EN standard 12299 for evaluation of

ride comfort for rail passengers

Dr Bjrn Kufver (Ferroplan)

Contents

Introduction / Background (1)

principles
p
in the CEN comfort standard ((1))
Basic p
Mean comfort standard method NMV (4)
Continuous comfort CCx, CCy and CCz (1)
Mean comfort complete methods NVA and NVD (2)
Comfort on discrete events PDE (5)
Comfort on curve transitions PCT (3)
Discussion and conclusions (5)
B Kufver - Ferroplan

Introduction - 1
Th
The European
E
Committee
C
itt for
f Standardization
St d di ti (CEN) Technical Committee TC256 - European standards for
the railway sector
1999 - a European prestandard for comfort evaluation
ENV 12299
The research was conducted by UIC (ORE) and BRR
Revision performed by experts from France,
Germany, Italy, Sweden
A new standard EN 12299 was p
published in 2009

B Kufver - Ferroplan

Basic principles - 1
Indirect measurements
meas rements
Accelerometers and gyros vehicle body
Vehicle conditions accelerometer positions test
speed test sections time intervals
Full scale tests ((and computer
p
simulations))
Low-pass or band-pass filtering
Statistical post-processing
Scales / interpretation of results
Ride comfort as such / vehicle assessment

B Kufver - Ferroplan

The mean comfort standard

method - 1

Validated for seated passengers (UIC / ORE)

Calculated for a 5-minute run
Measurements in the floor
Accelerations in x-, y-, and z-directions
Band-passed filtered signals 0.4-100 Hz
validated for fairly straight tracks
3 * 60 5-second rms-values
95 percentile (4th highest value) from each direction

N MV = 6 (a

) + (a

Wd
2
XP 95

) + (a

Wd
2
YP 95

Wb
2
ZP 95

B Kufver - Ferroplan

The mean comfort standard

method - 2
Evaluation scale for NMV:
NMV < 1.5
1.5 < NMV < 2.5
2.5 < NMV < 3.5
3.5 < NMV < 4.5
NMV > 4.5

Veryy comfortable
Comfortable
Medium
Uncomfortable
Veryy uncomfortable

B Kufver - Ferroplan

The mean comfort standard

method - 3
Certain similarities with ISO 2631 evaluation
The controversial point is the 95 percentiles
In each direction onlyy 1 (of
( 60)) 5-second rms-values is
used

Table 1: Three hypothetical five-minute vibration patterns for one

direction (each of sixty five-second rms values, m/s2).
Series A
Series B
Series C

First highest rms value

0.3
0.3
0.9

2nd
0.3
0.3
0.9

3rd
0.3
0.3
0.9

4th
0.3
0.3
0.3

5th
0.1
0.3
0.3

ith
0.1
0.3
0.3

60th
0.1
0.3
0.3

B Kufver - Ferroplan

The mean comfort standard

method - 4
The 95 percentiles in x-, y-, and z-directions,
respectively, may occur during three different 5second
d iintervals.
t
l
The final NMV-value cannot be well correlated to
l
local
l ttrack
k condition
diti ((since
i
th
the critical
iti l llateral
t l yvalue and the critical vertical z-value may be
located several kilometres apart
apart.

B Kufver - Ferroplan

Continuous comfort CCx, CCy and CCz

Since NMV is based on only 3 of 180 rms
rms-vales,
vales there is a
substantial loss of information.
The CEN working group recommends that all 180 values
are presented
t d in
i th
the ttestt report,
t as three
th
time
ti
series:
i
Continuous Comfort.
A preliminary scale is suggested for evaluation of individual
rms-values
l
Table 1: Preliminary scale for the CCy(t) and CCz(t) comfort indexes
indexes.
CCy(t), CCz(t) < 0.20 m/s2
0.20 m/s2 < CCyy((t),
) CCz((t)) < 0.30 m/s2
0.30 m/s2 < CCy(t), CCz(t) < 0.40 m/s2
CCy(t), CCz(t) > 0.40 m/s2

Very comfortable
Comfortable
Medium
Less comfortable

B Kufver - Ferroplan

The mean comfort complete

methods - 1
Th NVA method
The
th d ttakes
k vibrations
ib ti
b
both
th att th
the flfloor and
d att
the seat into account.
Floor: vertical direction
Seat pan: lateral and vertical directions
Seat back: longitudinal direction
Based on 95 percentiles
More
M
cumbersome
b
tto use, both
b th iin reall ttests
t and
d
computer experiments

B Kufver - Ferroplan

The mean comfort complete

methods - 2
The NVD method
th d iis validated
lid t d ffor standing
t di passengers.
Floor: x-, y-, z-directions, median values
Floor: lateral y-direction,
y direction maximum value
Too sensitive to outliers ? (ORE)

Maximum value replaced with 95 percentile (ORE)

Both complete methods NVA and NVD have the same

disadvantages as the standard method NMV

B Kufver - Ferroplan

Comfort on Discrete Events PDE - 1

Validated
V
lid t d for
f seated
t d and
d standing
t di passengers (BRR,
(BRR
additional tests conducted by UIC/ERRI)
Voting by test subjects on a scale
Very comfortable
Comfortable
Acceptable
Uncomfortable
U
f t bl
Very uncomfortable
Q
Quantifies
tifi the
th percentage
t
who
h voted
t d
Uncomfortable or Very uncomfortable
B Kufver - Ferroplan

Comfort on Discrete Events PDE - 2

Discomfort
Di
f was ffound
d on llarge track
k iirregularities
l i i (Di
(Discrete
Events; PDE) and on short transition curves (Curve
Transitions; PCT)
PDE is derived from conditions on straight track and circular
curves (based on a manual selection of peak
peak-to-peak
to peak
patterns of the lateral acceleration)
Mean lateral acceleration (due to curvature and cant)
Peak-to-peak lateral acceleration

PDE = max 16.62 &y&pp + 27

2 .01 &y&mean 37
3 .0;0

B Kufver - Ferroplan

Comfort on Discrete Events PDE - 3

ERRI suggested
d a more automatic
i evaluation
l i off Di
Discrete
Events (PDE) based on continuous evaluation of several
g
signals
t+

&y&2s (t ) =

1
T

T
2

*
&
&
y
P,
Wp ( )d
t

T
2

T
T
T
T
&y&pp (t )=max &y&P,* Wp ( ),
) t , t + , min &y&P,* Wp ( ),
) t , t +
2
2
2
2

PDE (t ) = max 16.62 &y&pp (t ) + 27.01 &y&2s (t ) 37.0;0

B Kufver - Ferroplan

Comfort on Discrete Events PDE - 4

For the
F
th assessmentt off a particular
ti l local
l
l eventt (which
( hi h will
ill affect
ff t th
the
two-second sliding window during more than 2 seconds), the local
maximum of PDE(t) shall be used)
PDE(t)
1,4

14
Lateral acc(t)

Acceleratio
on [m/s ]

12
1,2

12

2s average
2s peak-to-peak

10

PDE(t)

0,8

0,6

0,4

0,2

0
0

10

12

14

16

Time t [seconds]

B Kufver - Ferroplan

Comfort on Discrete Events PDE - 5

Originally,
O
i i ll th
the PDE functions
f
ti
were derived
d i d and
d validated
lid t d ffor circular
i l
curves and straight track only. PDE > 0 may be found in short
transition curves without large track irregularities.
PDE(t)

Lateral acc(t)

1,4

14

2s average
2s peak-to-peak
peak to peak

12
1,2

12

A
Acceleration
n [m/s ]

PDE(t)
1

10

08
0,8

0,6

0,4

0,2

0
0

10

12

14

16

Time t [seconds]

B Kufver - Ferroplan

Comfort on Curve Transitions

PCT - 1

PCT is
i d
derived
i d ffrom conditions
diti
on ttransitions
iti
curves off th
the
clothoid type, evaluation starting 1 seconds before the
transition curve to 1.6 seconds after the transition curve
Maximum lateral acceleration (averaged 1 second)
Maximum lateral jjerk ((averaged
g 1 second))
Maximum roll velocity (averaged 1 second)

PCT = max 28 .54 &y&1s

max

+ 20 .69 &y&&1s 11 .1);

) 0,
2 283
2.283
&
+ ( 27 .36 1s max )

B Kufver - Ferroplan

Comfort on Curve Transitions PCT - 2

t+

&y&1s (t )=

T
2

1
*
&y&Wp
( )d
T T
t

1
T
T
&y&&1s (t )= &y&1s (t + ) &y&1s (t )
2
2
T
t+

1
&1s (t )=
T

T
2

*
&

Wp ( )d
t

T
2

B Kufver - Ferroplan

Comfort on Curve Transitions PCT - 3

PCT (t ) = max {0; ( 28 .54 &y&1s (t ) ,

+ 20 .69 max (sign

i ( &y&1s (t )) &y&&1s ( ),
) ]t TA 2.6s, t ])},
+ ( 27 .36 max (&1s ( ) , ]t TA 1.6s, t ])) 2.283

The parameter
Th
t TA (seconds)
(
d ) should
h ld b
be chosen
h
llarge enough
h tto allow
ll
hi
high
h llateral
t l
jerk and high roll velocity to affect the evaluation even if they occur in the
beginning of a long transition curve,
but small enough in order to exclude these values when they do not belong to the
same transition as the lateral acceleration at the time t.

B Kufver - Ferroplan

Discussion - 1
The new EN 12299 (as well as the old ENV
12299) is based on research from
UIC/ORE/ERRI and BRR.
Missing knowledge #1: Monetary
assessment
Missing knowledge #2: Motion sickness

B Kufver - Ferroplan

Discussion - 2
The NMV, NVA and NVD methods:
Believed to be valid on fairly straight lines
(b t nott on curves))
(but
neglect up to 98.3% (59 of 60) of the
measured rms-values
((may)
y) combine longitudinal
g
((x),
), lateral (y)
and vertical (z) vibration values from three
different 5-second intervals.

B Kufver - Ferroplan

Discussion - 3
PCT and PDE methods:
PCT - clothoids and linear cant transitions
only
PCT - very short straight lines or circular
curves ?
Derived from the same tests and the using
almost but not exactly - the same postprocessing (see next slide)

B Kufver - Ferroplan

Discussion - 4
t+

T
t+
2

1
*
&y&1s (t )= &y&Wp
( )d
T T
t

&y&&1s (t )=

1
T
T
&
&
&
&
y1s (t + ) y1s (t )
T
2
2
t+

1
&1s (t )=
T

&y&2s (t ) =

1
T

T
2

*
&
&
y
P,Wp ( )d
t

T
2

T
T
&y&pp (t )=max &y&P,* Wp ( ), t , t + ,
2
2

T
T
min &y&P,* Wp ( ), t , t +
2
2

T
2

*
&

Wp ( )d
t

T
2

B Kufver - Ferroplan

Discussion - 5
Even if a new European standard has been
published,
there is still room for further research in
the area of ride comfort evaluation

B Kufver - Ferroplan