Date of Graduation

2001

Document Type

Thesis

Degree Type

MS

Committee Chair

Ismail Celik

Abstract

Droplet formation is a phenomenon that has great significance in multiphase flows, and it takes place in many natural and man-made applications. The application of interest in this study is the turbulent mixing of Diesel fuel and seawater during the refueling process of naval ships outfitted with compensated fuel ballast tanks (CFBT). Several phenomena take place in fuel tanks that cause turbulent mixing and thus droplet formation. These phenomena include, among other phenomena, a stratified shear flow of two immiscible fluids of different densities, and a densely buoyant vertical jet flow of a higher density fluid impinging on a quiescent layer of lighter fluid. Turbulent mixing due to the impingement of a water jet on a stable stratified immiscible water-fuel interface is a complex process governed by many flow parameters and fluid properties. Primary flow parameters are the jet velocity, initial jet diameter, interface surface tension, buoyancy force, and inertial (shear) force which can be expressed by dimensionless parameters such as the jet Reynolds number, interface Richardson number, and Weber number. These parameters affect the penetration depth of the jet into the fuel layer and govern the amount of mixing and entrainment as well as the size of droplets formed at the interface. Finding an engineering model to describe the droplet formation and entrainment due to the turbulent mixing at a water-fuel interface is a challenging task. It is also difficult to find an appropriate turbulence model that is able to predict the correct turbulence quantities at the interface. The turbulence model should particularly account for the damping effect of stable stratification. Although mostly turbulent, the interface may have laminar regions away from the jet inlet, and it may also exhibit a relaminarization process. The present study is a part of a larger effort to improve the design of CFBT in cooperation with researchers at the Naval Surface Warfare Center-Carderock Division (NSWC-CD) Bethesda, MD. Faculty and students from Johns Hopkins University, in Baltimore, Maryland, are part of this study, and their primary goal is to study the canonical flows which comprise turbulent mixing events in compensated fuel/ballast tanks used in U.S. naval surface ships. The overall goal of this project at West Virginia University is to develop sub-models for the prediction of the extent and location of mixing of fuel and water, and to estimate the total flow-through time for the fuel, as well as the amount of water hideout. Water hideout is the amount of water that remains inside the fuel tanks after the refueling process has been completed, and presents a concern with regard to efficiency. The fuel/water mixing, on the other hand, represents an environmental concern, as some of the fuel may become entrained in the compensating water that is forced overboard during refueling. The prediction of the size and distribution of the fuel droplets that form during mixing is an integral part of the overall effort, both for accurate predictions of mixing events, as well as in estimating the net amount of fuel entrainment into the water layer. Several versions of a droplet formation model have been developed by the CFD group at West Virginia University, as an effort of several researchers under the supervision of Dr. I. Celik. The latest droplet formation model (DFM) version 3.1, gave the best results for the droplet size predictions in the shear flow simulations. This model has been tested and calibrated for the shear flow, and it has been used throughout this study for impinging jet phenomena as the sub-model for droplet size prediction. The commercial code FLOW-3D, from Flow Sciences corporation in Los Alamos, New Mexico, has been used to assess the performance of the droplet formation model, DFM V3.1, for the vertical buoyant jet phenomenon that takes place inside the fuel tanks. The jet impingement phenomenon has been simulated in a 2-D axisymmetric geometry, and the results using two turbulence models, namely standard k-ε and the RNG k-ε models, have been analyzed and compared with the experimental results obtained at Johns Hopkins University by Friedman and Katz (1999). Both models predicted the variation of the jet penetration depth with the interface Richardson number (which is inversely proportional to the jet velocity) very well. The RNG k-ε model gave better results for the turbulence quantities and the droplets size predictions. The standard k-ε model had problems predicting the turbulence quantities near the wall and in regions where there is little or no turbulence. FLOW-3D was also used to model a half-scale model of the fuel tank in three dimensions. DFM-V3.1 together with the RNG-K-ε model was used to dynamically calculate droplets formed inside the tank. Results with first and second order advection schemes for the volume fraction were compared to experiments. The results agree well with the experimental results, especially for the second order scheme which was able to decrease the numerical diffusion significantly.

Download

Share

COinS