A Note On The Use of Four Post Test Rigs.: Doc. Ref. FSM-DS-98-465 Issue 4 Date 13.05.01. Title Rig Test Notes

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Doc. Ref.

FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

A NOTE ON THE USE OF FOUR POST TEST RIGS.

D.A.W.
29.12.98.

INTRODUCTION.

The purpose of this document is to introduce the concept of using a static test rig to help to optimise the
suspensions of wheeled vehicles, with particular emphasis on four wheeled competition vehicles.

The suspension of a conventional four-wheeled competition vehicle normally comprises a set of springs,
dampers and rigid links for each wheel. These connect the wheel hubs (unsprung masses) to the body of
the vehicle (sprung mass) and control the motion of the unsprung masses relative to the sprung mass.
Commonly, four springs and four dampers are used. Two additional springs are often used to influence the
relative movement of the unsprung masses.

The effect of the suspension upon the relative motion of the vehicle masses depends, of course, upon the
properties of the suspension elements but also upon suspension geometrical properties, the mass
distribution of the vehicle and tyre properties. As a minimum, the suspension is required to control three
sprung mass modes and the four unsprung masses (i.e. seven degrees of freedom) with only four dampers
and, perhaps, six springs. Clearly, spring and damper configurations (“settings”) cannot be set to control all
degrees of freedom independently. It follows that suspension settings represent a compromise and will, in
general, be sub-optimal for any particular degree of freedom.

Suspensions are also used to control other vehicle properties. Examples are steering response time
constant, the steady state and transient lateral stability of the vehicle, driver “feel”, powertrain “traction” and,
particularly for some competition vehicles, aerodynamic forces. As a result, a large number of variations
upon a conventional suspension layout have been devised. Some attempt specifically to increase the
number of suspension variables, whilst others are intended to address specific vehicle deficiencies. Many
are non-linear. The latter usually cause the character of the response of the vehicle to vary depending upon
the amplitude of the road disturbance and upon the history of road inputs. At best, such variations make the
management of a vehicle an even more difficult task.

The sum of the demands made upon a vehicle suspension makes arriving at the “best” compromise a
difficult task. The traditional method of arriving at a suspension set up is to use a combination of:

a) rules derived from simple theory,

b) intelligence obtained from rival teams,

c) driver preferences, and comparative lap times during circuit tests.

Simple theory can provide an adequate starting point provided that the suspension elements are
substantially linear, and the springs and dampers are “soft”. Simple theory is likely to be an inadequate
predictor of performance in presence of friction or other significant non-linearities, or if a “stiff” suspension
configuration is required (one which implies significant tyre deflection per unit road input).

Intelligence obtained from rival teams will, at best, achieve parity of performance.

Page 1 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Drivers are usually very good at sensing deficiencies in vehicle handling. Experienced drivers can
sometimes identify the cause(s) of handling deficiencies. As a rule, drivers will encourage suspension
adjustments that provide excellent feedback and short steering time constants. The effect of suspension
setting changes upon lap times is, sometimes, influenced by driver confidence. The ability of a suspension
to cope with road surface inputs is rarely assessed reliably by a driver – a vehicle with poor response driven
(relatively) slowly feels very like a vehicle with well controlled response driven quickly. These facts imply that
setting up a vehicle suspension using only driver feedback and comparative lap times rarely, if ever, yields
optimal suspension settings.

A multi-post test rig provides a tool for assessing and optimising the balance and energy dissipation
properties of a competition vehicle suspension.

SUSPENSION SETTING LIMITS.

If optimum use is to be made of the energy dissipation abilities of dampers used in a vehicle suspension,
then there are limits to the maximum values of both spring stiffnesses and damper settings for a particular
vehicle. Other vehicle and tyre considerations normally mean that minimum values also exist. The factors
that affect both extremes are discussed here.

It is true that the “best” control over vehicle response to vertical road inputs would be achieved with relatively
soft springs and dampers “matched” to the soft springs – hence also relatively soft. If this approach were
adopted, however, then the vehicle would require stiff anti-roll bars or geometry adjustments to contain
attitude changes in manoeuvres. Stiff bars, coupled with soft springs would lead to a loss of roll control
under limit manoeuvre conditions. Thus the lower limit to suspension settings is normally set by the
requirement to constrain attitude changes in manoeuvres whilst maintaining good control over transient
responses. The need to “work” the tyres so that they reach operating temperatures quickly is also a
consideration.

Conversely, it is possible to match dampers with relatively stiff springs to achieve satisfactory control over
vehicle response to road inputs. Such a strategy can achieve adequate control over road inputs with
relatively little damper motion, and hence with little motion of the vehicle sprung mass relative to its wheel
hubs. The strategy will, however, be limited by tyre characteristics. Specifically, tyre stiffnesses will impose
a practical upper bound to values of spring (and damper) stiffnesses. At the limit, with a locked suspension,
all vehicle motion of the sprung mass relative to the road will be caused by tyre deflection, and the vehicle
response will be severely under-damped.

An impure suspension layout, or "geometry" can cause undesirable vehicle transient motions at large
suspension excursions (“bump steer” would be an example). A characteristic like this, though rare in “pure”
competition vehicles, can be a reason for increasing suspension stiffness in order to reduce suspension
travel.

Aerodynamic down-force, when this is significant, is usually affected by ride height and pitch angle. Clearly
suspension deflection can couple with changes in the aerodynamic forces. In extreme cases, the coupling
can be so large that the vehicle becomes unstable above a critical speed. The requirements to control
aerodynamic forces may also force an increase in suspension stiffness, even when the critical speed is
beyond the capabilities of the vehicle. However, there may be an alternative solution to the problem, see
below.

Increasing the stiffness beyond the limit imposed by the tyre can be made to work, but the vehicle will be
under-damped, and will respond illogically to setting changes (increasing damper settings will, for example,
reduce vehicle damping). The behaviour of the vehicle will also be very sensitive to small changes in vehicle
parameters.

Page 2 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

To summarise, the range of spring and damper settings that can be used on a competition vehicle is limited
at the upper end by tyre stiffnesses and at the lower end by vehicle body control. Tyre temperatures,
suspension geometry and the requirement to control aerodynamic force coefficients may affect the lower
limit.

RIG TEST MEASUREMENTS.

Perhaps the most informative measurements recorded during a vehicle rig test are the four contact patch
loads. These can be used, at least on a comparative basis, to assess the ability of a suspension set-up to
maintain contact between the tyres and the road surface.

Vertically disposed accelerometers are attached to the simulated road surface, the wheel hubs and the
vehicle body, or sprung mass. In absolute terms, these can be used to define the road input and the
response of the vehicle to the road input. In relative terms, they can be used to estimate tyre deflection and
the deflection of the spring mass relative to the hubs.

Taken together, the measurements define, at least in principle, the simulated vertical road inputs to the
vehicle and the vertical components of the response of the vehicle to those inputs.

It is useful, though not essential, to be able to record the four damper displacements. These additional
measurements can be used to estimate the dynamic “velocity ratio” of the suspension, suspension
“installation stiffness” and can also be used to estimate overall front and rear suspension roll stiffnesses.

INFORMATION DERIVED FROM RIG TESTS.

In general, competition vehicle suspensions are non-linear. In addition, tyres have non-linear properties.
Both dampers and tyres have characteristics that vary (to some extent) with temperature. All vehicles have
additional characteristics that were not intended by the designer (for example, suspension friction and
backlash, and vehicle flexibility). These sometimes affect the performance of the suspension to a significant
extent.

Ideally, information recorded during a rig test should be processed to enable the performance of different
suspension configurations to be compared on a quantitative basis. It should also be possible to extract
information about unplanned vehicle characteristics such as those mentioned in the preceding paragraph.

The test rig data analysis suite includes facilities for processing recordings in several ways. These include:

• Time history plots of scaled measurements with or without average value removal. The starting
point and length of the plots are variable. The time histories can be further processed to extract
and plot maximum, minimum and mean values on a “per cycle” basis. No numerical information is
presented.

• The time history of one scaled parameter (measurement or weighted combination of


measurements) plotted against another. The X-axis parameter may be double integrated. Either
parameter may be differentiated. The average values of either parameter may be removed. A
linear trend in either parameter may be removed. The starting point and length of the plots are
both variable. Estimates of various properties of the relationship are appended to the plot as

Page 3 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

numerical values. Effectively, the numerical estimates represent linear approximations to the
(presumed) relationships between one parameter and the other.

• Frequency domain plots of various types. The basis of all frequency domain plots is the Discrete
Fourier Transform. The Transform describes any two scaled parameter time histories in terms of
the frequency components that describe the time histories. The frequency components are
estimated by averaging. Essentially, then, frequency domain plots represent averages (in the
least squared sense) of the chosen time histories. The most commonly used plot type is the
Frequency Response Function; this is the transform of one parameter divided by the transform of
the other. If the responses are linear, then the frequency response function is an estimate of the
Transfer Function of one parameter relative to the other. A Coherence Function plot type is
included to assess the validity of a Frequency Response Function.

• Reconstruction of a linear “bicycle” model of the vehicle.

The majority of the analysis facilities make the (often implicit) assumption that the vehicle under test has
linear properties. In reality, this is rarely (if ever) the case. However, a linear model can be interpreted
rather more easily than a non-linear model, provided it is remembered that the linear parameters are simply
a “view” of a complex non-linear vehicle. Much evidence exists to support the assumption that judgements
made from linear vehicle parameters do, in fact, translate reasonably well to the non-linear real vehicle.

RIG TEST SIMULATED ROAD INPUTS.

The analysis tools used to process test rig data are, in the main, aimed at characterising vehicle response to
road inputs, and making judgements based upon the results obtained. If the vehicle had linear response
characteristics, then the responses, and the judgements made about them would be independent of the time
history of the road input.

In reality, competition vehicles have non-linear characteristics, some of which would, in turn, be expected to
vary with temperature. Hence the ideal road input would be an exact representation of a real circuit.

It can be argued that a representation of a real circuit is not ideal for optimising the suspension settings of a
competition vehicle (see Appendix A). If the thesis is accepted, then a road input is required that will:

• input minimum total energy to the vehicle (and minimise energy dissipation and hence
temperature induced vehicle changes),

• generate representative vehicle response levels over a range of frequencies covering the principal
rigid body modes.

• allow the vehicle to be characterised in the minimum time.

A constant peak velocity swept sinusoidal input has the above characteristics. Phasing the four road
actuators appropriately permits responses to be recorded when the vehicle is subjected to pure heave, pitch,
roll or warp simulated road inputs. Of these, pure heave is the most useful, supported by pitch if the modal
responses are almost de-coupled. Pure warp allows the overall roll stiffness of the suspension to be
estimated. Roll input is the least used, partly because the roll response of the vehicle is affected by the fact
that the tyres are not rotating during rig tests.

Page 4 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

SIMULATED AERODYNAMIC DOWN-FORCE.

It will be necessary to apply simulated down-force for vehicles that generate significant aerodynamic down-
force in normal operation. This will load tyres in a realistic way, and will permit representative road inputs to
be applied to the vehicle without the tyres leaving the road actuators.

In reality, down-force will be expected to vary as a function of sprung mass ride height and attitude, and will,
of course, vary with airspeed. The aerodynamic derivatives (rate of change in force with position and
velocity) will interact with vehicle motion, and will affect the natural frequencies, the damping ratios and the
coupling of the heave and pitch modes. The “aeroelastic” problem is an interesting one because it can
cause vehicle instabilities at high speed. A multi-post test rig can provide an effective tool for investigating
the phenomenon; however, the aerodynamic derivatives must be mapped accurately, and the test rig must
be able to simulate down-force changes with high precision and without coupling with the structural modes of
the vehicle. When a vehicle is prone to aeroelastic instabilities, then it is probable that the best solution
would be to modify the aerodynamic, rather than the mechanical, characteristics of the vehicle.

When setting the suspension characteristics to optimise the vehicle for mechanical performance, the
preferred strategy is to simulate a constant down-force distribution appropriate for low to medium speed
corners, and to minimise the change in down-force with ride height and attitude. This can be achieved by
high pre-load, low rate springs or by pneumatic actuators. Each of these options has its own advantages
and disadvantages, but a common feature is that they require a relatively simple interface with the vehicle.

OPTIMISING DAMPER SETTINGS.

The objectives of a “generic” rig test are to:

• maximise the damping ratios of the linear vehicle heave and pitch modes,

• de-couple heave and pitch modal responses,

• minimise variations in the front and rear axle contact patch load per unit road disturbance, with
particular emphasis on the driven wheels.

High modal damping ratios will imply that vehicle motion following a disturbance will be dissipated quickly.
De-coupled modal responses will mean that the vehicle will respond consistently to disturbances and will, as
a result, generate consistent driver cues. Minimised front and rear contact patch loads implies that the effect
of disturbances on grip and traction will be minimised.

In general, it will not be possible to find a suspension configuration that will achieve optimal values of all the
above parameters simultaneously. A “cost function” or “performance index”, constructed from the
parameters, will enable suspension configurations to be compared. For a given combination of springs,
damper settings will exist that will yield a minimum value for the cost function.

Suspension non-linearities that limit suspension travel will affect cost function values and the way they
change with suspension changes. As a result, such devices are normally removed, or rendered inoperative,
at least during the initial stages of a rig test.

Page 5 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

The distribution of damping between bump and rebound, and between high and low speed affects the way
the vehicle will “jack” in response to road inputs, and will affect steering response and driver “feel”. In
general, more linear damper characteristics imply that a vehicle will respond in a more consistent way to
road inputs, but will feel more “lazy”. As a result, more linear dampers may make the vehicle faster, but the
driver may not be able to make use of the improved performance.

Spring stiffness values affect vehicle roll and pitch angles during manoeuvres and, crucially, the mechanical
mid-corner lateral balance of the vehicle. None of these can be assessed definitively during a rig test. As a
result, it is usual to adopt the strategy whereby spring settings will be specified by the team, and rig tests will
be used to optimise damper settings to match the chosen springs.

In view of the factors noted above, and the limitations of rig tests, it is normal to optimise dampers for two or
three spring configurations. In each configuration, damping distribution will be set to minimise vehicle
"jacking". If a team prefers “square” damper characteristics (stiff at low speed, soft at high speed), then an
alternative configuration with more linear damping characteristics will be suggested. The objective is to
provide teams with several configurations to evaluate during track tests.

SAMPLE RESULTS.

Figures 1 to 6, inclusive, of this document contain results analysed from a four post rig test carried out at
Thetford on a competition vehicle. The results were generated by analysing various measurements
recorded during a single run using Dynamic Suspensions’ analysis software. The following notes provide an
outline explanation of the plots shown.

The run comprised a swept sinusoidal heave input (all actuators in phase and the same amplitude). The
amplitude of the drive demand was ±30 mm. The demand signal was passed through a single pole low pass
filter set to 1 Hz. before being used to control the “road” actuators. The effect of the filter was to generate a
constant peak velocity of ±189 mm./sec.

Page 6 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Figure 1.

Figure 1 contains two frequency response functions (FRF) obtained from recorded data, and two frequency
response functions (the crosses) produced from linear filters fitted to the measured data. The first plot is
F(ront )Ax(le )V(ertical) C(ontact )P(atch )L(oad) per unit F(ront )Ax(le )V(ertical) DDZh (double differential of
vertical hub position, i.e. hub acceleration). The dependent variable has units of lbf, the independent
variable has units of gn. Thus, at low frequency, the FRF will asymptote to the weight (in N.) on each front
actuator pad (Newton’s Second Law). A linear vehicle, properly set up, would produce a plot which would, at
high frequencies, asymptote to the unsprung (i.e. hub & wheel) mass. Between the two extremes, the shape
of the FRF describes the response of the sprung mass, supported by the suspension springs and dampers,
“above” the hub. The inset legends show the computed filter parameters. In theory (but only in theory):

Amp is the spring mass*go (N) per “corner”,

fo is the sprung mass natural frequency (Hz.) on its suspension,

zeta is the damping ratio of the sprung mass, spring and damper combination,

Ofst is the (sprung + unsprung mass)*go (N) per “corner”.

Note that tyre characteristics do not affect the shape of these plots directly. The second plot is similar, but is
for the rear axle.

The filter is capable of describing adequately the characteristics of a “single corner” linear vehicle. It is,
therefore, valid for a four-wheeled (linear) vehicle if it is symmetrical about its longitudinal geometric axis,
and if the vehicle body (the sprung mass) does not pitch when excited by a heave road input.

Page 7 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Optimum values for fo and zeta will depend upon the stiffness ratios of the suspension springs and
corresponding tyres. As a target, values for zeta should probably be between 0.450 (for a stiffly sprung
suspension) and 0.650 (for a softly sprung suspension).

To summarise, figure 1 contains a description of the sprung mass acting on its suspension. The plots show
how well the struts (springs and dampers) are matched to the sprung mass of the vehicle.

Figure 2.

Figure 2 contains similar FRF plots, but this time the independent variables are road acceleration (DDZr),
rather than hub acceleration. Here the legends contain the frequency in Hz (Freq), amplitude in lb/gn (Amp)
and phase angle in degrees (Phse) of the FRF’s at the marked points shown on the plot. Transfer functions
of the CPL relative to road acceleration have a complicated structure, but broadly show the vehicle response
on its suspension and its tyres. As a consequence the damping ratio of the heave mode will be lower than
that computed from the plots shown in figure 1. Two additional plots have been added to figure 2. These
indicate the effect of asymmetries in the vehicle, its suspension, (and the rig) about the nominal longitudinal
axis of symmetry. Magnitudes of around 10 percent of the symmetric response are quite typical – much
greater than that should probably be investigated further.

It can be argued that, other things being equal, a set up with minimum value of “Amp” for the driven wheels
is to be preferred.

The concept of axle-based vehicle dynamics is simplistic and can be misleading. Figure 3 is the “modal”
equivalent of the previous plots, for the heave mode only. To be specific:

Page 8 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Vert CPL is the average of the four contact patch loads,

Vert DDZr is the average of the four “road” accelerations,

Vert DDZh is the average of the four hub accelerations.

Optimum peak values will depend upon static “weight on wheels” and upon the ratio of suspension spring
stiffness to sprung mass, but would be expected to be between 1.5 and 2.0 times the static weight on
wheels.

Figure 3.

The first FRF shown in figure 3 describes the heave response of the vehicle on its suspension and tyres; the
second describes the sprung mass of the vehicle on its suspension. The legends contain the optimised
parameter values for appropriate simple linear filters. The damping ratio (zeta) in the second legend is an
approximate estimate of the heave mode damping ratio. The estimate is an approximation because the filter
used has two poles (and is linear) - the real vehicle has (at least) 8 degrees of freedom and is non-linear.

Page 9 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Figure 4.

Figure 4 is a (reasonable) attempt to describe the performance of the installed suspension (springs, dampers
and geometry). The new “parameter” shown here is the difference between hub and body accelerations.
Formally,

FAxV DDZ(h-b) = FAxV DDZh - FAxV DDZb,

and FAxV DDZb is the F(ront )Ax(le )V(ertical) body (or sprung mass) acceleration.

If the new parameter is double integrated (as it was to generate the plot shown), then the FRF describes the
complex sum of spring force per unit displacement and damper force per unit displacement. If the damper
force is linear, then the plots describe the total (spring plus damper) stiffness of the front and rear axles. The
shapes of the actual plots are not atypical: the plots cannot be trusted at frequencies lower than about 3 Hz.
(dividing zero by zero, in the limit). Phase angles of around +50 degrees at the heave mode resonant
frequency indicate that the dampers and springs are well matched (slightly over-damped, or “damper
controlled” in the example shown).

The phase angles of the strut complex stiffnesses should asymptote to 90° at high frequencies. A phase
angle that peaks at some frequency and then decreases with increasing frequency, usually indicates that the
unsprung mass value used to correct contact patch load is incorrect (too large) or that an additional “spring”
exists in series with the strut. In the latter case, the additional spring constitutes an “installation” stiffness;
this can reduce the overall effectiveness of a suspension.

Page 10 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Figure 5.

Figure 5 contains two response functions. These represent, respectively, the ratio of tyre deflection per unit
strut deflection for the front and rear axles. The legends contain “spot” values of the functions taken at a
frequency approximating to the heave mode resonant frequency. Generally, the functions increase in
magnitude at frequencies below and above the heave mode frequency. Friction forces would be expected to
cause an increase in magnitude at low frequencies. The increase at high frequencies is caused by ever
increasing damper forces.

High spot values (close to unity) would suggest that suspension settings are too stiff for the tyres. Low spot
values (less than, say, 0.25) could suggest that the tyres would not be “worked” enough to reach operating
temperatures. When suspension settings are too stiff, then the effect of damper changes would be counter-
intuitive, i.e. increasing damper “stiffness” would decrease heave and (possibly) pitch mode damping ratios
and vice versa.

The two additional plots were designed to indicate the effect of asymmetries on displacement-related hub
responses. In the example front axle hub roll was zero, because the FLH hub accelerometer was non-
functional during the test, and its signals were replaced by those from the FRH accelerometer.

Page 11 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Figure 6.

Figure 6 contains three frequency response functions.

The first describes the heave mode hub acceleration with respect to the average “road” acceleration. The
function is an alternative description of how well the dampers control the heave mode. (All other things
being equal, optimum control is achieved when the peak value is a minimum.)

The second FRF describes the pitch mode hub acceleration with respect to the average “road” acceleration.
It is the natural companion to the first. Arguably. the maximum amplitude of this should be minimised for a
competition vehicle. A peak value less than 0.1 would be considered to be acceptable.

The third FRF describes the pitch mode body acceleration with respect to the average “road” acceleration. If
hub and body acceleration FRF’s can be minimised at the same time, then this suggests that both the tyres
and the suspension will be set up correctly for the centre of gravity of the sprung mass. This rarely happens;
some customers prefer to minimise hub acceleration, others prefer to minimise body acceleration.

Figure 7 is a recent addition to the “standard” plots. It indicates the dynamic variation of front (red) and rear
(green) axle ride heights per unit (heave) road input. The response functions can be useful for
“aerodynamic” race vehicles, where the change in position of the centre of pressure is an important
consideration.

Page 12 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Figure 7.

A linear representation of the vehicle can be identified from the measurements recorded during a rig test run.
Various constraints are available, including fixing hub masses to known values. Sprung mass estimates can
also be fixed in order to allow the effect of damper changes to be assessed with no interference from
changes to identified mass parameters. A typical results window is shown in figure 8. The “Performance
Index” (also called a “Cost Function”) shown is an empirical function designed to attain a degree of
consistency when comparing the effects of suspension changes on a vehicle.

Page 13 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Figure 8.

Page 14 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

MISCELLANEOUS NOTES.

It is important to remember that a “rig set-up” addresses mechanical (i.e. non-aerodynamic) vehicle
performance. Where a vehicle generates significant aerodynamic down-force, then a rig set-up must be
adapted in order to maintain control at high vehicle speeds and to make the variation is suspension
performance progressive with change in speed. An objective is to achieve rig set-up characteristics through
more critical low/medium speed corners, and to control the sprung mass attitude and ground clearance by
increasing wheel rate stiffness smoothly and progressively as the speed increases.

On occasion, a rig set-up does not improve a competition vehicle. There could be several reasons for this.

• The vehicle could already be configured optimally. After over three hundred rig tests, this has yet
to happen.

• A driver cannot, or refuses to, drive the revised vehicle. This has happened when a radical
departure from an existing set-up has been presented to a driver on a quick look basis.

• The revised vehicle really is inferior. Again, this has happened (once), although warning signs
were apparent at the start of the test. It must be remembered that suspension set-up is just one
of many coupled variables that define vehicle performance. If the vehicle has a deficiency, then it
is sometimes possible to mask the deficiency by compromising the suspension settings. In that
case, optimising the suspension may expose the deficiency.

• A race team fails to convert a rig set-up to an acceptable race set-up. This is the most common
cause of failure of a rig test. It occurs most often with a “down-force” vehicle. It seems to be a
fact that a very “tight” suspension set-up makes vehicle easy to manage and appears to be the
preferred option when track time is limited.

As a general comment, a stiff set-up can be made to work, and may well be preferred by a driver. However
grip levels will be relatively low (even on a smooth circuit), and may be further compromised by measures
required to control load distribution (front axle droop stops, for example). It is quite common for a rig
suspension set-up to feel slow to a driver whilst returning improved lap times.

Page 15 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

APPENDIX A.

A Note on Circuit Simulations.

A circuit simulation test is one in which the road actuators are driven with time history signals representing
the surface inputs of a specific circuit and vehicle down-force actuators (when fitted) are driven under load
control to obtain prescribed force and moment time histories. The two sets of actuators are driven
synchronously. Together they can simulate the motion of the vehicle and its suspension during a complete
lap of the circuit.

A real circuit can be simulated by “reverse fitting” vehicle responses on the test rig with similar responses
recorded during a lap of the desired circuit using an identical vehicle set-up. The process requires the
manoeuvring forces (braking, accelerating and cornering) and the aerodynamic forces acting on the vehicle
to be estimated accurately and removed from the track measurements. The residual forces, suspension
deflections and/or hub accelerations are assumed to be road surface inputs. A model of the vehicle,
deduced from trial rig tests, is then used to estimate the road input time histories (also known as drive files)
that will reproduce the track measurements.

The weak links in the process are the accuracy of the aerodynamic force coefficient “map”, and the fact that
the vehicle model used to generate the drive files is itself a linear approximation of the real vehicle. Despite
this, it can be argued (with some justification) that the vehicle model derived, after iteration, is the best
possible linear model (for the particular circuit).

Problems begin once the vehicle under test (or its suspension) has been changed in a significant way. After
such a change, it is probable that the speed profile of the vehicle will change. As a consequence the inertia
forces may change both in magnitude and as a function of distance travelled. In this case the load actuator
demand files will no longer be correct.

Perhaps the most serious problems will occur if rig data recordings for the revised configuration are to be
processed statistically. This might be required to compare frequency response functions of the original and
revised configurations or if a model for the revised configuration is to be identified (perhaps to compare the
two configurations numerically). Both types of processing can be effected in principle. However, the
recorded signals will be random in structure, and will therefore be subject to normal statistical variances. As
a conservative estimate, the averaging time required to achieve a frequency response function reliability of
ten percent would be over 8.5 minutes. Very few circuits are sufficiently long to satisfy this criterion.

To be fair, circuit simulations can be used to improve a vehicle configuration, provided the objectives are
sufficiently focused. For example, a “generic” suspension configuration will need to be modified for a
particular circuit, for example by moving damping between bump and rebound half cycles, by adjusting
spring pre-loads, and/or by adjusting suspension travel limits. A first cut at such relatively minor adjustments
can be carried out on a test rig using circuit simulations.

Page 16 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

APPENDIX B.

Parameter Definitions and Nomenclature.

Table B1 contains a list of the measurement included in the Dynamic Suspensions Test Rig Measurement
Frame. Hopefully, the channel titles used for the measurement frame are self-explanatory. For the record,
this frame is broadcast from the Test Rig Controller at a rate of 2,000 frames per second. The frame is
received by an Interface Unit, which extracts a specified sub-set of the basic frame, and decimates that sub-
set to achieve the specified recording data rate.

Generally, the analysis and presentation routines use parameters derived from the Measurement Frame.
This is so that small asymmetries in the vehicle and transducer mis-alignments can be averaged out. The
set of derived parameters used to process test rig recordings is defined in Table B2. The analysis program
assumes that a derived parameter is compiled from exactly four measurement channels, and is defined as:

4
Derived _ Parameter = ∑ (Measured _ Parameter [Pi ]⋅ W i )
i =1

where Pi is the Measurement Frame Address of the ith measured parameter,


and Wi is the corresponding weight, or multipying factor.

The Measurement Frame Address for each measured parameter is shown in Table B1, Column 1. The
“Title” column of Table B2 contains the mnemonic used to label plots generated by the analysis program. By
convention, the derived parameter computation is terminated when either the fourth parameter number and
associated weight have been read, or when a null Measurement Frame Address is encountered.

Table B3 defines the nomenclature used for the summary file.

Page 17 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Chan Title Scale Factor Offset


1 Frame Count 1.000 0.000
2 FLH Load -1.6256 0.000
3 FRH Load -1.6413 0.000
4 RLH Load -1.6844 0.000
5 RRH Load -1.6527 0.000
6 FLH Pad Accn 409.5 0.000
7 FRH Pad Accn 387.442 0.000
8 RLH Pad Accn 389.764 0.000
9 RRH Pad Accn 418.300 0.000
10 FLH Position -12.042 0.000
11 FRH Position -12.042 0.000
12 RLH Position -12.042 0.000
13 RRH Position -12.042 0.000
14 FLH Strut Pos -20.412 0.000
15 FRH Strut Pos -19.727 0.000
16 RLH Strut Pos -25.800 0.000
17 RRH Strut Pos -23.667 0.000
18 FLH Hub Accn 362.856 0.000
19 FRH Hub Accn 327.133 0.000
20 RLH Hub Accn 332.295 0.000
21 RRH Hub Accn 375.299 0.000
22 FLH Body Accn 365.922 0.000
23 FRH Body Accn 318.655 0.000
24 RLH Body Accn -481.805 0.000
25 RRH Body Accn -457.507 0.000
26 Dvr Vert Accn -452.456 0.000
27 Roving Accn -455.858 0.000
28 Frnt Roll Pos -97.365 0.000
29 Rear Roll Pos -190.560 0.000
30 FLH CP Load -1.6256 0.000
31 FRH CP Load -1.6413 0.000
32 RLH CP Load -1.6844 0.000
33 RRH CP Load -1.6527 0.000
34 Ext Input -409.6 0.000
35 PosDemF -61.776 0.000
36 PosDemR -74.360 0.000
37 Stable 1 0.000

Table B1.

Page 18 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Title P1 P2 P3 P4 W1 W2 W3 W4
FAxV CPL 30 31 0 0 0.5 0.5 0.0 0.0
RAxV CPL 32 33 0 0 0.5 0.5 0.0 0.0
FAxV DDZr 6 7 0 0 0.5 0.5 0.0 0.0
RAxV DDZr 8 9 0 0 0.5 0.5 0.0 0.0
FAxV DDZh 18 19 0 0 0.5 0.5 0.0 0.0
RAxV DDZh 20 21 0 0 0.5 0.5 0.0 0.0
Vert CPL 30 31 32 33 0.5 0.5 0.5 0.5
Pitch CPL 30 31 32 33 0.5 0.5 -0.5 -0.5
Vert DDZr 6 7 8 9 0.5 0.5 0.5 0.5
Ptch DDZr 6 7 8 9 0.5 0.5 -0.5 -0.5
Vert DDZh 18 19 20 21 0.5 0.5 0.5 0.5
Ptch DDZh 18 19 20 21 0.5 0.5 -0.5 -0.5
FAxVFst 30 31 18 19 0.5 0.5 -41.5 -41.5
RAxVFst 32 33 20 21 0.5 0.5 -33.5 -33.5
FAxV DDZ(r-h) 6 7 18 19 0.5 0.5 -0.5 -0.5
RAxV DDZ(r-h) 8 9 20 21 0.5 0.5 -0.5 -0.5
FAxV Zstr 14 15 0 0 0.5 0.5 0.0 0.0
RAxV Zstr 16 17 0 0 0.5 0.5 0.0 0.0
FAxV DDZb 22 23 0 0 0.5 0.5 0.0 0.0
RAxV DDZb 24 25 0 0 0.5 0.5 0.0 0.0
FAxV DDZ(h-b) 18 19 22 23 0.5 0.5 -0.5 -0.5
RAxV DDZ(h-b) 20 21 24 25 0.5 0.5 -0.5 -0.5
Vert DDZb 22 23 24 25 0.5 0.5 0.5 0.5
Ptch DDZb 22 23 24 25 0.5 0.5 -0.5 -0.5
FAxRFst 30 31 18 19 0.5 -0.5 -41.5 41.5
RAxRFst 32 33 20 21 0.5 -0.5 -33.5 33.5
FAxR Zstr 14 15 0 0 0.5 -0.5 0.0 0.0
RAxR Zstr 16 17 0 0 0.5 -0.5 0.0 0.0
FAxR DDZ(h-b) 18 19 22 23 0.5 -0.5 -0.5 0.5
RAxR DDZ(h-b) 20 21 25 25 0.5 -0.5 -0.5 0.5
Roll CPL 30 31 32 33 0.5 -0.5 0.5 -0.5
Warp CPL 30 31 32 33 0.5 -0.5 -0.5 0.5

Table B2.

Page 19 of 20 Pages
Doc. Ref. FSM-DS-98-465
Issue 4
Date 13.05.01.
Title Rig Test Notes

Mnemonic Description
Title Run title, usually describing the set-up change from the previous run
FileName 1 Name of the data file used to identify parameters
Ktf Front tyre stiffness (usually N/mm)
Ktr Rear tyre stiffness (usually N/mm)
Kwf Front suspension spring stiffness (usually N/mm)
Kwr Rear suspension stiffness (usually N/mm)
Cwf Front suspension damper stiffness at Dfreq.
Cwr Rear suspension damper stiffness at Dfreq.
Dfreq Frequency (Hz.) at which damper stiffnesses are true
Muf Front unsprung mass (usually Kg)
Mur Rear unsprung mass (usually Kg)
Mss Sprung mass (usually Kg)
Rcg Sprung mass c.g. position (fraction of wheelbase)
Rkk Sprung mass radius of gyration (fraction of wheelbase)
FoAxf Front axle model natural (i.e. undamped) frequency (Hz)
ZoAxf Front axle model damping ratio (fraction of critical)
FoAxr Rear axle model natural (i.e. undamped) frequency (Hz)
ZoAxr Rear axle model damping ratio (fraction of critical)
FoHv Model heave mode natural frequency (Hz.)
ZoHv Model heave mode damping ratio (fraction of critical)
FoP Model pitch mode natural frequency (Hz.)
ZoP Model pitch mode damping ratio (fraction of critical)
Max Ptch Maximum pitch amplitude per unit heave input
Pfreq Frequency of max pitch amplitude
Ld Varf Max. peak front contact patch load variation (fraction of static weight)
LdVarr Maximum peak rear contact patch load variation (fraction of static weight)
T/Sf Ratio of front tyre to front strut Deflection at DFreq
T/Sr Ratio of rear tyre to rear strut Deflection at DFreq
Hhve Peak value of average hub deflection per unit average platform acceleration
Hptch Peak value of hub “pitch” deflection per unit average platform acceleration
Bptch Peak value of sprung mass “pitch” deflection per unit average platform acceleration
Fdcg Front axle dynamic ground clearance
Rdgc Rear axle dynamic ground clearance
HtRat Fornt/rear tyre heat input ratio
Perf Cost function, or performance index.
Comf Comfort Rating
Sym Symmetry Rating

Table B3. Summary Data Nomenclature.

Page 20 of 20 Pages

You might also like