Investigating Scour Behaviour Around Culverts Using a Validated Numerical Model

One of the most prevalent damage mechanisms affecting waterway crossings is scour. Scour is the erosion of a waterway channel bed and banks due to the hydrostatic forces of the water flowing in the channel. Extensive damage to bridges, culverts and floodways during recent major flood events has been reported all across the Australia resulting in huge repair and replacement costs for road authorities in Australia. Scour is a complex mechanism and is dependent on local characteristics of water flows in a waterway and soil properties of the waterway’s bed and banks. Limited local research has been undertaken to study the effects of scour around culverts and also the numerical modelling of such effects. If accurate computer models can be developed then this would allow the scour processes to be studied more effectively. Modelling could reduce the need to undertake laboratory experimental work to study the scour mechanism. Future modelling could then be used to study the effects of different conditions applied to a culvert such as culvert geometry, flow rates, culvert blockages and protection treatments. This project aimed to develop a validated and accurate numerical model using ANSYS Fluent software to study the effects of scour around culverts. A series of laboratory experiments were developed to measure the scour around a small scale culvert model at different flow rates. Blockages of 0%, 50% and 100% were also applied to the culvert inlet to measure the effect of blockages to scour extent. These experiments provided measurable flow profiles and resulting bed scour profile from the flows. The small scale experiments showed that for unblocked and partial blocking of a culvert the resulting scour holes were more shallow and elongated to that of the fully blocked culvert. A fully blocked culvert resulted in sever scouring taking place closer to the end of the culvert at much lower flowrates to that of the unblocked and partially blocked scenarios. The data derived from these experiments was used as a basis to develop a numerical model using ANSYS Fluent software. Initial numerical modelling highlighted some issues with the ANSYS software using ‘multiphase’ models to determine flow profiles. ANSYS did not accurately model the water flow profiles compared to actual small scale experiment flow profiles. Findings indicated that the ANSYS Fluent software did not accurately represent the effect of friction loss due to bed roughness with the resulting profiles. More research was required to determine if ANSYS could model these profiles accurately. To enable a practical completion of the project within resource and time constraints, simplified single phase models were developed. Simplified 2D numerical models used to simulate the actual flow profiles. The flow profiles were created as fixed boundaries using the profile data from the small scale experiments. This would then ensure the flow profiles would be the same as that of the actual flows. Using these flows, the models were then used to measure values for velocity and bed shear stress that occurred along the bed of the channel. At critical values of velocity and bed shear stress, the soil particles that make up the bed would be overcome by the force of the water and movement of the particle would occur. This movement is the resulting erosive scour that occurs. The critical values of velocity and bed shear stress for movement to occur were calculated using empirical equations. The velocity and bed shear stress measured by ANSYS was then compared to the empirical values to determine the locations at which potential scour would occur. These locations were then compared to the actual locations to determine the accuracy of ANSYS predictions. The analysis demonstrated that based on the modelling, ANSYS could predict scour locations occurring in the vicinity of the actual locations but had a low reliability overall at predicting potential scour lengths. Larger ranges of data sets for flow and scour profiles are needed to determine the effectiveness of predicting scour extents using ANSYS in this way. Further refinement of the research done in this project is required to adequately assess the suitability of ANSYS Fluent software to simulate scour