Work Package 1 is devoted
to collection and analysis of railway network data on the large
space and time scales. The data consists of a wide range of
parameters including track design, geotechnical, maintenance, and
dynamic measurements. A major objective of this WP is to use the
collected data to identify the parameters that influence the
performance of the tracks. SNCF is the leader of this WP.
Sixteen sites of the Swedish, Spanish and French networks are
selected. For each of these sites, the following data concerning
track, traffic and geotechnical properties are collected
-
Track design: plan
design (curvature, alignment), slope, profile (embankment,
excavation)
-
Superstructure and
geotechnical properties: Type and mechanical properties of rail,
pad, fasteners, sleepers, sleepers spacing, ballast thickness,
“Los Angeles values and grain size distribution of ballast,
sub-ballast layer, form layer, underlying soil and substratum
characteristics,
-
Maintenance: maintenance
operations type (tamping, re-railing, grinding, etc.), geometrical
quality,
-
Traffic: maximum speed,
cumulated tonnage on a given time span,
-
Dynamic and static
measurements on several local sites: displacement, speed,
acceleration and constraints under train passing and long term
settlement evolution
Search
for correlation is based on three different types of statistical
analysis:
-
Person’s analysis method
in which correlations coefficients are calculated between
numerical values of the variates.
-
Spearman’s analysis
method in which correlations coefficients are calculated between
ranks of variates: the lowest value of the variate gets rank #0,
the following one ranks #1, and so on.
-
Multivariable factor
analysis
For
bi-variate analysis, the initial results have shown that, in spite
of some high values of correlation coefficients, relationships
between descriptive and maintenance variates are not significant.
Possible reasons could be:
-
Limited
number of the chosen track sites and compared to their high
heterogeneity
-
Important number of
variates
-
Insignificance of the
“tamping frequency” as maintenance indicator
|
