the slight displacement effects of the low pass filter,
constant and is employed as a zero transition reference of the
vibratory platform in later determination of phase shift.
For compensation of the vibratory platform the filtered
signal must now be corrected by the zero transition reference.
Specifically, it is 'shifted to the left', that is to say the
measured phase shift is added to the zero transition reference.
Now the phase shift is correctly determined in relation to the
zero transition reference of the vibratory platform.
The amplitude spectrum of the vibratory platform
comprises only effective fractions and no so-called dummy or
imaginary fractions. In order to eliminate the falsifying effect
of the vibratory platform this real fraction must be subtracted
from the real fraction of the measured amplitude spectrum.
From the resulting effective and imaginary fraction of
the compensated signal the actual, so-called compensated amount
and phase value of the oscillating one quarter vehicle is
calculated.
The aim of the evaluation is a qualitative assessment of
the measured running gear. More particularly, the quality of the
installed shock absorber is to be assessed. In connection with
such statement of quality a recommendation, on which the operator
of the test bed may rely, is provided with respect to shock
absorber replacement. The following three recommendations are
provided:
- 24 -
CA 022~23 1998-12-07
The shock absorber is completely in order (no
replacement required).
The reliability of the shock absorber is in the boarder
area (replacement is recommended).
The shock absorber is certainly defective (replacement
absolutely necessary).
The basis of valuation is the so-called degree of axle
damping. This degree of axle damping is then related to the
critical value degree of axle damping, which takes into account
the particular running gear properties and test conditions. In
particular, any attempts at manipulation are compensated for on
the basis of this critical value degree of axle damping. The
possibilities of manipulation are described in the following. The
critical value degree of axle damping may be defined restrictively
in a different fashion and is ultimately ascertained in a field
test after consultation with the vehicle manufacturers. The
characteristic curve of the critical value degree of axle damping
is only defined by two specific parameters. They then apply for
all types of vehicle and all test conditions.
The evaluation is based on a plurality of complex
individual steps, which are represented in the flow chart of
Figure 5. One feature of the evaluation method is a plausibility
test, which comes before the calculation and evaluation of the
degree of damping. This plausibility test has the purpose of
eliminating any extreme situations of the running gear, which can
possibly not be calculated. Such incalculable, extreme running
gear is not necessarily due to bad or defective shock absorbers.
Extremely good shock absorber performances, which may possibly be
- 25 -
-- ~ . ,
CA 022~23 1998-12-07
enhanced by a large fraction of external friction, offend against
the one quarter vehicle model for parameter calculation.
The characteristic frequencies for the tension and
compression stage are extracted separately for the measured
amplitude and phase spectrum. If one stage is defective, then the
shock absorber must be replaced.
The resonant frequency f3 is the same as the frequency,
at which the amplitude maximum is swept through. This extreme
value must be local and global. If for example the first value at
35 Hz is a global maximum value wo, same is not accepted as being
local and the series of measurements is rejected as being invalid.
The following problems impair the exact ascertainment of
the ~/2 frequency. In the course of recording the spectrums
scanning takes place in frequency-equidistant steps. As a rule the
frequency point is not exactly reached, at which a compensated
phase shift of exactly ~/2 becomes established. The disruption of
road adherence, already mentioned several times, which occurs
after passage through the upper resonant frequency, will
particularly falsify the phase spectrum. Owing to disruption of
road adherence the continuously extending phase function comprises
erratic values, which may lie above the ~/2 level, although the
good damping action means that ~/2 resonance will be reached much
later or even not at all, as illustrated in Figure 6.
In the same manner certainly reaching the ~/2 level may
be prevented owing to an erratic measurement. By interpolation of
the section of the phase function using a third order polynomial
there is both an elimination of the erratic points and also an
exact calculation of the ~/2 frequency f2. The support points for
- 26 -
CA 022~23 1998-12-07
the interpolation start here at the resonant frequency and
terminate at the local maximum of the phase function. The effect
of a good shock absorber performance of the frequency curve is
illustrated in Figure 7.
The frequencies f2 and f3 may be relied upon for a
plausibility test. In this respect two extreme cases should be
dealt with before the actual parameter calculation.
If the ~/2 frequency is not present, this means that the
~/2 frequency level was not reached. The damping of the running
gear is exceptionally good. A calculation of parameters would not
be possible and is unnecessary.
If the ~/2 frequency f2 is larger than or equal to the
resonant frequency f3 then in the oscillating system without
damping the frequencies f2 and f3 are theoretically equal. If this
behavior is found during the recording of the frequency curve,
then the damping of the running gear is extremely poor.
Calculation of parameters would not be possible and is
unnecessary. The effect of poor shock absorber performance of the
frequency curve is represented in Figure 8.
The two extreme cases could be provoked by manipulation
of the tire pressure. Checking of air pressure is necessary before
the plausibility test of the characteristic frequencies.
The estimation of parameters is a method, which
estimates the parameters of a predetermined complex frequency
curve on the basis of measured support points. The complex
frequency curve is based on the simple one quarter vehicle model:
- 27 -
CA 022~23 1998-12-07
G bo + ib,~ - b2~ 2 _ ib3~ 3 + b4~ 4
1 + ia~ - a2c~ 2 _ ia3~ 3 + a4cl~ 4
or, respectively, with the real parameters:
Fp = --mS(!3 --im d ~J~ 3 + mu mr ~ 4
k ( /~ + kR + k ~) -- im d ~1~ 3 + ml/ mF 4
wherein:
mF spring-suspended mass [kg]
mu non-spring-suspended mass [kg]
mS static overall mass [kg]
k spring constant of the suspension spring [N/m]
kR spring constant of the tire [N/m]
d damping constant of the shock absorber [kg/s]
circular frequency of the exciting oscillation
[l/S]
The basis of the linear equation system is a loss
function V, which calculates the departure of all support points
for the complex frequency curve to be assessed:
V = ~u~v[Rv(l -a2~2 +a4~4)- Iv(a~ -a3~3)-(bO -b2~2 +b4c~4)]2
+ylv[Rv(al~-a3~3)+lv(l-a2~2 +a4cl)~)-(b,(l~-b3~3)]2
wherein:
R~ is the measured real fraction of the support point
- 28 -
CA 022~23 1998-12-07
is the measured imaginary fraction of the support
point v
~v is the weight fraction of the support point v
N is the number of support points - 1.
This loss function is differentiated after each
coefficient to be estimated and set at zero. That is to say, every
coefficient is to be so estimated that the resulting loss runs
through a minimum.
For the nine coefficients to be estimated there are
consequently nine linear equations, which are linked in a linear
equation system. The coefficients a1 - b5 are determined by the
solution of this linear equation system. The application of the
method is extremely complex and is hence not particularly suitable
for determination of parameters. The absolute values of the
individual parameters of the frequency curve of the simple one
quarter vehicle model differ to a great extent. This means that
the accuracy of the individual parameters suffers. The parameters
bo and b1 are always zero in the case of the frequency curve of the
simple one quarter vehicle model. The parameter estimation method
however allocates values to these parameters and deals with the
departures by means of the higher coefficients. Each support point
is weighted with the function phiV. That is to say, its influence
on the loss function is set. In this respect the it is not the
absolute amount but the ratio of the weight functions of the
support points inter se which is relevant. By appropriate
definition of the weight function the band width of the solutions
may be practically varied at will. There is no unambiguous,
optimum weight valid for all measurement curves.
- 29 -
,
CA 022~23 1998-12-07
The particular feature of the frequency curve for the
simple one quarter vehicle model is the convergence of the
amplitude spectrum toward the spring constant of the tire kR at
high frequencies. This asymptote may be calculated by a modified
form of the parameter estimation method. Of the originally nine
parameters to be estimated the parameters bo and b1 are basically
zero and the parameter b2 is the static mass mS, which is also
known. The linear equation system is thus reduced to six linear
equations. For estimation 20 different weight functions are swept.
For each weight function the theoretical support points are
calculated. The weight function with the minimum mean departure in
the support points employed is then used.
For verification of the estimated spring constant values
kR of the tire the frequency curves of the tire are recorded under
different amounts of bias. Three results may be derived from these
results:
The spring constant is constant over the entire
frequency range from 35 Hz to 1 Hz.
The value of the spring constant is in agreement with
the estimated value with an accuracy of 2%.
The typical kR characteristic curves of a tire are
primarily dependent on the tire pressure, the dimensions of the
tire and the bias, that is to say the weight, with which the tire
is thrust against the ground. The specific manufacturer data such
as a rubber blend, type of carcass or tread are only of secondary
significance. Furthermore the wheel employed is without any
significant influence on the spring constant kR.
- 30 -
CA 022~23 1998-12-07
The spring constant kR corresponds to the degree of
coupling between the vibratory platform and the oscillating
spring-shock absorber system of the non-spring-suspended mass mu.
If the degree of coupling too high, a large amount of oscillation
energy will be supplied to the spring-shock absorber system. The
excess energy is then interpreted as insufficient shock absorber
performance. If the degree of coupling too low, too little
oscillation energy will be supplied to the spring-shock absorber
system. The deficit of energy is then attributed to the shock
absorber performance.
As explained below, this erroneous assessment will be
satisfactorily compensated for in the case of small air pressure
departures as regards the assessment of the shock absorber, unlike
other methods with the degree of axle damping.
The spring constant kR of the tire is suitable for
examination of the tire pressure. The basis of the plausibility
test is the specification of dimensions of the tire types (for
example 225/50 R15). The rated air pressures for a certain type of
dimension vary on use of this tire on different types of vehicle
only to an insignificant extent and are then coupled at least with
the weight of the vehicle. Heavy motor vehicles imply in this case
a higher rated tire pressure than lighter vehicles (BridgestoneTM
"Ratgeber Reifen" 1992).
225/50 R16 Rated tire pressure Rated tire
pressure
VA (bar) HA (bar)
Mercedes 500 3.0 3.3
Volvo 2.3 2.6
Porsche Carrera 2.0 2.5
... ~ . . . _ ~ ..
CA 022~23 1998-12-07
For the tire type 225/50 R16 as indicated a minimum
curve for the minimum value of 2.0 bar (for the front axle of the
Porsche Carrera) and a maximum curve for the maximum value of 3.3
(for the rear axle of Mercedes 500) would be stored in a data
base. Both of the curves represent the changes in the spring
constant value against biasing weight (Figure 9). The two extreme
working points are now connected by the rated kR curve. All other
vehicle types, which are fitted with these tires, then lie more or
less on a rated kR curve. For a fixed bias a change in the tire
pressure has a linear relationship to a change in the spring
constant kR. Accordingly a tolerance band of, for example, 0.5 bar
may be defined. If the estimated kR value is outside the
predetermined tolerance band, the measurement is not evaluated. An
instruction to check tire pressures is issued as well.
Figure 10 shows the results of measurement with the tire
type UniroyalTM 155/70 R15 on as tire test bed. In the case of a
test series with the Ford Escort the vehicle loading and the tire
pressure were varied. The plausibility tests did not indicate any
false assessment.
Attempts aimed at manipulating the final result by
changing the load are rendered more difficult by this plausibility
test. It is possible to see clearly whether an increased degree of
coupling is due to an excessively high tire pressure or an
extremely high load. A higher weight implies a higher bias and
accordingly a higher degree of coupling with the tire pressure
unchanged.
The actual curve of the amplitude and phase spectrum is
substantially in accord with the theoretical target or rated curve
of the simple one quarter vehicle model. It is only in the
- 32 -
. , .. . . _ . .
CA 022~23 1998-12-07
frequency range of disruption of road adherence that the simple
one quarter vehicle model cannot be utilized. However since on
reaching the disruption of road adherence the measurement is
terminated in any case, this can be neglected. Any other departure
of the spectrum is either due to defective running gear or to an
unusual property of the running gear. The so-called signal form
test is to classify irregularities in the amplitude spectrum.
In Figure 11 a satisfactory amplitude and phase spectrum
is to be seen. At the resonant point the amplitude curve smoothly
and unambiguously runs through its maximum. The phase function
smoothly converges toward zero with increasing exciting frequency.
In Figure 12 the spectrums of a manipulated shock
absorber are represented. The resonance maximum of the amplitude
spectrum is not unambiguous. The phase spectrum is practically
free of interference in this range. This leads to a spread of the
wrongly determined characteristic frequencies f2 and f3. The
resulting degree of axle damping is then clearly above or outside
the calculable value range. This case cannot be employed for
calculation.
Owing to camber angle some running gear sweeps through
a plurality of resonant points, in which the vehicle is subject to
powerful horizontal oscillations. Such a spectrum is illustrated
in Figure 13. In this case, such so-called camber flutter could
also be interpreted as due to a defective shock absorber. The
difference between this and the amplitude curve of a mechanically
defective shock absorber is however the fact that between the
characteristic frequencies f2 and f3 no local minimum is swept
through. Such series of measurements are made available for
further parameter calculation.
CA 022~23 1998-12-07
The parameter calculation is based on the complex
frequency curve of the simple degree one quarter vehicle model:
F -- mS ~ --imS ~d~ (j)3+ m / m~ 4
k ( li + ~F + k )(J) -- im ~ d ~j~3 + mu mF ~i) 4
whereln:
mF spring-suspended mass [kg]
mu non-spring-suspended mass [kg]
mS static overall mass [kg]
k spring constant of the suspension spring [N/m]
kR spring constant of the tire [N/m]
d damping constant of the shock absorber [kg/s]
circular frequency of the exciting oscillation
[l/S]
All the following calculation steps were performed using
computer algebra:
15 - calculate the amplitude curve A from the complex frequency
curve Fp.
- For calculation of the characteristic equation for the above
resonant point f3 differentiate the amplitude curve A and make
A' equal to zero.
- Solve this equation in terms of the damping constant d and
the following applies:
d3 =g(~3,ms,mulkR~k) ~-
- For calculation of the characteristic equation for the ~/2
frequency f2 calculate the real fraction of A and make A~e
equal to zero.
- Solve this equation in terms of the damping constant d and
the following applies:
- 34 -
. , ~ ,
CA 022~23 1998-12-07
d2 =h(C~,2, ,ms~mu~kRrk)
In both equations g and h the exciting frequencies omega
and the s.tatic weight mS are measured. The spring constant of the
tire kR iS determined in the parameter estimation method and is
accordingly known as well.
Two equations are now available for d2 and d3 with two
unknowns mu and kR. The same may however not be linked directly,
since the damping constant is frequency-dependent. The linking
together occurs in the amplitude curve A. Generate the amplitude
curves A2 (replace d by d2) and A3 (replace d by d3). Since the
amplitudes at both frequency points are also measured, there are
still two equations with two unknowns ml and k, which may be now
linked. It is now possible to calculate all parameters.
The calculation of the parameters is now able to be
resolved to several zero position problems as far as the fourth
order. If the parameters are not able to be calculated (for
example no reai zero point or division by zero), a series of
measurements is to be considered, which corresponds to an extreme
situation. In such cases evaluation on the basis of the degree of
axle damping is omitted. From an extreme situation with a valid
signal curve and with characteristic frequencies present it is
possible to conclude that there is either very good or very poor
running gear. For the classification in this special case recourse
is had to the ground standing value. If this value is relatively
large the running gear is in order. If the value is on the other
hand low, the running gear must be classified as poor and
replacement of~the shock absorber is recommended. If~ the ground
standing value is not able to be classified, an error message is
issued.
, . , ~ _ . .
CA 022~23 1998-12-07
Using the measurements described so far basically all
measurement series may be calculated, whose signal curves and
characteristic frequencies were satisfactory. If the above
mentioned error should be due to an unusual measuring situation,
then the original measurement values must be communicated to the
software partner for detailed investigations. The respective
saving of data is automatically performed when there is an error.
The results of the parameter calculation may not be
employed directly for an assessment of the measured running gear.
The reasons for this are the selection of the correct calculation
model, since the simple one quarter vehicle model does not
differentiate between the external friction of the suspension
parts and the internal friction of the damping liquid and the lack
of comparability of the parameters in the installed condition with
measured parameters in the dismounted condition. The dynamics of
both examination methods are extremely different. For instance,
when recording bell curves measurement is normally performed with
large strokes at relatively low exciting frequencies, whereas the
vibratory platform only implies extremely small strokes at high
frequencies at the shock absorber. If running gear is to be
assessed using the calculated parameters, then the comparison
values for this running gear must be held in a data base. In view
of the large number of types of running gear and installed shock
absorbers the production of a user friendly data base is not
possible for use in the workshop.
The aim is consequently the assessment of the complete
running gear situation, which is not tied to the type of vehicle
and takes into account attempts at manipulation or takes same into
CA 022~23 1998-12-07
account in assessment. The result of the assessment is to be a
recommendation to replace the installed shock absorber.
This aim is to be reached using the so-called degree of
axle damping (Achs~):
2 ~I(k+kR)*mu
The parameters employed here are the result of the so-
called model iteration.
For familiar reasons the results of the parameter
calculation may not be utilized directly for the determination of
the degree of axle damping. This will be clear, when the solution
quantity is represented in an mu-k field. Theoretically all
solution curves for the different characteristic frequencies
should intersect at a single (mu,k) point, as illustrated in Figure
14.
In fact there is no unambiguous solution point (mu,k).
In practice a solution field is defined by the three solution
curves, as shown in Figure 15. As will be seen, the solution
straight lines f2 and f, are spread apart and the solution straight
line fl is substantially above the target solution point (ml,k).
With the aid of computer algebra it is possible to show that
principally the influence of external friction is responsible for
these shifts. The solution range for the non spring-suspended mass
can however be limited to a range of under 10 kg.
In the case of vehicles, which are measured with shock
absorbers of varying quality, there is a shift of the entire
solution field in the mu dimension. In the case of running gear
- 37 -
, . . .~ .. ....
CA 022~23 1998-12-07
whose external friction is to be assumed to be large this effect
is particularly clear.
Non-spring-suspended mass Mu (kg) with different shock
absorber performances
good medium poor
BMW 740i 32.0 39.5 46.0
Ford Escort 29.0 30.5 29.5
For the calculation of the correct solution point a so-
called model iteration is swept through. Starting with the
original measured data the measured data are interactively
modified in order to eliminate the effects of the external
friction. The algorithm terminates, when all three solution
straight lines intersect precisely at one point and more
particularly the f1 iteration straight line converges to the f
target solution straight line.
Model iteration is difficult to handle in this form. The
resulting parameters are to a large extent in agreement with
manufacturers' data. A problematical parameter is the spring
constant k, which in the some vehicle types is subject to a
relative variation of 50~. Model iteration may require computing
times of up to one minute. In order to be able to reliably use the
terminating model iteration, the measurement of the resonant
frequency f1 of the spring-suspended mass is necessary). The amount
of time then required for measurement would then be doubled.
The resonant maximum of the resonant frequency of the
spring-suspended mass is not distinct in the case of some vehicle
types so that some search algorithms must be utilized, which find
- 38 -
CA 022~23 1998-12-07
the resonant frequency fl on the basis of empirically determined
facts.
The~work involved in determining the spri~g constant fl
bears no relation to its influence on the final result. Neither
the damping constant d nor the degree of axle damping are
significantly influenced by the varying spring constant k or,
respectively, the absence thereof. As an example some data for the
BMW 3.28 i are relevant:
BMW 3.28i front axle right
Parameter Target value measurement measurement
measurement series 1 series 2
Non-spring-suspended
mass mu (kg) ca. 40 40.7 41.5
Suspension spring k (N/mm) 19.2 18.8 24.6
Damping coefficient d(kg/s) ? 1202 1216
Degree of
axle dampiny without k (~) ? 13.2 13.3
Degree of axle damping
with k (~) ? 13.0 12.9
Owing to the small influence on the degree of axle
damping and the substantial disadvantages involved in the
determination of the spring constant k the latter is omitted so
that the degree of axle damping is equal to:
AChSDG 2*~k *m
- 39 -
CA 022~23 1998-12-07
The model iteration may be simplified to a great extent
and consequently performed more rapidly. The simplification is due
to the fact that the solution curve for the characteristic
resonant frequency f1 is replaced by a constant straight line at
k = 100000 N/m (Figure 16).
The non-spring-suspended mass applying here is
calculated using the customary iteration algorithm. This iteration
algorithm leads to the solution points (mu~loooo) with the least
measured value departure and the best cover of the damping
constants of the remaining solution curves f3.
The following statements may be made on the basis of the
degree of axle damping for an assessment of the running gear.
There is a high reproducibility in the determination of the
parameters d, kR and mu. There is no limitation t particular types
of vehicles. The larger k~ the higher the fraction of the coupled
oscillation energy in the spring-shock absorber system and the
higher must the shock absorber performance be. The larger mu, the
higher must the shock absorber performance be.
For purposes of comparison vehicles with manipulated
shock absorbers were measured. The degrees of axle damping of the
defective shock absorbers were then employed as basic data for the
determination of a critical or border value line. The critical or
border value line must take into account the characteristic
properties of the running gear, which are not inherent in the
degree of axle damping.
Test measurements have indicated that the ratio between
spring-suspended to non-spring-suspended mass mF/mU represents this
decisive property. Vehicles with a large mass ratio mF/mU made do
- 40 -
CA 022~23 1998-12-07
with substantially smaller degrees of axle damping than vehicles
with relatively small mass ratios, independently of the overall
weight of the vehicle.
This fact was included in a critical or border value
line for the mass ratio in the assessment. The critical value line
was in this case to possess a hyperbolic form with on the one hand
the minimum possible mass ratio as the pole point and on the other
hand the minimum degree of axle damping as the asymptote, as shown
in Figure 17.
On the basis of the basic data non-linear regression was
employed to explicitly represent the linear boundary value line.
The regression model utilized as a basis had to be a member of the
so-called growth models. These models are characterlzed by a
strictly monotonous, asymptotic form, which is in agreement with
the previously made observations. The model with the best
agreement is the so-called saturation model. This model renders
possible the description of the critical value line by the two
parameters a and b (AchsDGGrenz=critical value for degree of axle
damping):
a* m~
AChSDGgren2 = b+ m~
wherein:
a is the asymptote, that is to say the minimum degree
of axle damping; and
b is the pole point, that is to say the minimum mass
which is physically possible.
- 41 -
, .
CA 022~23 l998-l2-07
In Figures 18 and 19 the degrees of axle damping of
different measured vehicle types are represented for the
compression the tension stage. All types of vehicle were measured
in this case with their rated tire pressure and unloaded weight
5 and, respectively, with a driver.
The basic data of the critical or border value line are
the 50% shock absorber of the BMW 3.28, the 60% shock absorber of
the Ford Escort, the minimum mass ratio of 5 and the minimum
degree of axle damping of 7%. The running gear and axle designs in
an as-new condition were all found to be poor in the EUSAMA test,
as may be seen from the following table:
Running gear EUSAMA rating (%)
BMW 3.28i HA 22
BMW 5.28T HA 25
Audi A & VA 37
Audi A6 HA 32
Passat HA 39
Golf 3 HA 35
The degrees of axle damping are on the contrary
20 distinctly in the range applying for satisfactory running gear.
Problematical cases are only the rear axle of the BMW 5.23T and
the front axle of the Audi A6. In these two cases the running gear
the degree of axle damping is less than for example the 50%
damping of the BMW 3.28i . However both mass ratios are
25 substantially above that of the BMW 3.28i and hence substantially
above the critical value line.
In order to have further measurement points below the
critical value line the rear axle of the BMW 5.28 was measured
- 42 -
, ,
CA 022~23 1998-12-07
without the shock absorber. The vehicle could however only be
measured with a tire pressure decrease of -0.5 bar and -1.0 bar,
since at normal pressure the tire rose clear of the vibratory
platform.
The problem in the calculation of the parameters, in
which the non-spring-suspended mass mu fluctuates in accordance
with the quality of the installed shock absorbers, is however not
dealt with by using model iteration. However this is actually an
advantage in the case of the use of the mass value mu within the
degree of axle damping, as may be seen from the three measurement
points for the BMW 740i. Using an interface it was possible to
select between the shock absorber settings of hard, medium and
comfortable.
The poor shock absorber performance of the comfortable
setting implies a small mass ratio and displaces the degree of
axle damping to be nearer to the critical line.
The good damping performance with the hard setting
implies a large mass ratio and displaces the degree of axle
damping further away from the critical line. This effect occur to
a greater degree in the case of vehicles with a high degree of
external friction.
The rating in the running gear in separate stages and
made to apply for different vehicle type on the basis of the
degree of axle damping and including the mass ratio is a
substantial advance over the EUSAMA Method.
For maximum objectivity in rating of the test method it
is necessary additionally to compare the reactions to attempts at
- 43 -
CA 022~23 1998-12-07
manipulation with the new test method on the one hand with the
corresponding EUSAMA ratings on the other hand. The following
parameters were manipulated for this purpose:
Tire pressure - in set 0.5 bar steps from 1.0 bar under
the rated tire pressure to 1.0 above the rated tire
pressure ;
Loading - all vehicle were empty, some of them
additionally with a driver and one vehicle in its
predetermined design position;
Temperature - at the start of the development phase (in
winter 1996/97) one vehicle at 0 ~C and at 15 ~C;
Tires - one vehicle with three different tire designs.
Some vehicles with a special purpose wheel clip to vary
the non-spring-suspended mass.
The illustrations mentioned only constitute the tension
stage results. The corresponding values for the compression stage
hardly differ from those of the tension stage. The critical value
lines for the tension stage are extracted from the measurement
points, which were recorded at rated tire pressure and without a
load (at the most one driver, see Figure 19).
The values in brackets correspond to the ground standing
values at these measurement points. Same are necessary in order to
compare the reactions of the EUSAMA method with those of the new
method.
- 44 -
CA 022~23 1998-12-07
The tire pressure is one of the principal problem in the
measurement of running gear. The results of the assessment may be
manipulated to suit the particular intention. The preliminary
check of the tire pressure is to detect departures from the rated
pressure value and is the primary advantage of the new method of
testing. With the present stock of experience it is possible to
assume that departures in tire pressure of 0.5 bar may be detected
without difficulty. When further experience has been gained it may
be possible to detect a departure of 0.25 bar. Even smaller
intervals in the departure are theoretically not possible, since
the sum of the departures in the estimation of the parameters, the
preset rated value and the temperature drift constitute a natural
lower limit.
Figure 20 shows the measurement points for the rated
tire pressure and the measurement points for a pressure departure
of 0.5 bar. Major variations in tire pressure are not represented,
since same are certainly filtered out by the preliminary check.
The direction of the arrow indicates the increase in tire pressure
by 0.5 bar.
The following details may be derived from the above
graph. With an increase in tire pressure, the degree of damping
and the mass ratio decrease. Good running gear is more influenced
by the tire pressure than poor running gear. A change in the
rating does not occur. The influence of the tire pressure on the
EUSAMA values is on the contrary particularly great in the case of
poor running gear so that even a recommendation to replace may be
manipulated. The EUSAMA values were to some extent tripled by the
change in tire pressure. The variation in the rating points
(AchsDG,mF/mu) is certainly below 50%.
- 45 -
-
CA 022~23 1998-12-07
The influence of the tire pressure is not eliminated in
the result, but is made lower by a factor of 4 in comparison with
the EUSAMA method on average. Substantial departures from the
rated tire pressure can be detected. A manipulation of the
recommendation to replace on the basis of the tire pressure is not
possible. In fact poor running gear varies only around the error
of propagation as based on errors of measurement when the tire
pressure is changed.
The BMW 3.28i was measured in its design position, with
a load of ca. 230 kg and with a driver. The difference between the
empty vehicle and the vehicle with a driver was so slight that it
was not taken into account in the figure.
The following details may be gathered from Figure 21.
The degrees of axle damping were reduced by 2% owing to the load.
This is in contrast to the EUSAMA data, which increase slightly
with a load. The calculated value for the non-spring-suspended
mass mu is practically the same with and without a load. The weight
of the load means that the mass ratio is increased and the
assessment point is moved into the favorable field.
Assessment reacts to load of a vehicle with the better
classification of the running gear properties. The assessment of
this running gear situation is basically correct, but there is no
possibility of distinguishing the added load part of the mass from
the mass of the unloaded vehicle. The reaction is more significant
than with the EUSAMA method.
The manipulation of the recommendation to replace on the
basis of the loading is therefore possible in the case of running
gear in the environment of the critical value line.
- 46 -
. ~
CA 022~23 1998-12-07
The influence of the temperature of the surroundings was
investigated by making measurements at approximately 0~C and then
at the temperature of the surroundings of approximately 15~C. From
Figure 22 it will be seen that the degrees of axle damping
increases on cooling and that the mass ratio is not changed by
cooling down. The EUSAMA values remain practically unchanged.
Manipulation of the recommendation to replace on the basis of the
environmental temperature of the surroundings can consequently not
be excluded.
With a change in tires the tire spring constant for the
vehicle and the non-spring-suspended mass will be changed. In this
connection two test series were run:
- The BMW 3.28i was measured with the complete sets of tires in
the design position.~5 - Several vehicles were measured with a special purpose wheel
clip (5.6 kg) and, respectively, a screwed-on steel plate
(14.3 kg) on the wheel.
The following details may be gathered from Figure 23.
The rating points (AchsDG,mF/mu) are not changed in their
classification. The slightly improved degree of axle damping which
heavy tires is compensated for by the decrease in the mass ratio.
This effect is particularly clear in the case of small
low mass ratios. The EUSAMA values decrease with an increase in
the tire spring constant kR. A change in tires mainly causes
changes in the spring constant kR and thus in the degree of
coupling. The influence on the non-spring-suspended mass is within
the range of accuracy of measurement owing to the small
differences in measured values. The influence of the spring
- 47 -
CA 022~23 1998-12-07
constant kR is taken into account in the degree of axle damping,
whereas in the case of such an increase in coupling the EUSAMA
method provide poorer road adherence data. A manipulation of the
recommendation is not possible.
The following may be seen from Figure 24. The degrees of
axle damping are, like the road adherence values, slightly reduced
by the additional weight. The additional weight is fully taken up
in the non-spring-suspended mass and in part leads to a drastic
reduction in the mass ratio.
Bad running gear can be imitated by manipulation of the
non-spring-suspended mass. The EUSAMA values tend to react to
manipulation in the same manner but not to such a significant
extent. The result of this is that highly defective running gear
may still be classified as good.
By way of conclusion it is possible to say the following
on the possibilities of manipulation. The manipulation of the tire
pressure would be detected by the preliminary examination in view
of kR and mS. As compared with the EUSAMA method there is the
advantage that statement as regards the tire pressure is possible
and the sensitivity of the examination method for tire pressure is
less. Manipulation by way of the temperature of the shock absorber
is excluded by the warming up phases. To be possible manipulation
on the basis of the load would involve massive external
manipulation and may therefore also be excluded. A further point
is that the load affects the mass ratio, see the above mentioned
graph. Manipulation by way of changing the tires is not possible,
since there is a check in this respect. As compared with the
EUSAMA method there is the advantage that the influence of the
changed spring constant is removed by calculation.
- 48 -
. .
CA 022~23 1998-12-07
Manipulation of the mass mu could make it seem that the
running gear is poorer than is the actually case, something which
would have to involve massive manipulation from the outside and
can consequently be excluded. As compared with the EUSAMA method
there is the advantage that the load is included in the mass ratio
and is therefore covered by the test method. Figure 25 shows by
way of conclusion the directions of displacement of the effective
parameters in the field of degree of axle damping to the degree of
axle damping mF/mu. Manipulations are consequently either excluded
or they have a substantial effect of the results of measurement.
It is to be understood that the above description is
intended to be illustrative and not restrictive. Many embodiments
will be apparent to those skilled in the art upon reviewing the
above description. Those skilled in the art will recognize as an
equivalent or alternative method of tire testing and combining a
tire testing machine with a wheel balancing machine. The scope of
the invention should, therefore, be determined not as reference to
the above description, but should instead be determined with
reference to the appended claims along with the full scope of
equivalence to which such claims are entitled.
- 49 -
.