Bubbles induced drag reduction
A DAPRA sponsored project at reducing drag reduction on ship hulls.
We have studyed how micro-bubbles injected into turbulent flows may alter the near-wall flow. The injection of microbubble is known to reduce drag in experiments, but there has been little known theory about it. We developed a non-uniformed grid version of the front tracking code, improved the QUICK advection scheme, and implemented a non-conserved form to deal with the turbulent flow. Our numerical simulations reveal, for the first time, that slightly deformed bubbles can lead to significant reduction of the wall drag. The theory put forward there has subsequently been supported by experimental work in the Netherlands and has also been found to account well for observations made in the Navy’s Large Cavitations Channel in Tennessee. Paper on this work was published on the Physics of Fluids, and has been cited 123 times.
Bubbly flows in vertical channels
Sponsored by the Department of Energy (DOE)
we conducted simulations of nearly spherical or deformable buoyant bubbles in laminar or weakly turbulent channels, for both upflows and downflows. The most significant discovery here was the realization that bubbly flows in vertical channels have a relatively simple structure that greatly simplifies the predictions of the properties of such flows. It was a fundamentally new insight and has led to several papers published in the Physics of Fluids, the International Journal of Multiphase Flows, and Chemical Engineering Sciences. It also led to a follow-up funding from industry (Areva, Inc.) on related problems.
Nucleate Boiling Flows
Sponsored by Scandia National Laboratory
This project, funded by Scandia National Laboratory, is the big challenge for Direct Numerical Simulations of multiphase flows. We have successfully developed the algorithm of the front triangular meshes to deal with the phenomenon of breakup and coalescence of the vapor bubbles during the process of their generation, growth and departure from the wall. For the large density ratio of liquid-vapor system, we also improved the codes by using WENO scheme for the advection term and BiCGSTAB method for the pressure equation. We developed a simple multiscale model to capture the thin microlayer left behind when the vapor bubble expands. We have conducted simulations of the long-term dynamics of the repeated release of vapor bubbles from many nucleate sites.
Turbulent channel bubbly flows at a high Reynolds Number
Funded by the Consortium for Advanced Simulation of Light Water Reactor (CASL)
In this long-term project, which is funded by the Consortium for Advanced Simulation of Light Water Reactor (CASL), we have continued to simulated turbulent bubbly flows in vertical channels, but the friction Reynolds number for the channel is much higher (Re+=250). The flows included 140 bubbles and the domain was resolved by about 50 million fixed grid points in total. The simulations have been carried out using 192 processors. In addition to showing that the overall structure persists at higher Reynolds number, the results showed that the bubbles in the wall layer form horizontal clusters. A paper describing these results has been published in the Journal of Fluid Mechanics. We also started large scale simulations for a turbulent channel bubbly flow with Re+=500 and about 600 bubbles by using of 2048 processors on the Titan at ORNL. We have published a Dataset of turbulent bubbly channel flow at Re+=150, which can be download publicly through google drive at https://drive.google.com/drive/folders/0B9jFD9dpXXzOcjA4eTJMODliMm8?usp=sharing .
These direct numerical simulation (DNS) results are then used to help develop averaged models for bubbly flows. I worked with a graduate student to use a neural network to statistically learn the relationships between closure terms in averaged models and quantities that are available through DNS results. The model predictions are in reasonably good agreement with DNS results for laminar flows with and without walls. Two papers have been published in the Physics of Fluid and International Journal of multiphase flow.
Full Eulerian solver for the impact of a solid
project funded by NNSA’s Predictive Science Academic Alliance Program II (PSAAP II)
We have developed a finite volume implementation of the full Eulerian formulations for the macro-modeling of the impact of solid cylinder, incorporating the front tracking method. The governing equations include a single set of mass and momentum conservation equations for both the solid body and its surrounding air/fluid and evolution equations of inverse deformation tensors. A regular fixed structured grid is used for these equations, and a Weighted Essentially Non-Oscillatory (WENO) method is used for the advection terms and the standard centered difference for other terms in the equations. Time integration is done using a 3rd order Runge-Kutta method. A hyperelastic Cauchy stress model is used to connect the deformation gradient tensor to the momentum equation. The solid/fluid interface is tracked by the front-tracking method, where the interface is represented by marker points that are connected by lines (2D) or triangles (3D).
Simulated the synthesis of aluminum nanoparticles
In this project, simulation the synthesis of aluminum nanoparticles has been done by using the commercial CFD software, Fluent. We developed a nodal method to account for nucleation, coagulation and surface growth of nanoparticles. Then we implemented this method in Fluent by using serial/parallel user defined functions. We successfully implemented 2D axisymmetric simulations to study the optimum conditions for producing homogeneous size distribution of nanoparticles.
Other works
In addition to above simulations, other projects that I we have done involving numerical simulations of multiphase flows include:
(1). Written a parallel code for Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD) to reduce the high-dimensional data of turbulent bubbly flows.
(2). Used a PID controller or moving frame technique for a bubble/fluid system to calculate the lift and drag forces acting on a single bubbles rising in shear flows (Part of CASL project).
(3). Developed a 2D axisymmetric compressible front tracking code to examine the breakup of a fuel droplet in a supersonic nozzle.
(4). Combined the immersed boundary and the front tracking method for a consulting work on the modeling of flows in a Micro-nozzle for a startup company.
(5). Developed a topological change algorithm for a triangular mesh and 3D front-tracking code with coalescence and breakup of bubbles, and applied it to the simulations of boiling flows and turbulent bubbly flows.