Safety of railway tracks in case of strong crosswind

1
Safety of railway tracks in case of strong
crosswind
Florence SOURGET SNCF, Research and Technology
Department
2
Safety of railway track in case of strong
crosswind

• Higher potential risk of overturning has been
  identified on the Mediterranean line.
• Influent factors are
• - Geographical location in a windy area,
• High viaducts and Embankment built on the line,
• The operational train speed which reaches 300
  km/h.

Paris
LN1
Lyon
Valence
LN5
Avignon
Marseille
First, risk due to crosswind had to be defined
and characterized, then appropriate protection
measures had to be taken for the safety level on
the Mediterranean line to be equivalent to other
high speed lines.
Montpellier
3
Safety of railway track in case of strong
crosswind
Vehicle sensitivity to crosswind
Line sensitivity to crosswind
4
Risk due to crosswind Vehicle sensitivity
Aerodynamic Coefficients
Safety criterion
?Q/Q0 ? 0,9
Wind Model
Railway dynamics simulations
CWC Characteristic Wind Curves
Mechanical Model
Configuration (flat ground, embankment, curve,
speed, )
5
Risk due to Crosswind Line sensitivity
Partition of the line
Identification of critical sites
Wind Modelling
Monitoring System
Safety Level
Protection Strategy
Definition and scaling of the protections
6
Risk due to crosswind Dynamic Protection System
(SPD)
LN5 12 zones, 1 anemometer /zone
7
Risk due to crosswind Pattern of the safety
analysis
Hazard event Vehicle overturning due to high
wind gust anywhere on the line

• Risk expressed by train.
• Risk expressed by element
• Partition of the line into elements of 50 metres
• length of the gust
  longitudinal component .
• A train running on LN5, meets different gusts one
  after the other.

Hazard event instantaneous joint realization of
two independent events.
8
Risk due to crosswind Probability of overturning
Critical Gust wind gust greater or equal to the
threshold wind value VCWC defined by the
Characteristic Wind Curves (for a given running
speed VT) plus a fixed wind margin ?V (estimated
in wind tunnel test). (VOverturningVCWC ?V).
Therefore P(R) P(RVT300) PVT(300)
P(RVT170) PVT(170) P(RVT80) PVT(80)
P300 PVT(300) P170 PVT(170)
P80 PVT(80)
With PVT conditional probability for a wind gust
to be critical given the train running
speed Pvtrain(VT) probability for a train to run
at the speed VT.
9
Risk due to crosswind Modelling of wind transfer
10
Risk due to crosswind Meteorological database

• Cumulative distribution function of the mean
  wind at anemometer

• Database of observed mean wind at MeteoFrance
  stations
• Law fitting on the data (Weibull) with Hazard
  Plotting
• Estimation of Bootstrap Confidence Interval for
  the distribution
• Selection of upper quartiles of the distribution

• Transfer coefficients anemometer ? line

• Estimation of the coefficients from data
  collected on temporary anemometers on the line
• Analysis of the variations which are taken into
  account as gaussian noise.

11
Risk due to crosswind Probability of critical
gust PVT
The probability of critical gust is estimated
from the previous wind modelling
12
Risk due to crosswind Uncertainties in the
Characteristic Wind Curves

• Uncertainties due to the aerodynamic
  coefficients measurements in wind tunnel tests
  are taken into account

• Polynomial modelling under constraints on
  regularity and passage by 0.
• Estimation of the modelling error
• Simulation of the coefficients from a Student law

• A new dimension is added into the integrals
  calculation

13
Risk due to crosswind Probability of running
speed Pvtrain(VT)
The running speed of the train depends on the
forecasting values of the mean wind at
anemometer. The conditional distribution of the
forecasted wind Vpredit given the real mean wind
is known through the formula Vpredit Vm
erprediction where erprediction is the
forecasting error is
distributed as a normal variable N(bias,
sprediction)
14
Risk due to crosswind Tests and results

• Two calculation codes
• Analytical resolution
• Monte Carlo Simulation
• Numerous configurations are tested
• Presence or absence of fences
• Presence or absence of the dynamical protection
  
  system
• Two wind margins 0 or 10 km/h
• Different traffic

15
Risk due to crosswind Results analysis

• Results are analysed
• in terms of importance
• in a relative way

• Two configurations are analysed wind margin 10
  km/h
• Deterministic Characteristic Wind Curves and
  upper quartile used for the mean wind
  distribution
• Probabilistic Characteristic Wind Curves and
  usual mean wind distribution

• Whatever the configuration
• Fixed protection (fences) reduce the risk by
  half.
• Complete protection system reduces the risk by a
  factor of 6.
• With the complete protection system, return
  periods are greater than 200 years (target
  reached)

16
Risk due to crosswind Conclusion

• Development of a complete methodology of risk
  analysis which includes
• meteorology (gust modeling and wind data)
• aerodynamics
• railway dynamics
• probability and risk analysis
• Results allow to determine
• the most dangerous sites where fences are a
  necessity.
• the contribution of the different protections on
  the level of safety.
• the level of safety which is ensured on the line
  and its coherence with the safety target.