Bus Ride Index - A Refined Approach to Evaluating Road Surface

Irregularities

Teron Nguyena , Bernhard Lechnerb , Yiik Diew Wongc and Jun Yew Tand
a
TUMCREATE Ltd, 1 Create Way, #10-02 CREATE Tower, Singapore 138602; b Technische
Universität München (TUM), Baumbachstrae 7, 81245 Mnchen, c Nanyang Technological
University (NTU), N1-01b-51, 50 Nanyang Avenue, Singapore 639798, Singapore, d Samwoh
Innovation Centre , 51 Kranji Cres, Singapore 728661, Singapore.

ARTICLE HISTORY
Compiled September 1, 2019

ABSTRACT
The road surface quality can be assessed with ride comfort indices because of their
strong correlation. Many studies on ride comfort have focused on cars and trucks,
but their results are not applicable to buses, which are characterised by inherently
different vehicle dynamics. In this study, a quarter-vehicle simulation concept was
used to develop a Bus Ride Index (BRI) for evaluating the effect of road irregu-
larities on bus ride comfort. A BRI model was developed to optimise ride comfort
depending on seat configuration and air suspension and validated according to tech-
nical data. The results show a good regression relationship between BRI and the
International Roughness Index (IRI). New IRI thresholds with regard to ride com-
fort and bus operating speeds were established to serve as a benchmark to develop
better pavement maintenance strategies for bus lanes and to estimate road quality
based on acceleration data.

KEYWORDS
ride comfort, road longitudinal profile, quarter-car simulation, bus lane,
International Roughness Index

Word Count: 5199 words, 10 figures, 9 tables

1. Introduction

Road roughness (e.g., unevenness, irregularities) has a strong impact on ride comfort;
vehicles cannot operate at high speeds when traversing rough surfaces. Numerous re-
search studies have evaluated the effect of road roughness on vehicle performance and
hence ride quality by using quarter-vehicle (Sayers 1995; Prem, Ramsay, and McLean
2000; Khavassefat, Jelagin, and Birgisson 2015), half-vehicle (Sayers 1989; Múčka and
Granlund 2012) and full-vehicle models (Capuruço et al. 2005; Cantisani and Lo-
prencipe 2010) to represent vehicle system. Each representation has its advantages
and disadvantages; the quarter-car simulation (QCS) is the simplest model and is fre-
quently used in many studies. Nonetheless, all vehicle dynamic models to evaluate

CONTACT Teron Nguyen. Email: [email protected] (To cite this article: Teron Nguyen,
Bernhard Lechner, Yiik Diew Wong Jun Yew Tan (2019) Bus Ride Index a refined approach to evaluating
road surface irregularities, Road Materials and Pavement Design, DOI: 10.1080/14680629.2019.1625806)
road roughness so far have been focused on car and truck applications (Múčka 2016a);
therefore, their results are not directly applicable to the vehicle dynamics of buses.
Bus transit can be considered as a form of heavy transport that causes heavy damage
to the road surface/pavement. In many cities around the world, a bus rapid transit
(BRT) system requires a strong pavement structure and more frequent maintenance
than regular car lanes. Global BRTData (BRT+ Centre of Excellence and EMBARQ
2018) reported that BRT lanes around the world are surfaced by Portland cement
concrete (35%; around 1441 km) or asphalt concrete (65%; around 2635 km). Dense
bus networks in cities such as Singapore, London and Hong Kong are continuously
loaded with a high number of buses.
Even though buses operate at lower speeds than cars or trucks and bus lanes are
maintained well, there continues to be concerns and interest in bus ride comfort in
relation to road surface irregularities. The vibrations from road irregularities are trans-
mitted to the bodies of drivers and passengers via their seats, although bus drivers
typically have more comfortable seats than passengers. The conventional International
Roughness Index (IRI) cannot represent passenger ride comfort directly and is not
applicable to lower-speed road categories. This motivated Cantisani and Loprencipe
(2010) and Yu, Chou, and Yau (2006) to develop speed-related ride quality thresholds
based on IRI for car rides. However, there have been no studies on bus dynamics or
whether their work (Cantisani and Loprencipe; Yu, Chou, and Yau) applies to bus
lanes. Hence, the present study aimed to address this gap in knowledge and meet the
following objectives:
• To develop a vehicle dynamics model by calibrating a Bus Ride Index (BRI) for
evaluating the bus lane surface roughness and accurately capturing bus passenger
ride comfort, and
• To propose new IRI thresholds conditioned on BRI at lower speeds that are more
appropriate for bus operation, which will be beneficial for pavement monitoring
and maintenance to ensure bus passenger ride comfort.
BRI was analysed for both artificial and field-measured road profiles to find the
regression relationships between BRI and IRI.
The remainder of the paper is structured as follows. First, the relevant scientific
literature is summarised. The fundamentals of BRI and its characteristics are then
described. Road profiles for BRI analysis (i.e. artificial and measured road profiles)
are presented. Findings on the relationship between IRI and BRI at different speed
levels and the advantages of BRI are discussed. The conclusion sums up the study.

2. Literature review

Existing approaches to evaluate road irregularity can be classified in different cate-

gories with various indices (Ueckermann and Oeser 2015): geometry in the distance
domain (e.g. standard deviation, energy content), geometry in the spectral domain
(e.g. power spectral density), effect in the distance domain (e.g. IRI, ride number),
and geometry and effect in the distance domain (e.g. profilograph, bump integrator).
Each approach is related to a specific measurement equipment or specific numeri-
cal calculation method. Measurement approaches consider the dynamic response of
measurement devices or vehicle components. Hence, they are more appropriate for
evaluating specific vehicle-dedicated lanes and the effects related to such vehicles (e.g.
ride quality). Numerical calculation methods are more suitable for investigating the

2
road surface itself regardless of the vehicle dynamics.
Among the vehicle response approaches, IRI is recognised and used worldwide as
a standard index for road roughness evaluation (Múčka 2017). It is based on a two-
degree-of-freedom (DOF) QCS. The comfortable ride of a car driver has been proposed
as the reference roughness at various IRI levels. IRI can be interpreted as the output
of an idealised response-type measurement system known as the golden-car driving
at 80 km/h. This is expressed by accumulated suspension motions over the distance
travelled (e.g. mm/m). The algorithm was proposed by Sayers (1995) and has also
been implemented in the proposed standard CEN-PREN 13036-5 (2015) and ASTM
E1926-08 (2015).
Given its simplified model, IRI includes several disadvantages, such as being specific
to car dynamics, a single calculated speed level of 80 km/h, and an indirect relation to
ride comfort (Bridgelall 2014; Cantisani and Loprencipe 2010; Múčka 2016a). Hence,
several other approaches to quantifying road irregularity have been developed based on
QCS to include other vehicle types (e.g. trucks) and effects (e.g. ride comfort). Some
examples are the Truck Ride Index-TRI for truck driver comfort in cabin (Prem,
Ramsay, and McLean 2000), Heavy Articulated Truck Index-HATI that accesses poor
ride to occupants of heavy articulated trucks (Hassan, McManus, and Cossens 2006),
and Longitudinal Evenness Index-LWI for evaluating ride comfort, ride safety, dynamic
load of road and cargo (Ueckermann 2002). Yu, Chou, and Yau (2006) developed speed-
related ride quality thresholds using IRI for local street application, as given in Table 1.
Table 2 presents another IRI threshold at low operating speeds, where similar ranges
of values can be observed for vehicle speeds of 30-60 km/h. Cantisani and Loprencipe
(2010) conducted this study by using an 8-DOF full-car simulation.
Regarding passenger ride comfort, studies on human subjects have shown that the
human body mass has a vibration resonance frequency of 4-8 Hz at which it is most
sensitive to acceleration. To include all possible effects of vibration on health, com-
fort and perception, ISO2631-1 (1997) considers a frequency range of 0.5-80Hz. The
weighted root-mean-square (RMS) acceleration is calculated as follows:

 Z T  21
1
aw = a2w (t) dt (1)
T 0

where aw (t) is the weighted acceleration as a function of time (m/s2 ) and T is the
duration of the measurement (s).
The frequency-weighting curves in the vertical direction are applied to calculate
the frequency-weighted RMS vertical acceleration (awz ) related to road roughness
excitation. Table 3 gives the suggested scale of awz according to different levels of
user ride comfort on public transport vehicles, which is similar to that suggested by
British Standard BS6841:1987 (1999). These recommendations have been widely used
in numerous studies to evaluate passenger ride comfort with regard to vehicle dynamics
subjected to road excitation (Prem and Ayton 2005; Cantisani and Loprencipe 2010;
Múčka 2016b).
Múčka (2016a) summarises current approaches, which include four QCS models
(SEI, TRI, LWI, HI) related to ride comfort evaluation and three QCS models (PQI,
HATI, VRI) used to measure dynamic suspension load, as given in Table 4. The QCS
models are mainly differentiated by the vehicle type, either car (SEI, LWI, HI, PQI,
VRI) or truck (TRI, LWI, HATI). Some were derived from IRI model. For urban buses,

3
the ride comfort is crucial because a single vehicle can carry about 100 passengers,
which can lead to public concern if any uncomfortable incidents occur because of road
irregularities. Increasing bus ride comfort will contribute to the strategy of shifting
car users to public transport. A QCS model for buses is necessary to evaluate road
surface irregularities associated with ride comfort but has yet to be developed, which
was the motivation for the present study.
With regard to bus dynamics, there have been few studies in the literature. Sekulić,
D., & Dedović (2011) used a 3-DOF QCS to evaluate the effects of the stiffness and
damping of the suspension system on optimising the vibrational behaviour of a bus.
Sekulić et al. (2013) used a 10-DOF oscillatory bus model to analyse the effects of
vibration on the comfort of intercity bus users. However, these models used only one
artificially generated road profile and focused more on the vehicle dynamics instead
of evaluating road irregularities. Recently, Sekulić et al. (2016, 2018) built the same
full bus model with MSC.ADAMS and validated it with two road profiles (poor and
good asphalt concrete pavement) to establish equal oscillatory comfort zones in which
passengers seated in the bus overhang and over the front axle or in the rear overhang
and over the rear axle are exposed to a higher level of ride discomfort. These are also
the fundamental positions for the proposed BRI in the present study. Overall, there
have been limited field surveys on the relationship between a ride comfort index and
IRI except for the experimental study by Fichera, Scionti, and Garescı̀ (2007), who
used three buses (18, 9.46, and 4.6 t) running on 12 public roads to evaluate the driver
ride comfort in bus cabins. They also built a 6-DOF half-bus model to represent the
system dynamics of an 18 t bus, whose parameters were used as a reference to develop
BRI.
In this study, the proposed BRI for road roughness evaluation has several goals:
• to reflect the ride comfort of bus passengers;
• to reflect the speed variation of the bus fleet, and
• to provide a simple calculation procedure.

3. Methodology

A realistic ride comfort index to evaluate road roughness can be obtained by carefully
considering the following aspects of vehicle ride dynamics:
• properties of seat dynamics,
• vehicle suspension and associated damping characteristics, and
• human tolerance to whole-body vibrations.
These aspects are considered in the proposed QCS bus model and ride comfort
evaluation standard.

3.1. Development of the BRI QCS bus model

IRI calculation is based on the rear-axle of vehicle-body responses (as illustrated by
Cantisani and Loprencipe (2010); Sayers (1998)). IRI can be considered as an indirect
indicator of the ride quality, which is not related to human tolerance, seat configura-
tion, and seat properties. In contrast, the proposed BRI takes the occupants vibration
response to the rear axle as the ride comfort index. A 3-DOF QCS is proposed for
BRI that includes a 2-DOF quarter-bus model and 1-DOF human and seat model, as

4
shown in Figure 1. Instead of the driver seat, which is equipped with a pneumatic elas-
tic suspension and shock absorber, BRI considers a passenger seat, which has higher
stiffness and less damping from hard polyurethane foam. The selected parameters are
described below; available bus data from bus manufacturers were used (gross vehicle
weight, axle weight, air suspension type, and specifications) by referencing previous
studies on bus dynamics (passenger weight, seat stiffness and damping, tire stiffness,
and damping). Compared to other studies, the technical parameters for BRI consider
a comfortable ride for bus passengers, as given in Table 5.
• Sprung and unsprung mass:
The bus parameters of the 2-DOF quarter-bus model were referenced from
the existing bus fleet in Singapore; technical specifications were available for the
Volvo B9L, Mercedes Citaro, and MAN NL323F but unavailable for Volo B7RLE
and Scania K230UB. These are either low-floor or low-entry buses for urban
areas and have similar dimensions and dynamics characteristics. Among these
bus types, only the Mercedes Citaro (Mercedes-Benz 2018) provided enough
technical data for reference. This single-decker bus model was used for simulation
purposes to build the BRI QCS. The maximum bus weight is 19,000 kg (including
the chassis, axles, tyres, wheels, drive train, passengers, interior, and cowlings),
while an empty bus weighs 10,770 kg. The bus weight is not distributed evenly;
the front axle carries one-third of the gross vehicle weight, while the rear axle
carries two-thirds. For BRI, a sprung mass was proposed in the case of a fully
loaded bus.
• Suspension and damping:
The Mercedes Citaro bus is equipped with an independent front axle system
(RL 82 EC) and rigid rear axle system (AV 133); both are manufactured by ZF
GmbH (2009). For the 2-DOF model, only the rear axle parameters are consid-
ered. The suspension system is an air spring for ordinary buses in the city, and
the double convoluted spring #22 was chosen from the Firestone catalogue (Fire-
stone 2014) and Enviro 500 bus specifications Alexander (2014). The dynamic
spring rate is calculated based on the following equation from Firestone:
" "    1.38 # #
V1 1.38 V1
k = [prequired + 14.7] Ac − Ae − 14.7 (Ac − Ae ) (2)
Vc Ve

where V1 is the internal air spring volume (in3 ) at the design height, Ae is
the effective area (in2 ) 0.5 in above design height, Ac is the effective area (in2 )
0.5 in below design height, Ve & Vc are the corresponding inner air volume (in3 ),
prequired is the required inflation pressure (lbs/in2 ) (prequired = loadcurrent /Aef f
and Aef f = loadmax,def ault /pdef ault ), loadmax,def ault is the maximum static load
(lbs) at default pressure per air spring, and pdef ault is the default pressure
(lbs/in2 ).
For the damping parameter, a simple approximation of a digressive damping
curve has been proposed (Fakultaet fuer Maschinenwesen 2018; Jazar 2005) in
which the damping ratio for passenger cars should be between 0.25 and 0.3 for
the best compromise between safety and comfort while favouring ride comfort.
The equivalent stiffness and damping values of the air suspension were doubled
after calculation to represent the dual-wheel rear axle.
• Tyre parameters:
The Mercedes Citaro bus is equipped with tires and wheels with the dimen-

5
 sions of 275/70 R22.5. Because of their larger diameters and higher axle loads,
bus tyres have a larger contact area than passenger car tyres. Gillespie et al.
(1992) suggested using a filter with a length of 304.8 mm (1 ft) to represent
a 11R22.5 heavy-duty truck tyre. According to the stated axle loads and sug-
gested tyres by ZF Friedrichshafen AG, a single-decker bus can be compared to
a two-axle straight truck with the above tyres. Therefore, the base length of the
moving average filter was set to 0.304 m. The tire stiffness and damping were
referenced from (Sekulić, D., & Dedović 2011; Sekulić et al. 2013) because they
had the only available data. The equivalent tire stiffness and damping value were
doubled to represent the dual-wheel rear axle.
• Human seat model:
For the passenger 1-DOF seat model, Wei and Griffin (1998); Patten, Sha,
and Mo (1998) demonstrated that the stiffness and damping of a foam cushion
are dependent on the load and excitation level. The nonlinearity is accounted
for by the preload-dependent parameters of stiffness and damping. Note that
those values are for a vehicle passenger foam seat and that the stiffness of urban
bus seats may be higher than the proposed values. On the other hand, Sekulić
et al. (2016)’s parametric design-of-experiment study showed that the optimal
stiffness (Kp ) and damping (Cp ) for the lowest awz value of a passenger seat are
Kp = 10,000 N/m and Cp = 330 Ns/m. Hence, these parameters are proposed
for BRI model.

Table 5 presents the calculated and proposed BRI parameters based on the above
analysis on vehicle dynamics. BRI has similar parameters to other bus models from
the literature. For the specific Citaro bus model, its axle configuration, air spring stiff-
ness, and damping are calculated. Figure 1 shows the proposed BRI QCS model with
normalised values based on the sprung mass (Ms). Table 6 compares its values to those
of IRI and TRI. The calculated undamped natural frequencies of the seat and passen-
ger, sprung mass, and unsprung mass were 1.93, 1.47, and 10.14 Hz, respectively, and
the calculated damped natural frequencies were 1.89, 1.26, and 10.14 Hz, respectively.
These computed frequencies are close to the oscillation frequencies of a regular bus
service (Sekulić, D., & Dedović 2011); the air spring helps maintain a nearly constant
sprung mass natural frequency (Cao, Song, and Ahmadian 2011). The tire damping
effect is minimal, so it can be ignored in the simulation and calculation. The passenger
seat damping effect is also small compared to that of the driver seat from TRI. Table 6
describes all BRI parameters that differ from IRI and TRI. Bus speeds are much lower
than the car and truck speeds for calculation; v = 40km/h is the operating speed on
urban roads and v = 60km/h is the operating speed on highways. These values are
proposed based on field observations and measurements.
The vertical vibration of a bus passenger can be found by solving the differential
equation of motion in a matrix expression:

[M ] {z̈} + [C] {ż} + [K] {z} = [A] {zr } (3)

where [M ], [C ], and [K ] are the symmetric mass, damping and stiffness matrices; [A] is
the stiffness matrix of the input from road roughness; {z}, {ż}, and {z̈} are the column
vectors of the generalised coordinates, velocities and acceleration, respectively; {zr }
is the column vector of the road roughness vertical motion; and z = [zu , zs , zp ]T .
The differentiation of the IRI, TRI, and BRI models can be based on the transfer
functions (TFs) of the vehicle responses for both the temporal and spatial frequencies.

6
Fundamentally, IRI is calculated based on the displacement of the sprung mass to the
unsprung mass, whereas TRI and BRI consider the response of the sprung mass to
road profile inputs. The vibration of the sprung mass is then transferred to a seat and
the human body. The similarity and difference between the three TF curves are shown
in Figure 2 based on the parameters in Table 6.
In Figure 2a, the dominant peak of the IRI-car response is near 1 and 10 Hz, while
the TRI-truck body bounce frequency is near 2.57 Hz, and the BRI-bus response is near
1.54 Hz. TFs of BRI-bus are almost unchanged over different passenger loading factors
(n = 0-1) because of the nonlinear dynamics of the air-suspension. Figure 2b shows a
similar trend for the wavenumber: the IRI-car is sensitive to the road roughness for a
wavelength range of 30-1 m (a wavenumber range of 0.03-1 cycles/m), while the TRI-
truck and BRI-bus are sensitive to wavelengths greater than about 3 m (wavenumber
of 0.33 cycles/m). These wavenumber values are equivalent to a magnitude of 0.1 abs.
However, the BRI and TRI transfer function gains/magnitude show a clear difference;
the gains are higher for BRI from 0.01 to 0.08 cycles/m and higher for TRI from 0.08
to 0.33 cycles/m. The transfer functions for the spatial frequency were calculated at
a travel speed of 80 km/h for comparison purposes only. During normal operation,
buses travel at much lower speeds, and the stiffer and less damped seat configuration
for the BRI calculation results in a greater differentiation between the TRI and BRI
outputs.

3.2. Validation of the BRI model

To validate the BRI QCS model, field measurements were conducted by using a seat-
pad accelerometer (SV100A, Svantek), which was placed on a seat cushion (Figure 3)
and preloaded with a 70 kg human, along a 2 km bus lane of Jalan Bukit Merah
Street in Singapore. The bus lane longitudinal profile was scanned by using a multi-
laser road profiler with a sampling interval of 25 mm and GPS data at 1 Hz. The
SV100A seat-pad accelerometer was settled on a seat cushion above the rear axle,
controlled wirelessly by a smartphone, and collected vibration data at a frequency of
750 Hz to calculate awz , and GPS data at 1 Hz. Bus lanes IRI was 3.5-4.8 m/km. To
include different bus models in the city, the measurement was also conducted for the
single-decker bus model MAN NL323F driving on the same road segments.
The BRI QCS model was simulated in MATLAB/Simulink using the same road
input and bus speed from the field measurement. The developed numerical algorithm
included the speed changes along the bus route and matched GPS locations between
road profiles and bus vibration measurements. There were 12 road sections with dif-
ferent lengths for an average ratio of awz simulation /awz measurement = 1.02 (std=0.13).
Figure 4 compares of time-series frequency-weighted acceleration (awz (t)) and awz
between simulation and measurement at different velocities (m/s); section 8 was a
smooth road and section 5 was a rough road surface. The graphs show that the sim-
ulation output and measurement data matched quite well for different speed changes,
especially the peak values. Perfect matching could not be achieved because the speed
sampling based on GPS signal did not have a high level of accuracy, and the bus wheel
trajectories and road profiles were misaligned.

7
4. Road profiles for BRI analysis

4.1. Generation of road profiles

A road profile is the variation in road elevation with distance as measured by appro-
priate field devices. If the vertical displacement is assumed to be a centred, homo-
geneous, and Gaussian function of distance, the road profile can be described as the
power spectral density (PSD) of the spatial frequency Gd (n)(m2 /cycle/m) or angular
spatial frequency Gd (Ω)(m3 /rad). The roughness profile can be defined as follows:

 −2
n
Gd (n) = Gd (n0 ) . (4a)
n0

 −2
Ω
Gd (Ω) = Gd (n0 ) . (4b)
Ω0

Table 7 divides the values of Gd (n0 ) and Gd (Ω0 ) according to ISO 8608:2016 (2016)
into eight road categories, where their PSDs were calculated based on a spatial fre-
quency n0 of 0.1 cycles/m and angular spatial frequency Ω0 of 1 rad/m. The different
road classes are illustrated on a log-log scale as straight lines with a constant waviness
w of 2 for the lines slope. Múčka (2018) reviewed 27 scientific papers to show that,
for simulation purposes, road class A may represent a typical good- or average-quality
road (e.g. motorway, expressway, and first-class road), whereas classes B and C may be
appropriate for simulating low-quality paved surfaces (second- and third-class roads or
local highways). In Germany, road class B is often considered as a basic standard for
road roughness evaluation, in which G0 = 1, 3 and 9 cm3 are equivalent to the target
value, warning value and threshold value, respectively (FGSV-489 2006; Maerschalk,
Ueckermann, and Heller 2011; Ueckermann and Oeser 2015).
With ISO 8608:2016, road profiles can be generated from different mathematical
algorithms once all necessary parameters are known. Agostinacchio, Ciampa, and Olita
(2014) presented a widely used equation for generating artificial profiles:

N √  n 
0
X
h (x) = ∆n.2k .10−3 . .cos(2π.i.∆n.x + ϕi ) (5)
i.∆n
i=0

where x is the abscissa variable from 0 to L; ∆n=1/L; nmax =1/B;

N=nmax /∆n=L/B, B is the sampling interval; k is a constant value depending on
the ISO road profile classification and is assumed to be an integer value from 3 to
9 corresponding to classes A-H; n0 =0.1 cycles/m, and ϕi is a random phase angle
following a uniform probabilistic distribution within 0-2π.
Equation 5 was used to generate 200 road profiles in MATLAB as the input for
vehicle dynamics simulations. The road surface profiles Figure 5 were very good (ISO
A-B class with hmax = 15 mm), good (ISO B-C class with hmax = 25 mm), average
(ISO C-D class with hmax = 50 mm), and poor (ISO D-E class with hmax = 100 mm).
Road classes E-H primarily represent rough unpaved roads which were not present
in the network covered in this study. On the other hand, since road classes D to H

8
are often associated unpaved roads (Múčka 2018), only the very good and good road
profiles were used to present urban road network in this study.

4.2. Field-measured road profiles

Singapore has a resident population of 5.5 million in a small land area of 719 km2 , bus
ridership has a high share of public transit at 3.9 million out of 7.9 million passenger
trips per day (PTC 2016). Figure 6a visualises bus lane schemes according to data
from Singapore (LTA 2018) which highlight full-day bus lanes mostly paved with
Portland cement concrete in the city centre. Figure 6b shows the daily bus volumes
generated using the method in (Liu, Zhou, and Rau 2018). The high daily bus volumes
are concentrated along bus lanes, with 1500-3000 buses equivalent to 23-46 million
equivalent single axle loads in 10 tons (10t-ESALs) accumulated over a service life
of 30 years. These traffic volumes are classified as the highest road class Bk100 (>32
million ESALs) according to German standards (FGSV-499 2015).
The longitudinal road profiles were collected in May 2018 by using a multi-laser
road profiler with a sampling interval of 25 mm. The device has a unique ”Stop &
Go” functionality that enables it to collect data efficiently from both urban and rural
networks at traffic lights, stop signs, junctions, and roundabouts. Measurements were
performed for 31.2 km of normal bus-lanes and 17.96 km of mixed traffic lanes in
the city, where buses operate regularly and frequently. However, the full-day bus lanes
could not be measured because vehicles excluding public buses are not allowed to drive
on them.
IRI was calculated for both the left and right wheel tracks. For each survey lane,
elevation data is at an offset of 85 cm left and 85 cm right of the survey vehicle centre-
line. This is the maximum distance that the measurement device can offer. Figure 7
shows the IRI distribution for a total of 872 profile segments, each 100 m in length.
Most of the IRI values were distributed from 2.5 to 6.5. With regard to riding quality,
IRI < 2.5 is good, IRI = 2.5-3.5 is average, IRI = 3.5-4.5 is marginal, IRI = 4.5-6.0 is
poor and IRI > 6 is bad (LTA 2009). The field-measured road profiles showed a high
proportion of poor bus lane surfaces.

5. Results and discussion

5.1. Correlation between BRI and IRI

IRI is proportional to the vertical vibration awz and total three-dimensional vibration
aw . Many studies have been conducted to establish the following relationship between
these indices:

aw (awz ) = b1 IRI + b2 (6)

where b1 is the slope of the straight line and b2 is the intercept.

In (Múčka 2016b), the summary of nine experimental studies on different types of
vehicles at different speeds showed that b1 ranges from 0.071 to 0.192 for awz and from
0.002 to 0.369 for aw , while the b2 ranges from 0.005 to 0.49. In particular, for a heavy
bus with a weight of 18 t, awz = 0.088IRI + 0.054 (R2 = 0.882) at the drivers seat
according to Fichera, Scionti, and Garescı̀ (2007), although they did not report the
speed.

9
 In this study, BRI was calculated for all road profiles with a default length of 100
m based on the above calibrated BRI QCS bus model. The relationships between IRI
and BRI at different speeds of 20, 40, 60 and 80 km/h were established for artificially
generated road profiles (Figure 8) and field-measured road profiles (Figure 9). The
maximum IRI values range around 10 mm/m for both artificial and field-measured
road profiles, representing realistically urban city bus lanes in the very good ISO A-B
and good ISO B-C road categories.
Lower R-squared values were observed for the measured road profiles at a speed of
80 km/h: BRI80 = 0.26IRI. For car, Cantisani and Loprencipe (2010) established the
relationship awz = 0.22IRI at the same speed of 80 km/h based on 124 road profiles
with lengths of 320 m each. Table 8 presents the b1 coefficients of BRI-IRI calculated
at 100, 160 and 320 m interval as the most used intervals worldwide (Múčka 2017) and
car awz -IRI regression at different speeds. The formers were always higher than the
latter by around 0.3-0.6, which means that the same IRI can cause greater discomfort
to bus passengers than car passengers. For BRI-IRI, similar results of coefficient b1 are
observed between the three alternatives at different speeds, however, longer calculating
intervals have improved the correlation (R-squared) in the BRI-IRI relationship. While
IRI varied at longer interval, BRI could also change accordingly. The significantly high
R-squared is nearly to 1.0 for IRI − BRI80 in the case of artificially generated road
profiles can be explained by the relatively homogeneous distribution compared to
the field-measured profiles, which contained various local obstacles such as bumps,
potholes, and distresses. Herein, a given value of IRI resulted in a range of BRI values
(i.e. different ride comfort levels), as clearly shown in Figure 9b.
The vehicle speed was found to have a high impact on the users ride comfort, and
different linear regression relationships were observed at different speed levels. Given
the different correlation R-squared values between BRI and IRI in the two datasets,
caution must be exercised when relating artificial road profiles to ride comfort. Artifi-
cial profiles are useful for vehicle suspension design but are not adequate for correctly
evaluating road irregularities and their effects on users. This finding is consistent to
the investigation by Loprencipe and Zoccali (2017) in the difference excitation effects
between ISO 8608 approach and in situ measurements of road profiles.

5.2. Ride quality levels based on IRI and BRI

Based on the established IRI-BRI regression relationships (Figure 9a and Table 8),
IRI thresholds were suggested for different speed levels based on the ride comfort
thresholds from comf ortable to uncomf ortable. The ride quality for bus operation at
lower speeds is given in Table 9 and Figure 10. Compared to the recommendations in
Table 1 and Table 2, Table 9 suggests a higher IRI requirement for bus lanes even for
the same ride comfort levels established by Cantisani and Loprencipe (2010).
The correlation between BRI, IRI, and the bus speed provides suggestions for bus
operation and pavement maintenance to ensure passenger ride comfort. Different stake-
holders (bus operators, road transport authorities) can use these suggestions for dif-
ferent applications. Figure 10 serves as a reference for estimating the bus ride comfort
from road roughness conditions based on the IRI levels and bus speed. Bus speeds
should be regulated at certain levels of IRI to ensure ride quality. Meanwhile, road
conditions should be maintained according to the proposed IRI thresholds to align with
the designed bus operating speed and passenger ride comfort. For example, under ur-
ban conditions where buses operate at 40 km/h, the bus lanes should be maintained

10
at IRI of 2.25-4.5 m/km to ensure good/fair ride quality. However, IRI should be
1.66-3.32 m/km for buses running at higher speeds of 60 km/h on highways.
With regard to other applications, the road roughness of a bus lane network can
be inferred from the BRI value, which can be quickly measured by using a portable
accelerometer on the passenger seat (e.g. the SV100A accelerometer used in this study).
In this manner, the road surface can be assessed, and a maintenance strategy can
be proposed for a specific road segment before/without requiring an expensive road
profiler instrument (e.g. full-day bus lanes that do not allow vehicles other than buses).
As a result, maintenance cost will be reduced as compared to regular scanning of
the whole road network. Furthermore, road surface can be up-to-date assessed once
every bus is designed as a probe to detect roughness condition automatically. Similar
approaches have been proposed (regardless of passenger ride comfort) to detect road
anomalies using a smart-phone (Seraj et al. 2015; Eriksson et al. 2008; Alessandroni
et al. 2017), to estimate road roughness Power Spectral Density (PSD) (González et al.
2008) and IRI (Du et al. 2014) from the vehicle acceleration.

6. Conclusion

BRI was developed based on a well-calibrated 3-DOF QCS derived from existing tech-
nical parameters for operating buses. The new and simple BRI model can measure di-
rectly passenger ride comfort with the inclusion of air-suspension stiffness and damping
ratio to maximise ride quality. The nonlinear dynamics characteristic of BRI enables
it to keep the low vibration frequency (around 1.2-1.5Hz) regardless of changing the
sprung mass by passenger loads. A comparison between the transfer functions of TRI
and BRI showed that BRI is more appropriate for evaluating ride comfort with regard
to bus/BRT lanes given the inherent difference in vehicle configurations.
Regression relationships were established between IRI and BRI at different speeds
for both artificially generated and field-measured road profiles. The analysis of BRI-
IRI correlation at two lengths of 100, 160 and 320 m has shown that the correlation
coefficients are unchanged, therefore, BRI can be applicable for different calculating
intervals. Based on the BRI values from field-measured road profiles, a given input
value for IRI can represent a range of ride comfort levels, which means that the con-
cerns of passengers may vary. IRI thresholds according to different speed levels and
ride quality are suggested, which will be useful for bus lane pavement maintenance
and monitoring. The findings of this study serve as a benchmark to estimate road con-
ditions with an onboard accelerometer, which is an advanced method for monitoring
citywide bus lane network more efficiently than the conventional method of using road
profiler instrument.
However, this study was based on asphalt pavement field-measured road profiles
and passive suspension and damping vehicle dynamics. Further study can involve
developing an active suspension QCS bus model to be coupled with the measured
profiles of concrete pavement for BRI calculation. Half- and full-bus models should
also be developed to consider other dynamic effects such as pitch and roll. The usage
of the proposed IRI thresholds certainly deserves careful consideration by transport
authorities.

11
Acknowledgements

This work is a part of PhD study of the first author and financially supported by the
National Research Foundation Singapore under its Campus for Research Excellence
And Technological Enterprise (CREATE) programme. We thank Samwoh Innovation
Centre for providing the field-measured road longitudinal profiles.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Teron Nguyen: https://orcid.org/0000-0001-6822-0753

References

Agostinacchio, M., D. Ciampa, and S. Olita. 2014. “The vibrations induced by surface irregu-
larities in road pavements - a Matlab R approach.” European Transport Research Review 6
(3): 267–275.
Alessandroni, Giacomo, Alberto Carini, Emanuele Lattanzi, Valerio Freschi, and Alessandro
Bogliolo. 2017. “A study on the influence of speed on road roughness sensing: The smartroad-
sense case.” Sensors (Switzerland) 17 (2).
Alexander, Dennis 500. 2014. “Enviro500-Low floor high capacity 12m Double Deck Bus.”
Accessed 2017-10-30. https://www.alexander-dennis.com/products/double-deck-3-axle/.
ASTM E1926-08. 2015. “Standard practice for computing international roughness
index of roads from longitudinal profile measurements.” ASTM International
http://www.astm.org/cgi-bin/resolver.cgi?E1926-08(2015).
Bridgelall, Raj. 2014. “Connected Vehicle Approach for Pavement Roughness Evaluation.”
Journal of Infrastructure Systems 20 (1): 04013001.
BRT+ Centre of Excellence and EMBARQ. 2018. “Global BRT Data.” Accessed 2018-10-01.
https://brtdata.org/.
BS6841:1987. 1999. “British Standard Guide to Measurement and Evaluation of
Human Exposure to Whole-body Mechanical Vibration and Repeated Shock.”
https://shop.bsigroup.com/ProductDetail/?pid=000000000000171912.
Cantisani, Giuseppe, and Giuseppe Loprencipe. 2010. “Road roughness and whole body vi-
bration: Evaluation tools and comfort limits.” Journal of Transportation Engineering 136
(September): 818–826.
Cao, Dongpu, Xubin Song, and Mehdi Ahmadian. 2011. “Editors perspectives: road vehi-
cle suspension design, dynamics, and control.” Vehicle System Dynamics 49 (1-2): 3–28.
https://doi.org/10.1080/00423114.2010.532223.
Capuruço, Renato, Tarek Hegazy, Susan Tighe, and Sameh Zaghloul. 2005. “Full-Car Rough-
ness Index as Summary Roughness Statistic.” Transportation Research Record 1905 (1):
148–156.
CEN-PREN 13036-5. 2015. “Road and aireld surface characteristics test methods. Part 5:
determination of longitudinal unevenness indices.” European Committee for Standardization
https://standards.globalspec.com/std/10218779/pren-13036-5.
Du, Yuchuan, Chenglong Liu, Difei Wu, and Shengchuan Jiang. 2014. “Measurement of inter-
national roughness index by using Z -axis accelerometers and GPS.” Mathematical Problems
in Engineering 2014.

12
Eriksson, Jakob, Lewis Girod, Bret Hull, Ryan Newton, Samuel Madden, and Hari Balakrish-
nan. 2008. “The pothole patrol: using a mobile sensor network for road surface monitoring.”
Proceeding of the 6th international conference on Mobile systems, applications, and services
- MobiSys ’08 29. http://dl.acm.org/citation.cfm?id=1378600.1378605.
Fakultaet fuer Maschinenwesen. 2018. “Vorlesung Grundlagen des Kraftfahrzeugbaus:
Federung und Dämpfung.” Technical University of Munich, Munich.
FGSV-489. 2006. “ZTV ZEB-StB - Zusätzliche Technische Vertragsbedingungen und Richtlin-
ien zur Zustandserfassung und -bewertung von Straen.” Forschungsgesellschaft für Straen-
und Verkehrswesen. http://www.fgsv-verlag.de.
FGSV-499. 2015. “RStO - Guidelines for the standardisation of pavement structures of traffic
areas (English Edition).” Road and Transportation Research Association. http://www.fgsv-
verlag.de.
Fichera, Gabriele, Marco Scionti, and Francesca Garescı̀. 2007. “Experimental Correla-
tion between the Road Roughness and the Comfort Perceived In Bus Cabins.” (April).
http://papers.sae.org/2007-01-0352/.
Firestone. 2014. “Engineering Manual and Design guide.” https://firestoneip.com/-
/media/www/fsip/files/ManufacturingIndustrial/Design%20manuals/EMDG Imperial 2014.pdf.
Gillespie, Thomas D, Steven M Karamihas, Michael W Sayers, and Will Hansen. 1992. Effects
of heavy vehicle characteristics of pavement response and performance. Technical Report
August 1992. Michigan, USA: The University of Minchigan.
González, A., E. J. O’Brien, Y. Y. Li, and K. Cashell. 2008. “The use of vehicle acceleration
measurements to estimate road roughness.” Vehicle System Dynamics 46 (6): 483–499.
Hassan, Rayya A, Kerry McManus, and Ian Cossens. 2006. “Development of HATI-Heavy
Articulated Truck Index.” In Research into Practice: 22nd ARRB Conference, Canberra,
Australia.
ISO 8608:2016. 2016. “Mechanical vibration - Road surface profiles - Re-
porting of measured data.” International Organization for Standardization
https://www.iso.org/standard/71202.html.
ISO2631-1. 1997. “Mechanical vibration and shock- Evaluation of human
exposure to whole-body vibration-Part 1:General requirements.” 1–30.
https://www.iso.org/standard/7612.html.
Jazar, Reza N. 2005. Advanced vibrations: a modern approach. Berlin: Springer Science &
Business Media.
Khavassefat, Parisa, Dennis Jelagin, and Björn Birgisson. 2015. “Dynamic response of flexible
pavements at vehicleroad interaction.” Road Materials and Pavement Design 16 (2): 256–
276. https://doi.org/10.1080/14680629.2014.990402.
Liu, Xiaodong, Yuan Zhou, and Andreas Rau. 2018. “Smart card data-centric replication of
the multi-modal public transport system in Singapore.” Journal of Transport Geography
(March 2018): 1–11. http://dx.doi.org/10.1016/j.jtrangeo.2018.02.004.
Loprencipe, Giuseppe, and Pablo Zoccali. 2017. “Use of generated artificial road profiles in
road roughness evaluation.” Journal of Modern Transportation 25 (1): 24–33.
LTA. 2009. “Pavement Management System (PMS).” Land Transport Authority of Singapore.
LTA. 2018. “One Motoring.” Accessed 2018-10-03. https://www.onemotoring.com.sg.
Maerschalk, Günther, Andreas Ueckermann, and Slawomir Heller. 2011.
“Längsebenheitsauswerteverfahren” Bewertetes Längsprofil”: Weiterentwicklung der
Längsebenheitsbewertung der Zustandserfassung und-bewertung.” Berichte der Bunde-
sanstalt fr Straenwesen.
Mercedes-Benz. 2018. “The Citaro - Technical data and dimensions.” Accessed 2018-10-05.
https://www.mercedes-benz.com.sg.
Múčka, Peter. 2016a. “Current approaches to quantify the longitudinal road
roughness.” International Journal of Pavement Engineering 17 (8): 659–679.
http://dx.doi.org/10.1080/10298436.2015.1011782.
Múčka, Peter. 2016b. “Road roughness limit values based on measured vehicle vibration.”
Journal of Infrastructure Systems 23 (2): 04016029.

13
Múčka, Peter. 2017. “International Roughness Index specifications around the world.” Road
Materials and Pavement Design 18 (4): 929–965.
Múčka, Peter. 2018. “Simulated road profiles according to ISO 8608 in vibration analysis.”
Journal of Testing and Evaluation, ASTM International 46.
Múčka, Peter, and Johan Granlund. 2012. “Is the road quality still better?” Journal of Trans-
portation Engineering 138 (12): 349.
Patten, W. N., S. Sha, and C. Mo. 1998. “A vibration model of open celled polyurethane foam
automotive seat cushions.” Journal of Sound and Vibration 217 (1): 145–161.
Prem, H, E Ramsay, and J McLean. 2000. “A road profile based truck ride index (TRI).” In
International Symposium on Heavy Vehicle Weights and Dimensions, 6th, 2000, Saskatoon,
Saskatchewan, Canada, .
Prem, Hans, and Geoff Ayton. 2005. “Improved echniques for ssessing ride quality on concrete
pavements.” In 8th International Conference on Concrete Pavements, Colorado, USA, 733–
754.
PTC. 2016. Commuters at heart on recommendations to improve commuter’s journey. Tech-
nical Report. Singapore: Public Transport Council. https://www.ptc.gov.sg/.
Sayers, Michael W. 1989. “Two Quarter-Car Models for Defining Road
Roughness: Iri and Hri.” Transportation Research Record (1215): 165–172.
https://trid.trb.org/view.aspx?id=308315.
Sayers, Michael W. 1995. “On the calculation of International Roughness Index from longitu-
dinal road profile.” Transportation Research Record 1501: 1–12.
Sayers, Michael W. 1998. “The little book of profiling: basic information about measuring and
interpreting road profiles.” University of Michigan, Ann Arbor, Transportation Research
Institute .
Sekulić, D, and V Dedović. 2011. “Analysis of driver oscillatory comfort by simulation us-
ing a model with 6 DOF.” In Proceedings of the 23rd International Scientific–professional
Meeting: Science and Motor Vehicles, Special Conference for West Balkans Belgrade.
Sekulić, Dragan, Vlastimir Dedović, Srdjan Rusov, Aleksandar Obradović, and Slaviša Šalinić.
2016. “Definition and determination of the bus oscillatory comfort zones.” International
Journal of Industrial Ergonomics 53: 328–339.
Sekulić, Dragan, Vlastimir Dedović, Srdjan Rusov, Slaviša Šalinić, and Aleksandar Obradović.
2013. “Analysis of vibration effects on the comfort of intercity bus users by oscillatory model
with ten degrees of freedom.” Applied Mathematical Modelling 37 (18-19): 8629–8644.
Sekulić, Dragan, Sran Rusov, Vlastimir Dedović, Slaviša Šalinić, Dušan Mladenović, and Ivan
Ivković. 2018. “Analysis of bus users’ vibration exposure time.” International Journal of
Industrial Ergonomics 65: 26–35.
Sekulić, D., & Dedović, V. 2011. “The effect of stiffness and damping of the suspension system
elements on the optimisation of the vibrational behaviour of a bus.” International Journal
for Traffic & Transport Engineering 1 (4): 231–244.
Seraj, Fatjon, Berend Jan van der Zwaag, Arta Dilo, Tamara Luarasi, and Paul Havinga. 2015.
“RoADS: A road pavement monitoring system for anomaly detection using smart phones.”
In Big data analytics in the social and ubiquitous context, 128–146. Springer.
Ueckermann, Andreas. 2002. “The dynamic evenness index.” Forschung Strassenbau und
Strassenverkehrstechnik) 839.
Ueckermann, Andreas, and Markus Oeser. 2015. “Approaches for a 3D assessment of pave-
ment evenness data based on 3D vehicle models.” Journal of Traffic and Transportation
Engineering (English Edition) 2 (2): 68–80. http://dx.doi.org/10.1016/j.jtte.2015.02.002.
Wei, L., and J. Griffin. 1998. “The prediction of seat transmissibility from measures of seat
impedance.” Journal of Sound and Vibration 214 (1): 121–137.
Yu, Jianxiong, Eddie YJ Chou, and Jyh-Tyng Yau. 2006. “Development of speed-related ride
quality thresholds using international roughness index.” Transportation Research Record
1974 (1): 47–53.
ZF GmbH. 2009. “Product Overview: Axle and transmission systems for buses and coaches.”
20. https://www.zf.com/products/en/buses/products 40065.html.

14
Table 1. IRI thresholds at different speeds 10-60 km/h (Yu, Chou, and Yau 2006).

IRI Threshold at Different Speeds [m/km]

Ride quality 10 km/h 20 km/h 30 km/h 40 km/h 50 km/h 60 km/h
Very good < 11.44 < 5.72 < 3.80 < 2.86 < 2.28 < 1.90
Good 11.44-17.99 5.72-8.99 3.80-5.99 2.86-4.49 2.28-3.59 1.90-2.99
Fair 18.00-22.79 9.00-11.39 6.00-7.59 4.50-5.69 3.60-4.54 3.00-3.79
Mediocre 22.80- 32.32 11.40-16.16 7.60-10.80 5.70-8.08 4.55-6.25 3.80-5.40
Poor > 32.32 > 16.16 > 10.80 > 8.08 > 6.25 > 5.40

15
Table 2. IRI thresholds at different speeds 30-80 km/h (Cantisani and Loprencipe 2010).

IRI Threshold at Different Speeds [m/km]

Ride quality 30 km/h 40 km/h 50 km/h 60 km/h 70 km/h 80 km/h
Very good < 4.17 < 3.41 < 2.98 < 1.86 < 1.60 < 1.20
Good/Fair 4.17-8.34 3.41-6.83 2.98-5.95 1.87-3.73 1.60-3.20 1.42-2.84
Mediocre 8.34-11.92 6.83-9.75 5.95-8.51 3.73-5.33 3.20-4.58 2.84-4.06
Poor > 11.92 > 9.75 > 8.51 > 5.33 > 4.58 > 4.06

16
Table 3. RMS acceleration and levels of comfort accord-
ing to ISO 2631-1.

RMS vertical acceleration

Levels of comfort
awz (m/s2 )
less than 0.315 not uncomfortable
0.315 to 0.63 a little uncomfortable
0.5 to 1.0 fairly uncomfortable
0.8 to 1.6 uncomfortable
1.25 to 2.5 very uncomfortable
greater than 2.0 extremely uncomfortable

17
Table 4. Summary of road unevenness indices based on QCS (Múčka 2016a).

Unevenness Abbre- Vehicle Velocity Response

No
index viation model (km/h) category
100 - Ride comfort
Single-number
(car); - Ride safety
1 unevenness SEI 3-DOF QCS car
80 - Road dynamic load
indicator
(truck) - Cargo dynamic load
Truck Ride
2 TRI 3-DOF QCS truck 100 Ride comfort
Index
2-DOF QCS truck; 100 - Ride comfort
Longitudinal
5-DOF three-axle (car) - Ride safety
3 evenness LWI
semi-trailer; 80 - Road dynamic load
index
3-DOF QCS car (truck) - Cargo dynamic load
2-DOF QCS
4 Health Index HI 80 Ride comfort
(IRI)
Pavement 2-DOF QCS Suspension
5 PQI 80
quality index (IRI) dynamic load
Heavy
2-DOF QCS Suspension
6 articulated HATI 100
truck dynamic load
truck index
Vehicle
2-DOF QCS Suspension
7 response VRI Variable
(IRI) dynamic load
index

18
Table 5. Parametric review and analysis for quarter bus model at the rear axle.

2-DOF 6-DOF 3-DOF 10-DOF Proposed

65-DOF
Unit quarter half bus quarter full bus BRI
full buse
busa 18tb busc 18td (bus 19t)
Effective mass
kg n/a n/a 100 90 70 70
of seat and passenger
Seat suspension
N/m n/a n/a 25000 40000 20000 10000
stiffness
Seat suspension
N s/m n/a n/a 1000 220 220 330
damping
Sprung mass kg 4000 4355 4500 5300 n/a 5600
Suspension stiffness
N/m 320000 300000 300000 408650 n/a 664800
(equivalent)
Suspension damping
N s/m 10000 5000 20000 45973 n/a 63800
(equivalent)
Unsprung mass kg 550 250 500 677 n/a 660
Tire stiffness
N/m 1700000 1000000 1600000 2000000 2000000 2000000
(equivalent)
Tire damping
N s/m 150 300 300 300
(equivalent)
a (Agostinacchio, Ciampa, and Olita 2014), b (Fichera, Scionti, and Garescı̀ 2007), c (Sekulić and Dedović
2011), d (Sekulić et al. 2013), e (Sekulić et al. 2016), n/a: not available

19
Table 6. Parameter values for BRI compared to IRI and TRI.

Description Value of different vehicle types

Variables Unit
of parameter IRI - car TRI - truck BRI - bus
Effective mass of
seat and driver/passengera mp 0.067 0.01 -
Seat suspension
kp 8.26 1.78 s−2
stiffnessa
Seat suspension
cp 0.7 0.06 s−1
dampinga
Sprung massa Ms 1 1 1 -
Suspension stiffnessa K2 63.3 250 118.12 s−2
Suspension dampinga C 6 30 11.34 s−1
Unsprung massa µ 0.15 0.15 0.12 -
Tire stiffnessa K1 653 400 355.37 s−2
Tire enveloping
b 250 300 304 mm
(base length)
60 (urban) 40 (urban)
Travel Speed v 80 km/h
100 (rural) 60 (highway)
Ride comfort criteria BS6841:1997 ISO 2631-1
a Values have been normalised by the sprung mass, Ms.

20
Table 7. Gd (n0 ) and Gd (Ω0 ) according to ISO 8608.

Gd (n0 )(10−6 m3 ) Gd (Ω0 )(10−6 m3 )

Road class
Lower limit Upper limit Lower limit Upper limit
A - 32 - 2
B 32 128 2 8
C 128 512 8 32
D 512 2,048 32 128
E 2,048 8,192 128 512
F 8,192 32,768 512 2,048
G 32,768 131,072 2,048 8,192
H 131,072 - 8,192
n0 = 0.1 cycles/m Ω0 = 1 rad/m

21
Table 8. Different b1 values for BRI-IRI and awz -IRI regression in (Cantisani and Loprencipe 2010)

Coefficient b1 of Equation 6 at different speed levels (and R-squared value)

10 km/h 20 km/h 30 km/h 40 km/h 50 km/h 60 km/h 70 km/h 80 km/h
Car awz -IRI
(320m) - - 0.08 0.09 0.11 0.17 0.20 0.22
Bus BRI-IRI
(100m) 0.04 (0.73) 0.08 (0.79) 0.11 (0.82) 0.14 (0.82) 0.17 (0.81) 0.19 (0.82) 0.22 (0.82) 0.26 (0.80)
Bus BRI-IRI
(160m) 0.04 (0.73) 0.08 (0.80) 0.11 (0.85) 0.14 (0.80) 0.17 (0.85) 0.19 (0.80) 0.22 (0.84) 0.26 (0.76)
Bus BRI-IRI
(320m) 0.04 (0.73) 0.08 (0.84) 0.11 (0.86) 0.14 (0.86) 0.17 (0.86) 0.19 (0.85) 0.22 (0.85) 0.26 (0.83)

22
Table 9. BRI thresholds based on bus speed levels.

IRI thresholds at different speed levels [m/km]

Ride quality Ride comfort levels BRI (m/s2 )
10 km/h 20 km/h 30 km/h 40 km/h 50 km/h 60 km/h 70 km/h 80 km/h

Very good <7.88 <3.94 <2.86 <2.25 <1.85 <1.66 <1.43 <1.21 Comfortable <0.315
Little
Good/fair 7.88-15.75 3.94-7.88 2.86-5.73 2.25-4.50 1.85-3.71 1.66-3.32 1.43-2.86 1.21-2.42 0.315-0.63
Uncomfortable
Fairly
Mediocre 15.75-22.50 7.88-11.25 5.73-8.18 4.50-6.43 3.71-5.29 3.32-4.74 2.86-4.09 2.42-3.46 0.63-0.9
Uncomfortable
Poor >22.50 >11.25 >8.18 >6.43 >5.29 >4.74 >4.09 >3.46 Uncomfortable >0.9

23
Figure 1. 3-DOF QCS for BRI based on Citaro bus model.

24
 (a) As functions of temporal frequency. (b) As functions of spatial frequency (travel speed of
80km/h).

Figure 2. Transfer functions (TFs) of sprung mass over road input for BRI and TRI, and sprung mass over
the unsprung mass of IRI. BRI TFs are similar at different passenger loading factors (n = 0-1).

25
(a) On-cushion Svantek (b) The Citaro single-decker (c) The MAN NL323F.
SV100A. bus.

Figure 3. Field measurement setup.

26
 (a) Smooth road (section 8). (b) Rough Road (section 5).

Figure 4. Comparison of the simulated and measured awz (t) at different road sections.

27
(a) Artificially generated road profiles. (b) Road classification.

Figure 5. Examples of four artificial road profiles and their classification.

28
(a) Bus lane schemes with the highlight of full-day bus (b) Highly concentrated daily bus volumes.
lanes in city centre.

Figure 6. Singapore road network.

29
Figure 7. Distribution of IRI overall measured road profiles.

30
Figure 8. Relationship between IRI and BRI at different speeds for artificially generated profiles.

31
(a) BRI-IRI correlation at different speeds. (b) The scatter of BRI40 at speed of 40km/h.

Figure 9. Relationship between IRI and BRI for field-measured road profiles (100 m interval).

32
 Figure 10. Ride quality based on IRI and BRI.

33

View publication stats