THE PREDICTION OF TRACK PERFORMANCE UNDER DYNAMIC TRAFFIC LOADING

Track performance was assessed in terms of vertical track settlement, and the factors that contribute to these settlements at specific field sites. The sites consisted of: (1) in-service revenue track having concrete and wood cross-ties and (2) wood and concrete tie sections from the DOT Facility for Accelerated Service Testing. Field tests were conducted to assess the physical states of the ballast materials prior to a track maintenance operation. The physical state tests were repeated after the surfacing to determine the resulting changes in the ballast properties. Standard penetration tests, Dutch cone tests, and undisturbed tube sampling for laboratory tests were used to investigate the subgrade. The dynamic wheel load distributions at each site were measured. A method was presented to account for the mix of wheel loads that contributed to the track deformations. The stress-dependent resilient moduli of the ballast materials were derived from repeated load triaxial tests, with and without shear stress reversals. The resilient moduli of the subgrades were determined from repeated load triaxial tests and site-specific correlations with the standard and Dutch cone penetration tests. A three dimensional, nonlinear, elastic, multilayer model, GEOTRACK, was used to estimate the stresses and deformations in the track foundations. The model was validated by comparing the predicted elastic foundation responses with those measured at FAST. Comparisons of measured and predicted track modulus were also made for the revenue field sites. The inelastic behavior of the soil was investigated using repeated load triaxial tests and a ballast box device for simulating field conditions. A strain superposition method was developed to account for the deformations that resulted from repeated applications of the mixed field loading conditions. The predicted values of track settlement were compared to field measurements obtained using soil strain gages and surveying from benchmarks. The track modulus and settlement predictions were of the same order of magnitude as the measured values. However, some differences in trends were observed. The study concludes with a discussion of the factors that were not accounted for in the methodology, and the implications of these factors in track performance