Australian Road Research, 15(4), December 1985, pp. 249-263.

Heavy Vehicle Design Parameters and Dynamic Pavement

Loading

A.N. HEATH
M.C. GOOD
Department of Mechanical and Industrial Engineering,
University of Melbourne,
Parkville, 3052

ABSTRACT

This paper presents preliminary results from ARRB Project 387: 'Whole Vehicle Dynamics
Affecting Dynamic Pavement Loading'. The aims and modelling approach of the project are
described. Comparisons between model predictions and experimental data are made, showing
reasonable agreement. The effects of various parameter changes, such as centre of gravity location,
distribution of sprung mass, speed, roughness scale, as well as suspension stiffness and damping, are
investigated. The direction for future work in this project is outlined. Model predictions indicate that
whole vehicle configuration has an important effect on the level of pavement loading. It is therefore
concluded that pavement-protecting guidelines for vehicle suspension selection should have regard
to overall vehicle configuration. The form of calculated speed-load relationships is complicated and
variable between vehicles, which indicates that the pavement damaging effect of a particular vehicle
should be assessed over a speed range rather than at just one or two speeds. It is predicted that the
dynamic loading level of vehicles fitted with tandem suspensions which provide only poor damping
of pitch motions will be high over certain speed ranges, but that these high levels will be significantly
reduced when shock-absorbers are fitted between the chassis and axles.

Introduction and differences was unclear. Spectral analysis of the P264

Aims experimental results in the initial stages of P387 indicated
that a significant component of the measured load was
ARRB Project 387 is concerned with dynamic wheel load associated with vibrations of the whole vehicle. This
variation in heavy road vehicles. Dynamic wheel loads indicated that the large-scale dynamic features of the
have a number of important effects: they accelerate the vehicle, such as the distribution of its major masses, the
deterioration of roadway structures, they degrade the geometry of their interconnection and the distances
ability of the vehicle to turn and brake and they may cause between the axle group centres, could have a significant
component fatigue. The loads are also a major source of effect on the level of loading measured. P387 was initiated
excitation for truck vibration, so that an understanding of with the goal of determining the influence of these 'overall
the mechanism of wheel load generation is essential in vehicle configuration' effects.
many heavy vehicle dynamic studies. The following aims for P387 were defined.
The project arose from the experimental investigations (a) To model the heavy vehicle dynamic loading process
of ARRB project 264: 'Dynamic Truck Suspension with respect to the effects of whole vehicle configur-
Performance' (Sweatman 1983). Experiments were ation and suspension characteristics.
conducted on a variety of trucks, each fitted with a
different suspension system. The results demonstrated that (b) To study the interaction between these factors with a
large variations in dynamic loading occur between different view to determining whether pavement-protecting
vehicles under the same operating conditions and that the guidelines for vehicle suspension selection should
level of loading can be very high; in some cases peak have regard to overall vehicle configuration.
pavement forces exceeded twice the static value. This paper briefly describes some preliminary results
Variations in dynamic loading level between the from the project. In the next section two of the dynamic
vehicles were attributed principally to differences in models developed in the project are introduced, the
suspension design; however, the influence of other design general modelling philosophy is described and the
calculated responses are explained. In the third section,
ACKNOWLEDGEMENTS some preliminary model validation results based on the
The authors gratefully acknowledge the financial support provided by P264 experimental data and the predictions of one of the
the Australian Road Research Board and the valuable suggestions made introduced models are presented. The fourth section
by Dr Peter Sweatman, Dr Luan Mai and the reviewers of this paper. details some investigations into the effect of various

Australian Road Research, 15(4), December 1985 249

 HEAVY VEHICLES AND PAVEMENT LOADING

suspension and overall vehicle configuration design Larger excitation is provided to vibration in the
parameters. Future directions for work in the project are longitudinal plane of the vehicle (pitch and bounce) than
discussed in the fifth section. The final section is a to vibration out of the plane (roll) at usual operating
summary of the conclusions of the project to date. speeds, as may be deduced from the results presented by
Robson (1978). The relative strength of out-of-plane
vibration diminishes as the vehicle speed increases. For
this reason the planar models used here provide a
Modelling reasonable dynamic representation of the vehicle under
typical operating conditions. For symmetrical vehicles the
Dynamic Models in-plane and out-of-plane motions are dynamically
Lumped parameter, linear, dynamic models consisting of decoupled so that the effect of out-of-plane motions could
interconnected rigid masses, viscous dampers and massless be superimposed on the results from the present longi-
springs have been used in this project. The models do not tudinal plane models if such a refinement were considered
represent non-linear effects such as suspension stiction and necessary.
wheel lift, which may be important on smooth and rough
roads respectively. Linear models have the advantage that
they require less computation, so that the effect of model Modelling Philosophy and Calculated Responses
parameters can be investigated more thoroughly. It is Only the components of dynamic pavement load which
anticipated that these models will be able to predict the are induced by road roughness have been considered in
major components of dynamic load for most vehicles this project, because road roughness is the dominant
under typical operating conditions, although this hypothesis source of dynamic pavement loading excitation on all but
needs further testing. smooth roads (see the experimental results of Whittemore
et al. (1970)). The detailed time history of the dynamic
The energy dissipation properties of multi-leaf springs load generated by a vehicle travelling along a particular
have been represented by the imaginary parts of complex road is not generally required for road damage or vehicle
stiffness elements, such as are commonly used in structural vibration studies, and a more economical form of load
dynamics models. In contrast to viscous dampers, which description is desirable. A stochastic approach has been
are more commonly used, the energy per cycle of adopted for this project in which the road and load are
oscillation dissipated by these elements is independent of regarded as random processes which may be characterised
frequency — a characteristic which better approximates by certain statistics. The road surface is regarded as a
the behaviour of real leaf springs (see Fancher et al. member of an ensemble of a stationary, ergodic, Gaussian
(1980)). random function and the vehicle is considered to travel
along the road at constant speed.
Two of the developed models were used for the
calculations presented in this paper; they are shown When a stable dynamic model is excited under these
schematically in Fig. I. The models represent the pitch and conditions the resulting load function has corresponding
bounce motions of a rigid truck with a tandem drive stationarity and ergodicity properties. If, in addition, a
suspension and an articulated vehicle with tandem drive linear model is employed, the resulting load variable will
and tandem trailer suspensions. be normally distributed. Although real road surfaces can
never satisfy the conditions of stationarity and vehicles do
not travel at precisely constant speed, in most practical
applications the roughness characteristic of the road and
the speed of the vehicle vary with time constants that are
much longer than the longest time constants of interest in
the dynamic loading process, so that this description is
both adequate and useful.

The statistics used in this paper are all derived from

the normalised wheel load, L = L'/uL' , where L' is the
instantaneous load and ui is the mean wheel load. The
basic statistic used is sL, which is the standard deviation of
Fig. la — Rigid truck dynamic model
the variable L and may be compared with the P264
experimental quantity 'Dynamic Load Coefficient' (DLC).
The wheel loads calculated in this paper are all normally
distributed and for this case the standard deviation
provides a complete specification of the load probability
distribution.

The other statistic used in this paper is the power

spectral density function GLL(/) which may be regarded as
the breakdown of the variance of L (s2 L) into independent
frequency bands. The area under the curve is the variance
of the load. This statistic is particularly useful because it
Fig. I b — Articulated vehicle dynamic model may be related to the types of system vibration.

250 Australian Road Research, 15(4), December 1985

 HEAVY VEHICLES AND PAVEMENT LOADING

Truck vibration involving significant motion of the of the load measurement sensors leads to an attenuation of
whole vehicle occurs at frequencies of about 3 Hz. The the dynamic component of the load signal, as discussed by
area under the GLL curve in this frequency range may be Sweatman (1983). This does not cause any problems if the
regarded as the contribution of this form of vibration to the transducer data are used for model validation, because it is
load variance. The area in regions about 10 Hz is a comparatively simple matter to use the particular model
associated with localised suspension resonances such as to predict the load that was actually measured rather than
wheel hop, tramp and suspension pitch for tandem the load on the pavement. However, difficulties may arise
suspensions. The area at frequencies intermediate between if the transducer data are used for the ranking of the
these two is generally attributable to vibrations involving dynamic loading level of heavy vehicles in practical road
flexing of the vehicle chassis (not represented in the present tests. Modelling experience indicates that the measured
models). (See Elmadany, Dokainsh and Allan (1979) for a DLC value may underestimate the true DLC value by up
disucssion of these frequency ranges.) Note that periodic to 25 per cent (see the discussion under the sub-heading
components on the spectral density plots may be `Validation Studies'), and that the degree of under-
represented as delta functions and in experimental estimates estimation will depend on the suspension characteristics.
of spectral functions appear as sharp spikes. This problem can be overcome by the re-introduction of
the accelerometer signals provided in the original GM
design, as was noted by Sweatman (1983).

Preliminary Model Validation

P264 Data
ARRB Load Transducer System
Some sections of the P264 data were analysed to provide
This section briefly describes two sources of error in the experimental spectral density functions, for direct com-
GM-designed load transducer system used in P264, which parison with model predictions. The sections analysed
are important in the interpretation of the experimental were All nominally 60 km/h tests on non-articulated,
load spectra. A fuller discussion of error sources is given in three-axle vehicles over a road length with roughness
Heath and Good (1985). 142 c/km on the NAASRA scale. The vehicle models,
The first error source is the high-pass filtering drive suspension types and axle group loads are given in
(fc. = 0.16 Hz) of the transducer outputs which leads to Table I and schematic diagrams of the drive suspensions
significant distortion in a frequency band of width 0.32 Hz are given in Fig. 2. A test speed of 60 km/h was chosen for
centred on the wheel rotation frequency. To understand analysis because it involves a wheel rotation frequency of
the nature of this distortion, consider the transducer load about 5.5 Hz, which was unlikely to correspond with a big
experienced as a result of an out-of-balance force rotating peak in the load spectral density function, thereby
with the wheel. Such a force would be constant in the minimising one of the sources of error in the load
rotating reference frame of the transducer, and would thus transducer system discussed in the last section.
be completely removed from the transducer system output
The experimental spectral density functions were
by high-pass filtering. Similarly, a vertical force varying
computed using a direct FFT calculation procedure on
sinusoidally at the wheel rotation frequency can be
thought of as being generated by two contra-rotating 512-point time records sampled at 102.4 Hz, with 50 per
forces; the component rotating with the wheel would cent overlap averaging and a linear Hanning window.
again be filtered out, leaving a transducer output of only Nineteen averages, from data spanning a time interval of
half the true force amplitude. 50 s, were used in the preparation of the plots, which are
shown in Fig. 3. Note that the spectral function is
The existing ARRB load transducer system measures calculated for the load at the leading axle of the tandem
the load between the load transducer and rim, rather than drive suspension normalised by its static weight assuming
the true wheel load. The mass of the tyre and rim outboard perfect load sharing of the group. This variable is similar to

Table I
Test Vehicles
Vehicle Make and Drive Susp. Tyre Axle Group
Number Model Type Press. Load (t)
(kPa)
Ford Hendrickson 660 15.01
LNT 9000 RT340
2 Kenworth Hendrickson 750 15.09
K123 RT380
3 Kenworth Kenworth 720 14.42
W925S2 TBB
4 Kenworth Kenworth 750 15.27
K124 Big Six
5 International Ridewell 750 15.09
Harvester Dynalastic
3070A

Australian Road Research, 15(4), December 1985

251
 HEAVY VEHICLES AND PAVEMENT LOADING

0.020

0.01

0.012
Norm.
Centre pivot
Load
Spectrum
(a) Vehicles 1 and 2 0.008

0.004

0.00 tiJ
0 6 12 18 24 30
Freq. (Hz)

Fig. 3a — Experimental load spectrum for vehicle I

444
Torsion bar
0.05
II4"
\slc

0.04

(b) Vehicle 3 0.03

Norm.
Single upper radius rods Load
Spectrum
0.02

0.01

Dual lower radius rods

0.00 0
6 J j 12 18 24 30
(c) Vehicle 4 Freq. (Hz)

Rubber spring
Fig. 3b — Experimental load spectrum for vehicle 2
Centre pivot

0.010

0.008

Articulation pivots 0.006

Norm.
(d) Vehicle 5 Load
Spectrum
Fig. 2 — Schematic diagrams of the drive suspensions 0.004
(from Australian Department of Transport 1978)

0.002

the load variable used in P264, although that load measure

was normalised with the experimental mean load value 0.00
rather than the nominal one. 0 12 ~ 18
Freq. (Hz)
24 30

Experimental spectral estimates are subject to random Fig. 3c — Experimental load spectrum for vehicle 3
and bias errors (Bendat and Piersol 1971). Random errors
decrease as the the record length and the effective
resolution bandwidth of the calculation increase. Bias Some important features of the plots are as follows.
errors are greatest at sharply peaked regions of the (1) For all vehicles, the dominant contribution to the
spectrum — they act to reduce the dynamic range of the measured dynamic load variance comes from low
estimate and they diminish as the resolution bandwidth is frequency regions of the spectrum near 3 Hz. These
decreased. The effective resolution bandwidth used here regions are associated with vibration involving
was approximately 0.2 Hz, which provides acceptable significant motion of the whole vehicle and are
bias errors for the major features of the plots and yields a therefore affected by changes in overall vehicle
normalised random rms error for each spectral estimate of configuration, as well as the overall elastic and
approximately 0.3. dissipative force-deflection characteristic of the axle

252 Australian Road Research, 15(4), December 1985

 HEAVY VEHICLES AND PAVEMENT LOADING

0.020
(4) Some contribution to the load variance from extremely
low frequency components is evident in all plots.
These low frequency components are presumably the
0.016
result of load transfer due to road camber and slope as
well as drive torque reaction as the engine driving
0.012 moment is altered during the course of the run. The
Norm.
Load largest low frequency components occur in the case of
Spectrum
vehicle 3.
0.008

Validation Studies
0.004

The rigid truck model previously described was used to

predict the 'as measured' and actual load spectral density
0.00
0 6 12 18 24 30 functions for direct comparison with the experimental
Freq. (Hz) plots. Two major difficulties were encountered in the
Fig. 3d — Experimental load spectrum for vehicle 4
course of this work.

The first problem was that a complete set of

0.05
measured model parameters for each test vehicle was not
available. A number of parameters had to be estimated to
0.04
overcome this difficulty. These were the pitch moment of
inertia for the sprung mass and the various damping
coefficients in the model. The pitch moments of inertia
0.03 were estimated from photographs of the test vehicles.
Norm.
Load Damping coefficients were varied until a reasonable fit
Spectrum
with the experimental results was obtained.
0.02

The second difficulty was that no profile height

0.01
spectral density function was available for the road section
under test. It is essential that an estimate of the road
spectral characteristic be made before model predictions
0.00 can be calculated. A band-limited white noise slope model
0 6 12 18 24 30
(Freq. (Hz) was employed for the description of the road surface,
which is an adaptation of the frequently used white noise
Fig. 3e — Experimental load spectrum for vehicle 5 slope model (Karnopp 1978). The overall roughness level
was adjusted until reasonable agreement was obtained
with experimental results. It is anticipated that this model
group suspension. Suspension design features which is reasonably close to the actual road characteristic.
do not affect this gross force-deflection behaviour,
such as the unsprung mass and the details of the load The theoretical 'as measured' and actual load spectra
sharing mechanism, should have a relatively small for each test vehicle are shown in Fig. 4. There is a
influence on these important, low frequency reasonable agreement between these theoretical predictions
contributions. and the experimental measurements in Fig. 3. Curve
shapes and peak frequencies generally correspond fairly
(2) A significant contribution to the load standard closely, although the experimental data contain periodic
deviation is found at the suspension resonance and low frequency components which are not reproduced
frequencies in the case of vehicles 1, 2 and 5. These by the model. The only significant difference to emerge is
parts of the spectrum are thought to be associated for vehicle 1: a small peak at a frequency of about 3.9 Hz
with localised pitching of the tandem suspension appears in the theoretical prediction which is not evident
system without significant motion of the whole in the experimental curve. This peak is associated with
vehicle and should therefore be relatively insensitive whole vehicle pitching motions and its location and size is
to the overall vehicle configurational effects, but will sensitive to changes in the pitch moment of inertia, centre
be influenced by the details of suspensiot design. of gravity location and speed. It is believed that errors in
the estimates of these three parameters are responsible for
(3) Periodic components associated with excitation related this small discrepancy between model predictions and the
to wheel rotation are clearly present in Figs 2c and 2e,
experimental results.
and appear at about 5.5 Hz and its higher harmonics.
The largest periodic component is indicated for Fig. 4 also demonstrates the effect of the outboard
vehicle 5 at the fundamental wheel rotation frequency. mass in the ARRB load transducer. It shows that the
However, it should be recognised that due to the measured load signal is considerably attenuated in the high
distortions of the transducer system in the frequency frequency regions near the suspension resonance peak.
band centred on the wheel rotation frequency, it is The predicted 'actual' and 'measured' normalised load
impossible to state what the true periodic component standard deviation values are compared in Table IL It is
is. seen that the underestimation of the DLC is quite

Australian Road Research, 15(4), December 1985

253
 HEAVY VEHICLES AND PAVEMENT LOADING

0.020
0.020

as measured as measured
actual
0.016 actual
0.016

0.012
0.012
Norm. Norm.
Load Load
Spectrum Spectrum

0.008
0.008

0.004
0.004

0.00
0.00
0 5 10 15 20 25 30 0 10 15 20 25 30
Freq. (Hz) Freq. (Hz)

Fig. 4a — Predicted load spectrum for vehicle 1 Fig. 4d — Predicted load spectrum for vehicle 4

0.05 0.05

as measured as measured
actual actual
0.04 0.04

0.03 0.03
Norm. Norm.
Load Load
Spectrum Spectrum

0.02 0.02

0.01 0.01

0.00 0.00
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Freq. (Hz) Freq. (Hz)

Fig. 4b — Predicted load spectrum for vehicle 2 Fig. 4e — Predicted load spectrum for vehicle 5

0.010
Table II
as measured

0.008
actual Predicted Effect of Accelerometer Correction on
Normalised Load Standard Deviation (SD) at the
Leading Drive Axle
0.006
Norm. Vehicle Actual Load As Measured' % Under-
Load No. SD Load SD estimation
Spectrum

0.004 1 0.1612 0.1267 21

2 0.1910 0.1585 17
3 0.1096 0.0878 20
0.002
4 0.1395 0.1208 13
5 0.1918 0.1520 21
0.00
0 5 10 15 20 25 30
Freq. (Hz)

capacity to describe the general features of the experimental

Fig. 4c — Predicted load spectrum for vehicle 3 plots, which gives some confidence in the modelling
assumptions that have been made.
substantial and that the degree of underestimation is larger
for suspensions with low unsprung mass, high wheel hop
frequencies, and poorly damped suspension resonances.
Preliminary Parametric Study
It should be emphasised that the model validation
results presented here are only of a preliminary nature, This section discusses some preliminary investigations into
because of the rough parameter estimates that were made the effect of speed, roughness, suspension characteristics
and the approximate road model that was used. Never- and overall vehicle configuration on the dynamic loading
theless, they demonstrate that the rigid truck model has the level. A perturbation approach was adopted, in which a set

254 Australian Road Research, 15(4), December 1985

 HEAVY VEHICLES AND PAVEMENT LOADING

of basic model parameters was defined and the effect of (c) vehicle 3, with its P264 payload substituted by a
various changes was investigated. This approach does not standard trailer giving the same static load on the
yield a complete understanding of the effect of the various drive axle group.
parameters. However, it is a useful technique for identifying
their likely influence. A more global parameter search Note that there is a general upward trend in the load
technique will be used for later work in this project when a variance with speed for each vehicle, which is related to
new, more efficient variance calculation procedure has the following facts.
been implemented.
(1) The amplitude of the road surface roughness
Basic model parameters for the rigid truck model components rises rapidly with increasing wavelength.
were chosen to represent the P264 test vehicles 1 to 5. (2) The dynamic loading level of a heavy vehicle is
Basic model parameters for the articulated vehicle model sensitive to the amplitudes of road roughness
were chosen for a hypothetical case, where the payloads of
components only within a certain range of wave-
the P264 test vehicles 2 to 5 were replaced by a trailer of
lengths. As the speed rises the vehicle's sensitive range
standard typical geometry, yielding an identical static load
extends to longer wavelengths with greater roughness
at the drive axle group. Vehicle 1 was not investigated with
amplitudes, so that the dynamic loading level tends to
a trailer because it was a rigid truck with a long wheelbase
increase.
and it was never intended that it be used as a prime-mover
for an articulated vehicle. The road model used in the
Superimposed on this upward trend is a series of
model validation section was used for the calculations
oscillations which may be of considerable amplitude
made here, using the same roughness level as a basic
compared to the mean characteristic. The size of the
parameter.
oscillations and their position varies from vehicle to
vehicle. The oscillations increase in frequency as the speed
is decreased. This peculiar behaviour is expected from the
Speed and Roughness Effects theory of stationary processes; the oscillations are similar
in form to a damped sin(1/x) function. It should also be
The models predict that speed and roughness have a major noted that predictions of speed-DLC relationships reported
influence on the level of dynamic loading. Plots of the by Scarpatetti (1975) (using models similar to the ones
standard deviation of the load on the leading axle of the used here) show similar oscillating characteristics.
drive suspension vs speed are shown in Fig. 5 for three
cases: The oscillations in the curve are the result of time
delays inherent in the system (the roadway inputs to the
(a) vehicle 1 as tested, trailing axles are time delayed repetitions of those to the
leading axle). A simple insight into the way in which they
(b) vehicle 3 as tested, and arise is given by the example illustrated in Fig. 6.

0.30
Regression line for Vehicle 1
• Vehicle 1, speeds 41, 62 and 75 km/h
using speeds of 41, 62 and 75 km/h
▪ Vehicle 1, speeds 60, 80 and 100 km/h
Regression line for Vehicle 1
• Vehicle 3, speeds 43, 63 and 78 km/h using speeds of 60, 80, 100 km/h

Vehicle 1 as tested

0.20
Regression line for Vehicle 3 ---
Normalised using speeds of 43, 63, 78 km/h
Load
Standard
Deviation
Vehicle 3 as tested

Vehicle 3 with trailer

0.10

0 20 40 60 80 100 120 140

Speed km/h

Fig. 5 — Influence of speed on normalised load standard deviation

(load at leading drive axle)

Australian Road Research, 15(4), December 1985

255
 HEAVY VEHICLES AND PAVEMENT LOADING

Fig. 6a — The suspension pitch mode

Peak Exciting Components Zero Exciting Components

Fig. 6b — Different pitch mode excitation levels induced by various wavelength roughness components

The first part of Fig. 6 represents a mode of vibration rising and falling of the DLC value continues as the speed
of a walking beam tandem suspension, 'the suspension is dropped further, with the pitch resonance frequency
pitch mode', where the suspension axles move up and corresponding successively to the frequencies for the
down, out of phase, bouncing against the tyre springs at a second, third, and so on 'peak exciting' and 'zero exciting'
fixed natural frequency. When the wheels are forced in components. This is the basis for the oscillations in the
opposite directions with sinusoidal motions at the pitch speed-DLC relation and explains why they increase in
natural frequency, a large dynamic load results (resonance). frequency with decreasing velocity.
Excitation at higher or lower frequencies produces much
lower forces. The response of a vehicle is composed of the response
of many modes (many axle inputs may also be involved),
The second part of Fig. 6 illustrates how different which further complicates the speed-DLC characteristic.
wavelength components excite suspension the pitch mode Generally, the oscillations are not attenuated until the
to varying degrees. Roughness components with wave- factor V/ Wis considerably greater than the highest natural
lengths equal to (0.5 n)S, n= 0,1,2 . . ., (where S is the frequency associated with substantial contributions to the
axle spacing) lead to the largest excitation of pitch. dynamic loading level, where W is the largest distance
Components with wavelengths equal to an integral between axle centres. The velocity at which these
multiples of S produce no suspension pitch motion at all. oscillations die out is far greater than the highest
conceivable vehicle maximum speed, even for short
A roughness component of wavelength L is perceived wheelbase trucks. Theoretical arguments show that the
on the vehicle as an oscillation of frequency V/ L, where V characteristic is influenced by the distances between the
is the vehicle speed. At very high speeds, the frequency for wheel centres, the system natural frequencies and the
even the longest wavelength 'peak exciting' component is mode shapes. However, the relations involving these
much higher than the pitch natural frequency, so that there various factors are very complicated and their effects are
is no significant resonant response from the pitch mode. As not as yet well understood. It is possible to select model
the speed is dropped, the frequency for the first 'peak parameters for vehicles with only single axle groups in
exciting' component (L = S / 2) will become equal to the such a way that the oscillations are eliminated. However,
pitch mode frequency, exciting the resonance substantially the values that must be used are not representative of
and leading to a relatively high DLC value. When the current design practices.
speed is dropped further the frequency for the first 'zero
exciting' component (L = S) coincides with the pitch The fact that the predicted load variance v. speed
natural frequency — hence there is zero pitch excitation at relation is so complicated and variable indicates that it will
the natural frequency and a relatively low DLC value. The be difficult to rank the loading level of a group of different

256 Australian Road Research, 15(4), December 1985

 HEAVY VEHICLES AND PAVEMENT LOADING

heavy vehicles by a single index. Curves may cross and The further assumption of proportionality with speed
recross so that a vehicle which generates a lower loading underlies Sweatman's (1983) recommendation No. 6:
level than another at one particular speed may not
necessarily generate a lower loading level at a different A quantitative dynamic road loading requirement, which could
be based on limiting the dynamic load coefficient to 0.2 or less
speed. Thus it seems that the pavement damaging effect of
under roughness and speed coefficients given by V.R1/2 = 850,
a given vehicle should be assessed over a range of speeds,
should be considered as a Design Rule.
rather than at just one or two speeds. Note also that the
P264 experiments involved only three test speeds, which is
Using V.R1/2 alone to define the test condition 'severity'
not a sufficient number to detect the presence of the implies that regressions like those in Fig. 5 should pass
predicted oscillations. Theoretical arguments therefore through the origin. The present analysis indicates that this
suggest that experiments designed to determine the recommendation does not have a sound theoretical basis
speed-DLC relationship should use a greater number of with respect to the specification of test conditions.
test speeds at smaller intervals.
The analysis of the P264 experimental data described Assuming that the shape of the spectral characteristic
by Sweatman (1983) produced regression relationships is the same for all roads, a simple relationship between
between DLC, speed and roughness for the various test predicted loading level and road roughness exists for linear
vehicles of identical form. These are reproduced below for vehicle models, in contrast to its complicated relationship
the test vehicles 1 and 3: with speed: the load variance is directly proportional to the
slope variance of the pavement. If in addition it is assumed
Vehicle 1: DLC = — 0.0272 + 2.93 x 10-4 V.R1i2 (1) that the NAASRA meter has a linear response (i.e. the
roughness count increases in proportion to the geometric
Vehicle 3: DLC = — 0.0144 + 1.67 x 10-4 V.R1/2 (2) scale of the roughness), then the dynamic models predict
that DLC values will be directly proportional to the
Here V is the vehicle speed and R is the NAASRA NAASRA statistic R. However, regressions constructed in
roughness. The regressions were constructed for each P264 demonstrated highest correlation between DLC and
vehicle using only three test speeds. the factor V.R1 /2, which disagrees with the result above.
Further work needs to be done to determine to what extent
For interest, similar regression relationships were this disagreement is attributable to NAASRA meter and
constructed using the same test speeds on the theoretical heavy vehicle non-linearity, and the failings of the
plots. The points and regression lines are shown in Fig. 5, hypothesised spectral characteristics of the road.
and the resulting relationships for the normalised load
standard deviation s2L are as follows:
Effects of Suspension Characteristics
Vehicle 1: s2L = — 0.080 + 3.50 x 10-4 vR1/2 (3)
This section details some investigations into the effect of
Vehicle 3: s2L = + 0.018 + 1.16 x 10-4 vR1/2 (4) various suspension characteristics. Most calculations in
this section were performed for vehicle 1 in the as-tested
(Note that the measured value of R = 142 c/km was condition, running at the test speed of 62 km/h, over a
substituted directly in determining the above coefficients.) road with an identical roughness characteristic to that used
The correspondence between the two regression relation- in the model validation work. The response variable
ships is good considering the possible errors in the load presented here represents the load at the leading axle of the
measurement system, the differences between the DLC drive suspension group.
and load standard deviation variables, and the fact that the
model did not consider periodic and low frequency This vehicle was fitted with a rigid walking beam
components of the load. This gives further reason for drive suspension, which is free to pitch about its central
confidence in the developed models. pivot with very little damping. The experimental and
theoretical results for this truck presented previously
It should also be noted that the predicted oscillating showed a substantial contribution to the dynamic loading
behaviour of the loading level/speed characteristic suggests level associated with suspension pitching, which was in
that linear regressions are unlikely to provide reasonably large part responsible for the relatively high loading level
accurate representations of the load standard deviation of the vehicle at this speed. Experimental and theoretical
over anything but very small velocity ranges. This work reported by Sayers and Gillespie (1983) has also
indicates that the coefficients of regressions calculated identified the lightly damped pitching mode as being
from experimental DLC measurements will be quite responsible for high dynamic loading levels associated
sensitive to the selection of test speeds. To demonstrate this with this type of suspension.
effect a new regression line has been constructed for
vehicle 1 in Fig. 5, using calculated data for speeds of 60, Shock absorbers fitted between the vehicle chassis
80 and 100 km/h rather than the P264 test speeds of 41, and each of the drive axles are an effective way to improve
62 and 75 km/h. The two lines predict substantially the damping of this mode of vibration. Calculations were
different dynamic loading levels over most of the speed made to determine the effect on the overall loading level of
range. It is therefore concluded that linear velocity fitting two different sets of dampers: strong ones of equal
regressions should not be used to characterise the loading capacity to those fitted on vehicle 3, and weak ones of half
level of a particular vehicle. this strength. The results are presented in Table III. They

Australian Road Research, 15(4), December 1985 257

 HEAVY VEHICLES AND PAVEMENT LOADING

Table III
Effect of Auxiliary Dampers Fitted to Vehicles 1, 2 and 5
Damping Condition Normalised Leading Drive Axle
Load Standard Deviation
Veh. 1 Veh. 2 Veh. 5
(1) No dampers 0.1612 0.1910 0.1918
(2) Weak auxiliary dampers 0.1273 0.1549 0.1355
(3) Strong auxiliary dampers 0.1164 0.1401 0.1115

illustrate that the dynamic loading level of this vehicle is Vehicle 4 is a special case since its drive suspension
effectively reduced by this provision, which is in theory was not fitted with dampers, yet its experimental spectral
simple and relatively cheap. Reductions of the order of 30 plot (Fig. 3d) indicated no significant suspension pitch
per cent are achieved when the stronger dampers are fitted. response. For this drive suspension design, suspension
This makes the dynamic loading level of vehicle 1 pitch causes sliding between the spring pads and the leaf
comparable with the best of the simulated vehicles, vehicle spring (refer Fig. 2). It is believed that friction associated
3 (refer Table II). Included in Table III are corresponding with this sliding is responsible for the effective damping of
figures for vehicles 2 and 5 (these vehicles also have lightly the suspension pitch mode.
damped suspension pitch resonances), which demonstrate
Table IV presents some results for a number of
large reductions in load variance when dampers are fitted.
changes in drive suspension characteristics of vehicle 1. It
shows that substantial increases in suspension stiffness (of
Fig. 7 shows the effect of the dampers on the load the order of 30 per cent) lead to only fairly small increases
spectrum for vehicle 1. The greatest reduction is apparent in loading level (less than 5 per cent) at the test speed.
in the suspension pitch mode region. A useful reduction is While these increases are fairly small it should be
also apparent in the region of the first peak, as a result of emphasised that they are accompanied by considerable
better damping of the low frequency vibration modes. changes in the shape of the spectral density function,
indicating that the types of dominant vibration had
0.020
changed.
No auxiliary dampers

—————— Weak auxiliary dampers Small increases are also observed for a 10 per cent
0.016
Strong auxiliary dampers increase in tyre stiffness. The decrease in dynamic loading
level with tyre pressure reported in the P264 test results
0.012
was not reproduced in these calculations. However, when
Norm. the theoretical prediction of the measured load rather than
Load
Spectrum
the actual load was used, a much smaller increase in load
0.008 (less than 1 per cent) resulted. For revised estimates of the
model parameters, with a road model more accurately
representing the test road surface, it is conceivable that this
0.004 downward trend in the 'as measured' DLC could be
reproduced.
0.00 The Table shows that increased energy dissipation in
0 5 10 15 20 25 30
the leaf spring can lead to reductions in the load variance.
Freq. (Hz) The reductions are all confined to the low frequency
Fig. 7 — Predicted,effect of auxiliary dampers on the load spectrum of vehicle I regions of the spectrum.

Table IV
Effect of Various Drive Suspension Changes in Vehicle 1

Modification Normalised Leading % Change

Drive Axle Load From Standard
Standard Deviation

(1) Standard Vehicle 0.1612

(2) Drive Suspension Stiffness
(a) increased by 30% 0.1662 + 3.1
(b) decreased by 30% 0.1530 — 5.1
(3) Leaf Spring Damping Level
(a) increased by 30% 0.1560 — 3.2
(b) decreased by 30% 0.1691 + 4.9
(4) Tyre stiffness on drive axles
(a) increased by 10% 0.1635 + 1.4
(b) decreased by 10% 0.1573 — 2.4

Australian Road Research, 15(4), December 1985

258
 HEAVY VEHICLES AND PAVEMENT LOADING

In summary, the vehicle under investigation was the reduced static load. When the static load was
most sensitive to changes in system damping. Stiffness maintained constant, the dynamic load decreased as the
changes, while influencing the shape of predicted response CG was moved forward (and the sprung mass was
curves, did not lead to substantial changes in load variance. correspondingly increased). The wheelbase had a smaller
but still significant effect: increases in wheelbase led to
reductions in the normalised load variance.
Overall Vehicle Configuration Effects
A second investigation was made to determine the
In this section the influence of various 'overall' vehicle effect of replacing the P264 test load by a standard trailer
configuration effects are investigated. The first investigation yielding the same static load at the drive axle. The results,
was based on vehicle 1 in its 'as-tested' condition and given in Table VI, show that there are differences between
consisted of varying its pitch moment of inertia, the loading of the leading axle of the drive group, when a
location of its sprung mass centre of gravity (CG) and its semi-trailer replaces the P264 test payload, which are
wheel base. Variations in CG location were accomplished significant compared to the differences between the
in two ways: firstly with a constant sprung mass, so that the various suspension loading levels. (The maximum dif-
wheel static load changed; secondly by changing the ference between 'as tested' and 'with trailer' predicted load
vehicle payload in such a way that the drive axle static load standard deviation values for the same suspension is
remained constant. Wheelbase changes were made in a 0.0498, compared with the maximum difference between
fashion which kept the static axle loads constant. The suspensions in the 'as tested' condition of 0.0814.)
results are presented in Table V for the leading drive axle
load. They demonstrate that the dynamic loading In three of the four cases the semi-trailer caused a
coefficient for this vehicle at this speed is quite strongly substantial drop in the dynamic loading level. However, in
influenced by the CG location, but is only slightly affected one case there was a small rise. The rank order of the
by the pitch moment. For the rase of constant sprung mass, suspensions was not maintained, and the differences
the variance of the absolute dynamic load actually between the load values changed markedly, demonstrating
decreased as the CG was moved forward, but the a significant interaction effect. To illustrate the fundamental
normalised load standard deviation increased because of dynamic differences between the two rases, plots of the

Table V
Effect of Changes in Overall Vehicle Configuration for Vehicle 1
Modification Normalised Leading % Change
Drive Axle Load From Standard
Standard Deviation
(1) Standard Vehicle 0.1612
(2) Sprung Mass CG Position
Keeping Payload Constant
(a) 0.75 m forward 0.1676 + 4.0
(b) 0.75 m rearward 0.1561 - 3.2
(3) Sprung Mass CG Position With
Payload Changes to Keep Static
Drive Axle Load Constant
(a) 0.75 m forward 0.1490 - 7.6
(b) 0.75 m rearward 0.1706 + 5.8
(4) Pitch Moment of Inertia
(a) increased by 30% 0.1616 + 0.2
(b) decreased by 30% 0.1609 - 0.2
(5) Wheelbase
(a) 1 m longer 0.1545 - 4.2
(b) 1 m shorter 0.1648 + 2.2

Table VI
Effect of Replacing P264 Test Payload with 'Standard Semi-Trailer Giving an
Equivalent Drive Axle Static Load
Vehicle Normalised Leading Drive Axle % Change
Number Load Standard Deviation With Trailer
As Tested With Trailer
2 0.1910 0.1412 - 25.7
3 0.1096 0.1133 + 3.4
4 0.1395 0.1047 - 24.9
5 0.1918 0.1583 - 17.5

Australian Road Research, 15(4), December 1985

259
 HEAVY VEHICLES AND PAVEMENT LOADING

0.020
vehicle and drive suspension. Note that there is a direct
With P264 payload
correspondence between P264 test vehicle numbering and
With added semitrailer
0.016 the suspension and vehicle numbers used in the diagrams.
(For example, the combination suspension no. = 1,
vehicle no. = 4 represents the P264 test vehicle 4 with its
0.012
Norm. drive suspension replaced by the drive suspension of P264
Load
Spectrum vehicle 1.) Note also that the basic data for the two
diagrams are the same; they are just plotted in different
0.008
ways to separately reveal the influences of suspension and
overall vehicle configuration. Some features to note in the
0.004 diagrams are as follows.
(a) Suspension has a large effect on predicted dynamic
0.00
loading levels. (Variations in load standard deviation
0 5 10 15 20 25 30 of up to 50 per cent occur for a single vehicle fitted
Freq. (Hz)
with different suspensions.)
Fig. 8 - Predicted effect of replacing the P264 payload of vehicle 4 with a
semi-trailer giving an equivalent drive axle static load
(b) Overall vehicle configuration has a smaller but still
significant effect on predicted dynamic loading.
(Load standard deviation varies by up to 25 per cent
spectra for vehicle 4 in its articulated and non-articulated
configurations are given in Fig. 8. for a single drive suspension fitted to the various test
vehicles.)
The third investigation into the effect of overall
configuration effects was made using vehicle 3 in its (c) It is not possible to establish a consistent ranking for
hypothetical articulated form. The effects of varying the predicted dynamic load among the suspensions using
horizontal locations of the tractor CG and the fifth wheel an arbitrary selection of test vehicles (i.e. the ranking
(with constant masses), as well as the trailer CG height and established among suspensions using one set of
wheelbase were investigated. The results, shown in combinations of test vehicles and suspensions may be
Table VII, demonstrate that these parameters have their different to the ranking established when another set
largest effect on the load at the front axle. Smaller but, in a of combinations is employed). However, there is a
number of cases, still substantial variations occur for the separation of the characteristics of suspensions 1, 2
tractor drive and trailer axle loading level. and 5 from suspensions 3 and 4: suspensions in the
first group are predicted to always have higher
The next investigation roughly assesses the influence loading levels than suspensions in the second group,
of overall vehicle configuration on the results obtained in
regardless of the test vehicles chosen for comparison
the P264 runs at 60 km/h. Calculations were performed
purposes. It should be noted though that this separation
to consider the levels of dynamic loading that would have
breaks down when the figures for the hypothetical
been obtained if the drive suspensions of the P264 tests had
articulated vehicles presented above are included.
been fitted to each of the other .test vehicles (in the P264
tests they were fitted to only one vehicle). Results from this (d) Both Figures illustrate that suspension and overall
exercise are presented in Figs 9a and b, where the plotted vehicle configuration effects interact substantially in
points connected by solid lines represent the predicted their influence on predicted dynamic loading. This
dynamic loading levels for the various combinations of can be seen from the fact that the distances between

Table VII
Effect of Changes in Overall Vehicle Configuration for Vehicle 3 in Articulated Form
Normalised Load Standard Deviation
Front Leading Leading
Modification Axle Drive Axle Trailer Axle
(I) Standard Vehicle 0.0994 0.1133 0.1079
(2) Fifth Wheel Location
(a) 0.2 m forward 0.0929 0.1103 0.1051
(b) 0.2 m rearward 0.1060 0.1167 0.1123
(2) Tractor Sprung Mass
CG Position
(a) 1 m forward 0.0568 0.1204 0.1091
(b) 1 m rearward 0.1295 0.1183 0.1223
(3) Trailer CG height
(a) 0.5 m higher 0.1004 0.1131 0.1081
(b) 0.5 m lower 0.0986 0.1136 0.1079
(4) Trailer wheelbase
(a) 1 m longer 0.0937 0.1103 0.1055
(b) 1 m shorter 0.1007 0.1158 0.1077

260 Australian Road Research, 15(4), December 1985

 HEAVY VEHICLES AND PAVEMENT LOADING

0.25 0.25

0.20 0.20

0.15 0.15

Normalised Normalised
Load Load
Standard Standard
Deviation Deviation
0.10 0.10

+ Suspension 1 ▪ Vehicle 1
x Suspension 2 x Vehicle 2
L Suspension 3 6 Vehicle 3
0.05 O Suspension 4 0.05 o Vehicle 4
o Suspension 5 O Vehicle 5
(Solid linet connect combinations with (Solid lines connect combinations
the same drive suspension) using the same vehicle)

1
0 1 2 3 4 5 0 1 2 3 4 5
Vehicle Number Suspension Number
(a) Vehicle effects (b) Suspension effects

Fig. 9 — Influence of overall vehicle configuration

the lines in the diagrams change with vehicle number (a) When lightly-damped suspension resonances are
(Fig. 9a) or suspension number ((Fig. 9b). The eliminated by fitting comparatively weak shock
interactions are large enough to cause cross-overs in absorbers where required, the influence of suspension
the lines; the three cross-overs occurring in Fig. 9a design on predicted dynamic loading levels drops and
indicate that a ranking established among the becomes approximately equal to the influence of
suspensions by road testing each one fitted to the same overall vehicle configuration. (Both effects are
vehicle will depend on the characteristics of the associated with variation in predicted dynamic loading
vehicle employed (i.e. the ranking may change if a of up to 25 per cent.)
different vehicle is chosen).
(b) Interactions between suspension design and overall
A further feature of the plots is that the predicted vehicle configuration effects on predicted levels of
dynamic loading level for vehicle 3 fitted with any dynamic loading become relatively more important,
one of the suspensions is consistently lower than the leading to a greater number of cross-overs in Figs 10a
dynamic loading levels of the other vehicles fitted and b than in the corresponding Figs 9a and b.
with the same suspension. It is not clear, at this stage,
what characteristics of vehicle 3 contribute to its (c) Vehicle 3, as for the previous case, has a predicted
relatively low loading levels at this speed. dynamic loading performance which is consistently
superior to the other vehicles.
The three suspensions 1, 2 and 5, which had
relatively high predicted dynamic loading levels when To summarise the discussion of this section, overall
fitted to each of the P264 test vehicles, were also those vehicle configuration is predicted to have an infuence on
which had lightly damped suspension resonances (see the dynamic pavement loading which is significant compared
discussions under the sub-headings `13264 Data' and to the effect of suspension design and interacts substantially
`Effects of Suspension Characteristics'). To assess the with it. It is therefore concluded that pavement-protecting
relative influence of suspension and overall vehicle con- guidelines for vehicle suspension selection should have
figuration in the absence of suspension resonance effects, regard to overall vehicle configuration.
new results were calculated for the dynamic loading levels
of suspensions 1, 2 and 5, fitted with the previously
introduced 'weak auxiliary dampers', in combination with
the various test vehicles. (Note, as discussed before, that Future Work
these dampers are an effective way to improve the
damping of suspension resonances; they are also associated The work reported here has given some confidence in the
with some improvements in the predicted low frequency modelling approach adopted in this project. However, the
dynamic loading performance.) The results, shown in Figs accuracy of the models has not been rigorously tested and
10a and b together with the old figures for suspensions 3 it is desirable that further validation work be carried out, in
and 4, illustrate the following. conjunction with model development if required. It will be

Australian Road Research, 15(4), December 1985 261

 HEAVY VEHICLES AND PAVEMENT LOADING

o 0.25

Normalised Normalised
Load Load
Standard Standard
Deviation Deviation

■ Suspension 1 with weak added dampers • Vehicle 1

X Suspension 2 with weak added dampers X Vehicle 2
A Suspension 3 A Vehicle 3
0 Suspension 4 0 Vehicle 4
O Suspension 5 with weak added dampers O Vehicle 5

(Solid lines connect combinations with (Solid lines connect combinations with
the same drive suspension) the same vehicle)

0 0
1 2 3 4 5 1 2 3 4 5
Vehicle Number Suspension Number
(a) Vehicle effects (b) Suspension effects

Fig. 10 — Influence of overall vehicle configuration when lightly damped suspension resonances are removed

necessary to obtain a set of experimental data for which to be carried out and this requires experimental data
model parameters have been accurately measured if this is for which parameters have been accurately measured.
to be achieved. (2) Direct calculation with the dynamic models of this
project has indicated that suspension characteristics
The results in this paper have demonstrated that
and overall vehicle configuration both play an
dynamic loading in heavy vehicles is a complicated important role in determining the level of dynamic
phenomenon and that the level of loading is affected by a loading. It is therefore concluded that pavement-
large number of variables, with significant interacting protecting guidelines for vehicle suspension selection
effects. Suspension characteristics and overall vehicle should have regard to overall vehicle configuration.
configuration are both important, and need to be Further investigation is required to better understand
considered in any pavement-protecting vehicle design the vehicle factors involved.
guidelines.
(3) The predicted relationship between loading level and
To explore such a large parameter space it is essential speed is an extremely complicated one, not well
that efficient calculation procedures be adopted. The described by a linear function. Cross-overs in the
procedures for the load variance calculations made in this characteristics of different vehicles can occur, making
paper are based on the calculation of spectral density it difficult to rank the loading level of a group of heavy
curves and are slow and cumbersome compared with new vehicles by a single index. Care should be exercised in
methods that have been uncovered in the literature. The applying regression equations fitted to experimental
implementation of these new methods is being actively data measured at a limited number of test speeds.
pursued. (4) In all of the empirical data examined to date, the
dominant contributions to dynamic pavement loading
These new methods also express the load variance in have been associated with low-frequency, whole
terms of the solution of an algebraic equation, for which it body motions of the vehicle. These contributions are
may prove possible to derive useful closed-form expres- influenced by the overall elastic and dissipative force-
sions. This would allow the definition of analytic formulae deflection characteristic of the axle group, but are
which could be useful in guidelines to limit dynamic relatively insensitive to design features which do not
pavement loading. It is intended, as a future part of this affect this gross behaviour, such as unsprung mass and
project, to explore the analytic properties of these solutions the details of the load sharing mechanism.
to determine whether such useful formulae can be found. Vehicles fitted with tandem suspensions which have
(5)
poor damping of suspension pitch motions are
predicted to produce very large dynamic loading
levels over certain speed ranges. These high levels are
Conclusions substantially reduced when shock absorbers are fitted
between the vehicle chassis and the axles. This
(1) The work described in this paper has given a measure theoretically simple and relatively cheap modification
of confidence in the modelling approach adopted in would result in dynamic loading levels comparable
this project. However, further validation work needs with the better simulated vehicles.

262 Australian Road Research, 15(4), December 1985

 HEAVY VEHICLES AND PAVEMENT LOADING

References KARNOPP, D.C. (1978). Vehicle response to stochastic

roadways. Vehicle System Dynamics 7, pp. 97-109.
ROBSON, J.D. (1978). The role of the Parkhilovski model in
AUSTRALIAN DEPARTMENT OF TRANSPORT (1978). road description. Vehicle System Dynamics 7(3),
A guide to heavy vehicle suspension systems and acceptable pp. 153-62.
axle groups. (AGPS: Canberra.) SAYERS, M. and GILLESPIE, T.D. (1983). Dynamic
pavement/wheel loading for trucks with tandem sus-
BENDAT, J.S. and PIERSOL, A.G. (1971). Random Data:
pensions. Proc. 8th IAVSD Symp., Massachusetts Inst.
Analysis and Measurement Procedures. (Wiley-Interscience:
Technol., 15-19 August, pp. 517-33.
New York.)
SCARPATETTI, V. (1975). Uberlegung zur Forderung des
ELMADANY, M.M., DOKAINSH, M.A. and ALLAN, A.B. niederlandischen Vehrkehrsministerium, bei Tandem-
(1979). Ride dynamics of articulated vehicles — a literature Achskonstruktion einen mechanischen Ausgleich
survey. Vehicle System Dynamics 8(4), pp. 287-316. vorzusehen. Daimler Benz Int. Rep.
FANCHER, P.S., ERVIN, R.D., MACADAM, C.C. and SWEATMAN, P.F. (1983). A study of dynamic wheel forces in
WINKLER, C.B. (1980). The measurement and represen- axle group suspensions of heavy vehicles. Australian Road
tation of the mechanical properties of truck leaf springs. Research Board. Special Report, SR No. 27.
Soc. Auto. Eng. Paper 800905. WHITTEMORE, A.P., WILEY, J.R., SCHULTZ, P.C. and
HEATH, A.N. and GOOD, M.C. (1985). Heavy vehicle design POLLOCK, D.E. (1970). Dynamic pavement loads of
parameters and dynamic pavement loading. Australian heavy highway vehicles. NCHRP Rep. 105. Highway
Road Research Board. Internal Report, AIR 387-1. Research Board.

Andrew Heath completed a Bachelor of Malcolm Good is a Principal Research

Mechanical Engineering degree at Melbourne Scientist with the CSIRO Division of Manu-
University in 1982. He is now studying for a facturing Technology, having moved there
Ph.D. on an ARRB sponsored project at the recently from the Department of Mechanical
same university. His project is concerned with Engineering, University of Melbourne. He
whole vehicle dynamics affecting dynamic was educated at that university, receiving
pavement loading. bachelor's, master's and doctor's degrees in
mechanical engineering. Dr Good's research
interests have been in the fields of fluid
mechanics, vehicle dynamics, highway geo-
A.N. HEATH, M.C. GOOD, metrics, human factors of automobile and
B.E. B.E., M.Eng.Sc., Ph.D. motorcycle control vehicular impacts with
roadside structures and most recently, the
dynamics and control of industrial robots.
During 1975 he was a Visiting Research
Fellow at the Institute of Sound and Vibration
Research in Southampton, U.K. and at the
Highway Safety Research Institute of the
University of Michigan, U.S.A. He occupied a
similar position in the Control Theory and
Systems Program of the General Electric
Company's Corporate Research and Develop-
ment Center, Schenectady, N.Y. for ten
months during 1983/84.

Australian Road Research, 15(4), December 1985

263