Session D29: CFD: Applications II

3D simulation of liquid-gas flows with contact lines and applications to drop impact on a fiber

In this talk, we propose a numerical approach to simulate liquid-gas flows with contact lines, then, apply it to investigate drop impact on a horizontal fiber. This approach combines a conservative level set method to capture the interface, an immersed boundary method to represent the curved boundary, and a curvature boundary method to treat the contact lines. The simulation results are compared with experimental results. Two key aspects are investigated: the critical velocity for a drop to be captured by a fiber, and the topology of the impacting drop. Numerical results show good qualitative agreement with experiments in predicting the critical velocity. A regime map of drop topology based on the Weber number and drop-to-fiber diameter ratio is compared to the experimental results.   

High-performance sailboat hydrofoil optimization using vortex lattice methods, and the effects of free-stream turbulence

The identification of an optimized hydrofoil shape depends on an accurate characterization of both its geometry and the incoming, turbulent, free-stream flow. We analyze this dependence using the computationally inexpensive vortex lattice model implemented in AVL, coupled with the recently developed global, derivative-free optimization algorithm implemented in $\Delta-DOGS$. Particular attention will be given to the effect of the free-stream turbulence level --- as modeled by a change in the viscous drag coefficients --- on the optimized values of the parameters describing the three dimensional shape of the foil. Because the simplicity of AVL, when contrasted with more complex and computationally expensive LES or RANS models, may cast doubts on its usefulness, its validity and limitations will be discussed by comparison with water tank measurement, and again taking into account the effect of the uncertainty in the free-stream characterization.   

Application of RANS Simulations for Contact Time Predictions in Turbulent Reactor Tanks for Water Purification Process

California's current drought has renewed public interest in recycled water from Water Reclamation Plants (WRPs). It is critical that the recycled water meets public health standards. This project consists of simulating the transport of an instantaneous conservative tracer through the WRP chlorine contact tanks. Local recycled water regulations stipulate a minimum 90-minute modal contact time during disinfection at peak dry weather design flow. In-situ testing is extremely difficult given flowrate dependence on real world sewage line supply and recycled water demand. Given as-built drawings and operation parameters, the chlorine contact tanks are modeled to simulate extreme situations, which may not meet regulatory standards. The turbulent flow solutions are used as the basis to model the transport of a turbulently diffusing conservative tracer added instantaneously to the inlet of the reactors. This tracer simulates the transport through advection and dispersion of chlorine in the WRPs. Previous work validated the models against experimental data. The current work shows the predictive value of the simulations.   

Statistical data generated through CFD to aid in the scale-up of shear sensitive processes.

A number of industrial processes are considered shear-sensitive, where the product quality depends on achieving the right balance between mixing energy input and the resulting strain rate distribution in the process. Examples of such industrial processes are crystallization, flocculation and suspension polymerization. Scale-up of such processes are prone to a number of challenges including the optimization of mixing and shear rate distribution in the process. Computational Fluid Dynamics (CFD) can be a valuable tool to aid in the process scale-up; however for modeling purpose, the process will often need to be simplified appropriately to reduce the computational complexity. Commercial CFD tools with appropriate Lagrangian particle tracking models can be used to gather statistical data such as maximum strain rate distribution and maximum number of passes through a specific strain rate. This presentation will discuss such statistical tools and their application to a model scale-up problem   

Assess and improve the sustainability of water treatment facility using Computational Fluid Dynamics.

Fluids problems in water treatment industry are often simplified or omitted since the focus is usually on chemical process only. However hydraulics also plays an important role in determining effluent water quality. Recent studies have demonstrated that computational fluid dynamics (CFD) has the ability to simulate the physical and chemical processes in reactive flows in water treatment facilities, such as in chlorine and ozone disinfection tanks. This study presents the results from CFD simulations of reactive flow in an existing full-scale ozone disinfection tank and in potential designs. Through analysis of the simulation results, we found that baffling factor and CT10 are not optimal indicators of disinfection performance. We also found that the relationship between effluent CT (the product of disinfectant concentration and contact time) obtained from CT transport simulation and baffling factor depends on the location of ozone release. In addition, we analyzed the environmental and economic impacts of ozone disinfection tank designs and developed a composite indicator to quantify the sustainability of ozone disinfection tank in technological, environmental and economic dimensions.   

Transition to complex dynamics in the cubic lid-driven cavity

The cubic lid-drive cavity flow is simulated numerically by a Chebyshev collocation method, focusing on the onset of unsteadiness. The onset is directly to intermittent chaos. This has been reported by others in 2014, but they were unable to explain why, in such a simple geometry at quite modest Reynolds numbers ($Re \approx 1930$), the flow should go from being steady directly to intermittent chaos. In this presentation, we show that the reason has three components: instability of the steady flow is subcritical, breaking of the reflection symmetry about the spanwise midplane, and spanwise confinement. These ingredients lead to there being no locally attracting states for $Re$ beyond which the steady state loses stability, and the intermittent bursts are excursions shadowing unstable and stable manifolds of the (unstable) saddle local states. We show that this comes about because the instability of the steady state is close to a subcritical double Hopf bifurcation, and note that very recent theory on such bifurcations shows that they have associated dynamics that captures all the observed complexity in the cubic lid-driven cavity in the neighborhood of its primary instability.   

Numerical study and validation on a two-phase ejector flow using R134a refrigerant

An ejector is a pumping device that uses a low pressure jet flow to entrain a low-momentum secondary flow, and the two flows are mixed and pressurized in a mixing tube and a diffuser. When the ejector replaces an expansion valve in a standard refrigeration cycle, a compression work can be saved by the pumping effect and the efficiency of the cycle is known to be improved. However, the details of flow characteristics in the ejector are still unknown due to difficulties in experiments and complex flow phenomena. We numerically studied a supersonic ejector flow of R134a refrigerant, and validated the results against experimental data. As a results, we found that combinations of mixture, realizable k-epsilon, evaporation-condensation models, and energy equation are suitable to predict the ejector performance in a design point of view.   

Flow characteristics of infinite-span wings with wavy leading edges

Implicit LES computations are performed for an infinite-span wing based on the NACA0021 aerofoil section with a sinusoidal wavy leading edge (WLE). At $\text{Re}_\infty=1.2\times10^5$ and $\text{M}_\infty=0.3$, the computations performed in this study show that three-dimensional laminar separation bubbles (LSBs) form at troughs of the undulated wing. Prior to stall, LSBs can be found in all troughs. However, past the stall angle, LSBs tend to group together in a collocated fashion, leaving regions of complete separation in between groups where a separated shear layer (SSL) is formed. It is found that the size of the LSB group is highly dependent on the number of WLE wavelengths used in the spanwise-periodic domain. The LSB group formation process is investigated by means of simulations where the geometry is slowly pitched from an angle of attack of $\alpha=10^{\circ}$ to $\alpha=20^{\circ}$. The study also includes the analysis of instantaneous flow fields using Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD) techniques.