Application of Large Eddy Simulation technique to a Navier-Stokes equation (with the Smagorinsky subgrid-scale model) on prediction of sediment concentration in non-homogeneous turbulent flow

Abstract

The rapid advancement of computer speed and storage capacity in recent years has realized the application of the Large Eddy Simulation (LES) technique to simulate practical in-situ turbulent flows (e.g. coastal tidal flows) with complicated bathymetry. The LES technique has also been analogically applied to close the sediment transport equations. In this thesis, a two dimensional LES turbulence model (Deep Bay model) has been developed to simulate the sediment transport phenomenon inside Deep Bay, Hong Kong. A two dimensional model is adopted because the horizontal dimensions (approximately 15 km x 8 km) are three orders of magnitude greater than the water depth (on average 3 m). The methodology for developing the LES model, including the numerical discretization schemes and establishment of initial and open boundary conditions, has been presented in detail. A number of sensitivity tests have also been performed and the sensitivity of each parameter tested has been assessed and quantified by utilizing correlation coefficient and scattered plots. The sediment settling velocity and critical shear stress were found to be the most sensitive parameters out of the six parameters tested. A main difficulty associated with LES modelling is the verification of the modelling results since measurements are usually insufficient for point to point comparison. In this thesis, in addition to comparing modelling results with on-site measurements at a few locations, turbulent kinetic energy power spectra and suspended solid (SS) concentration variance spectra converted from modelling results have also been studied thoroughly against the properties of the theoretically known turbulence power spectrum. A unique feature of Deep Bay is that there is a large intertidal area at the inner bay. The LES model has therefore been developed specifically to deal with this feature. Numerical instability of the SS concentration in the intertidal area was encountered and found to be associated with the abrupt change in water depth and model grid arrangements. The causes of the problems arising in the intertidal area have been identified and solutions have been suggested. The modelling results obtained from the Deep Bay model with or without tackling the flows and sediment transport phenomenon in the intertidal area have also been compared and discussed. The water and sediment mass fluxes were found to be reduced significantly by 11% and 43% if a minimum depth of 0.5m was assigned. It was concluded that properly tackling the flows and sediment transport simulation in the intertidal area was important for Deep Bay. The Hong Kong Shenzhen Western Corridor to be completed across Deep Bay has been incorporated in the Deep Bay model to further assess the accuracy of the model by comparing with the results stated in the Environmental Impact Assessment (EIA) report for the Western Corridor project approved by the Environmental Protection Department. The reduction in the flow flux across the alignment of the Western Corridor obtained from the modelling results was in good agreement with that stated in the EIA report. Assessment of the flushing ability of Deep Bay after construction of the Western Corridor by using the Deep Bay model indicated a negative result and hence the existence of the intertidal area of Deep Bay may be concluded to be shortened in a long run. These discussions and assessments are presented in Chapter 8. In conclusion, the Deep Bay model is verified to be able to generate turbulent kinetic energy spectrum and SS concentration variance spectrum in agreement with the theoretically known power spectrum. Similar overall SS concentration patterns inside Deep Bay are also able to be generated in comparison with satellite images. The modelling results at several locations have been plotted against on-site measurements. Same order of magnitude and similar trend have been achieved at some locations. These results are discussed and presented in detail in this thesis. Finally, recommendations for improving the Deep Bay model and for future research on computer simulation of sediment transport in coastal waters have been made. Several aspects have been suggested mainly including drying and wetting techniques, non-uniform grid system, particle size distribution and non-uniform critical shear stress.