Compositional gravity current flows produced by the instantaneous release of a finite-volume, heavier lock fluid in a rectangular horizontal plane channel are investigated using large eddy simulation. The first part of the paper focuses on the evolution of Boussinesq lock-exchange gravity currents with a large initial volume of the release during the slumping phase in which the front of the gravity current propagates with constant speed. High-resolution simulations are conducted for Grashof numbers $\sqrt {Gr}$ = 3150 (LGR simulation) and $\sqrt {Gr}$ = 126000 (HGR simulation). The Grashof number is defined with the channel depth h and the buoyancy velocity u_b = $\sqrt {g'h}$ (g′ is the reduced gravity). In the HGR simulation the flow is turbulent in the regions behind the two fronts. Compared to the LGR simulation, the interfacial billows lose their coherence much more rapidly (over less than 2.5h behind the front), which results in a much faster decay of the large-scale content and turbulence intensity in the trailing regions of the flow. A slightly tilted, stably stratified interface layer develops away from the two fronts. The concentration profiles across this layer can be approximated by a hyperbolic tangent function. In the HGR simulation the energy budget shows that for t > 18h/u_b the flow reaches a regime in which the total dissipation rate and the rates of change of the total potential and kinetic energies are constant in time. The second part of the paper focuses on the study of the transition of Boussinesq gravity currents with a small initial volume of the release to the buoyancy–inertia self-similar phase. When the existence of the back wall is communicated to the front, the front speed starts to decrease, and the current transitions to the buoyancy–inertia phase. Three high-resolution simulations are performed at Grashof numbers between $\sqrt {Gr}$ = 3 × 10⁴ and $\sqrt {Gr}$ = 9 × 10⁴. Additionally, a calculation at a much higher Grashof number ( $\sqrt {Gr}$ = 10⁶) is performed to understand the behaviour of a bottom-propagating current closer to the inviscid limit. The three-dimensional simulations correctly predict a front speed decrease proportional to t^−α (the time t is measured from the release time) over the buoyancy–inertia phase, with the constant α approaching the theoretical value of 1/3 as the current approaches the inviscid limit. At Grashof numbers for which $\sqrt {Gr}$ > 3 × 10⁴, the intensity of the turbulence in the near-wall region behind the front is large enough to induce the formation of a region containing streaks of low and high streamwise velocities. The streaks are present well into the buoyancy–inertia phase before the speed of the front decays below values at which the streaks can be sustained. The formation of the velocity streaks induces a streaky distribution of the bed friction velocity in the region immediately behind the front. This distribution becomes finer as the Grashof number increases. For simulations in which the only difference was the value of the Grashof number ( $\sqrt {Gr}$ = 4.7 × 10⁴ versus $\sqrt {Gr}$ = 10⁶), analysis of the non-dimensional bed friction velocity distributions shows that the capacity of the gravity current to entrain sediment from the bed increases with the Grashof number. Past the later stages of the transition to the buoyancy–inertia phase, the temporal variations of the potential energy, the kinetic energy and the integral of the total dissipation rate are logarithmic.

Lock-exchange gravity current flows produced by the instantaneous release of a heavy fluid are investigated by means of high-resolution two-dimensional (2D) large-eddy simulation (LES) using a dynamic subgrid scale model for Grashof numbers $\left(\mathsf{Gr}\right)$ up to ${10}^{12}$ . The model is validated using 2D direct numerical simulation results for the classical lock-exchange gravity current flow in an infinite channel with no-slip walls. Then the model is applied to study lock-exchange flows for the case in which the heavier (lock) fluid is initially situated in between a vertical end wall and the lock barrier. Three cases with different aspect ratios of the initial lock fluid are considered for Grashof numbers of the order of ${10}^9–{10}^{10}$ . It is found that 2D LES is able to capture most of the physics observed in experiments including the evolution of the head, as well as the formation of coherent billow structures at the apex of the head due to growth of the Kelvin-Helmholtz (KH) instabilities. However, due to the fact that three-dimensional instabilities cannot develop in 2D simulations, the coherence of these KH billows in the latter stages of the evolution of the current is found to be stronger compared to the experiment where mixing is sensibly higher at the interface between the current and the surrounding fluid. The LES simulations accurately capture the front and bore velocities in the slumping phase and predict that during the inviscid self-similar phase the front speed decay is proportional to $t^{-1∕3}$ , which is consistent with theory, where $t$ =time. A LES simulation at a Grashof number closer to the inviscid range $(\mathsf{Gr}={10}^{12})$ shows that the evolution of the current and the structure of the head region are qualitatively similar to those observed at lower Grashof numbers $(\mathsf{Gr}>{10}^8)$ . The energy balance shows that the higher front velocity in the $\mathsf{Gr}={10}^{12}$ simulation is due to the fact that the amount of mixing (dissipation) decays with the Grashof number. This allows a larger fraction of the potential energy to be converted into kinetic energy. The flow fields are also used to obtain the distributions of the bed shear stress which determine the amount of sediment entrained by a compositional gravity current propagating over a loose bed.

The Singapore Regional Waters (SRW) is one of the more complex tidal regions in the world. This complexity is caused by various factors including the interaction of the Indian and Pacific oceans with their mainly semi-diurnal and diurnal tide, respectively, complicated coastline geometry, small islands and sharply varying bottom topography. Tidal data analysis is hampered by the lack of reliable coastal stations with long-term water level records while numerical tidal modelling studies suffer from lack of accurate high-resolution bathymetry data and uncertainty in the prescription of the tidal open boundary forcing. The present study combines numerical modelling with available along-track satellite altimetry data and a limited set of reliable coastal stations. It proposes a structured approach to study the sensitivity of tidal propagation and interactions to parameters like the prescription of tidal forcing at the open ocean boundaries, local depth information and seabed roughness. To guide and facilitate this analysis, the open-source software environment OpenDA for sensitivity analysis and simultaneous parameter optimisation is used. In a user-controlled way, the vector difference error in tidal representation could so effectively be reduced by ~50%. The results confirm the benefit of using OpenDA in guiding the systematic exploration of the modelled tide and reducing the parameter uncertainties in different parts of the SRW region. OpenDA is also shown to reduce the repetitive nature of simultaneous parameter variation. Finally, the behaviour of the tide in the region and its sensitivities to changes in tidal boundary forcing and to local depth and friction variation in the narrow regions of the Malacca Strait is now much better understood. With most of the systematic errors reduced in the numerical model as a result of the sensitivity analysis, it is expected that the model can be applied to study tide-surge interaction and is much better suited for later application in combination with data assimilation techniques such as Kalman filtering for which systematic model errors should be minimal.

The formation, evolution, and structure of intrusive Boussinesq gravity currents propagating into a two-layer fluid with a sharp interface are investigated using highly resolved large eddy simulation at Reynolds numbers defined with the front velocity in the range of ${\mathrm{Re}}_f=3000–5000$ . Two cases are studied in which the density of the lock fluid is equal to the depth-weighted mean of the densities inside the two layers of ambient fluid. In the first one (case SC) the depths of the two layers of ambient fluid are equal. In the second one (case NSC) the depths of the two layers are different. The propagation speed in the slumping phase and the evolution of the currents are found to be close to both experiments and theory. Experimental observations showed the front continued to propagate for some time after the bore has caught the head region with the same constant velocity observed during the slumping phase. The present simulations show the reason is that unmixed lock fluid is still left inside the head region at the end of the slumping phase. This is in contrast to the case of gravity currents propagating over no-slip surfaces. It is also confirmed that even after the end of the slumping phase the local dissipation rate inside the region of unmixed fluid present inside the head remains very small compared to typical values observed in the interfacial layers behind the head. However, the size of the head and the region of unmixed fluid inside the head and dissipative-wake continue to decrease after the end of the slumping phase. The growth of three-dimensional (3D) instabilities at the front of the intrusion currents is observed to generate large-scale structures similar to the lobes observed in the case of gravity currents propagating over a no-slip surface. The dominant wavelength present in the initial growth stages of the 3D front instability in case SC is about half the one predicted by stability theory for a current propagating over a no-slip surface at the same Grashof number. The length scales of the front structures during the slumping phase are found to be smaller than the ones observed for currents propagating over no-slip surfaces at similar Grashof numbers.