Professional Documents
Culture Documents
TJRT A 16163291
TJRT A 16163291
TJRT A 16163291
net/publication/333051941
CITATION READS
1 709
3 authors, including:
Yanran Jiang
Monash University (Australia)
1 PUBLICATION 1 CITATION
SEE PROFILE
All content following this page was uploaded by Yanran Jiang on 04 August 2019.
1. Introduction
Train or rail transportation is favoured in many countries due to its high efficiency,
large capacity, comfort, and safety. There is an increasing number of passengers
choosing trains as their preferred mode of transportation. Thus, it is important to
improve passenger satisfaction on a continuous basis. Ride comfort is a vital metric
when evaluating the quality or experience of a passenger on board a train. The ride can
be affected by various factors, such as vibration, acoustic noise, smells, temperature,
humidity, visual stimuli and design layout [1–3]. In general, vibration caused by the
train’s motion is taken as the primary concern since it can be experienced by the
passenger through supporting surfaces, such as seats, backrests, and the floor while
standing or lying down [4]. The primary measure of vibration appears to be accelera-
tion and the ratings of ride comfort reflect the tolerance of passengers to acceleration,
while the tolerance for vibration exposure is related to duration and frequency [5]. The
longer the exposure time to vibration, the greater the discomfort of passengers.
Moreover, even if the intensities for all frequencies of vibration are equal, the sensitivity
of humans to vibration varies in different directions of motion in different parts of the
body [4,5]. One common method is to evaluate the level of ride comfort for the
passenger by using a weighted acceleration curve that considers human comfort
sensitivity [1].
Several methods have been introduced to evaluate the ride comfort of railway
vehicles and vibration in different countries and regions, such as ISO 2631, EN
12,299 and the Sperling’s method [2,6]. It is important for rail operators to comply
with prescribed standards to meet passenger expectations and maintain a high-quality
ride satisfaction. However, it is difficult to establish a universal set of comfort require-
ments because many factors may affect passengers’ perception of comfort, including
track conditions and vehicle operations which may vary in different countries [1,2,6–9].
Ride comfort indices vary depending on the existing standards or methods and the way
in which acceleration data are measured and processed. It is important to understand
the relationships between various ride comfort indices as well as the benefits and
limitations of different methods of evaluation.
In this paper, existing standards and methods are critically reviewed and a new
method based on previous experimental work is proposed. The R.M.S-based method
of evaluation is specified in ISO 2631–1, which is also the basis of the statistical
method developed in EN 12,299 [10–13]. During a train ride, there can be large
fluctuations in both acceleration and frequency levels experienced by occupants. ISO
2631–1 is suitable for evaluation of a motion environment with small variations in
levels while the statistical method introduced in EN 12,299 addresses the fluctuations
and variation associated with passengers [14]. The statistical method also avoids
sensitivity to artefactual extremes. Thus, compared with ISO 2631, EN 12,299 is
considered more precise and has been adopted by most countries for evaluation of
the levels of ride comfort [6,15]. An alternative method of evaluation of ride comfort
was proposed by Sperling, which is fundamentally different from the methods based
on ISO 2631. The Sperling method is more appropriate in some cases, especially
when comparing two or more different railway vehicles [8,16]. The Sperling’s ride
index (Wz) is determined for each direction using the frequency-weighted accelera-
tions which are different in three directions [16,17]. A suitable method for evaluating
ride comfort should take a range of conditions experienced by the test subjects into
consideration.
Limited research has been done to correlate various evaluation methods by using
different vibration models [1]. The ride comfort indices defined in ISO 2631 and EN
12,299 are commonly adopted in favour of the Sperling’s method is seldom applied
and discussed. Dumitriu and Leu evaluated ride comfort in railway vehicles on
a track with vertical irregularities by implementing two different comfort indices,
Nmvz and WZ, corresponding to the EN 12,299 and Sperling’s method respectively
[18]. Since there is no scale for the partial comfort index Nmvz to rate the comfort
level, the vertical ride comfort is assessed based on the range of Nmv which may lead
to inaccurate results [6]. Tengku compared the ride comfort level of passengers in
two positions, sitting and standing by using both the EN 12,299 (Nva, Nvd) and
Sperling’s method (WZ) [2]. The Ride Comfort Index discussed in both studies is
the Mean Comfort Index (Nmv, Nva, Nvd). Another frequently used Ride Comfort
Index in EN 12,299 is called the Continuous Comfort Index (CCx, CCy, CCz). This
index uses a quadratic average (r.m.s) of the frequency weighted accelerations
INTERNATIONAL JOURNAL OF RAIL TRANSPORTATION 3
measured to evaluate the Mean Comfort [10]. Since the mean comfort is determined
in the longitudinal, lateral, and vertical directions, respectively, and it has simila-
rities to the Sperling’s index (WZ).
The strengths and limitations of the Continuous Comfort indices, Mean Comfort
indices and the Sperling’s index in three directions are examined in this paper. The
outcome of this study will help identify the most appropriate comfort evaluation
method for various conditions and establish relationships between these ride comfort
indices. A new index based on Sperling method which combines an overall Sperling’s
index independent of the directions is proposed and validated. Here, an assessment is
conducted using data obtained from instrumented ride comfort wagons travelling on
a section of V/Line’s extensive train network.
2. Methodology
To assess the levels of ride comfort, the dynamic response of a passenger the vehicle
body along a typical train journey need to be measured by sensors. The measured
time-domain accelerations would be processed with the frequency weighting curves
corresponding to the different evaluation methods. The various ride comfort indices,
namely Continuous Comfort index (CC), Mean Comfort index (Nmv) and Sperling’s
index (WZ) were calculated by the frequency weighting accelerations according to
the related expressions. In this study, the Instrumented Passenger Vehicle (IPV) [15]
with a continuous ride monitoring system was used to collect the dynamic response
over a test section. This kind of test is well established in ISO 2631 and EN 12,299
[6]. A detailed description of the IPV, evaluation procedures, the frequency weight-
ing curves and the ride comfort indices are presented in the following section.
train station. Noise recognition algorithms were used to check the integrity of those
trip data so that the locations with high noise response could be identified.
Figure 1 shows the layout of the installed hardware system on the V/Line N-series
IPVs including tri-axial accelerometers, bogie suspension displacement, unsprung side
frame acceleration, various sensors, and differential GPS.
(Sperling’s methods 0.5-30 Hz, EN12999 0.4-80 Hz, ISO 1-80 Hz). The frequency
referenced accelerations were then weighted with appropriate frequency factor corre-
sponding to various ride comfort indices. The general evaluation procedures for
Sperling index (WZ), Mean Comfort index (Nmv) and Continuous Comfort index
(Cc) are shown in Figure 2. In addition, following the trip, each recorded event is
mapped against track chainage using the event time stamp and BRN43’s onboard GPS.
It is helpful to examine the track irregularities if the calculated ride comfort indices
exceed the thresholds. The evaluation method of long-term riding comfort over an
entire section is fundamentally based on ISO 2631 by the European Railroad Research
Institute [13,20] which evaluates human exposure to whole-body vibration.
H ð f Þ ¼ Hh ð f Þ � Hl ð f Þ � Ht ð f Þ � Hs ð f Þ (2)
" 2
#12
1:911f 2 þ ð0:25f 2 Þ
Bv ð f Þ ¼ 0:588 (8)
ð1 0:277f 2 Þ2 þ ð1:564f 0:0368f 3 Þ2
Thus,
Bh ð f Þ ¼ 1:25Bv ð f Þ (9)
Furthermore, the weighting factors are substantial at some specific frequencies, which
means human beings are more sensitive at these frequencies, as shown in Figure 5 [3].
The sensitive ranges for vertical and lateral vibrations in ISO 2631 are 4–12 Hz and
0.6–2 Hz, respectively. In UIC513R, it is 4–16.5 Hz and 0.6–2 Hz, respectively, for
vertical and lateral (EN 12,299 was developed based on UIC 513R). In Sperling’s
method, the range (3-7Hz) are the same for both directions.
Figure 5. Comparison of frequency weight functions for the (a) vertical direction and (b) lateral
direction.
CCx ðtÞ ¼ aW
XP ðt Þ ðLongitudinalÞ
d
(10)
CCy ðt Þ ¼ aW
YP ðt Þ ðLateralÞ
d
(11)
CCz ðt Þ ¼ aW
ZP ðt Þ ðVerticalÞ
b
(12)
where P denotes on the floor interface, X, Y and Z denotes the longitudinal, lateral,
vertical direction respectively. Wd is the frequency weighted value in the lateral/
longitudinal direction. Wb is the frequency weighted value in the vertical direction.
The frequency weighting r. m. s accelerations are expressed as
� �� �2 ��0:5
1 t
aW
Xi ðt Þ ¼
i
ò a ðt Þ dt ; T ¼ 5s (13)
T t T Wi
A preliminary scale to evaluate the ride comfort in the individual y and z directions are
given based on certain experiences (Table 2) with the index reported in two decimals [13].
The Mean Comfort Index (Nmv) is defined as the 95th percentile of five-minute
frequency weighted r. m. s values on the floor level, using the similar method as the
Continuous Comfort. The accelerations are measured in the longitudinal (x), lateral (y)
and vertical (z) directions, and then the five-second weighting r. m. s accelerations are
calculated for each direction over the entire tested track. The 95th percentile (i.e. the 4th
highest value) is selected and used for further processing. Finally, the 95th percentiles of
the weighted accelerations in the three directions are combined with an r. s. s (root-sum
-square) calculation as shown below [13]:
INTERNATIONAL JOURNAL OF RAIL TRANSPORTATION 11
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
�2 �2 �2
Nmv ¼ 6 � aW XP95
d
þ aW YP95
d
þ aW ZP95
b
(14)
�
Nmvy ¼ 6 � aW
YP95 ðLateralÞ
d
(16)
�
Nmvz ¼ 6 � aW
ZP95 ðVerticalÞ
k
(17)
A scale for the comfort index Nmv is given in Table 3 with the index reported for each
test zone in one decimal [13].
Sperling [5,22] introduced an alternative evaluation method which is different from
the ISO 2631 based methods. The ride comfort index is WZ, a frequency weighted r. m.
s value of acceleration evaluated over defined time intervals or a defined track section
[17]. For each frequency, WZ can be calculated using Equation (18).
�� 3 � �101
a
WZ ¼ 0:896 Fð f Þ (18)
f
where a is the amplitude of acceleration (cm/s2), f is the excitation frequency in Hz and
F ð f Þ is the frequency weighting function that expresses human vibration sensitivity.
This relation is only valid if the vibration is harmonic.
In reality, the vibration of vehicles is stochastic in nature, thus usually the electronic
instrumentations are used to evaluate the ride index. Equation (18) was rewritten concern-
ing the electronic parameters while without changing their contents, as shown below.
�6;67
1
WZ ¼ a2 Bð f Þ2 (19)
Since the vibration spectrum of the vehicle is not discrete, but a continuous function of
frequency, the total WZ can be obtained for a continuous spectrum by integration over
the given range of frequencies [5].
hf i6;67
1
WZ ¼ ò f1 Gð f ÞBð f Þ2 df
2
(21)
where f1 and f2 are the lower and upper limits of the range of frequencies considered.
Usually, the energy of vibration is limited to the interval between 0.5 and 30Hz,
ie. f1 ¼ 0:5; f2 ¼ 30. Gð f Þ is a two-sided square acceleration(cm/s2)2.
12 Y. JIANG ET AL.
3.1. Evaluation and comparison of the ride comfort level by different indices
The perceptions of ride comfort along the test line can be quantified based on the evaluation
procedures and functions described in the Methodology section. The Figure 6–8 show the
changes of Ride Comfort level along the test line when evaluated by WZ, Nmv, and CC in x,
y, z directions, respectively. It can be seen that vertical (z-direction) vibrations affect
passengers most while the discomfort brought by the longitudinal vibration in most cases
can be neglected. Generally, the vertical vibration is caused by the track irregularities.
Figure 6. Comparison of three different ride comfort indices in longitudinal (x) direction.
INTERNATIONAL JOURNAL OF RAIL TRANSPORTATION 13
Figure 7. Comparison of three different ride comfort indices in lateral (y) direction.
According to the preliminary scale for CCy and CCz, as shown in the last section (Table 2),
one location exceeding 0.4m/s2 may be classified as ‘less comfortable’ (uncomfortable) ride.
While there is no relationship between the partial ride comfort index (Nmvx, Nmvy, Nmvz) and
passengers’ feeling, the evaluation scale for Nmv could be used, i.e. passengers may feel
uncomfortable when the calculated ride comfort index in the location exceeds 3.5. A WZ
value of 2.5 is often compared with CC of 2.5 m/s2. As for motions and vibrations, WZ is often
considered as an acceptable value for ride comfort on trains [2].
When comparing the differences of Continuous Comfort index and partial Mean
Comfort index over a same operation distance, it can be found that the trip was
‘uncomfortable’ at some locations due to the lateral and vertical vibrations in terms
of the Continuous Comfort Index, while it was ‘comfortable’ over the whole trip in
terms of the Partial Mean Comfort Index. It should be noted that the partial Nmv and
CC do not always match, which has been highlighted by other researchers [10]. The
characteristic of the Mean Comfort index is to choose of 95th percentiles that is the 4th
highest value, which may lead to doubtful consequences in some cases. For example,
Table 5 provides the three series of hypothetical 5-minute vibration responses. They are
considered equally comfortable if only the 4th value is considered, but this result seems
unlikely. Thus, the continuous comfort index would be more appropriate for evaluating
the momentary riding at limited points.
14 Y. JIANG ET AL.
Figure 8. Comparison of three different ride comfort indices in vertical (z) direction.
Table 5. Three hypothetical five-minute vibration patterns for one direction (each of sixty-five-
second r. m. s values, m/s2).
First highest
r. m. s value 2th 3th 4th 5th ith 60th
Series A 0.5 0.5 0.4 0.4 0.1 0.1 0.1
Series B 0.5 0.5 0.4 0.4 0.4 0.4 0.4
Series C 0.9 0.9 0.9 0.4 0.4 0.4 0.4
When comparing the differences of Partial Mean comfort and Sperling’s index, it can
be found that Sperling’s index WZ is more sensitive to the level of ride comfort. The
level of ride comfort is considered as ‘comfortable’ when it is evaluated using Nmvz at
some locations in the vertical direction, while it is opposite to WZ. Besides, the results of
the Continuous Comfort index and Sperling’s index are not always matched as well.
It should be noted that in the Figures 7, 8 and 9, each bar of Continuous Comfort index
represents the comfort level over 5s while the Mean Comfort index and Sperling’s index
characterize the comfort on trains over 5 mins. Both Sperling and Mean Comfort criteria
give measures for ride comfort averaged over certain time or distance intervals, which
provide an overview of the variations in ride comfort over the complete journey and different
track sections along the line. However, the resulting comfort experienced at a single distinct
location cannot be measured. In this case, the Continuous Comfort Index (CC) is a good
INTERNATIONAL JOURNAL OF RAIL TRANSPORTATION 15
Figure 9. Evaluate the ride comfort level via the compound WZ and Nmv.
choice. Additionally, the Sperling’s index is an average of values for the different time or
distance intervals, which may fail to take the high values, i.e. the unpleasant events into
consideration. Such activities may occur at singular track defects, at switches and a curve
transition, etc. One advantage of WZ is that the scale is a pure number, which is convenient
to compare two or more different conditions [23]. Both WZ Ride Index and the Continuous
Comfort Index are determined for each direction. Even if the partial Mean Comfort can also
evaluate the ride comfort level in different directions, there is no precise scale for this index.
Furthermore, the Mean Comfort (Nmv) can be considered as the resultant expression of
Continuous Comfort CC in three directions, while there is no consequent index of three
directions (x, y, z) in Sperling’s method [21]. Therefore, there is a potential scope to develop
a new index with consideration of more than one direction based on Sperling’s index.
This new Sperling’s index is the root mean square of the values of Sperling’s index
under x, y, z directions over the same period. The test line of V/Line can be assessed by
the compound Sperling’s index and the Mean Comfort index rather than the partial
Mean Comfort indices, as shown in Figure 9. The results show that it is comfortable
along the line under the compound Sperling’s index, while it is ranked as uncomfor-
table at some locations under the Mean Comfort index.
In z-direction,
3:456
Nmvz ¼ 0:09899 � Wzz (25)
In the three directions,
Nmv ¼ 0:2881 � Wz 3:07 (26)
INTERNATIONAL JOURNAL OF RAIL TRANSPORTATION 17
These equations can be readily applied to assist the railway operators to assess the
trains comfortable level by using two different indices. The fitted curves were formed
to be reliable with 95% confidence. In the Figure 10(a-d), each point represents
a relation between one Mean Comfort index and one Sperling’s index over every
5-minute intervals. There is only a narrow range of the distributed points near the
fitted curves.
4. Conclusions
Methods for evaluating passenger Ride Comfort i.e. Continuous Comfort index, Mean
Comfort index and Sperling’s index have been examined. The results highlight both
benefits and limitations of employing various evaluating methods. Both of the Mean
Comfort index (Nmv) and Continuous Comfort index (CC) focussed on the local comfort,
while the Mean Comfort index only takes the 4th highest value in 5 mins. The selection of
Mean Comfort can help avoid extremes but may also lead to some doubtful consequences.
As such, the Continuous Comfort index is recommended for the momentary riding at
limited points. Apart from the Nmv and CC, the Sperling’s index (WZ) measures for ride
comfort averaged over certain time or distance intervals. This index is also an average of
values for the different time or distance intervals and as such, it does not take high values
into consideration. Since the original existing Sperling’s index is determined in individual
directions respectively, a new compound Sperling’s index expression was proposed to take
into consideration the three directions (x, y, z). The relationships between the Mean
Comfort index and Sperling’s index were established by correlating the data using the
least square method. The results of this work are readily applicable to rail operators for
assessing the ride comfort levels of trains on various tracks and routes. The new compound
Sperling’s index can be used to evaluate vehicles with different suspensions and seats since
the value represents an overall vibration level. The new compound Sperling’s index also
provides an added measure of ride comfort with reference to established indices that
correlate with ride quality. The maintenance and improvement plans could be designed
around calculated passenger ride comfort levels in which problematic track sections with
excessive vibration can be identified and upgraded.
Acknowledgments
The authors acknowledge the support from the Institute of Railway Technology, Monash
University.
Disclosure statement
�
No potential conflict of interest was reported by the authors.
References
[1] Kim Y-G, Kwon H-B, Kim S-W, et al. Correlation of ride comfort evaluation methods for
railway vehicles. Proc Inst Mech Eng F J Rail Rapid Transit. 2003;217:73–88.�
18 Y. JIANG ET AL.
[2] Munawir TIT, Samah AAA, Rosle MAA, et al. A comparison study on the assessment of
ride comfort for LRT passengers. IOP Conf Ser Mater Sci Eng. 2017;226:012039.
[3] Suzuki H. Effects of the range and frequency of vibrations on the momentary riding
comfort evaluation of a railway vehicle. Jap Psychol Res. 2002;40:156–165.
[4] Griffin MJ. Handbook of Human Vibration.� Londaon: Academic Press; 2012. �
[5] Knothe K, Stichel S. Rail vehicle dynamics [Internet].�Cham: Springer International
Publishing; 2017 [cited 2019 Mar 29]. Available from: https://www.springer.com/gp/
book/9783319453743 �
[6] Dumitriu M, Leu M. Correlation between ride comfort index and Sperling’s index for
evaluation ride comfort in railway vehicles. Appl Mech Mater. 2018;880:201–206.
[7] Choi I-Y, Um J-H, Lee JS, et al. Comparison of ride comfort between Korean and
European high speed railway by using continuous comfort method. J Korean Soc Railw.
2015;18:203–211.
[8] Gangadharan KV. Experimental and analytical ride comfort evaluation of a railway coach.
�
2003.
[9] Suzuki H. Research trends on riding comfort evaluation in Japan. Proc Inst Mech Eng
F J Rail Rapid Transit. 1998;212:61–72.
[10] Kufver B, Persson R, Wingren J. Certain aspects of the CEN standard for the evaluation of
ride comfort for rail passengers. Beijing, China: 2010 cited 2019 Mar 29. p. 605–614.
Available from: http://library.witpress.com/viewpaper.asp?pcode=CR10-056-1 �
[11] B153/RP1-RP20. Application of ISO standard to railway vehicles: comfort index Nmv -
comparison with the ISO/SNCF comfort note and with the Wz. ERRI; 1981.�
[12] UIC 513R. Guidelines for evaluating passenger comfort in relation to vibration in a railway
vehicle. International Union of Railways. UIC; 1994.�
[13] EN 12299:2009. Railway applications-ride comfort for passengers-measurements and eva-
luation. Brussels: CEN; 2009 April.
[14] Budiwantoro B, Masyhur AH, Suganda I. Testing of dynamic characteristic and comfort of
Indonesia automated people mover from Bandung. Surakarta, Indonesia: 2018. p. 040015.
doi:10.1063/1.5042985. �
[15] Thompson C, Hardie G, Shamdani A, et al. Continuous measurement of passenger ride
comfort and track condition on the V/Line network. Core 2018. 2018:622.
[16] Evaluation of passenger ride comfort of Indian rail and road vehicles with ISO 2631-1
Standards Part 2 - Simulation | scinapse | academic search engine for paper [Internet].
Scinapse. [cited 2019 Mar 29]. Available from: https://scinapse.io/papers/2464891650
[17] Narayanamoorthy R, Khan S, Berg M, et al. Determination of activity comfort in Swedish
passenger trains.:10.�
[18] Pradhan S, Samantaray AK, Bhattarcharyya R. Evaluation of ride comfort in a railway
passenger vehicle with integrated vehicle and human body bond graph model. Volume 12:
transportation Systems [Internet]. Tampa, Florida, USA: ASME; 2017. p. V012T16A013.
doi:10.1115/IMECE2017-71288
[19] Welcome to V/Line [Internet]. V/Line - regional public transport for Victoria. cited 2019
Apr 5. Available from: https://www.vline.com.au/.
[20] ISO 2631-4:2001. Mechanical vibration and shock-evaluation of human exposure to
whole-body vibration-part 1: general requirements. ISO; 2013.�
[21] Olmos JM, Astiz MÁ. Improvement of the lateral dynamic response of a high pier viaduct
under turbulent wind during the high-speed train travel. Eng Struct. 2018;165:368–385.
[22] Cheng Y-C, Hsu C-T. Running safety and comfort analysis of railway vehicles moving on
curved tracks. Int J Struc Stab Dyn. 2014;14:1450004.
[23] Kumar V, Rastogi V, Pathak P. Simulation for whole-body vibration to assess ride comfort
of a low–medium speed railway vehicle. Simulation. 2017;93:225–236.