Effect of rail unevenness correlation on the prediction of ground-borne vibration from railways
Invited paper
University of Southampton
Tuesday 2 june, 2015, 14:00 - 14:20
0.1 London (90)
Abstract:
Track and ground vibration is generated by the unevenness of the rails and
wheels. Usually in predicting ground vibration the two rails are assumed to
provide the same input. In practice their unevenness is not identical and
may be only partially correlated. A study is presented of the influence of
rail unevenness correlation on the predicted track and ground vibration.
This is based on an integrated railway model with varying complexity
describing the dynamic system of a ballasted track on layered half-space.
The model is formulated in the frequency-wavenumber domain and uses as
inputs the moving train axle loads and the PSD of the rail unevenness in
terms of the wavenumber along the railway track. In its basic formulation,
the track is coupled with the ground by assuming that the tractions are
uniformly distributed beneath the track, whereas the continuity of the track
and ground displacements is required only along the centreline of the track.
In order to investigate how ground vibration levels are influenced by taking
into account different correlation levels between the two rails, the
traction variation across the track-ground interface is included and the
track submodel is discretised laterally including both rails separately and
allowing for the pitching and/or bending motion of the sleepers. The focus
of the investigation is placed on the effect of the i) description of the
coupling between the track and the half-space, ii) modelling fidelity of the
railway track system and iii) statistical properties of the dynamic loadings
due to the irregularity of both rails on the predictions of the dynamic
response of the track and surrounding soil. The paper presents the effect of
the different modelling approaches on the response predictions and compares
the dynamic response calculated for a range of model/excitation parameters
highlighting the computational effort needed by each simulation.