چكيده به لاتين
One of the most important challenges in railway transportation is vibration caused by passing railway vehicles in adjacent residential, historical, and also sensitive buildings. These vibrations can lead to Residents' dissatisfaction, malfunction of sensitive devices, cracking and even damage to historical buildings. Nowadays due to the advancement of computer technology, numerical models are playing a crucial role in the prediction of railway-borne induced vibrations. In order to investigate the efficient numerical model, the model was validated in two steps, first by a previous study that was presented in the literature and second by experimental field tests. Then, a parametric study was carried out. Parameters of the parametric study were chosen based on the fact that one of the most serious challenges in ground-borne vibration modeling is the reflection of vibration from the boundaries of models. Following that, the effect of boundary conditions such as viscous boundary, infinite element boundary and fixed boundary, the effect of the width of the model and mesh sizes on computational cost and the model’s accuracy were investigated. Moreover, the application of asymmetric models has been investigated. Lastly, some contours were developed based on the parametric study to predict the optimal parameters. Results from the parametric study indicate that in time domain studies, the most accurate models were the models with viscous boundaries, and in the frequency domain the most accurate models were the models with infinite boundary conditions. Moreover, by increasing the width of models, the reduction rate of errors for fixed boundary models tends to decrease more in comparison with a model with other types of boundary conditions. This indicates the importance of the model's width in fixed boundary models. For instance, by increasing the width of the models from 35 to 55m (each side) a reduction of 110% and 10% were observed in models with fixed boundary models and infinite models, respectively. In models with a width of more than 60m (each side), the models with fixed boundary conditions can be associated with similar and even more accuracy to models with other boundary conditions. Changing the model's width is not impose a significant effect on models with viscous boundary conditions. Effects of mesh sizes are more significant in models with fixed boundaries, infinite boundaries and viscous boundaries, respectively. Furthermore, by increasing mesh sizes, the reduction rate of computational costs is less in models with fixed and infinite boundaries. Comparison between asymmetric and symmetric models shows that asymmetric models are associated with decent accuracy in time domain. However, when frequency domain response considers, the asymmetric models' responses are acceptable only when the models have infinite boundary conditions. Moreover, the models with fixed boundaries and infinite boundaries are much more predictable in terms of accuracy and computational costs.
