Back-to-back MSE walls are a novel use of reinforced soil technology, and they are frequently implemented for bridge approaches and width-restricted highway and railway embankments. Urbanization has, however, led to an increase in the construction of transportation infrastructures. An investigation on model back-to-back MSE walls supporting railways has been carried out on a strong clay foundation. The foundation soil was clay with the desired shear strength. The model was conducted with a scale of 1/10th supporting railway tracks. Geogrid was used as a reinforcement, and wooden blocks were used as modular blocks for wall facings. The effects of different overlapping methods and distance between both walls on wall behavior have been evaluated. The scarcity of usable natural backfill soil for construction has been an alarming concern. Thus the recycled waste coal mine over dump was used as subballast/backfill soil. The coal mine overburden dump was used as a sustainable alternative to natural backfill/subballast. Cyclic loading simulating train loadings have been simulated in the model tests. Connected case of the model test was conducted in the laboratory. A finite element comparison of the model tests has also been conducted. A parametric study was carried out on back-to-back MSE walls subjected to heavy axle loads. Artificial intelligence-based ensemble models were used to predicted the geogrid tensile forces obtained from the parametric study.

Back-to-back mechanically stabilized earth walls (BBMSEWs) are a specialized type of reinforced soil structure widely employed in the stabilization of embankments, roads, and bridge abutments. Despite their prevalent use, the technical understanding of these structures, particularly their seismic performance, remains limited. Given the inherent randomness of earthquakes, the seismic response of structures is often evaluated probabilistically, with fragility curves serving as a popular tool for assessing the likelihood of varying degrees of damage or failure. In this study, a specific BBMSEW configuration is simulated and validated using FLAC2D software. Following this, nonlinear dynamic analyses are performed to develop both scalar and vector fragility curves, based on peak ground acceleration (PGA) and peak ground velocity (PGV), under far-field and near-field seismic conditions. The study further investigates the influence of metal strip overlap length on the vulnerability of these walls. The results not only facilitate the prediction of wall vulnerability across different seismic intensities but also reveal that increasing the overlap length of the metal strips from 0.65 to 0.85 times the wall height can reduce the probability of seismic damage by up to 35% in far-field earthquakes and up to 50% in near-field earthquakes. Moreover, the study finds that vector fragility curves provide a more realistic assessment compared to scalar fragility curves.