Modeling of Subcooled Flow Boiling and Other Heat Transfer and Fluid Flow Phenomena by Matthew P. Wilcox A Thesis Submitted to the Graduate Faculty of Rensselaer Polytechnic Institute in Partial Fulfillment of the Requirements for the degree of MASTER OF SCIENCE Major Subject: MECHANICAL ENGINEERING Approved: _________________________________________ Ernesto Gutierrez-Miravete, Thesis Adviser Rensselaer Polytechnic Institute Hartford, Connecticut March 2013 i © Copyright 2013 By Matthew P. Wilcox All Rights Reserved ii TABLE OF CONTENTS TABLE OF CONTENTS ................................................................................................. iii LIST OF TABLES ............................................................................................................. v LIST OF FIGURES .......................................................................................................... vi ABSTRACT .................................................................................................................... vii 1. INTRODUCTION ....................................................................................................... 1 1.1 RESEARCH ....................................................................................................... 2 1.2 CONTENT ......................................................................................................... 4 2. HEAT TRANSFER AND FLUID FLOW: THEORY ............................................... 5 2.1 GOVERNING EQUATIONS ............................................................................ 5 2.2 NUMERICAL METHODS................................................................................ 7 2.3 NATURAL CONVECTION .............................................................................. 9 2.4 LAMINAR FLOW ........................................................................................... 11 2.5 TURBULENT FLOW ...................................................................................... 12 2.5.1 CALCULATING TURBULENCE PARAMETERS .......................... 14 2.6 TWO-PHASE FLOW ...................................................................................... 16 2.7 BOILING HEAT TRANSFER ........................................................................ 19 3. HEAT TRANSFER AND FLUID FLOW: MODELING ........................................ 21 3.1 NATURAL CONVECTION ............................................................................ 21 3.1.1 HORIZONTAL CYLINDER ............................................................... 21 3.1.2 VERTICAL PLATE ............................................................................ 28 3.2 LAMINAR FLOW ........................................................................................... 34 3.3 TURBULENT FLOW ...................................................................................... 38 3.4 3.3.1 TURBULENT FLOW WITHOUT HEAT TRANSFER ..................... 38 3.3.2 TURBULENT FLOW WITH HEAT TRANSFER ............................. 43 TWO-PHASE FLOW ...................................................................................... 47 3.4.1 GAS MIXING TANK .......................................................................... 47 iii 3.5 3.4.2 BUBBLE COLUMN ............................................................................ 51 3.4.3 BUBBLE COLUMN WITH POPULATION BALANCE MODEL ... 57 BOILING FLOWS ........................................................................................... 61 3.5.1 POOL BOILING .................................................................................. 61 3.5.2 SUBCOOLED BOILING .................................................................... 66 4. DISUSSION AND CONCLUSIONS ........................................................................ 79 REFERENCES ................................................................................................................ 81 APPENDIX A: MULTIPHASE FLOW MODELS ........................................................ 83 A.1 Volume of Fluid Model .................................................................................... 83 A.2 Mixture Model ................................................................................................. 83 APPENDIX B: POPULATION BALANCE MODEL................................................... 84 B.1 EQUATION FORMULATION ....................................................................... 85 B.1.1 Particle State Vector ......................................................................................... 85 B.1.2 Population Balance Equation ............................................................... 86 APPENDIX C: SUBCOOLED BOILING ..................................................................... 87 iv LIST OF TABLES Table 2.4.1-1: Turbulent Flow Input ............................................................................... 15 Table 2.5.1-2: Turbulence Parameter Calculation ........................................................... 15 Table 3.1.1-1: Horizontal Cylinder Input ........................................................................ 22 Table 3.1.1-2: Horizontal Cylinder Water Density ......................................................... 22 Table 3.1.1-1: Mesh Validation for Horizontal Cylinder ................................................ 27 Table 3.1.2-1: Vertical Plate Input .................................................................................. 28 Table 3.1.2-2: Vertical Plate Water Density .................................................................... 29 Table 3.1.2-1: Mesh Validation for Vertical Plate .......................................................... 33 Table 3.2-1: Laminar Flow Input..................................................................................... 34 Table 3.2-2: Laminar Flow Water Density ...................................................................... 34 Table 3.2-3: Mesh Validation for Laminar Flow ............................................................. 37 Table 3.3.1-1: Turbulent Flow Without Heat Transfer Input .......................................... 38 Table 3.3.2-1: Turbulent Flow With Heat Transfer Input ............................................... 43 Table 3.3.2-2: Turbulent Flow Water Density................................................................. 44 Table 3.3.2-3: Mesh Validation for Turbulence With Heat Transfer .............................. 46 Table 3.4.1-1: Gas Mixing Tank Input ............................................................................ 48 Table 3.4.1-2: Mesh Validation for Gas Mixing Tank .................................................... 50 Table 3.4.2-1: Bubble Column Input ............................................................................... 52 Table 3.4.2-2: Mesh Validation for Bubble Column ....................................................... 56 Table 3.4.3-1: Population Balance Model Input .............................................................. 57 Table 3.4.3-2: Bubble Size Distribution – Surface Tension of 0.072 N/m ..................... 60 Table 3.4.3-3: Bubble Size Distribution – Surface Tension of 0.0072 N/m ................... 60 Table 3.5.1-1: Pool Boiling Input .................................................................................... 61 Table 3.5.1-2: Pool Boiling Water Density ..................................................................... 62 Table 3.5.1-3: Mesh Validation for Pool Boiling ............................................................ 65 Table 3.5.2-1: Subcooled Flow Boiling Input ................................................................. 68 Table 3.5.2-2: Subcooled Boiling Water Properties ........................................................ 69 Table 3.5.2-3: Boiling Model Case Input ........................................................................ 70 Table 3.5.2-4: Boiling Model Case Results ..................................................................... 72 Table 3.5.2-5: Input Parametric Case Matrix .................................................................. 74 v Table 3.5.2-6: Input Parametric Case Results ................................................................. 74 Table 3.5.2-7: Axial Height Liquid Volume Fraction ..................................................... 75 Table 3.5.2-8: Absolute Impact on Liquid Volume Fraction .......................................... 77 Table 3.5.2-9: Relative Impact on Liquid Volume Fraction ........................................... 78 Table 3.5.2-10: Mesh Validation for Subcooled Boiling ................................................ 78 vi LIST OF FIGURES Figure 2.5-1: Example of Turbulent Flow ....................................................................... 12 Figure 2.6-1: Flow Regimes ............................................................................................ 16 Figure 2.6-2: Baker Flow Pattern .................................................................................... 17 Figure 2.7-1: Boiling Heat Transfer Regimes ................................................................. 19 Figure 3.1.1-1: Horizontal Cylinder Temperature ........................................................... 21 Figure 3.1.1-2: Horizontal Cylinder Density ................................................................... 23 Figure 3.1.1-3: Horizontal Cylinder Velocity Vector ...................................................... 24 Figure 3.1.1-4: Interference Fringes Around a Heated Horizontal Cylinder ................... 24 Figure 3.1.1-5: Temperature at θ = 30° Vs. Radial Distance .......................................... 25 Figure 3.1.1-6: Temperature at θ = 90° Vs. Radial Distance .......................................... 26 Figure 3.1.1-7: Temperature at θ = 180° Vs. Radial Distance ........................................ 26 Figure 3.1.2-1: Vertical Plate Temperature ..................................................................... 29 Figure 3.1.2-2: Vertical Plate Velocity Vector Plot ........................................................ 30 Figure 3.1.2-3: Interference Fringes Around a Heated Vertical Plate ............................. 31 Figure 3.1.2-4: Dimensionless Temperature as a Function of Prandtl Number .............. 32 Figure 3.1.2-5: Dimensionless Temperature as a Function of Prandtl Number .............. 32 Figure 3.2-1: Laminar Flow Velocity Profile Vs. Positon............................................... 35 Figure 3.3.1-1: Turbulent Flow Velocity Magnitude Vs. Position .................................. 39 Figure 3.3.1-2: Wall Shear Stress Vs. Position ............................................................... 40 Figure 3.3.1-3: Radial Velocity ...................................................................................... 40 Figure 3.3.1-4: dAxial-Velocity/dx Vs. Position ............................................................. 40 Figure 3.3.1-5: Flow Results for Mass Flow Rate of 0.5 kg/s ......................................... 41 Figure 3.3.1-6: Flow Results for Larger Mass Flow Rate of 1.5 kg/s ............................. 41 Figure 3.3.1-7: Turbulent Kinetic Energy ....................................................................... 42 Figure 3.3.1-8: Production of Turbulent Kinetic Energy ................................................ 42 Figure 3.3.2-1: Temperature ............................................................................................ 44 Figure 3.3.2-2: Radial Velocity ....................................................................................... 44 Figure 3.3.2-3: Velocity Magnitude Vs. Position ............................................................ 45 Figure 3.3.2-4: Wall Shear Stress Vs. Axial Position...................................................... 45 Figure 3.4.1-1: Gas Volume Fraction .............................................................................. 49 vii Figure 3.4.1-2: Liquid Vector Velocity ........................................................................... 49 Figure 3.4.1-3: Gas Vector Velocity................................................................................ 50 Figure 3.4.2-1: Bubble Column Reactor.......................................................................... 51 Figure 3.4.2-2: Instantaneous Gas Volume Fraction ....................................................... 53 Figure 3.4.2-3: Instantaneous Liquid Velocity Vectors................................................... 54 Figure 3.4.2-4: Instantaneous Gas Velocity Vectors ....................................................... 55 Figure 3.4.2-5: Instantaneous Gas Volume Fraction (10 cm/s) ....................................... 56 Figure 3.4.3-1: Instantaneous Gas Volume Fraction with PBM ..................................... 58 Figure 3.4.3-2: Bubble Column Liquid Vector Velocity with PBM ............................... 59 Figure 3.4.3-3: Bubble Column Liquid Vector Velocity with PBM ............................... 60 Figure 3.5.1-1: Instantaneous Gas Volume Fraction ....................................................... 63 Figure 3.5.1-2: Instantaneous Liquid Velocity Vectors................................................... 64 Figure 3.5.1-3: Instantaneous Gas Velocity Vectors ....................................................... 64 Figure 3.5.1-4: Volume Fraction of Vapor on Heated Surface ....................................... 65 Figure 3.5.2-1: Case 1 - Temperature (K) ....................................................................... 70 Figure 3.5.2-2: Case 1 - Liquid Volume Fraction ........................................................... 71 Figure 3.5.2-3: Case 1 - Mass Transfer Rate (kg/m3-s) ................................................... 71 Figure 3.5.2-4: Liquid Volume Faction Vs. Position for Cases 1-6 ................................ 73 Figure 3.5.2-5: Liquid Volume Faction Vs. Position for Cases 7-12 .............................. 76 Figure 3.5.2-1: Void Fraction in Various Boiling Regimes ............................................ 87 viii ABSTRACT Investigations into various fluid flow and heat transfer regimes were modeled numerically to better understand the phenomena that occur during subcooled flow boiling. The theory of each fluid flow and heat transfer regime that occurs during subcooled flow boiling is discussed in detail and followed by the development a numerical model. Numerical models to analyze natural convection, laminar flow, turbulent flow with and without heat transfer, two-phase flow, pool boiling and subcooled flow boiling were created. The commercial software Fluent was used to produce the models and analyze the results. Different modeling techniques and numerical solvers were employed depending on the scenario. The results of each model were compared to experimental data when available to prove its validity. Although numerous heat transfer and fluid flow phenomena were analyzed, the primary focus of this research was subcooled flow boiling. The impact different boiling model options have on liquid volume fraction was investigated. Three bubble departure diameter models and two nucleation site density models were analyzed using the same initial conditions. The bubble departure diameter models did not show any relationship with liquid volume fraction; however, the Kocamustafaogullari-Ishii nucleation site density model tended to predict a greater liquid volume fraction, meaning less vapor production, than the Lemmert-Chawla nucleation site density model. A second study on how initial conditions impact the liquid volume fraction during subcooled flow boiling was explored. Cases were analyzed that increased or decreased the heat flux, the inlet temperature or the mass flow compared to a base case. The differences in liquid volume fraction between the cases were compared and relationships between heat flux, inlet temperature and mass flow rate with respect to liquid volume fraction were developed. Overall, the inlet temperature had the greatest impact on liquid volume fraction, the wall heat flux had the second greatest impact and mass flow rate had the smallest impact. ix 1. INTRODUCTION Since the 19th century, the world’s standard of living has greatly increased primarily due to the generation and distribution of electricity. Over 80% of the world’s electricity production is generated by converting thermal energy, from a fuel source into electrical energy. The Rankine Cycle is an energy conversion process where fuel is burned to heat water and form steam. The steam is used to turn a turbine which spins an electric generator. Electricity production involves numerous engineering processes but is primarily based around heat transfer and fluid flow. There are many different fuel sources available to electrical power plants such as coal, oil, natural gas and uranium. The fuel source in focus here will be uranium or nuclear fuel. Nuclear power plants harness energy released during fission to heat water. This water is then pumped through a heat exchanger to produce steam. The heat transfer mechanisms at work within a nuclear reactor core are extremely complex. All three major forms of heat transfer are at work; conduction, convection and radiation. The fluid flow through the reactor core is also complex due to the intense energy transfer and phase change. In Pressurizer Water Reactors, the water surrounding the reactor core is prevented from bulk boiling because it is highly pressurized; however, a small amount of localized boiling does occur. This is also known as subcooled flow boiling. This research focuses on the convective heat transfer and fluid flow phenomena that occur during subcooled flow boiling. Specifically, topics on turbulence, two-phase flow and phase change are discussed. Subcooled boiling occurs when an under-saturated fluid comes in contact with a surface that is hotter than its saturation temperature. Small bubbles form on the heated surface in locations called nucleation sites. The number of bubbles that form is heavily dependent on fluid inlet temperature, pressure, mass flow, heat flux and microscopic features of the surface. After the bubbles form on the heated surface, they detach and enter the bulk fluid. When this occurs, saturated steam is dispersed in a subcooled liquid which is where the term subcooled boiling originates. 1 1.1 RESEARCH Subcooled flow boiling is characterized by the combination of convection, turbulence, boiling and two-phase flow. Determining the amount of subcooled boiling that occurs is challenging and has become a topic of great interest in recent years. A number of mechanistic models for the prediction of wall heat flux and partitioning have been developed. One of the most commonly used mechanistic models for subcooled flow boiling was developed by Del Valle and Kenning. This model accounts for bubble dynamics at the heated wall using concepts developed initially by Graham and Hendricks for wall heat flux partitioning during nucleate pool boiling. More recently, a new approach to the partitioning of the wall heat flux has been proposed by Basu et al. The fundamental idea of this model is that all the energy from the wall is transferred to the liquid adjacent to the heated wall. Then, a fraction of the energy is transferred to vapor bubbles by evaporation while the remainder goes into the bulk liquid. [1] Additionally, focus has been put towards accurately modeling the three most impactful parameters in subcooled flow boiling. These parameters are the active nucleation site density (Na), departing bubble diameter (dbw) and bubble departure frequency (f). The two most common nucleation site density models were developed by Lemmert and Chwala and Kocamustafaogullari and Ishii. Both of these models are available in Fluent. Many correlations have been developed to determine the bubble departure diameter. Tolubinsky and Kostanchuk proposed the most simplistic correlation which evaluates bubble departure diameter as a function of subcooling temperature. Kocamustafaogullari and Ishii improved this model by including the contact angle of the bubble. Finally, Unal produced a comprehensive correlation which includes the effect of subcooling, the convection velocity and the heater wall properties. All three of these bubble departure diameter correlations are available in Fluent. The most common bubble departure frequency correlation for computational fluid dynamics was developed by Cole. It is based on a bubble departure diameter model and a balance between buoyancy and drag forces. The Cole bubble departure frequency model is available in Fluent. Another improvement that has been made in recent years to the modeling of subcooled flow boiling is the use of population balance equations (PBEs) to better 2 determine how swarms of bubbles interact after detaching from the heated surface. This is a relatively new way of investigating subcooled flow boiling that was recommended by Krepper et. al. [2] and investigated by Yeoh and Tu [1]. Population balance equations have been introduced in several branches of modern science, mainly areas with particulate entities such as chemistry and materials. These equations help define how particle size populations develop in specific properties over time. Population balance equations are available in Fluent but not in combination with the boiling model. 3 1.2 CONTENT This thesis produced an investigation on subcooled flow boiling using Fluent. Fluent is a widely accepted commercial computational fluid dynamics code that can simulate complex heat transfer and fluid flow regimes. This thesis had three major objectives. The first objective was to gain an understanding of the phenomena that occur during subcooled flow boiling. The second objective was to determine how the boiling model options described in Section 1.1 impact the liquid volume fraction at different axial locations. The third objective was to evaluate how heat flux, inlet temperature and mass flow rate impact the liquid volume fraction at different axial locations. Due to its complexity, development of the subcooled flow boiling model was performed in stages. With the development of each model, a more complex fluid flow or heat transfer scenario was analyzed. The first and simplest model created was for natural convection. The theory of natural convection is described in Section 2.3 and the analytical modeling results are presented in Section 3.1. Two natural convection geometries were analyzed. The first was a horizontal cylinder suspended in an infinite pool and the second was a vertical plate suspended in an infinite pool. The second model developed was a laminar flow model. The theory of laminar flow is described in Section 2.4 and the analytical modeling results are discussed in Section 3.2. The third model developed was a turbulent flow model. The theory of turbulent flow is described in Section 2.5 and the analytical modeling results are displayed in Section 3.3. Section 3.3 contains two turbulent flow scenarios; turbulent flow without heat transfer and turbulent flow with heat transfer. The fourth model developed was a water / air twophase flow model. The theory of two-phase flow is described in Section 2.6 and the analytical modeling results for the two scenarios analyzed are shown in Section 3.4. The first is a gas mixing tank and the second is a bubble column. The final and most complex model created includes a phase transformation (vaporization and condensation) model. The theory of boiling heat transfer is described in Section 2.7 and the analytical modeling results are presented in Section 3.5. Two models were created, the first is for pool boiling and the second is for subcooled flow boiling. After each model was created, a mesh validation was performed and the results were compared to known experimental data when possible to validate the information generated by Fluent (CFD). 4 2. HEAT TRANSFER AND FLUID FLOW: THEORY This section discusses basic theory behind some common heat transfer and fluid flow scenarios. It is meant to provide a background on the phenomena involved in subcooled flow boiling. 2.1 GOVERNING EQUATIONS Conservation equations are a local form of conservation laws which state that mass, energy and momentum as well as other natural quantities must be conserved. A number of physical phenomena may be described using these equations [3]. In fluid dynamics, the two key conservation equations are the conservation of mass and the conservation of momentum. Conservation of Mass in Vector Form (continuity equation): ππ β β πv + (∇ β)= 0 ππ‘ Conservation of Mass in Cartesian Form: ππ π π π (ππ£π₯ ) + (ππ£π ) + (ππ£π§ ) = 0 + ππ‘ ππ₯ ππ¦ ππ§ Conservation of Momentum in Vector Form: π π·v β β π + π∇ β 2v = −∇ β + ππ π·π‘ Conservation of Momentum in Cartesian Form: ππ£π₯ ππ£π₯ ππ£π₯ ππ£π₯ ππ π 2 π£π₯ π 2 π£π₯ π 2 π£π₯ π( + π£π₯ + π£π¦ + π£π§ )=− +π( 2 + + ) + πππ₯ ππ‘ ππ₯ ππ¦ ππ§ ππ₯ ππ₯ ππ¦ 2 ππ§ 2 π( ππ£π¦ ππ£π¦ ππ£π¦ ππ£π¦ π 2 π£π¦ π 2 π£π¦ π 2 π£π¦ ππ + π£π₯ + π£π¦ + π£π§ )=− +π( 2 + + ) + πππ¦ ππ‘ ππ₯ ππ¦ ππ§ ππ¦ ππ₯ ππ¦ 2 ππ§ 2 ππ£π§ ππ£π§ ππ£π§ ππ£π§ ππ π 2 π£π§ π 2 π£π§ π 2 π£π§ π( + π£π₯ + π£π¦ + π£π§ )=− +π( 2 + + ) + πππ§ ππ‘ ππ₯ ππ¦ ππ§ ππ§ ππ₯ ππ¦ 2 ππ§ 2 5 In subcooled flow boiling, as in many other instances of fluid dynamics, energy is added or removed from the system. In this situation, the conservation of energy equation is important. Conservation of Energy in Vector Form: ππΆΜπ π·π π ln π π·π β β π) − ( = −(∇ ) π·π‘ π ln π π π·π‘ Conservation of Energy in Cartesian Form: ππ ππ ππ ππ πππ₯ πππ¦ πππ§ π ln π π·π ππΆΜπ ( + π£π₯ + π£π¦ + π£π§ ) = − ( + + )−( ) ππ‘ ππ₯ ππ¦ ππ§ ππ₯ ππ¦ ππ§ π ln π π π·π‘ 6 2.2 NUMERICAL METHODS After the conservation laws governing heat transfer, fluid flow and other related processes are expressed in differential form (shown above), they can solved using numerical methods to determine pressure, temperature, mass flux, etc. for various situations and boundary conditions. Each differential equation represents a conservation principle and employs a physical quantity as its dependent variable that is balanced by the factors that influence it. Some examples of differential equations that may be solved through numerical methods are the conservation of energy, conservation of momentum and time averaged equation for turbulent flow. [4] The goal of computational fluid dynamics is to calculate the temperature, velocity, pressure, etc. of a fluid at particular locations within a system. Thus, the independent variable in the differential equations is a physical location (and time in the case of unsteady flows). Due to computational limitations, the number of locations (also known as grid points or nodes) must be finite. By only focusing on the solution of the differential equations at discrete locations, the need to find an exact solution to the differential equation is not necessary. The algebraic equations (also known as discretization equations) involving the unknown values of the independent variable at chosen locations (grid points) are derived from the differential equations governing the independent variable. In this derivation, assumptions about the value of the independent variable between grid points must be made. This concept is known as discretization. [4] A discretization equation is an algebraic relationship that connects the values of the dependent variable for a group of grid points within a control volume. This type of equation is derived from the differential equation governing the dependent variable and thus expresses the same physical information as the differential equation. The piecewise nature of the profile (or mesh) is created by the finite number of grid points that participate in a given discretization equation. The value of the dependent variable at a grid point thereby influences the value of the dependent variable in its immediate area. As the number of grid points becomes very large, the solution of the discretization equations is expected to approach the exact solution of the corresponding differential equation. This is true because as the grid points get closer together, the change in value between neighboring grid points becomes small and the actual details of the profile 7 assumption become less important. This is where the term “mesh independent” originates. If there are too few grid points (coarse mesh), the profile assumptions can impact the solution results and the discretization equation solution will not match the differential equation solution. To ensure that the discretization equation results are not dependent on the profile assumptions, the solution should be checked for mesh independence. [4] One of the more common procedures for deriving discretization equations is using a truncated Taylor series. Other methods for deriving the discretization equations include variational formulation, method of weighted residuals and control volume formulation. In the iterative process for solving a discretization equation, it is often desirable to speed up or to slow down the changes, from iteration to iteration, in the values of the dependent variable. The process of accelerating the rate of change between iterations is called over-relaxation while the process of slowing down the rate of change between iterations is called under-relaxation. To avoid divergence in the iterative solution of strongly nonlinear equations, under-relaxation is a very useful tool [4]. Fluent allows for manipulation of the relaxation constants for many independent variables to improve convergence ability. Fluent offers numerous spatial discretization solvers for the various independent variables such as pressure, flow, momentum, turbulence, and energy. Fluent implements the control volume formulation with upwinding which was first proposed by Courant, Isaacson, and Rees in 1952. Other options include QUICK, power law and third-order MUSCL. 8 2.3 NATURAL CONVECTION Convection is the transport of mass and energy by bulk fluid motion. If the fluid motion is induced by some external force, it is generally referred to as forced convection. Natural convection is a transport mechanism, in which the fluid motion is not generated by any external source (like a pump, fan, suction device, etc.) but driven by buoyancy-induced motion resulting from internal body forces produced by density gradients. The density gradients can arise from mass concentration gradient and or temperature gradients in the fluid [5]. For example, in a system where a heated surface is submersed in a cooler fluid, the cooler fluid will absorb energy from the heated surface and become less dense. Buoyancy effects due to body forces will cause the heated fluid to rise. At this point, the surrounding, cooler fluid will move in to take its place. The cooler fluid is then heated and the process continues forming a convection current that continuously removes energy from the heated surface. In nature, natural convection cells occur everywhere from oceanic currents to air rising above sunlight-warmed land. Most weather patterns are created by natural convection. Natural convection also takes place in many engineering applications such as home heating radiators and cooling computer chips. The amount of heat transfer occurring due to natural convection in a system is characterized by the Grashof, Prandtl and Rayleigh numbers. The Grashof number, Gr, is a dimensionless parameter that represents the ratio of buoyancy to viscous forces acting on a fluid; and is defined as: πΊπ = ππ½(ππ − π∞ )πΏ3 (π ⁄π)2 where β is the thermal expansion coefficient: 1 ππ π½=− ( ) π ππ π The Prandtl number, Pr, is a dimensionless parameter that represents the ratio of momentum diffusivity to thermal diffusivity; and is defined as: Pr = 9 Cp μ k The Rayleigh number, Ra, is a dimensionless parameter that represents the ratio of buoyancy and viscosity forces times the ratio of momentum and thermal diffusivities; and is defined as: Ra = GrPr When the Rayleigh number is below a critical value for a particular fluid, heat transfer is primarily in the form of conduction; when it exceeds the critical value, heat transfer is primarily in the form of convection. Like forced convection, natural convection can either be laminar or turbulent. Rayleigh numbers less than 108 indicate a buoyancy-induced laminar flow, with transition to turbulence occurring at about 109. [6] In many engineering applications, convection is mixed meaning that both natural and forced convection occurs simultaneously. The importance of buoyancy forces in a mixed convection flow can be measured by the ratio of the Grashof and Reynolds numbers: Gr gβΔTL = Re2 V2 When this number approaches or exceeds unity, there are strong buoyancy contributions to the flow. Conversely, if the ratio is very small, buoyancy forces may be ignored. 10 2.4 LAMINAR FLOW Single-phase fluid flow can be grouped into two categories, laminar or turbulent flow. Laminar flow implies that the fluid moves in sheets that slip relative to each other. Laminar flow occurs at very low velocities where there are only small disturbances and little to no local velocity variations. In laminar flow, the motion of the fluid particles is very orderly and is characterized by high momentum diffusion and low momentum convection. The Reynolds number is used to characterize the flow regime. The Reynolds number, Re, is a dimensionless number that represents the ratio of inertial forces to viscous forces; and is defined as: Re = ρVA μ This quantity helps to quantify the relative importance of these two types of forces for given flow conditions. For internal flow, such as within a pipe, laminar flow is characterized by a Reynolds number less than 2300. The velocity of laminar flow in a pipe is can be calculated by [5]: π’= ππ 2 ππ π2 (− ) (1 − 2 ) 4π ππ₯ ππ Or, in terms of the mean velocity, V: π2 π’ = 2π (1 − 2 ) ππ The energy equation for flow through a circular pipe assuming symmetric heat transfer, fully developed flow and constant fluid properties is [5]: ππ 1π ππ π 2π π’ = πΌ[ (π ) + 2 ] ππ₯ π ππ ππ ππ₯ 11 2.5 TURBULENT FLOW In fluid dynamics, turbulence is a flow regime characterized by chaotic and stochastic changes. Turbulent flows exist everywhere in nature from the jet stream to the oceanic currents. Turbulent flows are highly irregular and random which makes a deterministic approach to turbulence problems impossible. They have high diffusivity, meaning there is rapid mixing and increased rates of momentum, heat and mass transfer. Because of these properties, turbulent flows are very important to many engineering applications. Turbulent flows involve large Reynolds numbers and contain three- dimensional vorticity fluctuations. The unsteady vortices appear on many scales and interact with each other generating high levels of mixing. Also, like laminar flows, turbulent flows are dissipative. Because turbulence cannot maintain itself, it depends on its environment to obtain energy. A common source of energy for turbulent velocity fluctuations is shear in the mean flow; other sources, such as buoyancy, exist too. If turbulence arrives in an environment where there is no shear or other maintenance mechanisms, the turbulence will decay and the flow tends to become laminar. [7] In flows that are originally laminar, turbulence arises from instabilities at large Reynolds numbers. For internal flows, such as within a pipe, turbulent flow is characterized by a Reynolds number greater than 4000. For flows with a Reynolds number between 2300 and 4000, both laminar and turbulent flows are possible. This is called transition flow. [7] A common example of the transition of laminar flow to turbulent flow is smoke rising from a cigarette. Figure 2.5-1: Example of Turbulent Flow 12 As the smoke leaves the cigarette, it travels upward in a laminar fashion as shown by the single stream of smoke. At a certain distance, the Reynolds number becomes too large and the flow begins to transition into the turbulent regime. When this happens, the flow of the smoke becomes more random and rapidly mixes with the air causing the smoke to dissipate. Exact modeling of turbulent flow requires the exact solution of the Continuity and Navier-Stokes equations which can be extremely difficult and time consuming due to the many scales involved. To reduce the complexity, an approximation to the NavierStokes equation was developed by Osborne Reynolds called the Reynolds-averaged Navier–Stokes equations (or RANS equations). This method decomposes the instantaneous fluid flow quantities of the Navier-Stokes equations into mean (timeaveraged) and fluctuating components. The RANS equations can be used with approximations based on knowledge of the turbulent flow to give approximate timeaveraged solutions to the Navier–Stokes equations. [8] For the velocity terms: π’π = π’Μ π + π’π′ where π’Μ π and π’π′ are the mean and fluctuating velocity components respectively. Similarly, for scalar quantities: π = πΜ + π ′ where π denotes a scalar such as energy, pressure, or species concentration. Substituting expressions of this form for the flow variables into the instantaneous continuity and momentum equations and taking a time-average yields the time-averaged continuity and momentum equations [8]. These are written in Cartesian tensor form as: πΏπ πΏ (ππ’Μ π ) = 0 + πΏπ‘ πΏπ₯π πΏ πΏ πΏπ πΏ πΏπ’π πΏπ’π 2 πΏπ’π πΏ ′ ′ Μ Μ Μ Μ Μ Μ (ππ’Μ π ) + (ππ’Μ π π’Μ π ) = − + [π ( + − πππ )] + (−ππ’ π π’π ) πΏπ‘ πΏπ₯π πΏπ₯π πΏπ₯π πΏπ₯π πΏπ₯π 3 πΏπ₯π πΏπ₯π The two above equations are called the RANS equations. They have the same general form as the instantaneous Navier-Stokes equations, with the velocities and other solution variables now representing time-averaged values. The RANS equations can be used with approximations based on knowledge of the turbulent flow to give approximate 13 ′ ′ Μ Μ Μ Μ Μ Μ time-averaged solutions to the Navier–Stokes equations. An additional term,(−ππ’ π π’π ), known as the Reynolds stress appears in the equation as a results of using the RANS method. [8] One way that the Reynolds stress is evaluated in practice is through the k-Ο΅ turbulence model. The k-Ο΅ model was first introduced by Harlow and Nakayama in 1968 [9]. The k-Ο΅ model has become the most widely used model for industrial applications because of its overall accuracy and small computational demand. In the k-Ο΅ model, k represents the turbulent kinetic energy and Ο΅ represents its dissipation rate. Turbulent kinetic energy is the average kinetic energy per unit mass associated with eddies in the turbulent flow while epsilon (Ο΅) is the rate of dissipation of the turbulent energy per unit mass. In the derivation of the k-Ο΅ model, it was assumed that the flow is fully turbulent, and the effects of molecular viscosity are negligible. As the strengths and weaknesses of the standard k-Ο΅ model have become known, modifications have been introduced to improve its performance. These improvements have helped create many, new, more accurate models, among these, the realizable k-Ο΅ model which differs from the standard k-Ο΅ model in two important ways. First, the realizable model contains an alternative formulation for the turbulent viscosity. Second, a modified transport equation for the dissipation rate, Ο΅, has been derived from an exact equation for the transport of the meansquare vorticity fluctuation. The term “realizable” means that the model satisfies certain mathematical constraints on the Reynolds stresses, consistent with the physics of turbulent flows. [8] 2.5.1 CALCULATING TURBULENCE PARAMETERS All of the computational fluid dynamic models discussed in this thesis use the k-Ο΅ turbulence model when applicable. In the Fluent code, turbulence models require certain parameters to be established prior to initialization to properly set the initial and boundary conditions for the flow. For instance, based on the conditions in Table 2.5.1-1, the equations in Table 2.5.1-2 [8] were used to determine the boundary and initial condition inputs for the turbulent flow models presented in Section 3.3. 14 Table 2.5.1-1: Turbulent Flow Input Input Parameter Mass Flow Rate (πΜ) Pipe Diameter (D) Viscosity (μ) Density (ρ) Turbulence Empirical Constant (Cμ) Numerical Value 1.0 kg/s 0.03 m 0.001003 kg/m-s 998.2 kg/m3 0.09 [8] Table 2.5.1-2: Turbulence Parameter Calculation Variable Hydraulic Diameter (Dh) Flow Area (A) Average Flow Velocity (uavg) Reynolds Number (ReDh) Turbulence Length Scale (l) Equation Numerical Value 4∗π΄ π·β = π π· 2 π ∗ (2 ) = =π· 4∗π∗π· π· 2 π΄ =π∗( ) 2 0.03 π 2 =π∗( ) 2 πΜ π’ππ£π = π∗π΄ 0.5 ππ/π = ππ 998.2 3 ∗ 0.00070686 π2 π πΜπ·β π ππ·β = ππ΄ ππ 0.5 π ∗ 0.03 m = ππ 0.001003 π − π ∗ 0.00070686 π2 π = 0.07 ∗ π·β = 0.07 ∗ 0.03 π − Turbulent Intensity (I) Turbulent Kinetic Energy (k) Dissipation Rate (Ο΅) 0.00070686 m2 1.41726 m/s 42314 0.0021 m 1 πΌ = 0.16 ∗ π ππ· 8 β = 0.03 m 4.22483 % 1 0.16 ∗ 42314−8 3 2 π = (π’ππ£π ∗ πΌ) 2 2 3 π = (1.41726 ∗ 0.0422483) 2 π 3/2 3/4 k ε = Cπ π 0.00537853/2 = 0.093/4 0.0021 15 0.0053785 m2/s2 0.030859 m2/s3 2.6 TWO-PHASE FLOW Fluid flows that contain two or more components are referred to as multiphase flow. The flow components can be of the same chemical substance but in different states of matter such as water and steam, be different chemical substances but the same state of matter such as water and oil or finally be of different chemical substance and different states of matter such as water and air. This section focuses on two-phase flow involving water and air while Section 2.7 focuses on two-phase flows involving water and steam. Depending on the volume fraction of each component in the two-phase flow, different flow patterns can exist. Understanding the flow pattern of a two-phase flow is important because pressure drops and heat transfer rates are heavily impacted by the flow type. The characteristic flow patterns for two phase flow, in order of increasing gas volume fraction from liquid to gas, are bubbly flow, plug flow, stratification flow, wavy flow, slug flow, annular flow and spray flow. A schematic representation of each of these flow patterns is shown in Figure 2.5-1 [10]. Figure 2.6-1: Flow Regimes The flow patterns shown in Figure 2.6-1 can be classified into three categories, bubbly flow, slug flow and annular flow. Bubbly flow is when the liquid phase is continuous and the vapor phase is discontinuous such that the vapor phase is distributed 16 in the liquid phase in the form of bubbles. This flow pattern occurs at low gas volume fractions. Subcooled boiling is classified as bubbly flow. Slug flow occurs when there are relatively large liquid slugs surrounded by vapor. This flow occurs at moderate gas volume fractions and relatively low flow velocities. Annular flow is when the liquid phase is continuous along the wall and the vapor phase is continuous in the core. This flow pattern occurs at high gas volume fractions and high flow velocities. Although not considered to be a flow regime, film boiling is the opposite of annular flow (the vapor phase is continuous along the wall and the liquid phase is continuous in the core). Film boiling occurs when the heat flux is relatively large compared to the mass flux. Film boiling is discussed further in Section 2.7. As stated previously, knowing the flow pattern is important to determine the pressure drop and heat transfer rate within a system. The flow pattern changes as a function of gas volume fraction and flow velocity. The flow pattern of a system can be determined using the Baker flow criteria shown in Figure 2.6-2 [10, Figure 3-4]. Figure 2.6-2: Baker Flow Pattern 17 Two-phase flows obey the same basic laws of fluid mechanics that apply to single phase flows; however, the equations are more complicated and more numerous. Two-phase flows are more complicated because there are more equations to solve due to the secondary phase and additional phenomena to account for such as mass transfer, phase-interface interactions (slip and drag). Three common multiphase flow models available in Fluent are Volume of Fluid, Mixture and Eulerian, each with varying strengths and computational demand. The Volume of Fluid model solves a single set of momentum equations for two or more fluids and tracks the volume fraction of each fluid throughout the domain. The Mixture model solves for the momentum equation of the mixture and prescribes relative velocities to describe the dispersed phase. The Eulerian model solves momentum and continuity equations for each of the phases, and the equations are coupled through pressure and exchange coefficients. These are discussed in detail in Appendix A. 18 2.7 BOILING HEAT TRANSFER Boiling is defined as a mode of heat transfer that occurs when saturated liquid changes to saturated vapor due to heat addition. It is normally characterized by a high heat transfer capacity and a low wall temperature which is made possible due to the large amount of energy required to cause a phase change. This is essential for industrial cooling applications, such as nuclear reactors and fossil boilers. Due to its importance in industry, a significant amount of research has been carried out to study the capacity and the mechanism of boiling heat transfer. There are two basic types of boiling, pool boiling and flow boiling. Pool boiling occurs when heat is added to a stagnant fluid while flow boiling occurs when heat is added to a moving fluid. Both types of boiling heat transfer can be separated into four regimes which are shown in Figure 2.7-1 [11]. Figure 2.7-1: Boiling Heat Transfer Regimes The first regime of boiling, up to point A, is known as natural convection boiling. During this regime, no bubbles form; instead, heat is transferred from the surface to the 5/4 bulk fluid by natural convection. The heat transfer rate is proportional to π₯ππ ππ‘ [10]. The second regime of boiling, from point A to point C, is called nucleate boiling. During this stage, vapor bubbles are generated at certain preferred locations on the heated surface called nucleation sites. Nucleation sites are often microscopic cavities or cracks in the surface. When the liquid near the wall superheats, it evaporates forming 19 bubbles at the nucleation sites. When the liquid evaporates, a significant amount of energy is removed from the heated surface due to the latent heat of the vaporization. Vaporization also increases the convective heat transfer by mixing the liquid water near the heated surface. There are two subregimes of nucleate boiling. The first subregime is when local boiling occurs in a subcooled liquid (subcooled boiling). In this situation, the bubbles form on the heated surface but tend to condense after leaving the heated surface. The second subregime is when local boiling occurs in a saturated liquid. In this case, the bubbles do not collapse. It is possible for both subregimes to take place between points A and C. Nucleate boiling is characterized by a very high heat transfer rate and a small temperature difference between the bulk fluid and the heated surface. For this reason, it is considered to be the most efficient heat transfer boiling regime. [10] As the heated surface increases in temperature, more and more nucleation sites become active. As more bubbles form at the nucleation sites, they begin to merge together and form columns and slugs of gas, thus decreasing the contact area between the bulk fluid and the heated surface. The decrease in contact area causes the slope of the line to decrease until a maximum is reached (point C). Point C is referred to as the critical heat flux. When the critical heat flux is reached, the vapor begins to form an insulating blanket around the heated surface which dramatically increases the surface temperature. This is called the boiling crisis or departure from nucleate boiling. [11] As the temperature delta increases past the critical heat flux, the rate of bubble generation exceeds the rate of bubble separation. Bubbles at the different nucleation sites begin to merge together and boiling becomes unstable. The surface is alternately covered with a vapor blanket and a liquid layer, resulting in oscillating surface temperatures. This regime of boiling is known as partial film boiling or transition boiling and takes place between points C and D. [10] If the temperature difference between the surface and the fluid continues to increase, stable film boiling is achieved. During stable film boiling, there is a continuous vapor blanket surrounding the heated surface and phase change occurs at the liquid-vapor interface instead of at the heated surface. During this regime, most heat transfer is carried out by radiation. [11] 20 3. HEAT TRANSFER AND FLUID FLOW: MODELING 3.1 NATURAL CONVECTION Two examples of natural convection were examined in this section. The first was a heated horizontal cylinder and the second was a heated vertical plate, both were submerged in an infinite pool of liquid. These examples were chosen because of their simplicity, because they are commonly found in nature and because they have been previously studied and results are available for validation of the numerical computations. 3.1.1 HORIZONTAL CYLINDER The results from modeling a cylinder with a constant surface temperature submerged in an infinite pool of liquid were analyzed in this section. The cylinder was slightly warmer than the surrounding fluid and therefore energy passed from the cylinder to the nearby fluid causing its temperature to increase. Table 3.1.1-1 lists the important input needed to replicate the results shown in this section. The liquid temperature field after 20 seconds is shown in Figure 3.1.1-1. Figure 3.1.1-1: Horizontal Cylinder Temperature 21 Table 3.1.1-1: Horizontal Cylinder Input Input Geometry Cylinder Diameter Pool Height Pool Width 2D Space Solver Time Time Step Size Type Velocity Formulation Gravity Models Energy Viscous Density Initial Conditions Cylinder Surface Temperature Initial Fluid Temperature Material Properties (Water) Specific Heat Thermal Conductivity Viscosity Density Solution Methods Scheme Gradient Pressure Momentum Energy Transient Formulation Value 0.02 m 0.28 m 0.24 m Planar Transient Pressure Based Relative -9.8 m/s2 (Y-direction) On Laminar Boussineq 310 K 300 K 4182 J/kg-K 0.6 W/m-K 0.001003 kg/m-s See Table 3.1.1-2 PISO Least Square Cell Based PRESTO! Second Order Upwind Second Order Upwind Second Order Implicit Table 3.1.1-2: Horizontal Cylinder Water Density Density (kg/m3) 999.9 994.1 974.9 958.4 Temperature (K) 273 308 348 373 22 As the temperature increases, the fluid expands and its density decreases causing the fluid to rise due to buoyancy forces. Even fluid that is not in direct contact with the heated cylinder experiences a density change, shown in Figure 3.1.1-2. The density gradient is shown by the color transition surrounding the cylinder from least dense (blue) to most dense (red). This is caused by energy transfer via conduction to the bulk fluid. Figure 3.1.1-2: Horizontal Cylinder Density As the fluid rises, it separates from the cylinder allowing new, cooler fluid takes its place. When the warm fluid rises, it loses energy to the surrounding bulk fluid which causes the buoyancy driving head to diminish which causes the fluid to climb more slowly until it eventually stops. At this point, the fluid is pushed to the left or right by the fluid travelling upwards below it and fluid recently pushed aside begins to sink. This motion creates a small convection cell to the left and the right of the rising plume about 3 cm above the heated cylinder. This process continues indefinitely as long as there is a temperature gradient between the cylinder and the bulk fluid. Figure 3.1.1-3 is a liquid velocity vector plot that shows how the liquid moves within the control volume. The cycle of fluid energy absorption and replacement around the cylinder and the two convection cells above the cylinder are more visible in this figure. 23 Figure 3.1.1-3: Horizontal Cylinder Velocity Vector To verify that the model produced realistic results, the solution was compared to experimental data. Figure 3.1.1-4 shows interference fringes surrounding a heated horizontal cylinder in natural convection. Each interference fringe can be interpreted as a band constant temperature. (a) (b) Figure 3.1.1-4: Interference Fringes Around a Heated Horizontal Cylinder (a) is from [12] and (b) shows isotherms from Fluent 24 Figure 3.1.1-4 shows comparable results between experimental data and the results determined by Fluent. Both have isotherms that extend away from the cylinder and grow in distance away from one another as they get farther from the heated surface. Quantitative experimental data from Ingham [13] was compared to the Fluent results to provide model validation. Figure 3.1.1-5, Figure 3.1.1-6 and Figure 3.1.1-7 show a comparison of dimensionless temperature versus dimensionless distance for four dimensionless times at an angle of 30°, 90° and 180°, respectively, from the positive x-axis. Dimensionless temperature is T = (T’ – T0) / (Twall – T0) where T’ is the actual fluid temperature, T0 is the bulk fluid temperature and Twall is the wall temperature. Dimensionless time is t = t’ * (βgΔT/a)1/2 where t’ is real time, ΔT is (Twall – T0), β is the coefficient of thermal expansion and a is the diameter of the cylinder. (a) (b) Figure 3.1.1-5: Temperature at θ = 30° Vs. Radial Distance (a) is from [13] and (b) is from Fluent 25 (a) (b) Figure 3.1.1-6: Temperature at θ = 90° Vs. Radial Distance (a) is from [13] and (b) is from Fluent (a) (b) Figure 3.1.1-7: Temperature at θ = 180° Vs. Radial Distance (a) is from [13] and (b) is from Fluent 26 The heated horizontal cylinder model developed in Fluent showed good agreement compared to experimental data at the three different radial locations. This comparison provided confidence that the information gathered from the Fluent model was accurate. To ensure that the mesh had no impact on the results, a mesh validation was performed. The mesh validation was performed by comparing the results shown in this section (“Analysis Value” in Table 3.1.1-1) to a second case with an increased number of nodes and elements (“Mesh Validation” in Table 3.1.1-1). The second case was exactly the same as the case used in this section except that its mesh was refined. The results from the mesh validation are shown in Table 3.1.1-1, and prove that the results are mesh independent. Table 3.1.1-1: Mesh Validation for Horizontal Cylinder Number of Nodes Number of Elements Max Velocity (m/s) Max Total Temperature (°F) Min Density (kg/m3) Analysis Value 19716 38688 0.01627 309.9239 993.1765 27 Mesh Validation 23636 46400 0.01621 309.9531 993.1625 Difference 19.88 % 19.93 % -0.37 % 0.01 % 0.00 % 3.1.2 VERTICAL PLATE The results from modeling a vertical plate with a constant surface temperature submerged in an infinite pool of liquid were analyzed in this section. Like the cylinder, the plate was also slightly warmer than the surrounding fluid and therefore energy passed from the plate to the fluid causing its temperature to increase. Table 3.1.2-1 lists the important input needed to replicate the results shown in this section. Table 3.1.2-1: Vertical Plate Input Input Value Geometry Plate Height Plate Width Pool Height Pool Width 2D Space Solver Time Time Step Size Type Velocity Formulation Gravity Models Energy Viscous Density Initial Conditions Plate Surface Temperature Initial Fluid Temperature Material Properties (Water) Specific Heat Thermal Conductivity Viscosity Density Solution Methods Scheme Gradient Pressure Momentum Energy Transient Formulation 28 0.18 m 0.01 m 0.20 m 0.13 m Planar Transient Pressure Based Relative -9.8 m/s2 (Y-direction) On Laminar Boussineq 310 K 300 K 4182 J/kg-K 0.6 W/m-K 0.001003 kg/m-s See Table 3.1.2-2 PISO Least Square Cell Based PRESTO! Second Order Upwind Second Order Upwind Second Order Implicit Table 3.1.2-2: Vertical Plate Water Density Density (kg/m3) 999.9 994.1 974.9 958.4 Temperature (K) 273 308 348 373 The liquid temperature field after 20 seconds is shown in Figure 3.1.2-1. When energy is exchanged between the plate and the fluid, a thermal boundary layer is created. Thermodynamic equilibrium demands that the plate and the fluid in direct contact with it be at the same temperature. The region in which the fluid temperature changes from the plate surface temperature to that of the bulk fluid is known as the thermal boundary layer. The teal color in Figure 3.1.2-1 shows the growth of the thermal boundary layer. It is relatively small at the bottom of the plate because there has been little heat addition but the thermal boundary layer grows (teal color expands away from the plate) as the fluid reaches the top of the plate. Although not visible in Figure 3.1.2-1, as the thermal boundary layer expands, so does the momentum boundary layer which means that there is more fluid motion due to heat addition from the hot plate. Figure 3.1.2-1: Vertical Plate Temperature 29 Figure 3.1.2-2 shows the fluid velocity in vector form. The growth of the momentum boundary layer is more visible in this figure (the teal colored arrows expand away from the plate). The figure shows that the velocity is primarily vertical with a magnitude that increases with elevation. The increase in fluid velocity is caused by longer contact time with the heated surface creating a greater temperature gradient and therefore a larger buoyancy force. Figure 3.1.2-2: Vertical Plate Velocity Vector Plot Comparing Figure 3.1.2-2 (vertical plate velocity vectors) with Figure 3.1.1-3 (horizontal cylinder velocity vectors) produces interesting results. Because of the larger heated region, it was expected that the vertical plate would produce a greater maximum fluid velocity compared to the horizontal cylinder. The vertical plate produced a maximum fluid velocity of 0.0149 m/s while the horizontal cylinder produced a maximum fluid velocity of 0.0177 m/s. Although the difference is small, it is notable. The horizontal cylinder produced a larger maximum velocity because the buoyancy driving head does not fight against the drag force generated by the heated surface. Although the plate continued to heat the fluid as it traveled upward, the velocity is limited by friction which caused the plate scenario to have a smaller maximum velocity. To ensure that the model was giving realistic results, the solution was compared to experimental data. Figure 3.1.2-3 shows interference fringes surrounding a heated 30 vertical plate in natural convection. Each interference fringe can be interpreted as a band constant temperature. (a) (b) Figure 3.1.2-3: Interference Fringes Around a Heated Vertical Plate (a) is from [12] and (b) is from Fluent The model of a vertical plate submerged in an infinite pool was in qualitative agreement to experimental data. Figure 3.1.2-3 shows that the experimental data and model solution have isotherms that extend away from the plate and grow in distance away from one another as they get farther from the heated surface. Quantitative experimental data from Ostrach [14] was compared to the Fluent results to assess the accuracy of the model. Figure 3.1.2-4 and Figure 3.1.2-5 show a comparison of dimensionless temperature versus dimensionless distance for five different Prandtl numbers. Figure 3.1.2-4a shows theoretical values and Figure 3.1.2-4b compares some of the theoretical values to experimental data. The information contained in Figure 3.1.2-5 was calculated by Fluent. Dimensionless temperature is T = (T’ – T∞) / (T0 – T∞) where T’ is the actual fluid temperature, T∞ is the bulk fluid temperature and T0 is the wall temperature. Dimensionless distance is η = (Grx / 4)1/4 * (Y / X) where Grx is the Grashof number, Y is the vertical height and X is the distance from the plate. 31 (a) (b) Figure 3.1.2-4: Dimensionless Temperature as a Function of Prandtl Number (a) Theoretical Values and (b) Experimental Values [14] Figure 3.1.2-5: Dimensionless Temperature as a Function of Prandtl Number 32 The heated vertical plate model developed in Fluent produced similar temperature results to the experimental data for five different Prandtl numbers. This comparison provided confidence that the information gathered from the Fluent model was accurate. To ensure that the mesh had no impact on the results, a mesh validation was performed. The mesh validation compared the results shown in this section (“Analysis Value” in Table 3.1.2-1) to a second case with an increased number of nodes and elements (“Mesh Validation” in Table 3.1.2-1). The second case was exactly the same as the case used in this section except that its mesh was refined. The results from the mesh validation are shown in Table 3.1.2-1, and prove that the results are mesh independent. Table 3.1.2-1: Mesh Validation for Vertical Plate Number of Nodes Number of Elements Max Velocity (m/s) Max Total Temperature (°F) Min Density (kg/m3) Analysis Value 12310 23572 0.01376 309.8089 993.2319 33 Mesh Validation 18081 35168 0.01380 309.7991 993.2365 Difference 46.88 % 49.19 % 0.29 % 0.00 % 0.00 % 3.2 LAMINAR FLOW A simple axisymmetric flow model was developed to gain a better understanding of laminar flow in a pipe. The Reynolds number was 352, based upon the selected initial conditions, which is well within the laminar regime. Table 3.2-1 lists the important input needed to replicate the results shown in this section. Table 3.2-1: Laminar Flow Input Input Value Geometry Pipe Diameter Pipe Length 2D Space Solver Time Type Velocity Formulation Gravity Models Energy Viscous Material Properties (Water) Specific Heat Thermal Conductivity Viscosity Density Initial Conditions Pipe Wall Surface Temperature Fluid Inlet Temperature Fluid Inlet Velocity Solution Methods Scheme Gradient Pressure Momentum Energy 0.03 m 0.50 m Axisymmetric Steady Pressure Based Relative -9.8 m/s2 (X-direction) On Laminar 4182 J/kg-K 0.6 W/m-K 0.001003 kg/m-s See Table 3.2-2 305 K 300 K 0.05 m/s Coupled Least Square Cell Based Second Order Second Order Upwind Second Order Upwind Table 3.2-2: Laminar Flow Water Density Density (kg/m3) 999.9 994.1 Temperature (K) 273 308 34 A noteworthy characteristics of laminar flow is the parabolic shape of the velocity profile. Fluid velocity within the pipe slowly decreases as distance from the pipe centerline increases. This is vastly different from turbulent flow which has a very flat velocity profile and is described in more detail in Section 3.3. Figure 3.2-1 shows the velocity magnitude versus position (distance from the pipe centerline) at various distances from the pipe entrance. For example, “line-10cm” shows the velocity profile 10 cm from the pipe entrance. As the flow develops, the entrance effects dissipate, the velocity profile becomes more and more parabolic until it reaches a steady state at 45 cm from the entrance. Figure 3.2-1: Laminar Flow Velocity Profile Vs. Positon Another characteristic of laminar flow is the lack of mixing that occurs within the fluid as it travels through the pipe. The radial velocity within the pipe is basically zero and each fluid molecule or atom remains about the same distance from the centerline as it travels through the pipe. Figure 3.2-2 shows the radial flow velocity. As expected, the radial velocity for most of the pipe is near zero and is less than 10 -3 times the average axial velocity. Radial velocity spikes near the entrance of the pipe due to pipe boundary conditions and entrance effects but these have little impact on system as a whole. 35 Figure 3.2-2: Laminar Flow Radial Velocity Figure 3.2-3 shows the temperature profile of the laminar flow analyzed. Diffusion and conduction are the primary forms of heat transfer within the fluid. The growth of the thermal boundary layer as the fluid travels down the pipe is shown in Figure 3.2-3 by the expansion of the teal colored region. Figure 3.2-3: Laminar Flow Temperature As in natural convection, laminar flow generates a momentum boundary layer but its development is not visible pictorially. The momentum boundary layer is created by drag forces produced by the wall. Figure 3.2-4 shows the wall shear stress as a function of distance from the pipe entrance. Figure 3.2-4: Laminar Flow Wall Shear Stress 36 Figure 3.2-4 shows that the wall stress is much larger in the first 10 cm which is caused by entrance effects. Once the entrance effects dissipate, the wall shear stress slowly decreases as the flow reaches a steady state. To ensure that the mesh had no impact on the results, a mesh validation was performed. The mesh validation was performed by comparing the results shown in this section (“Analysis Value” in Table 3.3.2-1) to a second case with an increased number of nodes and elements (“Mesh Validation” in Table 3.3.2-1). The second case was exactly the same as the case used in this section except that its mesh was refined. The results from the mesh validation are shown in Table 3.3.2-1, and prove that the results are mesh independent. Table 3.2-3: Mesh Validation for Laminar Flow Number of Nodes Number of Elements Max Velocity (m/s) Min Radial Velocity (m/s) Max Dynamic Pressure (Pa) Max Temperature (K) Analysis Value 26320 25353 0.079561 -0.003293 3.15925 304.6503 37 Mesh Validation 31000 29970 0.079507 -0.003528 3.155022 304.6855 Difference 17.78 % 18.21 % -0.07 % 7.12 % -0.13 % 0.01 % 3.3 TURBULENT FLOW 3.3.1 TURBULENT FLOW WITHOUT HEAT TRANSFER A simple axisymmetric flow model was developed to gain a better understanding of turbulent flow in a pipe. The Reynolds number was 42314, based upon the selected initial conditions, which is well within the turbulent regime. Table 3.3.1-1 lists the important input needed to replicate the results shown in this section. Table 3.3.1-1: Turbulent Flow Without Heat Transfer Input Input Value Geometry Pipe Diameter 0.03 m Pipe Length 0.50 m 2D Space Axisymmetric Solver Time Steady Type Pressure Based Velocity Formulation Relative Gravity -9.8 m/s2 (X-direction) Models Energy Off Viscous Realizable k-ε Turbulence Model Near Wall Treatment Enhanced Turbulent Intensity* 4.22483 % Initial Conditions Fluid Mass Flow Rate 1.0 kg/s Material Properties (Water) Density 998.2 kg/m3 Viscosity 0.001003 kg/m-s Solution Methods Scheme Coupled Gradient Least Square Cell Based Pressure Second Order Momentum Second Order Upwind Turbulent Kinetic Energy Second Order Upwind Turbulent Dissipation Rate Second Order Upwind * Calculation shown in Table 2.5.1-2. 38 Figure 3.3.1-1 shows the velocity magnitude versus position (distance from the pipe centerline) at various distances from the pipe entrance. The velocity profile of turbulent flow differs significantly in two ways compared to the velocity profile of laminar flow (Section 3.2). First, turbulent flow velocity profiles are much flatter. Therefore, the fluid velocity doesn’t decrease significantly until close to the pipe wall. Second, entrance effects dissipate much quicker in turbulent flow [5] and thus the fluid velocity reaches a steady state velocity profile in a shorter distance. Figure 3.3.1-1 (turbulent flow) shows that flow reached a steady profile about 10 cm from the pipe entrance. Figure 3.2-1 (laminar flow) shows that flow reached a steady profile about 45 cm from the pipe entrance. This qualitatively matches experimental data well. Figure 3.3.1-1: Turbulent Flow Velocity Magnitude Vs. Position Figure 3.3.1-2 shows the wall shear stress versus distance from the pipe entrance. The shear stress is very large at the pipe entrance and decays to the steady state value after about 10 cm (same location where the velocity profile reaches steady state). The large increase in shear stress at the beginning of the pipe (~1-2 cm from the inlet) is caused by the entrance effects. Figure 3.3.1-3 shows that that maximum absolute radial velocity occurs near the pipe entrance. Conservation of momentum requires that the axial velocity decrease near the entrance due to the increase in radial velocity. Figure 3.3.1-4 shows that the greatest reduction in axial velocity occurs near the pipe 39 entrance which matches expectations. Since shear stress is related to change in velocity perpendicular to the wall (axial velocity), the increase in wall shear stress is reasonable. Figure 3.3.1-2: Wall Shear Stress Vs. Position Figure 3.3.1-3: Radial Velocity Figure 3.3.1-4: dAxial-Velocity/dx Vs. Position 40 To further investigate the impact of entrance effects, two additional cases were run using a mass flow rate of 0.5 kg/s (Figure 3.3.1-5) and 1.5 kg/s (Figure 3.3.1-6). (a) (b) (c) Figure 3.3.1-5: Flow Results for Mass Flow Rate of 0.5 kg/s (a) radial velocity (b) wall shear stress vs. position (c) daxial-velocity/dx vs. position (a) (b) (c) Figure 3.3.1-6: Flow Results for Larger Mass Flow Rate of 1.5 kg/s (a) radial velocity (b) wall shear stress vs. position (c) daxial-velocity/dx vs. position 41 Figures 3.3.1-5 and 3.3.1-6 show that wall shear stress and maximum radial velocity are directly related to mass flow rate. At a certain distance from the entrance, the change in axial velocity as a function of position reaches zero and the wall shear stress reaches a constant value. The pipe length necessary to reach a steady state shear stress is directly related to the mass flow rate. A larger mass flow rate requires a greater length. Figure 3.3.1-7 and Figure 3.3.1-8 show the turbulent kinetic energy and the production of turbulent kinetic energy as a function of distance. Figure 3.3.1-7: Turbulent Kinetic Energy Figure 3.3.1-8: Production of Turbulent Kinetic Energy Most of the turbulent kinetic energy is located near the pipe wall due to shear stress. The trend of Figure 3.3.1-8 is similar to that of Figure 3.3.1-2 because shear stress, created by the wall, produces turbulent kinetic energy. 42 3.3.2 TURBULENT FLOW WITH HEAT TRANSFER The turbulent flow model described in Section 3.3.1 was modified to include heat transfer from the pipe wall to the fluid. Table 3.3.2-1 lists the important input needed to replicate the results shown in this section. Table 3.3.2-1: Turbulent Flow With Heat Transfer Input Input Value Geometry Pipe Diameter 0.03 m Pipe Length 0.50 m 2D Space Axisymmetric Solver Time Steady Type Pressure Based Velocity Formulation Relative Gravity -9.8 m/s2 (X-direction) Models Energy On Viscous Realizable k-ε Turbulence Model Near Wall Treatment Enhanced Turbulent Intensity* 4.22483 % Initial Conditions Fluid Mass Flow Rate 1.0 kg/s Fluid Inlet Temperature 300 K Wall Heat Flux 450 kW/m2 Material Properties (Water) Specific Heat 4182 J/kg-K Thermal Conductivity 0.6 W/m-K Viscosity 0.001003 kg/m-s Density See Table 3.3.2-2 Solution Methods Scheme Coupled Gradient Least Square Cell Based Pressure Second Order Momentum Second Order Upwind Turbulent Kinetic Energy Second Order Upwind Turbulent Dissipation Rate Second Order Upwind Energy Second Order Upwind * Calculation shown in Table 2.5.1-2. 43 Table 3.3.2-2: Turbulent Flow Water Density Density (kg/m3) 999.9 994.1 974.9 Temperature (K) 273 308 348 Figure 3.3.2-1 shows the fluid temperature change caused by energy addition from the pipe walls. The radial temperature distribution in Figure 3.3.2-1 is significantly more uniform than the radial temperature distribution in Figure 3.2-3 (laminar flow). Uniform temperature distribution is a characteristic of turbulent flow and made possible due to the chaotic nature of the flow regime. Figure 3.3.2-1: Temperature The radial velocity shown in Figure 3.3.2-2 is very similar to that shown in Figure 3.3.1-3 which is expected since the heat addition has a negligible impact on fluid velocity. If the heat transfer rate to the fluid was increased sufficiently such that flow velocity was significantly impacted, then the radial velocity between the two scenarios would also differ. Figure 3.3.2-2: Radial Velocity Comparing the velocity profiles for the two turbulent flow scenarios (Figure 3.3.1-1 and Figure 3.3.2-3) reveals that the velocity magnitude is slightly larger for the case with heat transfer. The heat transfer caused the density of the fluid to decrease and therefore the velocity increased slightly to maintain a constant mass flow through the pipe. 44 Figure 3.3.2-3: Velocity Magnitude Vs. Position As expected, the wall shear stress shown in Figure 3.3.2-4 is similar to the wall shear stress shown in Figure 3.3.1-2. Figure 3.3.2-4: Wall Shear Stress Vs. Axial Position To ensure that the mesh had no impact on the results, a mesh validation was performed. The mesh validation was performed by comparing the results shown in this section (“Analysis Value” in Table 3.3.2-1) to a second case with an increased number of nodes and elements (“Mesh Validation” in Table 3.3.2-1). The second case was exactly the same as the case used in this section except that its mesh was refined. The 45 results from the mesh validation are shown in Table 3.3.2-1, and prove that the results are mesh independent. Table 3.3.2-3: Mesh Validation for Turbulence With Heat Transfer Number of Nodes Number of Elements Max Velocity (m/s) Max Temperature (°F) Min Density (kg/m3) Max Dynamic Pressure (Pa) Analysis Value 31031 31000 1.502045 317.6659 989.4604 1122.853 Mesh Validation 35739 34624 1.500343 318.1447 989.2305 1119.909 Difference 15.17 % 11.69 % -0.11 % 0.15 % -0.02 % -0.26 % Comparing the velocity magnitude plots, radial velocity contours and wall shear stress plots from Section 3.3.1 and Section 3.3.2 shows that the addition of heat transfer has a negligible impact the fluid flow profile. This is reasonable since the heat flux is relatively small and does not create any localized phase change. Thus, the relationships developed in Section 3.3.1 are applicable to turbulent flows with heat transfer as long as the impact due to heat transfer is small. 46 3.4 TWO-PHASE FLOW 3.4.1 GAS MIXING TANK In many branches of engineering, gas injection techniques have been extensively utilized to enhance chemical reaction rates, homogenize temperature and chemical compositions, and remove impurities. In the steel industry, the advancements made in mixing have increased the level of control available over the steelmaking process which has improved the quality of steel produced. To mix the molten metal, gas is pumped through a porous plug located at the bottom of the mixing tank. The porous plug controls the velocity and bubble diameter of the gas. Buoyancy forces cause the injected gas to move quickly through the molten metal and drag forces causes mixing. Table 3.4.1-1 lists the important input needed to replicate the results shown in this section. After 5 seconds of gas injection, Figure 3.4.1-1 shows the gas volume fraction, Figure 3.4.1-2 shows the liquid vector velocity and Figure 3.4.1-3 shows the gas vector velocity. Midway through the liquid volume in Figure 3.4.1-1, the air jet begins to become wavy. The wavy behavior is explained by Rayleigh instability which states that surface tension tends to minimize surface area. Thus, after a certain distance the air jet will transform into air bubbles with the same volume but less surface area. The length required for the jet to breakup is dependent upon the air velocity and gas / liquid surface tension. The liquid and gas velocities shown in Figure 3.4.1-2 and Figure 3.4.1-3, respectively, are similar which indicates that the drag force between the two phases is strong. The maximum gas velocity is greater than the inlet velocity (0.5 m/s); therefore, buoyancy forces are significant. Figure 3.4.1-2 shows that there is a number of small eddies, created by the injected gas, that provide a significant amount of mixing within the liquid. 47 Table 3.4.1-1: Gas Mixing Tank Input Input Geometry Tank Width Tank Height Porous Plug Width 2D Space Solver Time Time Step Size Type Velocity Formulation Gravity Models Energy Viscous Multiphase Drag Turbulence Model Near Wall Treatment Turbulent Kinetic Energy Turbulent Dissipation Rate Initial Conditions Water Level Gas Velocity Bubble Diameter Material Properties (Water) Density Viscosity Material Properties (Air) Density Viscosity Surface Tension Solution Methods Scheme Gradient Momentum Volume Fraction Turbulent Kinetic Energy Turbulent Dissipation Rate Transient Formulation 48 Value 0.30 m 0.60 m 0.02 m Planar Transient 0.001 s Pressure Based Relative -9.8 m/s2 (Y-direction) Off Standard k-ε Eulerian Schiller-Nauman Standard 0 m2/s2 0 m2/s3 0.40 m 0.5 m/s 0.001 m 998.2 kg/m3 0.001003 kg/m-s 1.225 kg/m3 1.7894E-05 kg/m-s 0.072 N/m Phase Coupled SIMPLE Least Square Cell Based Second Order Upwind QUICK Second Order Upwind Second Order Upwind Second Order Implicit Figure 3.4.1-1: Gas Volume Fraction Figure 3.4.1-2: Liquid Vector Velocity 49 Figure 3.4.1-3: Gas Vector Velocity To ensure that the mesh had no impact on the results, a mesh validation was performed. The mesh validation was performed by comparing the results shown in this section (“Analysis Value” in Table 3.4.1-2) to a second case with an increased number of nodes and elements (“Mesh Validation” in Table 3.4.1-2). The second case was exactly the same as the case used in this section except that its mesh was refined. The results from the mesh validation are shown in Table 3.4.1-2, and prove that the results are mesh independent. Table 3.4.1-2: Mesh Validation for Gas Mixing Tank Number of Nodes Number of Elements Max Liquid Velocity (m/s) Max Gas Velocity (m/s) Max Static Pressure (psia) Max Liquid Total Pressure (psia) Max Liquid Volume Fraction Analysis Value 30625 30256 1.539086 2.046923 3925.424 4775.512 1.000000 50 Mesh Validation 36045 35644 1.453488 2.086285 3894.616 4732.633 1.000000 Difference 17.70% 17.81% -5.56% 1.92% -0.78% -0.90% 17.70% 3.4.2 BUBBLE COLUMN A bubble column reactor is an apparatus primarily used to study gas- liquid reactions. This apparatus is a vertical column of liquid with gas introduced continuously at the bottom through a sparger. The bubble column contains gas dispersed as bubbles in a continuous volume of liquid. Per Section 2.6, the flow is considered to be bubbly. Bubbles form and travel upwards through the column due to the inlet gas velocity and buoyancy. The gas introduced through the sparger provides mixing, similar to the gas mixing tank in Section 3.4.1 but much less intense. This method of mixing is less invasive and requires less energy than mechanical stirring. Bubble column reactors are often used in industry to develop and produce chemicals and fuels for use in chemical, biotechnology, and pharmaceutical processes. Figure 3.4.2-1 shows a schematic representation of a bubble column reactor. Figure 3.4.2-1: Bubble Column Reactor In all gas-liquid flows, the bubbles can increase or decrease in size due to coalescence or breakup. Coalescence occurs when two or more bubbles collide and the thin liquid barrier between them ruptures to form a larger bubble. Bubbles breakup occurs when a bubble collides with a turbulent eddy approximately equal to its size. The method to calculate the change in bubble size due to turbulent eddies is discussed in Section 3.4.3. Table 3.4.2-1 lists the important input needed to replicate the results shown in this section. 51 Table 3.4.2-1: Bubble Column Input Input Geometry Column Width Column Height 2D Space Solver Time Time Step Size Type Velocity Formulation Gravity Models Energy Viscous Multiphase Drag Turbulence Model Near Wall Treatment Turbulent Kinetic Energy Turbulent Dissipation Rate Initial Conditions Water Level Gas Flow Rate Bubble Diameter Material Properties (Water) Density Viscosity Material Properties (Air) Density Viscosity Surface Tension Solution Methods Scheme Gradient Momentum Volume Fraction Turbulent Kinetic Energy Turbulent Dissipation Rate Transient Formulation 52 Value 0.10 m 0.75 m Planar Transient 0.001 s Pressure Based Relative -9.8 m/s2 (Y-direction) Off Standard k-ε Eulerian Schiller-Nauman Standard 0 m2/s2 0 m2/s3 0.50 m 0.05 m/s 0.005 m 998.2 kg/m3 0.001003 kg/m-s 1.225 kg/m3 1.7894E-05 kg/m-s 0.072 N/m Phase Coupled SIMPLE Least Square Cell Based Second Order Upwind QUICK Second Order Upwind Second Order Upwind Second Order Implicit Figure 3.4.2-2 shows a comparison between gas volume fraction 1 second and 5 seconds after gas has begun flowing through the bubble column. At both time points the gas tends to flow in slugs. After 5 seconds, the gas has reached the top of the liquid and caused the surface to change shape. The liquid level is also higher after 5 seconds by about 5 cm. The level increase is known as gas holdup and caused by phase drag forces and displacement. Figure 3.4.2-2b shows that most of the gas travels along the wall in a quasi-annular flow type regime. (a) (b) Figure 3.4.2-2: Instantaneous Gas Volume Fraction After (a) 1 Second and (b) 5 Seconds Figure 3.4.2-3 shows a comparison between the liquid velocity vectors 1 second and 5 seconds after the gas has begun flowing through the bubble column. Distinct paths of liquid movement, primarily along the walls, can be seen at both time points. Due to buoyancy and phase drag forces, the largest liquid velocities coincide with the regions of greatest gas volume fraction. 53 (a) (b) Figure 3.4.2-3: Instantaneous Liquid Velocity Vectors After (a) 1 Second and (b) 5 Seconds Figure 3.4.2-4 shows a comparison between the gas velocity vectors 1 second and 5 seconds after gas has begun flowing through the bubble column. The white region two-thirds up the bubble column in Figure 3.4.2-4a is a region where the gas has not reached. It is noteworthy that the original gas-liquid interface is not flat but consists of two parabolas. This is occurs because most of the gas travels close to the wall (shown in Figure 3.4.2-2). Figure 3.4.2-4b shows that the greatest gas velocities occur near the walls which are also areas of greatest gas volume fraction. Higher gas volume fractions lead to greater buoyancy forces which cause greater gas velocities. A second case was completed to better understand the impact that gas inlet velocity has on gas holdup. This case is the same as the case described in Table 3.5.1-1 except that the gas inlet velocity was increased to 10 cm/s. Figure 3.4.2-5 shows the gas volume fraction 1 second and 5 seconds after gas has begun flowing through the bubble column. Comparing Figure 3.4.2-5a and Figure 3.4.2-5b reveals that the injected gas caused the water level to rise about 15 cm due to gas holdup. This is a much larger 54 increase than the gas hold shown in Figure 3.4.2-2, which employed a gas inlet velocity of 5 cm/s. (a) (b) Figure 3.4.2-4: Instantaneous Gas Velocity Vectors After (a) 1 Second and (b) 5 Seconds To ensure that the mesh has no impact on the results, a mesh validation was performed on the original case (gas velocity of 5 cm/s). The mesh validation was performed by comparing the results shown in this section (“Analysis Value” in Table 3.4.2-2) to a second case with an increased number of nodes and elements (“Mesh Validation” in Table 3.4.2-2). The second case was exactly the same as the case used in this section except that its mesh was refined. The results from the mesh validation are shown in Table 3.4.2-2, and prove that the results are mesh independent. 55 Figure 3.4.2-5: Instantaneous Gas Volume Fraction (10 cm/s) After (a) 1 Second and (b) 5 Seconds Table 3.4.2-2: Mesh Validation for Bubble Column Number of Nodes Number of Elements Max Liquid Velocity (m/s) Average Gas Velocity (m/s) Max Liquid Volume Fraction Max Static Pressure (Pa) Analysis Value 7006 6750 0.625945 0.313947 0.998733 4929.094 56 Mesh Validation 8785 8500 0.63157 0.308535 1.00000 4920.58 Difference 25.39 % 25.93 % 0.90 % 1.72 % 0.13 % -0.17 % 3.4.3 BUBBLE COLUMN WITH POPULATION BALANCE MODEL The bubble column model discussed in Section 3.4.2 was expanded to include a population balance model (PBM) so that the growth, coalescence, and breakup of the bubble swarm within the column can be tracked. For additional information on population balance and the model implemented within Fluent, see Appendix B. A population balance model with three discrete bubble sizes was added to the Section 3.4.2 bubble column model. Table 3.4.3-1 lists the input used to create the population balance model implemented in this section. Table 3.4.3-1: Population Balance Model Input Input Value Discrete 3 0.0075595 m 0.0047622 m 0.0030000 m Method Number of Bins Bin-0 Bin-1 Bin-2 Bin Distribution Bin-0 Bin-1 Bin-2 Aggregation Kernel Model Surface Tension Breakage Kernel Model Surface Tension Formulation 25 % 50 % 25 % Luo 0.072 N/m Luo 0.072 N/m Hagesather Figure 3.4.3-1 shows a comparison between the gas volume fraction at 1 second and 5 seconds after gas has begun flowing through the bubble column. When comparing Figure 3.4.3-1 to Figure 3.4.2-2, there are significant differences. One of the more obvious differences is the distribution of the gas phase at the two time points. With the population balance model implemented, Figure 3.4.3-1, the gas phase distribution is much more uniform without any large areas of high gas volume. This is most noticeable at the bottom of the bubble column. 57 (a) (b) Figure 3.4.3-1: Instantaneous Gas Volume Fraction with PBM After (a) 1 Second and (b) 5 Seconds Figure 3.4.3-2 shows a comparison between the liquid velocity vectors at 1 second and 5 seconds after gas has begun flowing through the bubble column. Figure 3.4.3-2b shows that the top of the bubble column has the largest liquid velocities. This is not as noticeable in Figure 3.4.2-3 where the liquid velocity is more uniform from top to bottom because the bubbles remain at a constant diameter. Greater liquid velocities are achieved at the top of the bubble column with the population balance model because of bubble coalescence. Table 3.4.3-1 shows that there are more, larger bubbles at the top of the column than at the bottom. The larger bubbles have more surface area which cause more drag between the liquid and gas phases. The larger bubbles also attain higher velocities because the buoyancy forces are larger. The combination of higher gas velocities and larger drag forces cause the liquid velocity to be greater. Figure 3.4.3-3 shows a comparison between the gas velocity vectors at 1 second and 5 seconds after gas has begun flowing through the bubble column. Similar to 58 Figure 3.4.2-4, the shape of the gas as it initially climbs the bubble column is made up of two adjacent parabolas; however, it is much more severe in Figure 3.4.3-3a. Figure 3.4.3-3b shows a uniform gas velocity distribution throughout the bubble column where there are no sections of little or no movement. This is different from Figure 3.4.2-4b where areas of no movement (in the center of the column) are prevalent. (a) (b) Figure 3.4.3-2: Bubble Column Liquid Vector Velocity with PBM After (a) 1 Second and (b) 5 Seconds The population balance model calculates the bubble size distribution at each axial height using the Luo breakup and coalescence model. Table 3.4.3-2 shows the bubble size population fraction at the inlet and outlet of the bubble column. This table shows that there is a strong bias for the smaller bubbles to coalesce into larger bubbles. Thus, surface tension is a strong driver to reduce surface area and there is very little turbulence to cause the bubbles to break apart. The impact that surface tension has on bubble size distribution was tested by reducing the surface tension by a factor of ten to 0.0072 N/m. Table 3.4.3-3 shows the 59 bubble size distribution at the inlet and outlet of the bubble column with the reduced surface tension. The smaller surface tension reduces the driving force for bubbles to coalesce and significantly reduces the average bubble diameter. Figure 3.4.3-3: Bubble Column Liquid Vector Velocity with PBM After (a) 1 Second and (b) 5 Seconds Table 3.4.3-2: Bubble Size Distribution – Surface Tension of 0.072 N/m Bin-0 (0.76 cm) Bin-1 (0.48 cm) Bin-2 (0.30 cm) Inlet (Fraction) 0.250 0.500 0.250 Outlet (Fraction) 0.865 0.117 0.018 Net (Fraction) +0.615 -0.383 -0.232 Table 3.4.3-3: Bubble Size Distribution – Surface Tension of 0.0072 N/m Bin-0 (0.0076 m) Bin-1 (0.0048 m) Bin-2 (0.0030 m) Inlet (Fraction) 0.250 0.500 0.250 Outlet (Fraction) 0.495 0.335 0.170 60 Net (Fraction) +0.245 -0.165 -0.080 3.5 BOILING FLOWS 3.5.1 POOL BOILING Pool boiling occurs when a liquid transforms to vapor due to energy absorption in a fluid that is stagnant. When the surface temperature of the heated surface sufficiently exceeds the saturation temperature of the liquid, vapor bubbles nucleate on the heated surface. The bubbles grow on the surface until they detach and move out into the bulk liquid. While rising is the result of buoyancy, the bubbles either collapse or continue to grow depending upon whether the liquid is locally subcooled or saturated. Pool boiling involves complex fluid motions initiated and maintained by the nucleation, growth, departure and collapse of bubbles, and by natural convection. [10] Table 3.5.1-1 lists the important input needed to replicate the results shown in this section. Table 3.5.1-1: Pool Boiling Input Input Geometry Column Width Column Height 2D Space Solver Time Time Step Size Type Velocity Formulation Gravity Models Energy Viscous Multiphase Drag Slip Mass Transfer Initial Conditions Bubble Diameter Initial Fluid Temperature Heater Temperature (Bottom) Backflow Temperature (Top) Backflow Volume Fraction (Top) 61 Value 0.01 m 0.05 m Planar Transient 0.002 s Pressure Based Relative -9.8 m/s2 (Y-direction) On Laminar Mixture Schiller-Nauman Manninen et al. Evaporation-Condensation 0.0002 m 372 K 383 K 373 K 0 Material Properties (Water) [15] Density Specific Heat Thermal Conductivity Viscosity Heat of Vaporization Material Properties (Vapor) [15] Density Specific Heat Viscosity Thermal Conductivity Surface Tension Solution Methods Scheme Gradient Pressure Momentum Volume Fraction Energy Transient Formulation See Table 3.5.1-2 4182 J/kg-K 0.6 W/m-K 0.001003 kg/m-s 2.418379E+08 J/kgmol 0.5542 kg/m3 2014 J/kg-K 1.34E-05 kg/m-s 0.0261 W/m-K 0.072 N/m PISO Least Square Cell Based Body Force Weighted Second Order Upwind QUICK Second Order Upwind Second Order Implicit Table 3.5.1-2: Pool Boiling Water Density Density (kg/m3) 974.9 958.4 Temperature (K) 348 373.15 Figure 3.5.1-1 shows the instantaneous gas volume fraction after 0.9 seconds and 1.7 seconds of heating. These two time points were chosen because the first time point shows steam releasing from the heated surface and entering the bulk fluid which is the driving force behind all fluid motion. The second time point was chosen because it shows how the fluid and vapor interact at a high level. The evolution of steam generation, upward movement (due to buoyancy) and liquid refill is shown in Figure 3.5.1-1 through Figure 3.5.1-3. Figure 3.5.1-1a shows that the entire bottom of the control volume is heated and some steam has formed (two areas of significant steam generation are shown in green). Figure 3.5.1-1b shows the vapor moving upward (teal region) and liquid taking its place (blue area at the bottom). 62 (a) (b) Figure 3.5.1-1: Instantaneous Gas Volume Fraction After (a) 0.9 Seconds and (b) 1.7 Seconds Figure 3.5.1-2 and Figure 3.5.1-3 display the liquid and gas velocities, respectively, at the two time points. Comparing these two figures indicates that the largest upward liquid and vapor velocities occur in generally the same regions. These regions also coincide with the areas of largest gas volume fraction (Figure 3.5.1-1). As vapor is formed on the heated surface, it eventually detaches and enters the liquid above. Due to buoyancy, the vapor travels upward through the liquid. Drag forces between the two phases cause the liquid to also travel upwards but at a slower rate due to slip. Other areas of high liquid velocity occur between the two swells of upward moving vapor and along the walls. The liquid being of greater density flows downward to refill the void created by the steam. This causes large velocity gradients and mixing. 63 (a) (b) Figure 3.5.1-2: Instantaneous Liquid Velocity Vectors After (a) 0.9 Seconds and (b) 1.7 Seconds (a) (b) Figure 3.5.1-3: Instantaneous Gas Velocity Vectors After (a) 0.9 Seconds and (b) 1.7 Seconds 64 Figure 3.5.1-4 shows the volume fraction of vapor on the heated surface after 2 seconds. Vapor is being produced significantly at two locations (vapor volume fraction is at a maximum), 0.0008 m and 0.0095 m. The vapor volume fraction is at a minimum at approximately 0.005 m which is where liquid is taking the place of the recently created vapor. Figure 3.5.1-4: Volume Fraction of Vapor on Heated Surface To ensure that the mesh had no impact on the results, a mesh validation was performed. The mesh validation was performed by comparing the results shown in this section (“Analysis Value” in Table 3.5.1-3) to a second case with an increased number of nodes and elements (“Mesh Validation” in Table 3.5.1-3). The second case was exactly the same as the case used in this section except that its mesh was refined. The results from the mesh validation are shown in Table 3.5.1-3, and prove that the results are mesh independent. Table 3.5.1-3: Mesh Validation for Pool Boiling Number of Nodes Number of Elements Min Mixture Density (kg/m3) Max Mixture Velocity (m/s) Min Liquid Volume Fraction Max Static Pressure (Pa) Max Phase Transfer (kg/m3-s) Analysis Value 26645 26208 754.389 0.059396 0.787011 452.2354 2.169675 65 Mesh Validation 32481 32000 742.115 0.062788 0.774197 452.2388 2.190905 Difference 21.90% 22.10% -1.63% 5.71% -1.63% 0.00% 0.98% 3.5.2 SUBCOOLED BOILING Subcooled boiling involves intense interactions between the liquid and vapor phases and therefore modeling can be a challenge. The Eulerian multiphase model is most appropriate because it is the only Fluent option capable of modeling multiple separate, yet interacting phases. When modeling subcooled boiling, there are three parameters of great importance. These parameters are the active nucleation site density (Na), departing bubble diameter (dbw) and bubble departure frequency (f) [1]. As discussed previously, nucleation sites are preferential locations where vapor tends to form. They are usually cavities or irregularities in a heated surface. However, not all sites are active and the number of nucleation sites per unit area is dependent on fluid and surface conditions. The departing bubble diameter is the bubble size when it leaves the heated surface and is dependent on the amount of subcooling and a balance of surface tension and buoyance forces. The bubble departure frequency is the rate at which bubbles are generated at an active nucleation site and it is dependent on heat flux and buoyancy and drag forces. The heat transfer rate from the wall to the fluid greatly impacts the number of active nucleation sites, bubble diameter and bubble departure frequency. The amount of energy transferred to the fluid changes based on the amount of vapor on the heated surface. Since the vapor area is constantly changing due to formulation, growth and departure of bubbles, the use of a correlation is necessary. Del Valle and Kenning created a mechanistic model to determine the area of the heated surface influenced by vapor during flow boiling. The most common active nucleation site density relationship was developed by Lemmert and Chwala. It is based on the heat flux partitioning data generated by Del Valle and Kenning [1]: ππ = [π(ππ ππ‘ − ππ€ )]π According to Kurul and Podowski, the values of m and n are 210 and 1.805 respectively. Another popular correlation for nucleation site density was created by Kocamustafaogullari and Ishii. They assumed that the active nucleation site density correlation developed for pool boiling could be used in forced convective system if the 66 effective superheat was used rather than the actual wall superheat. This correlation accounts for both the heated surface conditions and fluid properties and can be written as [1]: −4.4 1 2πππ ππ‘ ππ€ πππ ππ βππ ππ = π2 [βπ ] π(π∗ ) π(π∗ ) = 2.157 ∗ 10−7 ∗ π∗−3.2 ∗ (1 + 0.0049π∗ ) π∗ = ( ππ −ππ ππ ) Determining the lift off bubble diameter is crucial because the bubble size influences the interphase heat and mass transfer through the interfacial area concentration and momentum drag terms. Many correlations have been determined; however, the three discussed are applicable at low pressure, subcooled flow boiling. The first correlation was proposed by Tolubinsky and Kostanchuk. It establishes the bubble departure diameter as a function of the subcooling temperature [1]: πππ€ = πππ [0.0006 ∗ exp (− ππ π’π 45 ) ; 0.00014] On the basis of the balance between the buoyancy and surface tension forces at the heating surface, Kocamustafaogullari and Ishii modified an expression by Fritz that involved the contact angle of the bubble [1]: πππ€ = 2.5 ∗ 10−5 ( ππ − ππ π ) π√ ππ π ∗ (ππ − ππ ) A more comprehensive correlation was proposed by Unal which included the effect of subcooling and the convection velocity and heater wall properties [1]: πππ€ = 2.42 ∗ 10−5 ∗ π0.709 ∗ π √ππ· where π= (ππ€ − βππ π’π )1/3 ππ 2πΆ 1/3 βππ √πππ ⁄ππ πππ ππ √ ππ€ ππ€ πππ€ ππ π’π ;π = ππ ππ πππ 2[1 − (ππ − ππ )] 67 πΆ= βππ ππ [πππ ⁄(0.013βππ ππ 1.7 )] 3 π √π(π − π ) π π (π’π ) 0.47 πππ π’π ≥ 0.61 π/π Φ = {0.61 1.0 πππ π’π < 0.61 π/π The most common bubble departure frequency correlation for computational fluid dynamics was developed by Cole. It is derived from the bubble departure diameter and a balance between buoyancy and drag forces [1]: π=√ 4π(ππ − ππ ) 3ππ πππ€ The subcooled flow boiling model developed in Fluent uses the inputs listed in Table 3.5.2-1 to understand the impact that different boiling models and initial conditions have on axial liquid volume fraction. Table 3.5.2-1: Subcooled Flow Boiling Input Input Value Geometry Pipe Diameter Pipe Length 2D Space Solver Time Type Velocity Formulation Gravity Models Energy Viscous Near Wall Treatment Turbulent Intensity* Multiphase Drag Lift 68 0.03 m 0.50 m Axisymmetric Steady Pressure Based Relative -9.8 m/s2 (X-direction) On Realizable k-Ο΅ Enhanced 4.2079 % Eulerian Schiller-Nauman Boiling-Moraga Heat Transfer Mass Transfer Correlations Interfacial Area Bubble Diameter Initial Conditions Mass Flow Rate Inlet Fluid Temperature Wall Heat Flux Ranz-Marshall RPI Boiling See Table 3.5.2-3 Ia-Symmetric Sauter-Mean 0.3 kg/s 370 K 90000 W/m2 Material Properties (Water) Density See Table 3.5.2-2 Specific Heat See Table 3.5.2-2 Thermal Conductivity See Table 3.5.2-2 Viscosity See Table 3.5.2-2 Heat of Vaporization See Table 3.5.2-2 Material Properties (Vapor) Density 0.5542 kg/m3 Viscosity 1.34E-05 kg/m-s Thermal Conductivity 0.0261 W/m-K Surface Tension 0.072 N/m Solution Methods Scheme Coupled Gradient Least Square Cell Based Momentum Second Order Upwind Volume Fraction QUICK Turbulent Kinetic Energy Second Order Upwind Turbulent Dissipation Rate Second Order Upwind Energy Second Order Upwind * Calculated using equations from Table 2.5.1-2. Table 3.5.2-2: Subcooled Boiling Water Properties 368 K 370 K Density (kg/m ) 961.99 960.59 Specific Heat (J/kg-K) 4210.0 4212.1 Viscosity (kg/m-s) 0.0002978 0.0002914 Conductivity (W/m-K) 0.6773 0.6780 Heat of Vaporization (J/kgmol) N/A N/A Surface Tension (N/m) N/A N/A * Saturation temperature at atmospheric pressure. 3 69 373.15 K* 958.46 4215.5 0.0002822 0.6790 40622346 0.0589 To investigate the impact that the boiling models have on liquid volume fraction, a set of cases using the inputs listed in Table 3.5.2-1 and Table 3.5.2-3 were analyzed. Based on the modeling options in Fluent, six combinations were possible. The liquid volume fraction at different axial heights and the average liquid volume fraction between cases were compared. Table 3.5.2-3: Boiling Model Case Input Case Number 1 2 3 4 5 6 Bubble Departure Diameter Model Tolubinski-Kostanchuk KocamustafaogullariIshii Unal Tolubinski-Kostanchuk KocamustafaogullariIshii Unal Nucleation Site Density Model Lemmert-Chawla Lemmert-Chawla Frequency of Bubble Departure Model Cole Cole Lemmert-Chawla KocamustafaogullariIshii KocamustafaogullariIshii KocamustafaogullariIshii Cole Cole Cole Cole Plots of temperature, liquid volume fraction and mass transfer rate for Case 1 are shown in Figures 3.5.2-1, 3.5.2-2 and 3.5.2-3, respectively. Although these figures are specific to Case 1, their trends can be applied to all of the cases analyzed. Figure 3.5.2-1 shows how the liquid temperature increases. Note that the maximum bulk liquid temperature is about 373 K which is the fluid saturation temperature and therefore maximum temperature achievable by the liquid. Figure 3.5.2-1: Case 1 - Temperature (K) Figure 3.5.2-2 shows that the liquid volume fraction decreases as the fluid travels down the pipe. The phase change is caused by energy transferred from the walls to the flowing liquid and the small amount of subcooling at the pipe entrance. 70 Figure 3.5.2-2: Case 1 - Liquid Volume Fraction Figure 3.5.2-3 is of particular interest because it shows both the generation and destruction of steam bubbles. The light blue and teal areas next to the heated wall show that steam is being generated. After the bubbles grow in size they detach and join the bulk fluid. A small distance towards the pipe centerline away from the heated wall is a dark blue region. In this region the steam bubbles lose energy to the surrounding subcooled liquid and condense back into liquid. The generation and destruction of steam bubbles is characteristic of subcooled flow boiling. Figure 3.5.2-3: Case 1 - Mass Transfer Rate (kg/m3-s) The volume weighted liquid volume fraction for the six cases described in Table 3.5.2-3 are shown in Table 3.5.2-4. Case 4 predicted the largest liquid volume fraction while Case 2 predicted the smallest liquid volume fraction; however, the difference between the two cases is only about 1.6%. Therefore, the choice in boiling model has only a small impact on the overall liquid volume fraction for the conditions examined. The results also shows that the Kocamustafaogullari-Ishii nucleation site density model tends to predict a greater liquid volume fraction, meaning less vapor production, than the Lemmert-Chawla nucleation site density model. Cases 4 through 6 have a smaller liquid volume fraction range than Cases 1 through 3. This means that when the Kocamustafaogullari-Ishii nucleation site density model is employed, the choice of the bubble departure diameter model has less of an impact than if the Lemmert-Chawla nucleation site density model was employed. Comparing the results from a bubble departure diameter model perspective reveals that no one model has a 71 tendency to predict more or less of a liquid volume fraction. Thus, the nucleation site density model has a greater impact on liquid volume fraction than the bubble departure diameter model. Table 3.5.2-4: Boiling Model Case Results Case Number 1 2 3 4 5 6 Volume-Weighted Liquid Volume Fraction 0.91078539 0.90031346 0.90856631 0.91649488 0.91612881 0.91241595 Figure 3.5.2-4 shows the liquid volume fraction at nine axial heights for the six cases described in Table 3.5.2-3. Although Table 3.5.2-4 shows that the models predict similar liquid volume fractions within the entire control volume, Figure 3.5.2-4 shows that there are noticeable differences. First, there is a significantly higher liquid volume fraction 5 to 10 cm from the pipe inlet in Case 4 through 6 compared to Cases 1 through 3. Therefore, vapor formation using the Kocamustafaogullari-Ishii nucleation site density model requires more energy addition. Second, the liquid volume fraction 8 mm from the pipe centerline is significantly less in Cases 1 through 3 than in Cases 4 through 6. This is due to the smaller vapor production rate at the pipe wall. A second parametric study using the subcooled boiling model described in Table 3.5.2-1 was used to understand how inlet temperature, mass flow and heat flux impact liquid volume fraction. For this set of cases, the active nucleation site density was determined by the Lemmert and Chwala correlation, the bubble departure diameter was determined by the Tolubinsky and Kostanchuk correlation and the bubble departure frequency was determined by the Cole correlation. The liquid properties at three different temperatures are shown in Table 3.5.2-2 [15]. Six cases were analyzed in total as part of this parametric study. Case 1 from the first study is used as the nominal case to which the other six are compared. Cases 7 through 12 increase or decrease the heat flux, the inlet temperature or the mass flow rate relative to the Case 1 value. The input for the six cases analyzed is documented in Table 3.5.2-5. 72 (a) Case 1 (b) Case 2 (c) Case 3 (d) Case 4 (e) Case 5 (f) Case 6 Figure 3.5.2-4: Liquid Volume Faction Vs. Position for Cases 1-6 73 Table 3.5.2-5: Input Parametric Case Matrix Case Number 1 (base) 7 8 9 10 11 12 Inlet Temperature (K) 370 370 370 372 368 370 370 Mass Flow (kg/s) 0.30 0.30 0.30 0.30 0.30 0.33 0.27 Heat Flux (kW/m2) 90 100 80 90 90 90 90 The volume weighted liquid volume fraction for the six cases described in Table 3.5.2-5 are displayed in Table 3.5.2-6. Table 3.5.2-6 shows that the maximum and minimum liquid volume fractions occur in Case 8 and Case 9, respectively. The large impact that inlet temperature has on liquid volume fraction is attributed to the large specific heat of water (4212 J/kg-K). If the specific heat was smaller, the difference in liquid volume fraction between these two cases and the base case would be less. It is the specific heat property of water that makes it so useful in energy conversion cycles. Comparing the three cases that cause a decrease in liquid volume fraction from the base case (Cases 7, 9 and 12) to the three cases that cause an increase in liquid volume fraction from the base case (Cases 8, 10 and 11) demonstrates that the liquid volume fraction decreases more than it increases for the same delta change in initial conditions. Therefore, the initial conditions do not proportionally impact liquid volume fraction. Table 3.5.2-6: Input Parametric Case Results Case Number 1 7 8 9 10 11 12 Volume-Weighted Liquid Volume Fraction 0.91078539 0.87799626 0.93408281 0.57124303 0.96969908 0.92067945 0.89072032 Table 3.5.2-7 shows the liquid volume fraction at nine axial heights for the cases described in Table 3.5.2-5. This table allows for a finer comparison of liquid volume fraction between the cases analyzed. Table 3.5.2-7 verifies the relationships developed 74 using Table 3.5.2-6 and making observations based on overall liquid volume fraction is acceptable. Table 3.5.2-7: Axial Height Liquid Volume Fraction Height 0 cm 5 cm 10 cm 15 cm 20 cm 25 cm 30 cm 35 cm 40 cm Case 1 1.00000 0.99168 0.97680 0.96151 0.93987 0.91595 0.89784 0.88190 0.85984 Case 7 1.00000 0.98880 0.97050 0.95036 0.92220 0.89812 0.87830 0.85540 0.80840 Case 8 1.00000 0.99397 0.98231 0.97201 0.95598 0.93589 0.91644 0.90250 0.89019 Case 9 1.00000 0.95129 0.87624 0.80165 0.71266 0.57694 0.42719 0.31823 0.25132 Case 10 1.00000 0.99785 0.99390 0.98885 0.98427 0.97895 0.96784 0.95471 0.93927 Case 11 1.00000 0.99348 0.97938 0.96537 0.94748 0.92264 0.90098 0.88448 0.86812 Case 12 1.00000 0.98907 0.97482 0.95578 0.93222 0.91180 0.89572 0.87680 0.83296 Figure 3.5.2-5 shows the liquid volume fraction of the axial heights in Table 3.5.2-6 with respect to distance from the centerline. The x-axis is position, or distance from the centerline. The pipe wall is located at 0.015 m. The impact that inlet temperature (Case 8 and Case 9) has on liquid volume fraction is highly visible in Figure 3.5.2-5. Case 9 shows significant voiding in the centerline after 0.25 m from the pipe inlet due to the high inlet temperature (subcooling of about 1 K). Case 8 shows the opposite where 0.400 m from the pipe inlet there is no voiding 0.010 m from the pipe centerline. The liquid volume fraction at the nine axial heights from Cases 7 through 12 were compared to the liquid volume fraction of the base case (Case 1) at the same axial height using the following three equations for heat flux, inlet temperature and mass flow, respectively, where i stands for the axial height location. β(ππππ πΉππππ‘πππ) πΆππ π ππππ πΉππππ‘ππππ − π΅ππ π ππππ πΉππππ‘ππππ = β(π»πππ‘ πΉππ’π₯) πΆππ π π»πππ‘ πΉππ’π₯π − π΅ππ π π»πππ‘ πΉππ’π₯π β(ππππ πΉππππ‘πππ) πΆππ π ππππ πΉππππ‘ππππ − π΅ππ π ππππ πΉππππ‘ππππ = β(πΌππππ‘ ππππ. ) πΆππ π πΌππππ‘ ππππ.π − π΅ππ π πΌππππ‘ ππππ.π β(ππππ πΉππππ‘πππ) πΆππ π ππππ πΉππππ‘ππππ − π΅ππ π ππππ πΉππππ‘ππππ = β(πππ π πΉπππ€) πΆππ π πππ π πΉπππ€π − π΅ππ π πππ π πΉπππ€π 75 (a) Case 7 (b) Case 8 (c) Case 9 (d) Case 10 (e) Case 11 (f) Case 12 Figure 3.5.2-5: Liquid Volume Faction Vs. Position for Cases 7-12 76 The results of comparing the values from Table 3.5.2-7 using the three equations are shown in Table 3.5.2-8. For example, at an axial height of 10 cm, by increasing the heat flux from 90 kW/m2 to 100 kW/m2 (Case 1 to Case 7) the liquid volume fraction decreased by 0.0063 or 0.00063 per kW/m2. This calculation was carried out for Cases 7 and 8 since they both alter heat flux. The change in liquid volume fraction for each axial location was averaged to produce the overall impact that heat flux has on liquid volume fraction. The same process was followed for inlet temperature (Cases 9 and 10) and mass flow rate (Cases 11 and 12). Table 3.5.2-8: Absolute Impact on Liquid Volume Fraction Height 0 cm 5 cm 10 cm 15 cm 20 cm 25 cm 30 cm 35 cm 40 cm Average Case 7 Case 8 0.00000 0.00000 -0.00029 -0.00023 -0.00063 -0.00055 -0.00112 -0.00105 -0.00177 -0.00161 -0.00178 -0.00199 -0.00195 -0.00186 -0.00265 -0.00206 -0.00514 -0.00304 -0.00154 (kW/m2)-1 Case 9 Case 10 0.00000 0.00000 -0.02020 -0.00309 -0.05028 -0.00855 -0.07993 -0.01367 -0.11361 -0.02220 -0.16951 -0.03150 -0.23533 -0.03500 -0.28184 -0.03641 -0.30426 -0.03972 -0.08028 (K)-1 Case 11 Case 12 0.00000 0.00000 0.06000 0.08700 0.08600 0.06600 0.12867 0.19100 0.25367 0.25500 0.22300 0.13833 0.10467 0.07067 0.08600 0.17000 0.27600 0.89600 0.17178 (kg/s)-1 Table 3.5.2-8 shows the average impact that changing the heat flux, inlet temperature and mass flow rate have on the overall liquid volume fraction. Evaluating which of the three inputs is most impactful on liquid volume fraction is difficult in absolute terms (i.e. a 1 kg/s increase in mass flow rate is a larger percentage increase than a 10 kW/m2 increase in heat flux). Therefore, the values in Table 3.5.2-8 were compared on a percentage basis to provide further insight. Table 3.5.2-9 shows the liquid volume fraction change expected for a 1% change in each initial condition (heat flux, inlet temperature and mass flow rate). The second column of Table 3.5.2-9 repeats the initial conditions used in Case 1 (from Table 3.5.2-1), the third column shows 1% of the Case 1 value (for example, 90 kW/m2 * 0.01 = 0.9 kW/m2), the fourth column shows the results from Table 3.5.2-8, and the fifth column shows the outcome when columns three and four are multiplied together. 77 Table 3.5.2-9: Relative Impact on Liquid Volume Fraction Initial Condition Heat Flux Temperature Mass Flow Case 1 Value 90 kW/m2 370 K 0.3 kg/s 1% of Case 1 Value 0.9 kW/m2 3.70 K 0.003 kg/s Table 3.5.2-8 Results -0.00154 (kW/m2)-1 -0.08028 (K)-1 0.17178 (kg/s)-1 Equivalent Liquid Volume Fraction -0.00139 -0.29704 0.00052 Table 3.5.2-9 illustrates that a 1% increase in heat flux causes the average liquid void fraction to reduce by 0.00139, a 1% increase in temperature causes the average liquid void fraction to reduce by 0.29704 and a 1% increase in mass flow rate causes the average liquid void fraction to increase by 0.00052. It is understood that a 1% increase in the inlet temperature from the Case 1 condition would be greater than the saturation temperature at atmospheric pressure and therefore impossible; however, this exercise was performed to show the impact of the initial conditions in a more revealing manner. Table 3.5.2-9 indicates that inlet temperature has the greatest impact on liquid volume fraction, the wall heat flux has the second greatest impact and mass flow rate has the smallest impact. To ensure that the mesh had no impact on the results, a mesh validation was performed. The mesh validation was performed by comparing the results shown in this section (“Analysis Value” in Table 3.5.2-10) to a second case with an increased number of nodes and elements (“Mesh Validation” in Table 3.5.2-10). The second case was exactly the same as the case used in this section except that its mesh was refined. The results from the mesh validation are shown in Table 3.5.2-10, and prove that the results are mesh independent. Table 3.5.2-10: Mesh Validation for Subcooled Boiling Number of Nodes Number of Elements Max Liquid Velocity (m/s) Max Gas Velocity (m/s) Min Liquid Volume Fraction Max Phase Transfer (kg/m3-s) Analysis Value 25000 23976 0.8181624 0.9972627 0.4876771 24.87638 78 Mesh Validation 31000 29970 0.8199201 0.9982293 0.4853158 26.22442 Difference 24.00% 25.00% 0.21% 0.10% -0.48% 5.42% 4. DISUSSION AND CONCLUSIONS This thesis provided theoretical background and development of computational fluid dynamic models for various fluid flow and heat transfer phenomena. The areas explored include natural convection, laminar flow, turbulent flow with and without heat transfer, two-phase flow, pool boiling and subcooled flow boiling. Natural convection models of a heated horizontal cylinder and a heated vertical plate were developed in Section 3.1. These models implemented the Boussineq approximation to calculate density and thus temperature gradients and buoyancy forces. The heated horizontal cylinder model predicted a greater maximum velocity compared to the heated vertical plate even though the two models used the same surface and bulk fluid temperatures. The heated vertical plate had a lower maximum velocity due to drag forces invoked by the surface. Both natural convection models showed good agreement qualitatively and quantitatively with experimental data. Laminar flow within a pipe was investigated in Section 3.2. The parabolic velocity profile that is characteristic of laminar flow matched well qualitatively with experimental data. Also, as expected, the radial velocity for most of the pipe was near zero and was less than 10-3 times the average axial velocity. Two models involving turbulent flow within a pipe were created as part of Section 3.3. As expected, the velocity profiles calculated where flat and velocity magnitude didn’t decrease until very close to the pipe wall which matched well qualitatively with experimental data. The wall shear stress reached a maximum a slight distance from the entrance due to entrance effects causing a surge in radial velocity which caused a dramatic reduction in axial velocity. The turbulent flow model with the energy equation employed showed a very small increase in the fluid velocity magnitude due to the constant mass flow rate and the reduction in density due to the added energy. Two-phase flow involving water and air was examined as part of Section 3.4. The first model was a mixing tank using an air jet to stir the liquid. Effects of Rayleigh instability were observed. Before the jet broke the surface of the water, it became wavy and surface tension started to transform the jet into bubbles to reduce surface area. The second model created was a bubble column reactor. XXXXXXX was observed to occur. 79 Gas holdup was observed due to phase drag forces and displacement. The amount of gas holdup was found to be related to inlet gas velocity however the relationship was not linear. A population balance model was employed for two bubble column cases. The model predicted that the air bubbles would coalesce and grow in size as they traveled up the bubble column due to surface tension. When the surface tension was reduced, the number of bubbles that grew in size dramatically reduced. Section 3.5 discussed phase transformation due to heat addition in both stagnant and flowing liquids. The pool boiling model showed the progression of vapor formation on the heated surface, detachment and liquid refill. Drag forces between the two phases causes the liquid to travel upwards with the rising vapor but at a slower rate due to slip. The second phase transformation model developed and the focus of this research was the subcooled flow boiling model. The impact that different boiling model options have on liquid volume fraction was investigated. Three bubble departure diameter models and two nucleation site density models were analyzed using the same initial conditions. The bubble departure diameter models did not show any relationship with liquid volume fraction; however, the Kocamustafaogullari-Ishii nucleation site density model tended to predict a greater liquid volume fraction, meaning less vapor production, than the Lemmert-Chawla nucleation site density model. A second study on how initial conditions impact liquid volume fraction was explored. Numerous cases were analyzed that increased or decreased the heat flux, the inlet temperature or the mass flow compared to a base case. The differences in liquid volume fraction between the cases were used to develop relationships between heat flux, inlet temperature and mass flow rate with respect to liquid volume fraction. Overall, the inlet temperature had the greatest impact on liquid volume fraction, the wall heat flux had the second greatest impact and mass flow rate had the smallest impact. 80 REFERENCES 1. Yeoh, G. H; Tu, J. Y., “Modelling Subcooled Boiling Flows,” Nova Science Publishers, Inc., 2009. 2. Krepper, E.; Koncar, B.; Egorov, Y., “CFD Modeling of Subcooled Boiling – Concept, Validation and Application to Fuel Assembly Design,” Elsevier B.V., 2006. 3. Bird, R. B., Steward, W. E., Lightfoot, E. N., “Transport Phenomenon,” Wiley & Sons Inc., 2nd Edition, 2007. 4. Patankar, S. V., “Numerical Heat Transfer and Fluid Flow,” Hemisphere Publishing Co., 1st Edition, 1980. 5. Kays, William, Crawford, Michael, Bernhard, Weigand, “Convective Heat and Mass Transfer,” McGraw-Hill, 4th Edition, 2005. 6. Incropera, DeWitt, Bergman, Levine, “Introduction to Heat Transfer,” Wiley & Sons Inc., 5th Edition, 2007. 7. Tennekes, H; Lumley, J. L., “A First Course in Turbulence,” The MIT Press, 1972. 8. ANSYS, Inc., “ANSYS Fluent 14.0 Theory Guide,” 2012. 9. F. H. Harlow; P. I. Nakayama, “Transport of Turbulence Energy Decay Rate,” Los Alamos Sci. Lab., LA-3854, 1968. 10. Tong, L. S. “Boiling heat Transfer and Two-Phase Flow,” Wiley & Sons Inc., 2nd Edition, 1965. 11. Faghri, A.; Zhang, y.; Howell, J., “Advanced Heat and Mass Transfer,” Global Digital Press, 2010. 81 12. Eckert, E. R. G., “Introduction to the Transfer of Heat and Mass,” 1st Edition, 1950. 13. Ingham, D. B., “Free-Convection Boundary Layer on an Isothermal Horizontal Cylinder,” Journal of Applied Mathematics and Physics, Vol. 29, 1978. 14. Ostrach, S., “An Analysis of Laminar Free-Convection Flow and Heat Transfer About a Flat Plate Parallel to the Direction of the Generating Body Force,” Report 1111 – National Advisory Committee for Aeronautics. 15. NIST/ASME Steam Properties, Database 10, Version 2.11, 1996. 16. Krepper, E.; Rzehak, R., “CFD for Subcooled Flow Boiling: Simulation of DEBORA Experiments,” Elsevier B.V., 2011. 82 APPENDIX A: MULTIPHASE FLOW MODELS A.1 VOLUME OF FLUID MODEL The VOF model can model two or more immiscible fluids by solving a single set of momentum equations and tracking the volume fraction of each of the fluids throughout the domain. Typical applications include the prediction of jet breakup, the motion of large bubbles in a liquid, the motion of liquid after a dam break, and the steady or transient tracking of any liquid-gas interface. [8] A.2 MIXTURE MODEL The mixture model is a simplified multiphase model that can be used in different ways. It can be used to model multiphase flows where the phases move at different velocities, but assume local equilibrium over short spatial length scales. It can be used to model homogeneous multiphase flows with very strong coupling and phases moving at the same velocity and lastly, the mixture models are used to calculate non-Newtonian viscosity. The mixture model can model multiple phases (fluid or particulate) by solving the momentum, continuity, and energy equations for the mixture, the volume fraction equations for the secondary phases, and algebraic expressions for the relative velocities. Typical applications include sedimentation, cyclone separators, particle-laden flows with low loading, and bubbly flows where the gas volume fraction remains low. The mixture model is a good substitute for the full Eulerian multiphase model in several cases. A full multiphase model may not be feasible when there is a wide distribution of the particulate phase or when the interphase laws are unknown or their reliability can be questioned. A simpler model like the mixture model can perform as well as a full multiphase model while solving a smaller number of variables than the full multiphase model. [8] 83 APPENDIX B: POPULATION BALANCE MODEL Many industrial fluid flow applications including subcooled boiling involve a secondary phase with a size distribution. The size distribution of particles may include solid particles, bubbles, or droplets that evolve in a multiphase system. Thus, in multiphase flows involving a size distribution, a balance equation is required to describe the changes in the particle size distribution, in addition to momentum, mass, and energy balances. This balance is generally referred to as the population balance. To make use of this modeling concept, a number density function is introduced to account for the different sizes in the particle population. With the aid of particle properties (for example, particle size, porosity, composition, and so on), different particles in the population can be distinguished and their behavior can be described. [8] The population balance model gives the ability to track steam bubbles on a particle size basis after they have detached from a heated wall. The fate of a steam bubble traveling in a subcooled bulk fluid is not well understood. There are a number of possibilities that can occur which include breakup into smaller steam bubbles due to turbulent eddies, coalescence of multiple bubbles into one larger bubble or shrinkage due to transfer of energy from the bubble to the surrounding fluid. The growth rate is based on particle volume and therefore surface area. In nucleate boiling, the bulk fluid is subcooled. When steam bubbles form on the heated surface and eventually detach, they travel within the subcooled bulk fluid loosing energy through the steam-liquid interface causing the bubbles to shrink. The birth and death of particles can occur due to breakage and aggregation processes. In the case of subcooled nucleate boiling, mixing caused by turbulence plays an important role. Particle birth (and death) is caused by the breakage of a single large bubble into multiple smaller bubbles due to liquid turbulence eddies. Particle death (and birth) is due to the coalescence of multiple small bubbles into one larger bubble. In boiling applications, another way that bubbles are born is through phase change. Bubbles form on the heated wall at preferential locations called nucleation sites. The number of potential nucleation sites is dependent on the surface condition of the heated wall. A very smooth surface has a low number of cavities and therefore a low 84 number of potential nucleation sites. A rough surface has a large number of cavities and therefore a large number of potential nucleation sites. However, just because a heated surface has a high number of potential nucleation sites it does not mean that they are all active. An empirical formula governing the population of active sites is: Μ = π0 exp (− π πΎ 3 ) ππ€πππ Where N0 and K represent the liquid and surface conditions [10]. It can be seen that the population of active sites is a strong function of wall temperature and therefore heat flux. [10] B.1 EQUATION FORMULATION The goal of this section is to present an overview of the theory and governing equations used to calculate particle growth and nucleation. [8] B.1.1 PARTICLE STATE VECTOR The particle state vector is characterized by a set of external coordinates (π₯), which denote the spatial position of the particle and “internal coordinates” (φ), which could include particle size, composition, and temperature. From these coordinates, a number density function π(π₯, φ, t) can be postulated where φ Ο΅ πΊπ , π₯ π πΊπ₯ . Therefore, the average number of particles in the infinitesimal volume πππ₯ πππ is π(π₯, φ, t) πππ₯ πππ . The total number of particles in the entire system is ∫ ∫ ππππ₯ πππ ππ₯ β ππ The local average number density in physical space (that is, the total number of particles per unit volume is given by π(π₯, π‘) = ∫ ππππ πΊπ The total volume fraction of all particles is given by 85 πΌ(π₯, π‘) = ∫ π π(π) πππ πΊπ Where π(π) is the volume of a particle in state φ. B.1.2 POPULATION BALANCE EQUATION Assuming that φ is the particle volume, the transport equation for the number density function is given as: π ππ‘ [π(π, π‘)] + ∇ β [π’ β π(π, π‘)] + ∇π β [πΊπ π(π, π‘)] = π ∫ π 2 0 1 (π − π ′ , π ′ ) π (π − π ′ , π‘) π (π ′ , π‘) ππ ′ ∞ Birth due to Aggregation − ∫0 π (π, π ′ ) π (π, π‘) π (π ′ , π‘) ππ ′ Death due to Aggregation + ∫πΊ ππ (π ′ ) π½ (π|π ′ ) π (π ′ , π‘) ππ ′ Birth due to Breakage −π (π) π (π, π‘) Death due to Breakage π The boundary and initial conditions are given by π (π, π‘ = 0) = ππ ; π(π = 0, π‘) πΊπ = πΜ 0 Where πΜ 0 is the nucleation rate in particles / m3-s. 86 APPENDIX C: SUBCOOLED BOILING Subcooled flow boiling is a very efficient form of heat transfer that is described as having high heat transfer rates and low levels of wall superheat. Figure 3.5.2-1 [10] shows the various boiling regimes as a function of void fraction as the fluid travels along the heated surface. The void fraction in Region I is small and the level of voiding is mainly dependent on surface flux conditions. This region is known as wall voidage. Region II is known as the bubble detachment region and is mainly dependent upon the bulk flow characteristics. Eventually bulk boiling begins to occur and the MartinelliNelson curve can be used to determine void fraction. Figure 3.5.2-1: Void Fraction in Various Boiling Regimes If the heat flux from a heated wall into a subcooled fluid is slowly increased for a set of initial conditions, a point will be reached, known as the onset of nucleate boiling, where the transition from single-phase convection to subcooled flow boiling occurs. During nucleate boiling, heat transfer rates increase dramatically due to bubbles formation on the heated surface. As the bubble generation rate increases, heat carried by bubbles becomes a larger portion of the total energy transferred. If the wall heat flux is 87 allowed to increase further, the transition from subcooled flow boiling to saturated flow boiling will occur when the bulk fluid temperature reaches the saturation point. Although saturated flow boiling is an important form of heat transfer, the primary topic of this section is subcooled flow boiling. The efficient heat transfer mechanism provided by vapor generation in subcooled flow boiling is limited to the point where vapor generation exceeds the rate at which the liquid can replace it on the heated surface which leads to a greater portion of the heated surface being covered by vapor. This is known as the critical heat ο¬ux where the heat transfer coefο¬cient begins to decrease with increasing temperature leading to an unstable situation. In this event, the temperature of the heated surface increases rapidly which can lead to melting or destruction of the heater. The critical heat flux is dependent upon the working fluid, the mass flux, the inlet temperature and the saturation pressure. The veriο¬cation of design improvements and their inο¬uence on the critical heat ο¬ux requires expensive experiments. Therefore, the supplementation of experiments by numerical analyses is of high interest in industrial applications. [16] 88