3. Method employing track measurement data to evaluate the internal condition of embankments

Rain water infiltration around boundary sections of ground and civil structures can cause loosening inside embankments. Repair work can be carried out if boring investigations from the front face of an abutment into the embankment show alteration, but these inspections are both time and money consuming. Consequently, a method has been proposed which can evaluate the inner condition of embankments from track measurement data and axle box (vertical) acceleration data.

Analysis of axle box acceleration data collected in a section where significant track subsidence had occurred due to internal alteration of an embankment on a boundary section showed that the PSD (power spectral density) of wave lengths in the region of 1-5 m differed before and after repair work.

In sum, it is believed that there are large localized shifts in the state of vertical track support in the rail direction in areas where the embankment has suffered internal alteration, which manifest themselves as significant changes in the track wave length.

Based on the above, the various sets of data were compared and verification was obtained that in zones where embankments had suffered internal alteration, there was significant vehicle response to this state in 5m chord longitudinal level irregularities and in axle box acceleration data. These indicators were used to propose a method for evaluating the internal condition of embankments. It was particularly clear in zones where internal embankment weakening had occurred that 5m chord longitudinal level growth in irregularity was significant and coincided with large standard deviation and maximum growth values. Therefore, focusing on the growth of 5m chord longitudinal level irregularity should provide an accurate means to determine presence of or not of embankment deterioration (Fig.1). It was also shown that if growth of 5m chord longitudinal level irregularity is expressed in terms of shift over time, it is possible to estimate the time at which alteration occurred (Fig.2).