Sedimentation has been one of the existing challenges in the sustainability of hydraulic structures. Proper modelling of sediment transport is, therefore, an important task for a designer or a project planner. Research has shown the complexity of using a 3D model because of data and knowledge requirements, computing demand, time and cost associated to it etc. At the same time, there are some limitations to a 1D model which demands the optimum use of 2D softwares for modelling. This study investigates sediment transport capability of HEC-RAS 2D 6.2 and the possibility of using 2D modelling tools for bed load transport. With the proper study of newly added feature in HEC-RAS 2D, the study asks: to what extent can the software represent transport and bed changes that occur in a river channel and its bends?A physical model was set up at Norsk Hydroteknisk Laboratorium, NTNU to evaluate the pattern of bed changes, efficiency of sediment bypass channel and the change in efficiency with the installation of guide walls. Several tests were made at varied flow conditions through inlet and through the bypass channel. Observations were made for proper evaluation and comparison.Based on the observations made in the physical model, a calibrated hydraulic model was set up. Using the calibrated model, sediment transport modelling was done for three different cases. An extra simulation was performed to check the effect of increased discharge through the sediment bypass channel. Analyses indicate that the simulations show a good agreement with observed model in the case of transport through river channel without any structures. It also demonstrates how the efficiency of a bypass channel can be increased using structures like inlet gate and guide walls. On this basis, proper transport modelling can be done with HEC-RAS for design and planning purposes. Improvements in limitations mentioned in the study could result in better simulations. Further research is needed to identify parameters and methods to strengthen the effectiveness and reduce uncertainties in modelling.