Finite Element Modeling and Simulation of the Effect of Water Injection on Gas Turbine Combustor NOx Emissions and Component Temperature Joseph M. Basile 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: _________________________________________ Dr. Sudhangshu Bose, Thesis Adviser _________________________________________ Dr. Ernesto Gutierrez-Miravete, Committee Member _________________________________________ Dr. Norberto Lemcoff, Committee Member Rensselaer Polytechnic Institute Troy, New York November 2014 (For Graduation December 2014) © Copyright 2014 by Joseph M. Basile All Rights Reserved ii CONTENTS NOMENCLATURE .......................................................................................................................v LIST OF TABLES ...................................................................................................................... vii LIST OF FIGURES .................................................................................................................... viii ACKNOWLEDGMENT ................................................................................................................x ABSTRACT ................................................................................................................................. xi 1. INTRODUCTION ........................................................................................................1 1.1. Gas Turbine Operation ................................................................................................1 1.2. Nitrogen Oxide Formation ...........................................................................................3 1.2.1. Thermal Nitric Oxide .......................................................................................4 1.2.2. Prompt Nitric Oxide .........................................................................................4 1.2.3. Nitrous Oxide Mechanism................................................................................5 1.2.4. Fuel Nitrous Oxide ............................................................................................6 1.3. Water Injection .............................................................................................................6 2. METHODOLOGY .......................................................................................................7 2.1. Theory ............................................................................................................................7 2.1.1. Fluid Flow and Momentum .............................................................................8 2.1.2. Energy Balance .................................................................................................9 2.1.3. Equation of State.............................................................................................10 2.1.4. Turbulence.......................................................................................................10 2.1.5. Discretization of Model Equations ................................................................11 2.2. Modeling ......................................................................................................................11 2.2.1. Baseline Selection ............................................................................................11 2.2.2. Combustor Modeling ......................................................................................12 2.2.3. Prediction of NOx Emissions ..........................................................................21 2.2.4. 2D Axisymmetric Simulation of Combustion Process .................................22 2.2.5. 2D Axisymmetric Simulation of Water Injection ........................................27 iii 2.2.6. 3D Simulation of Combustion Process ..........................................................35 2.2.7. 3D Simulation of Water Injection .................................................................40 3. RESULTS AND DISCUSSION .................................................................................43 3.1. 2D Axisymmetric Models ...........................................................................................43 3.2. 3D Models ....................................................................................................................47 4. CONCLUSIONS .........................................................................................................52 5. REFERENCES............................................................................................................53 APPENDIX A: Determination of Boundary Conditions ........................................................54 Appendix A-1: 2D axisymmetric model, no water injection ............................................54 Appendix A-2: 2D axisymmetric model, with water injection .........................................55 Appendix A-3: 3D model, no water injection.....................................................................58 Appendix A-4: 3D model, with water injection .................................................................60 iv NOMENCLATURE ∇ a Cp CR 𝛿𝑖𝑗 𝜕 𝜕𝑡 𝜕 𝜕𝑛 𝐷 𝐷𝑡 𝐸𝑡 𝛾 e ε f f 𝐟𝐢 FA ℎ𝑓̇ 𝐻𝑓 Hw i J k k K kg N N2O NO NO2 NOx m 𝑚̇ 𝑝 ρ Pa ∏ij 𝒒 Gradient Ambient Heat capacity at constant pressure (J/kg) Compression ratio Kronecker delta Partial derivative with respect to time Partial derivative with respect to some variable n Derivative with respect to time Total energy per unit volume (J/m3) Ratio of specific heats (1.4) Natural exponent (~2.71) Turbulent dissipation rate (J/kg/s) Fuel Body force per unit mass (N) Body force per unit mass in the i direction (N) Fuel to air ratio Heat of combustion (W) Heating value of fuel (J/kg) Latent heat of evaporation for water (J/kg) Problem domain inlet Joules (kg*m2/s2) Thermal conductivity (W/m/K) Turbulent kinetic energy (J/kg) Degrees Kelvin Kilograms Newtons (kg*m/s2) Nitrous oxide Nitric oxide Nitrogen dioxide Nitrogen oxides Meters Mass flow rate (kg/s) Pressure (Pa) Density (kg/m3) Pascals (N/m2) Stress tensor (Pa) Heat flux vector (J/s) v 𝑄 R s T 𝐮 𝑢 ui 𝜇 𝜇𝑇 𝑣 w 𝑤 W External heat addition per unit volume (J/m3) Specific gas constant for air (287 J/kg/K) Seconds Temperature (K) Velocity Vector (m/s) Velocity magnitude in the x direction (m/s) Velocity component in the i direction (m/s) Viscosity Turbulent viscosity (Pa*s) Velocity magnitude in the y direction (m/s) water Velocity magnitude in the z direction (m/s) Watts (J/s) vi LIST OF TABLES Table 1: Values used for independent turbulence parameters ..................................................... 11 Table 2: Engine Performance: Key Parameters [9] ..................................................................... 18 Table 3: Problem domain inlet boundary conditions and target flow rates ................................. 21 Table 4: NOx emissions dependence on combustor exit temperature [10] ................................. 21 Table 5: Operating characteristics for the combustor, without water injection .......................... 27 Table 6: Operating characteristics for the combustor, with water injection ............................... 34 Table 7: Target flow rates and boundary conditions, no water injection .................................... 37 Table 8: Operating characteristics for the 3D combustor model, without water injection ......... 38 Table 9: 3D model target flow rates and boundary conditions, with water injection ................. 41 Table 10: Operating characteristics for the 3D combustor model, without water injection........ 42 Table 11: Operating characteristics - 2D axisymmetric models ................................................. 46 Table 12: 2D NOx production with and without water injection ................................................ 46 Table 13: Operating characteristics - 3D models ........................................................................ 50 Table 14: 3D NOx production with and without water injection ................................................ 50 Table 15: 2D model iterative convergence toward target mass flow rates, no water injection ... 54 Table 16: 2D model iterative convergence toward target mass flow rates, with water injection 55 Table 17: 3D model iterative convergence toward target mass flow rates, no water injection ... 58 Table 18: 3D model iterative convergence toward target mass flow rates, with water injection 60 vii LIST OF FIGURES Figure 2: Section view showing the four thermodynamic states in a gas turbine engine [2] ........ 2 Figure 3: Brayton Cycle P-V Diagram .......................................................................................... 3 Figure 4: Combustion Chamber Dimensional Analysis [9] ........................................................ 13 Figure 5: Full 3D combustor model ............................................................................................ 13 Figure 6: Combustion Chamber Simplified Model - Isometric View ......................................... 14 Figure 7: Combustion Chamber Model - Side View ................................................................... 15 Figure 8: Engine Air Flow Region .............................................................................................. 16 Figure 9: Air flow region and combustion chamber in COMSOL.............................................. 17 Figure 10: 2D Axisymmetric air flow region and combustion chamber in COMSOL ............... 18 Figure 11: 2D axisymmetric combustor model with combustion region .................................... 23 Figure 12: 2D axisymmetric mesh, no water injection ............................................................... 24 Figure 13: Air mass flow rate dependence on backpressure with quadratic curve fit ................. 25 Figure 14: Fuel mass flow rate dependence on fuel inlet pressure with linear curve fit ............. 26 Figure 15: 2D axisymmetric model temperature distribution, no water injection ...................... 27 Figure 17: 2D axisymmetric mesh, with water injection ............................................................ 30 Figure 18: Compressor air flow rate as a function of combustor backpressure .......................... 31 Figure 19: Fuel mass flow rate as a function of fuel inlet pressure............................................. 32 Figure 20: Water mass flow rate as a function of water inlet pressure ....................................... 33 Figure 21: 2D axisymmetric model temperature distribution, with water injection ................... 34 Figure 22: 3D combustion chamber model with combustion region .......................................... 35 Figure 23: Fuel flow from the fuel injector through the combustion region ............................... 36 Figure 24: 3D problem domain mesh, no water injection ........................................................... 37 Figure 25: 3D streamlines showing flow from the compressor outlet to combustor outlet, no water injection ............................................................................................................................. 38 Figure 26: Temperature profile of flow through the domain, no water injection ....................... 39 Figure 27: 3D combustion chamber model with heat absorbing region ..................................... 40 Figure 28: 3D problem domain mesh, with water injection ........................................................ 41 Figure 29: 3D streamlines showing flow from the compressor outlet and water inlet to combustor outlet .......................................................................................................................... 42 Figure 30: Temperature profile of flow through the domain, with water injection .................... 43 viii Figure 31: 2D absolute pressure results without water injection (left) and with (right) ............. 44 Figure 32: 2D Velocity magnitude results without water injection (left) and with (right) ......... 44 Figure 33: 2D Temperature results without water injection (left) and with (right) .................... 45 Figure 34: 2D wall temperature without water injection (left) and with (right) ......................... 47 Figure 35: 3D absolute pressure results without water injection (left) and with (right) ............. 48 Figure 36: 3D Velocity magnitude results without water injection (left) and with (right) ......... 48 Figure 37: 3D Temperature results without water injection (left) and with (right) .................... 49 Figure 38: 3D wall temperature without water injection (left) and with (right) ......................... 51 Figure 39: 2D model curve fit and predicted backpressure value to achieve the target inlet air mass flow rate, no water injection ............................................................................................... 54 Figure 40: 2D model curve fit and predicted fuel line pressure value to achieve the target fuel mass flow rate, no water injection ............................................................................................... 55 Figure 41: 2D model curve fit and predicted backpressure value to achieve the target inlet air mass flow rate, with water injection ............................................................................................ 56 Figure 42: 2D model curve fit and predicted fuel line pressure value to achieve the target inlet air mass flow rate, with water injection ...................................................................................... 57 Figure 43: 2D model curve fit and predicted water injection pressure value to achieve the target inlet air mass flow rate ................................................................................................................ 58 Figure 44: 3D model curve fit and predicted backpressure value to achieve the target inlet air mass flow rate, no water injection ............................................................................................... 59 Figure 45: 3D model curve fit and predicted fuel line pressure value to achieve the target fuel mass flow rate, no water injection ............................................................................................... 59 Figure 46: 3D model curve fit and predicted backpressure value to achieve the target inlet air mass flow rate, with water injection ............................................................................................ 60 Figure 47: 3D model curve fit and predicted fuel line pressure value to achieve the target inlet air mass flow rate, with water injection ...................................................................................... 61 Figure 48: 3D model curve fit and predicted water injection pressure value to achieve the target inlet air mass flow rate ................................................................................................................ 62 ix ACKNOWLEDGMENT Type the text of your acknowledgment here. x ABSTRACT Gas turbine engines are airbreathing engines which use the expansion of high temperature exhaust through an axial turbine to generate shaft work and, if the application requires it, thrust. Applications include electrical generating equipment and aircraft engines. These engines typically operate at temperatures high enough to spontaneously generate nitrogen oxides through various chemical mechanisms. These combustion byproducts are considered a significant source of environmental pollution and reduction of such emissions is a topic of contemporary research. In addition to pollution, the high temperature operation of these engines also necessitates the use of heat resistant materials for engine components. High temperatures can damage or wear engine components, increasing maintenance costs. Several numerical models have been developed to show that the production of nitrogen oxides can be reduced while simultaneously protecting engine components. The models and simulations used show that the surfaces of internal engine components can be maintained at low temperatures by spraying atomized liquid water into the flow stream. The temperature is reduced because heat from the combustion process will be absorbed by the water as it evaporates. The addition of water will also increase the overall mass flow rate through the engine and reduce the resultant flame temperature, thereby lowering the emission of nitrogen oxides. By injecting water into an appropriate part of the engine, these benefits can be realized simultaneously, thus increasing the advantage to implementing a water injection system. A 2D axisymmetric model of a gas turbine interior and combustion chamber showed general correlation to available test data and demonstrated the feasibility of the concept. A 3D model considering a 30° segment of the engine interior and combustion chamber showed improved accuracy with respect to available test data and clearly demonstrated the benefits of careful selection of water injection location. While a 2D axisymmetric model can provide general guidance for design decisions, a full 3D model is necessary for accurately predicting the best injection site. xi 1. INTRODUCTION 1.1. Gas Turbine Operation Gas turbine engines are airbreathing engines which use the expansion of high temperature exhaust through an axial turbine to generate shaft work and, if the application requires it, thrust. Typically, an axial compressor stage at the intake raises the temperature and pressure of incoming working fluid (typically air). From the compressor it enters the combustion chamber, where gaseous or atomized fuel is injected. The combustion process is typically started with an independent ignition source, but is self sustaining with adequate intake air and fuel flow rates. From the combustion chamber it flows through a turbine stage where work is extracted and then is expanded out the engine exhaust. Since their initial development in the early and mid twentieth century, gas turbine engines have evolved into a high power, efficient means of producing power and thrust. They have been used in a wide variety of applications, most visibly as the primary power source for commercial aviation aircraft. [1] A typical gas turbine engine cutaway, a General Electric J85-GE-17A turbojet, is shown in Figure 1. Figure 1: Sectioned Turbojet Gas Turbine Engine (air flow is left to right) [2] 1 More specifically, gas turbine engines operate based on an open loop Brayton Cycle. There are four different thermodynamic processes that occur between the four distinct states in the cycle. The different regions to which these states correspond are shown in Figure 2. Figure 2: Section view showing the four thermodynamic states in a gas turbine engine [2] The following is a description of the ideal Brayton Cycle, and ignores losses that would be found in a real engine. State 1 is the ambient gas which enters the engine intake. Between states 1 to 2, the gas compressed isentropically by an axial-flow rotating compressor. Therefore, at state 2 it is at elevated temperature and pressure. From the compressor stage it enters the combustion chamber, where the burning fuel heats it isobarically from states 2 to 3. From states 3 to 4, the gas expands isentropically through the turbine. For a jet engine, work is extracted as a torque on the turbine shaft and a reaction force (thrust) acting toward the intake. The turbine shaft is directly coupled to the compressor and thus work extracted by the turbine stage is used to compress the gas at the intake. This makes the cycle self-sustaining. From states 4 to 1, the gas is cooled isobarically back to ambient conditions, typically outside of 2 the engine housing. Figure 3 shows a typical pressure vs. volume diagram for an ideal Brayton Cycle. 3 Pressure (N/m2) 2 1 4 Volume (m3) Figure 3: Brayton Cycle P-V Diagram The thermodynamic efficiency of a real engine operating on the Brayton Cycle is dependent on the temperature change from State 1 to State 2 during compression. This means that a higher compression ratio will yield higher combustor inlet temperatures and, therefore, a higher cycle efficiency. 1.2. Nitrogen Oxide Formation A side effect of the operation of gas turbine engines is the formation of pollutants in the exhaust. Nitrogen oxides are one of the most harmful pollutants created during the combustion process. Nitrogen oxides generally include nitric oxide (NO) and nitrogen dioxide (NO2), and emissions generally refer to the sum total of NO and NO2 simply as NOx. Generally, high 3 combustion temperatures causes a reaction between the oxygen and nitrogen in the air flow and combustion products, forming NOx. The formation of NOx increases as the compression ratio and flame temperature of the engine increase. Gas turbine development has had a clear trend toward higher compression ratios and corresponding higher turbine inlet temperatures. This trend results in engines with increased performance and efficiency, but has also led to increasing nitrogen oxide emissions. Nitrogen oxides are formed when gaseous nitrogen from the intake oxidizes at high temperatures. Nitrogen oxide formation increases rapidly as flame temperature increases, especially for temperatures in excess of 1800K. [6] The U.S. government has identified nitrogen oxides as an air pollutant and taken steps to control its emission from gas turbine engines. [3] There are four distinct mechanisms through which NOx forms during the combustion process. NOx can form through the thermal nitric oxide route, the prompt nitric oxide route, the nitrous oxide (N2O) mechanism, and the fuel nitric oxide route. [4] [5] 1.2.1. Thermal Nitric Oxide The thermal nitric oxide route forms NO based on the Zel'dovich mechanism. This mechanism is described by the following reactions: 𝑂2 → 2𝑂 𝑂 + 𝑁2 → 𝑁𝑂 + 𝑁 𝑁 + 𝑂2 → 𝑁𝑂 + 𝑂 𝑁 + 𝑂𝐻 → 𝑁𝑂 + 𝐻 The first two reactions can be interpreted as a decomposition of the oxygen and nitrogen molecules in air generating NOx at elevated temperatures. However, in gas turbine combustors, the residence times of these components in the flame region are generally insufficient for the flow to reach chemical equilibrium. [4] 1.2.2. Prompt Nitric Oxide 4 The prompt nitric oxide route occurs in the combustion region much faster than NOx formed by the thermal NO route. This mechanism occurs when hydrocarbon radials react with N2 to form hydrocyanic acid (HCN), which then reacts for form NO. This is described by the following reaction: 𝑁2 + 𝐶𝐻 → 𝐻𝐶𝑁 + 𝑁 Although other reactions take place, this is the primary path for HCN, and therefore NO formation. Formation of NO from HCN occurs according to the following reaction sequence: 𝐻𝐶𝑁 + 𝑂 → 𝑁𝐶𝑂 + 𝐻 𝑁𝐶𝑂 + 𝐻 → 𝑁𝐻 + 𝐶𝑂 𝑁𝐻 + 𝐻 → 𝑁 + 𝐻2 𝑁 + 𝑂𝐻 → 𝑁𝑂 + 𝐻 These reactions have a far lower activation energy than the thermal NO mechanism, and therefore the prompt NO route is one of the primary means of NOx formation at temperatures below 2000 K. [5] 1.2.3. Nitrous Oxide Mechanism N2O forms from the reaction: 𝑁2 + 𝑂 = 𝑁2 𝑂 Another reaction which can occur during combustion is the formation of NO from N2O and O. This takes place according to: 𝑁2 𝑂 + 𝑂 → 2𝑁𝑂 𝑁2 𝑂 + 𝐻 → 𝑁𝐻 + 𝑁𝑂 𝑁2 𝑂 + 𝐶𝑂 → 𝑁𝐶𝑂 + 𝑁𝑂 5 At fuel lean conditions, and at lower flame temperatures, the thermal and prompt formation mechanisms are greatly reduced. The N2O mechanism is therefore is the primary mechanism through which NOx forms at low combustion temperatures. [4] 1.2.4. Fuel Nitrous Oxide Many hydrocarbon fuels contain fuel bound nitrogen, and oxidation of this fuel bound nitrogen is another source of NOx emissions. These emissions are generally dependant on the fuel being burned and are only loosely dependant on specific combustion conditions such as flame temperature. [4] Because fuel nitrous oxide is a direct result of the combustion of fuel, it is generally not feasible to limit this source of NOx emissions. The other NOx formation mechanisms can be reduced by lowering the flame temperature through injection of non-reactive fluid, often water, which absorbs some of the heat of generated by the reaction. 1.3. Water Injection One of the most effective means of reducing NOx emissions from gas turbine engines is the injection of water or steam into the flow stream. [3] This has the effect of the lowering flame temperature by increasing the mass flow rate of fluid through the engine without increase the heat generated by combustion. This effect is increased dramatically if liquid water is used, due to the latent heat of evaporation as the water becomes steam within the engine. Because the trend in higher gas turbine compression ratios and, therefore, increased NOx generation is likely to continue, research into a reduction of nitrogen oxide formation in gas turbine combustion chambers is warranted. Recent studies have shown that injecting atomized water directly into the combustion chamber can potentially cut nitrogen oxide emissions by roughly 50% on a commercial airliner during the takeoff/climbout phase of flight. [6] Substantial research into the best method for water injection has been pursued. Studies have considered the injection of atomized water into the compressor intake, the combustor intake, or other various places within the air flow stream. [6] Mixing of water directly into the incoming fuel has also been studied. [3] This thesis will describe a method to model and simulate the 6 injection of water into a gas turbine combustor to provide direct cooling for engine components while simultaneously reducing NOx emissions. 2. METHODOLOGY The goal of this study was to create a numerical model of a gas turbine combustion chamber and use it to simulate the effect of atomized liquid water on engine component temperature and NOx emissions. This section explains the methodology for modeling and simulating the baseline gas turbine combustion chamber within the interior of the engine. First, a baseline model was created showing nominal operation. This was then compared to a second model simulating water injection along the inner wall of the engine. This allowed a comparison of the operation of the engine with and without water injection. This process was completed using a 2D axisymmetric model to minimize processing time and then using a 3D model to maximize accuracy. The working hypothesis was that the injection of atomized water into the engine can be used to protect engine components from excessive temperatures caused by fuel combustion while simultaneously lowering NOx emissions. Such a strategy for water injection would be especially useful in the aviation industry, as both NOx emissions and engine component temperatures are highest during the takeoff and climbout phases of flight. [6] The flow through the combustor, the combustion process, and water injection were modeled in COMSOL 4.3. The software uses the principles of fluid dynamics to solve flow problems numerically, a process called computational fluid dynamics (CFD), using finite element discretization. The following sections describe the theory, the model, and the methodology used to simulate the combustor in steady state operating conditions. 2.1. Theory The flow through the gas turbine combustor was modeled using CFD. CFD methods use a set of partial differential equations (PDEs) to describe the properties of the fluids (and solids) in the problem domain. These PDEs are solved to determine the values of the dependant variables. 7 In the combustion chamber models, the problem domain encompasses the interior of the engine between the compressor outlet and turbine inlet, including the combustion chamber. Boundary conditions establish the pressure and temperature of incoming fluids and the pressure at the outlet. Conditions are also set within the problem domain to add or subtract heat at a predetermined rate to simulate combustion products or phase changes. A set of PDEs is then used to solve for the pressure, flow velocity, temperature, and other dependant variables within the interior of the engine. The solution to these PDEs form a complete description of the simulated fluid flow and heat fluxes through the interior of the problem domain. 2.1.1. Fluid Flow and Momentum The flow of fluid through the problem domain is governed by the conservation of mass and the conservation of fluid momentum. These principles are described by the following equations: 𝜕𝜌 + ∇ ∙ (𝜌𝑉) = 0 𝜕𝑡 𝜌 𝐷𝑽 𝜕 𝜕𝑢𝑖 𝜕𝑢𝑗 2 𝜕𝜇𝑘 = 𝜌𝐟 − 𝛁𝑝 + [𝜇 ( + ) − 𝛿𝑖𝑗 𝜇 ] 𝐷𝑡 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑖 3 𝜕𝑥𝑘 (1) (2) Equation 1 is the continuity equation. The continuity equation ensures that mass is conserved through the flow domain. It describes the rate of increase in density in a differential control volume and the rate of mass flux passing into or out of a differential control volume. When applied to a flow domain, a steady state solution will always show that the net mass flux across the boundaries of the system is zero. [7] Equation 2 is the Navier-Stokes equation. The Navier-Stokes equation describes the motion of viscous, compressible, Newtonian fluids. In Cartesian coordinates, the Navier-Stokes equations are: (3) 8 𝜌 𝐷𝒖 𝜕𝑝 𝜕 2 𝜕𝑢 𝜕𝑣 𝜕𝑤 𝜕 𝜕𝑢 𝜕𝑣 𝜕 𝜕𝑤 𝜕𝑢 = 𝜌𝐟𝐱 − + [ 𝜇 (2 − − )] + [𝜇 ( + )] + [𝜇 ( + )] 𝐷𝑡 𝜕𝑥 𝜕𝑥 3 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑦 𝜕𝑦 𝜕𝑥 𝜕𝑧 𝜕𝑥 𝜕𝑧 𝜌 𝐷𝒗 𝜕𝑝 𝜕 𝜕𝑣 𝜕𝑢 𝜕 2 𝜕𝑣 𝜕𝑢 𝜕𝑤 𝜕 𝜕𝑣 𝜕𝑤 = 𝜌𝐟𝐲 − + [𝜇 ( + )] + [ 𝜇 (2 − − )] + [𝜇 ( + )] (4) 𝐷𝑡 𝜕𝑦 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 3 𝜕𝑦 𝜕𝑥 𝜕𝑧 𝜕𝑧 𝜕𝑧 𝜕𝑦 𝜌 𝐷𝒘 𝜕𝑝 𝜕 𝜕𝑤 𝜕𝑢 𝜕 𝜕𝑣 𝜕𝑤 𝜕 2 𝜕𝑤 𝜕𝑢 𝜕𝑣 = 𝜌𝐟𝐳 − + [𝜇 ( + )] + [𝜇 ( + )] + [ 𝜇 (2 − − )] (5) 𝐷𝑡 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑧 𝜕𝑦 𝜕𝑧 𝜕𝑦 𝜕𝑧 3 𝜕𝑧 𝜕𝑥 𝜕𝑦 Equations 3 through 5 described the motion of viscous, compressible, laminar Newtonian fluids in each of the 3 principle directions. They take into account viscous stresses and fluid pressure acting on a differential control volume. [7] 2.1.2. Energy Balance The first law of thermodynamics, also known as the law of conservation of energy, can be applied to a differential control volume to describe the internal energy through the flow domain. 𝜕𝐸𝑡 𝜕𝑄 + 𝛁 ∙ 𝐸𝑡 𝐮 = − 𝛁 ∙ 𝒒 + 𝜌𝐟 ∙ 𝐮 + 𝛁(∏𝐢𝐣 ∙ 𝐮) 𝜕𝑡 𝜕𝑡 (6) Equation 6 described the change in internal energy over time through a differential control volume. The term 𝒒 represents the rate of heat transfer out of the control volume as a result of heat conduction. The rate of heat transfer is determined by the following equation: 𝜕𝑇 𝜕𝑇 𝜕𝑇 𝒒 = −𝑘 ( + + ) 𝜕𝑥 𝜕𝑦 𝜕𝑧 (7) Equation 7 describes the flow of heat through the fluid as well as any solid structures within the domain. [7] 9 2.1.3. Equation of State The relationships between the variables in Equations 1 through 7 must be established in order to fully define the flow. This is done with a series of equations that establish the thermodynamics state of the system. 𝑝 = 𝜌𝑅𝑇 (8) Equation 8 is the ideal gas law. It describes the relationship between the density, temperature, and pressure of a gas based on a constant, 𝑅. [7] 2.1.4. Turbulence Various methods have been developed for modeling turbulence in both compressible and incompressible flow problems. One of the most common methods involves solving for the turbulent kinetic energy (k) and the dissipation rate (ε). To reduce the complexity of the flow model, a Reynolds Averaged Navier-Stokes (RANS) 𝑘-ε model was used. The RANS model uses a density weighed time-average for the governing equations. This greatly reduces the complexity of the calculation required to solve turbulent, compressible flow problems. [7] The two equation 𝑘-ε model used to solve the flow through the combustor is as follows: 𝜌(𝐮 ∙ ∇)𝑘 = ∇ ∙ [(𝜇 + 𝜇𝑇 2 2 ) ∇𝑘] + 𝜇 𝑇 [∇𝐮: (∇𝐮 + (∇𝐮)T ) − (∇ ∙ 𝐮)2 ] − (𝜌𝑘∇ ∙ 𝐮) − 𝜌𝜀 𝜎𝑘 3 3 𝜇 𝜀 2 (9) 2 𝜌(𝐮 ∙ ∇)𝜀 = ∇ ∙ [(𝜇 + 𝜎𝑇 ) ∇𝜀] + 𝐶𝑒1 𝑘 [𝜇 𝑇 (∇𝐮: (∇𝐮 + (∇𝐮)T ) − 3 (∇ ∙ 𝐮)2 ) − 3 (𝜌𝑘∇ ∙ 𝐮)] − 𝑒 𝐶𝑒2 𝜌 𝜀2 (10) 𝑘 The turbulent viscosity 𝜇 𝑇 is defined as: 𝜇 𝑇 = 𝜌𝐶𝜇 10 𝑘2 𝜀 (11) Research into modeling of turbulent flows has resulted in widespread adoption of the turbulence parameter values given in Table 1: Table 1: Values used for independent turbulence parameters 𝐶𝑒1 1.44 Turbulence Model Parameters 𝐶𝜇 𝐶𝑒2 𝜎𝑘 1.92 0.09 1.0 𝜎𝑒 1.3 2.1.5. Discretization of Model Equations The solution to the equations governing the flow through the engine cannot be directly solved in a continuous domain. One method for solving a set of differential equations across a continuous domain is by dividing the domain into discrete sub-domains which can be solved using numerical methods. The solutions with these sub-domains are subsequently recombined into a complete solution across the complete problem domain. [8] The process of creating a set of sub-domains within the problem domain is called meshing. Meshing is a critical step in solving a CFD problem across a domain. For a solution to be accurate, the sub domains must be small enough to model the phenomena present. Accuracy generally increases as the average sub-domain size decreases. [8] The finite element method evaluates the error at the boundaries between elements for each dependant variable and then employs a convergent iteration scheme to reduce the error. If the mesh is sufficiently dense to model the phenomena in the domain, the solution will converge to within an acceptable degree of error over a finite number of iterations. [8] 2.2. 2.2.1. Modeling Baseline Selection To ensure that any numerical model produced realistic results, an existing engine with viable operational data was required as a baseline. The engine selected is one research article designed and operated by the University of Padova in Padova, Italy. [9] The engine is a small gas turbine with a single stage compressor and a single stage turbine. The engine is operated on a bench and the available source material does not describe operation 11 outside of a lab setting. It is designed to operate with an average turbine inlet temperature below 1000K, as this eliminates the need for high temperature alloys or ceramics in construction of the engine. Detailed design parameters for this engine were provided, enabling a combustor with similar operation to be modeled. [9] However, the low temperature operation of the engine also creates difficulty when attempting to predict NOx emissions because most publications on NOx emissions only concern flame temperatures above 1500K. This problem was solved by the University of Padova when a subsequent paper on the NOx emissions from this engine was published in 2009. This second source of data allowed a relationship between turbine inlet temperature and NOx mass fraction to be developed. [10] 2.2.2. Combustor Modeling The dimensions and drawings provided by the University of Padova were used to create a 3D model of a combustion chamber and a flow region representing the interior of the engine. A section view of the engine components provided by the University of Padova only included some overall dimensions. To estimate the size and locations of key features of the engine geometry, this images were imported into Microsoft Paint where a pixel count could be conducted. By measuring the distance between dimensioned features on the figure, a pixel-permillimeter scale was determined. This scale was then used to estimate the size and location of key geometry features in the engine. The end result of this process is shown in Figure 4. Dimensions shown in blue were from the original image while dimensions in red were added based on the estimated drawing scale. 12 Figure 4: Combustion Chamber Dimensional Analysis1 [9] From these estimated dimensions, a detailed 3D model of the combustion chamber was created in SolidWorks 2012. This is shown below in Figure 5: Figure 5: Full 3D combustor model 1 Modified from Figure 10 of Reference [9] to color code the flow regions and show estimated dimensions 13 Initial work on simulating the full engine were unsuccessful due to the geometric complexity. As a result, the final models used in the simulations were not meant to exactly replicate the engine, but rather to create a simplified version with similar operating characteristics. Many detailed features, such as the small flared tubes used to induce swirling flow, were omitted to simplify the calculation. This final model considered a 30 degree slice of the chamber, as the hole pattern and other features were assumed to be axially symmetric. Many of the holes were enlarged to simplify the meshing process and calculation. The model also includes the fuel inlet. No holes were included on the inner wall of the chamber. The model is shown in Figure 6 and Figure 7 . Figure 6: Combustion Chamber Simplified Model - Isometric View 14 Figure 7: Combustion Chamber Model - Side View The air flow region encompasses the region within the interior of the engine between the compressor outlet and the turbine inlet. The air flow region represents the overall boundaries of the problem domain. The combustion chamber and flow region were modeled using the same coordinate system, location, and orientation to eliminate the need for any geometry transformations after importing them into CFD software. After importing the geometry, the combustion chamber was already located at the proper location within the air flow region. The air flow region is shown in Figure 8. 15 Figure 8: Engine Air Flow Region These models were then imported into COMSOL to create a complete representation of the domain. No additional steps were required beyond a direct import of the geometry files because the separate entities were modeled within the same region of the coordinate system. The domain as it appears in COMSOL is shown in Figure 9. 16 Fuel Injector Air Flow Region Combustion Chamber Figure 9: Air flow region and combustion chamber in COMSOL The combustion chamber and the air flow region make up the extent of the domain that was studied. However, modeling this domain in full 3D takes requires significant computer power and time, and results in significant lag time between simulations. As a result, the flow region and combustion chamber geometries were also modeled as 2D axisymmetric objects. This resulted in a simpler model with greatly reduced time requirements between simulations. The original 3D models were bisected to create 2D cross sections. These were then further simplified and imported into COMSOL to create the axisymmetric flow domain, with the axis of rotation at r=0. The holes were modified into slits to better represent the total cross sectional area of flow through the combustor. The 2D axisymmetric flow domain is shown in Figure 10. 17 Fuel Injector Air Flow Region Combustion Chamber Figure 10: 2D Axisymmetric air flow region and combustion chamber in COMSOL The operation of the engine was based on the parameters given by Benini and Giacometti. Key parameters that were used in the simulation of the engine are shown in Table 2. Table 2: Engine Performance: Key Parameters [9] Property Ambient Air Pressure Ambient Temperature Compression Ratio Fuel to Air Ratio Fuel Heating Value Target Air Mass Flow Rate Value 101,300 Pa 288.15 K 2.66 0.0137 42,700,000 J/kg 0.53 kg/s Since the domain of the simulation does not include the compressor, the compressor outlet properties were predicted as a steady state input to the interior of the engine. The density of the air under ambient conditions was determined using the ideal gas law: 𝜌𝑎 = 𝑃𝑎 𝑅 ∗ 𝑇𝑎 18 (12) 𝐽 Where the assumed value for R was 287 𝑘𝑔∗𝐾. The density of the ambient air is 1.225g/m3. The density of the air at the compressor outlet (the problem domain air inlet i) was then determined according to: 𝜌𝑖 = 𝐶𝑅 ∗ 𝜌𝑎 (13) The density at the compressor outlet was 3.2583 kg/m3. The temperature of the air at the compressor outlet was found using an isentropic relation: 𝑇𝑖 𝜌𝑖 𝛾−1 =( ) 𝑇𝑎 𝜌𝑎 (14) Equation 14 was then solved for Ti to find that the compressor outlet temperature was 426.15 K. The ideal gas law was used to determined the compressor outlet pressure: 𝑃𝑖 = 𝜌𝑖 𝑅𝑇𝑖 (15) The compressor outlet pressure was found to be 398,515 Pa. These properties were used as the conditions at the air inlet of the domain in all simulations. The heat produced by the combustion of fuel in the air flow was determined based on the heating value provided in Table 2 and the fuel mass flow rate determined in accordance with: 𝑚̇𝑓 = 𝐹𝐴 ∗ 𝑚̇ 𝑖 (16) This resulted in a target mass flow rate of fuel of 0.007261 kg/s. The fuel was assumed to burn completely within the engine, and the heat of combustion was determined in accordance with: ℎ𝑓̇ = 𝑚̇𝑓 ∗ 𝐻𝑓 (17) The heat generated assuming complete combustion of the fuel within the flow domain was found to be 310,044.7 watts. In order to simulate a reasonable amount of water being injected into the engine, a flow rate consistent with that studied by Benini was chosen. The target mass flow rate for the atomized water being injected into the engine was 100% of the target fuel mass flow rate of 0.007261 19 kg/s. [10] However, this mass flow rate was simulated as air entering the annular region near the combustor exit along the inner wall of the flow domain. This was done to simplify the numerical model and avoid the need for an accurate model of species transport. This simplification is reasonable because the mass flow rate of water is two orders of magnitude smaller than the overall fluid flow rate through the engine and the heat absorbing properties can be corrected. To accomplish this, an assumption was made that the air in the simulation will absorb an equivalent amount of heat as a proportional amount of atomized water being injected into the engine. To determine the equivalent mass flow rate of atomized liquid water, the ratio of the heat capacities for air and water vapor was considered. The equivalent water flow rate was determined in accordance with: 𝐶𝑝𝑤 𝑚̇ 𝑎 = 𝑚𝑤 ̇ 𝐶 (18) 𝑝𝑎 The values used for the heat capacities for water vapor and air are 1901 𝐽 𝑘𝑔∗𝐾 and 1020 𝐽 𝑘𝑔∗𝐾 , respectively. The target mass flow rate for water entering the water injection inlet results in an equivalent air flow rate of 0.013533 kg/s. This value does not consider the latent heat of evaporation for the water entering the engine. The value used for the latent heat of evaporation of water is 2,257,000 J/kg. The water will absorb heat during evaporation according to: ℎ𝑤̇ = 𝐻𝑤 ∗ 𝑚𝑤 ̇ (19) This results in a total heat absorption rate due to evaporation of 16,388.1 watts for the target mass flow rate of water. The k-ε turbulence model used in the simulations requires the turbulent intensity and turbulent length scales to be defined as boundary conditions at the inlets. According to the COMSOL documentation, a turbulent intensity between 5% and 10% is typical for fully turbulent flows. The value used for all inlets is 5%, or 0.05. The hydraulic diameter at each inlet is used as the turbulent length scale. The boundary conditions and target flow rates in the problem domain are defined by the flow properties at the compressor outlet. These are shown in Table 3: 20 Table 3: Problem domain inlet boundary conditions and target flow rates Property Air Inlet Pressure Air Inlet Temperature Air Inlet Turbulent Length Scale Inlet Turbulent Intensity (All Inlets) Fuel Inlet Temperature Fuel Inlet Turbulent Length Scale Water Inlet Temperature Water Inlet Turbulent Length Scale Value 398,515 Pa 426.15 K 0.03525 m 0.05 426.15 0.007 m 373.15 K 0.0105 m Heat of Combustion 310,044.7 watts Target Fuel Flow Rate Target Water Flow Rate Target Equivalent Air Flow Rate Latent Heat of Evaporation for Injected Water 0.007261 kg/s 0.007261 kg/s 0.013533 kg/s 16,388.1 watts 2.2.3. Prediction of NOx Emissions NOx emissions increase rapidly as combustion temperature increases, particularly when the temperature exceeds 1800 K. [6] Most available data on NOx emissions only covers temperatures in excess of 1500 K, while the engine modeled herein in designed to operate at about 900 K. NOx production data according to Benini was used to determine a relationship between combustor outlet temperature and NOx emissions as a fraction of the air compressor . [10] The data was extracted from plots of experimental data and compiled into a table. Table 4: NOx emissions dependence on combustor exit temperature [10] Combustor Outlet Temperature 921 K 926 K 932 K NOx Mass Fraction 0.0001 0.000105 0.00012 The NOx production data were then used to create an exponential curve fit for temperatures in the range of expected operation. An exponential curve fit was chosen because it was expected to be approximate NOx production below the temperature range reported by Benini. Even as combustion temperature is greatly reduced, NOx production from the fuel is expected to remain 21 constant, an effect which cannot be accurately represented by a linear or polynomial fit. An exponential fit, in contrast, provides an estimate for a minimum level of nitrous oxide emissions. Only the 2D axisymmetric described in the following sections resulted in temperatures significantly below the values given by Benini. The NOx produced by the engine was approximated according to: 𝑁𝑂𝑥 = 1.950280 ∗ 10−11 ∗ 𝑒 0.01676∗𝑇 2.2.4. (20) 2D Axisymmetric Simulation of Combustion Process To model the combustion process, the 2D axisymmetric geometry shown in Figure 10 was altered to include a region between the fuel injector nozzle and the combustor exit where combustion was expected to occur. This region was used as a heat source to represent the heat generated by the combustion of fuel. The flow domain including the combustion region is shown in Figure 11. 22 Combustor Outlet / Turbine Inlet Fuel Inlet Combustion Region (Heat Addition) Air Flow Inlet / Compressor Outlet Figure 11: 2D axisymmetric combustor model with combustion region The combustion region and air flow domain were defined by the air material model provided by COMSOL. The material properties were defined by curve fits to data for thermodynamic properties. The purposed of these temperature dependant properties was to attempt to represent real air across a wide range of temperatures and pressures. The combustion chamber components were defined as AISI 4340 Steel, another material model provided by COMSOL. This model accurately represented the thermodynamic properties of a typical steel alloy as was used in construction of the research engine by Benin and Giacometti. [9] The domain was then meshed using the COMSOL meshing algorithm. For the problem domain being considered, the mesh generated by COMSOL using a "normal" physics controlled setting 23 was found to provide sufficient mesh quality to solve the problem. The mesh consisted of 12,690 polygonal elements, including boundary layers to accurately predict flow behavior along the solid surfaces within the domain. The 2D axisymmetric mesh without water injection is shown in Figure 12: Figure 12: 2D axisymmetric mesh, no water injection To model the flow through the combustion chamber, the "High Mach Flow" physics modules was applied to the problem domain. This physics module combines a compressible formulation for fluid flow with a k-ε turbulence model and a simple heat transfer model which is sufficient to model heat transfer across solids. The boundary condition at the compressor outlet (the inlet of the problem domain) called for a specified inlet Mach number. However, repeated simulations revealed that this input was used only as a starting condition and had no impact on the problem solution after convergence. As a result, the flow rates for the various fluids entering 24 the problem domain had to be controlled by varying the boundary conditions at each of the inlets and the outlet. The flow field was first solved assuming no fuel flow through the engine and no heat from combustion. The combustor exit was assumed to have a backpressure of 371,000 Pa, resulting in a mass flow rate of 0.98254 kg/s. This was significantly higher than the 0.53 kg/s given in Table 2. The backpressure was then evaluated at 380,000 Pa, 370,000 Pa, and 395,000 Pa, showing decreasing mass flow rates at each value. These results were then plotted as shown in Figure 13: 1.2 Mass Flow Rate (kg/s) 1 0.8 0.6 0.4 y = -9.10150668E-11x2 + 5.12916471E-05x - 5.51933686E+00 R² = 9.99900428E-01 0.2 0 365000 370000 375000 380000 385000 390000 395000 400000 Backpressure (Pa) Figure 13: Air mass flow rate dependence on backpressure with quadratic curve fit The curve fit shown in Figure 13 was generated using Microsoft Excel. It was then used to predict the backpressure required to produce a mass flow rate through the engine of 0.53 kg/s. The required backpressure was predicted to be 395,585 Pa. The simulation was run with this backpressure resulting in a mass flow rate of 0.5302 kg/s. This mass flow rate was considered to be within an acceptable margin from the baseline engine design. Equation 16 was then used to predict the mass flow rate of fuel that would be required to approximate the operation of the baseline engine. This was found to be 0.007261 kg/s. While holding the air inlet and combustor outlet backpressure constant, the fuel inlet pressure was 25 varied and plotted to determine the correct fuel pressure to achieve a mass flow rate of approximately 0.007261 kg/s of fuel. The data is plotted in Figure 14: 0.03 Mass Flow Rate 0.025 0.02 y = 0.00000633x - 2.49602687 R² = 0.99999276 0.015 0.01 0.005 0 395000 395500 396000 396500 397000 397500 398000 398500 399000 Fuel Pressure (Pa) Figure 14: Fuel mass flow rate dependence on fuel inlet pressure with linear curve fit This process resulted in a mass flow rate of air from the compressor of 0.52905 kg/s and a fuel mass flow rate of 0.00695 kg/s, compared with the target values of 0.53 kg/s and 0.007261 kg/s, respectively. These mass flow rates were considered to be within an acceptable deviation from the target values, as they were both within 5%. An additional model was run with the heat generated from the combustion of fuel in the engine entering the flow domain uniformly throughout the combustion region shown in Figure 11. This was done by adding in the "Heat Transfer in Fluids" physics module in COMSOL and coupling the temperature variable T within the software. This physics module is formulated to account for the effects of moving fluid on heat transfer within the problem domain. By coupling the temperature variable between the two physics models, a solution for the flow field across the problem domain can be found with an accurate prediction for temperature variation resulting from the combustion process. The boundary conditions determined without heat transfer effects were used as the initial boundary conditions. However, the increased temperature caused the flow rates from each of the 26 inlets to change. As a result, the same process was used once again to determine the correct pressures at each inlet, although fewer simulations were required because the boundary conditions were still close to the correct values. The final boundary conditions and flow rates are shown in Table 5: Table 5: Operating characteristics for the combustor, without water injection Variable Air Inlet Combustor Exit Fuel Inlet Pressure (Pa) 398,515 394,295 395,440 Mass Flow Rate (kg/s) 0.53075 0.53929 0.00766 The temperature distribution in the flow domain is shown in Figure 15. The average exit temperature was found to be 708.7 K, while the peak combustion temperature was 741.06 K. This was somewhat below the expected values, but still reasonable for the problem being considered. Full results are reported in section 3.1. Figure 15: 2D axisymmetric model temperature distribution, no water injection 2.2.5. 2D Axisymmetric Simulation of Water Injection To model the injection of atomized liquid water into the combustor, an additional heat generating region was added. This is shown in as the thin rectangular region on the inner wall of 27 the problem domain as shown in Figure 16. This region was used to subtract the heat of fusion of water corresponding to evaporation of a percentage of the water droplets being injected. Water Inlet Water Evaporation Region Figure 16: 2D axisymmetric combustor model with combustion and water evaporation regions The water was assumed to be saturated liquid droplets entering the engine at 373.15 K. The evaporation process was modeled by adjusting the heat being absorbed by this region while maintaining the inlet temperature of 373.15 K throughout the evaporation region. This resulted in a small percentage of the water evaporating in this region. The rest of the droplets were assumed to evaporate in the combustion region. It should be noted that in high temperature operation, or when the water injection flow rate is sufficiently low, all of the water droplets being injected into the engine will evaporate before 28 reaching the end of this region. In that case the heat absorbing region should be resized so that the temperature is maintained at 373.15 K throughout. The location for water injection was chosen in order to investigate the cooling effect of the water on the inner wall of the chamber. Rotating machinery within the center of the engine is used to transfer momentum from the turbine to the compressor during operation. By cooling the inner wall of the engine, the machinery in the center can be protected from the high temperatures created by combustion. As with the model without water injection, the problem domain was meshed using the COMSOL physics controlled meshing algorithm on a "normal" density setting. This resulted in a total of 13,793 polygonal elements, including boundary layers along the walls. The mesh is shown in Figure 17: 29 Figure 17: 2D axisymmetric mesh, with water injection The flow field was first solved based on the inlet and outlet pressures found while modeling the engine without water injection. The initial set of simulations did not include the heat of combustion or the heat of evaporation of the water entering the engine. They were used to establish the boundary conditions required for an accurate simulation of engine performance. To determine the required boundary conditions, simulations were first run with various combustion chamber backpressures. The combustor backpressure was varied from 380,000 Pa to 390,000 Pa and 395,000 Pa, resulting in the range of mass flow rates from the compressor shown in Figure 18: 30 Compressor Mass Flow Rate (kg/s) 0.9 0.8 0.7 0.6 0.5 y = -0.00001668x + 7.13147429 R² = 0.99925412 0.4 0.3 0.2 0.1 0 375000 380000 385000 390000 395000 400000 Backpressure (Pa) Figure 18: Compressor air flow rate as a function of combustor backpressure The relationship between backpressure and compressor flow rate was found to be nearly linear for the range being considered. This resulted in the linear curve fit shown in Figure 18. The relationship was used to predict the required backpressure of 395,772 Pa that was used for the remainder of the 2D axisymmetric simulations. The next step was to find the required fuel flow inlet pressure. Since the compressor outlet pressure and combustor backpressure were now known, only the fuel inlet pressure was varied. Fuel inlet pressures of 395,500 Pa, 395,800 Pa, and 395,600 Pa were simulated and resulted in the relationship to fuel mass flow rate shown in Figure 19: 31 0.009 Fuel Mass Flow Rate (kg/s) 0.008 0.007 y = 0.0000064357x - 2.5395764286 R² = 0.9999963045 0.006 0.005 0.004 0.003 0.002 0.001 0 395450 395500 395550 395600 395650 395700 395750 395800 395850 Fuel Inlet Pressure (Pa) Figure 19: Fuel mass flow rate as a function of fuel inlet pressure The relationship between fuel inlet pressure and fuel mass flow rate was also found to be nearly linear for the values considered. This resulted in the linear curve fit shown Figure 19. The relationship was used to predict the required fuel inlet pressure of 395,736 Pa. Now that they compressor backpressure and fuel inlet pressure were known, only the water inlet pressure needed to be found. This was done by simulating water inlet pressures of 398,000 Pa and 401,000 Pa, resulting in the relationship shown in Figure 20: 32 0.009 Water Flow Rate (kg/s) 0.008 0.007 0.006 0.005 y = 0.0000022200x - 0.8824100000 R² = 1.0000000000 0.004 0.003 0.002 0.001 0 397500 398000 398500 399000 399500 400000 400500 401000 401500 Water Inlet Pressure (Pa) Figure 20: Water mass flow rate as a function of water inlet pressure The linear curve fit shown in Figure 20 was used to determine the required water inlet pressure of 401,700 Pa. These values were then used to simulate the flow through the combustor with the heat of combustion and the heat of evaporation of the water droplets being considered. The latent heat of evaporation for the water being injected into the engine was found to be 16,388.08 watts for the target water flow. The simulation considering combustion and water injection resulted in different flow rates from the simulation without these effects, and thus the process of determining the correct boundary conditions once again had to be repeated. This resulted in the operating characteristics shown in Table 6: 33 Table 6: Operating characteristics for the combustor, with water injection Variable Air Inlet Combustor Exit Fuel Inlet Water Inlet Equivalent Water Flow Rate Pressure (Pa) 398,515 394,148 395,200 401,386 401,386 Mass Flow Rate (kg/s) 0.53081 0.55285 0.00726 0.01353 0.00726 The simulation was then run multiple times, and each time the amount of heat being absorbed by the evaporating water droplets was varied until the average temperature in the water evaporation region was found to be approximately 373.15 K. This occurred when about 6% of the water was evaporated in the region, corresponding to a heat absorption rate of 983.3 watts. This evaporation rate is considered valid because the heat being conducted through the wall of the combustion chamber into this region is perfectly balanced by the latent heat of evaporation for the water that evaporates in this region. The heat flows are therefore balanced. The temperature distribution in the flow domain is shown in Figure 21. The average combustor exit temperature was found to be 686.97 K, while the maximum combustion temperature was 728.7 K. Full results are reported in section 3.2. Figure 21: 2D axisymmetric model temperature distribution, with water injection 34 2.2.6. 3D Simulation of Combustion Process The limitations of the 2D axisymmetric model are clearly evident in the combustion temperature results. Specifically, the values obtained are several hundred degrees Kelvin under the experimental values. [10] To improve accuracy, a full 3D simulation was also completed. The simulation is of a 30° slice of the problem domain. As with the axisymmetric models, a region was added downstream of the fuel injectors to approximate the region where combustion is expected to take place. The geometry used to model the combustion process in the baseline engine is shown in Figure 22: Fuel Inlet Air Inlet / Compressor Combustion Outlet Region Combustor Exit / Turbine Inlet Figure 22: 3D combustion chamber model with combustion region As in the 2D axisymmetric models, the fluid in the problem domain was defined as air while the combustion chamber was AISI 4340 steel. The boundaries at either side of the 30° slice were defined as symmetry boundaries in the software, while the outer walls and combustion chamber surfaces were modeled with wall functions. The air inlet from the compressor stage is defined using the boundary conditions shown in Table 3. The combustion region is shown in Figure 22 as the 3D region just downstream of the fuel inlet. This region was created after an initial simulation showed that the fuel flow from the injector would swirl around this region before exiting. The shape of the region is intended to be a general representation of where combustion is likely to occur within the engine. The fuel is assumed to burn fully and uniformly within this region, which is defined as a heat source with a 35 total thermal power equal to 8.33% (1/12) of the value given in Table 3. Unheated fuel flow from the injector through this region is shown as streamlines in Figure 23: Figure 23: Fuel flow from the fuel injector through the combustion region The domain was meshed using the built in meshing algorithm in COMSOL. First, a mesh was generated using the "Normal" density setting. This resulted in an acceptable mesh throughout most of the problem domain, but the software was unable to generate the default series of 5 boundary layers. The mesh was then incrementally adjusted by reducing the boundary layer growth rate and eventually reducing the number of layers from 5 to 4. The final mesh consisted of 380,052 volumetric elements, as shown in Figure 24: 36 Figure 24: 3D problem domain mesh, no water injection It is important to note that like the heat of combustion, the target mass flow rates for the 3D models are 1/12 of the total values for the engine, as the 30° slice represents 1/12 of the engine. The target flow rates and heats rates for the 3D models are shown in Table 7: Table 7: Target flow rates and boundary conditions, no water injection Property Value Heat of Combustion 25837.1 watts Target Compressor Air Flow Rate Target Fuel Flow Rate 0.044 kg/s 6.051E-4 kg/s Initially, the boundary conditions for the combustor exit backpressure and fuel inlet found during development of the 2D axisymmetric model were used in this model. The calculated mass flow rates were initially close to the target values, but some additional refinement to the boundary conditions was undertaken to minimize error. The process used was similar to the iteration technique used to solve for the required boundary conditions in the 2D axisymmetric models. The primary limitation in refining the boundary conditions to precise mass flow rates was simulation time. The boundary conditions and mass flow rates for the 3D model without water injection are as shown in Table 8: 37 Table 8: Operating characteristics for the 3D combustor model, without water injection Variable Air Inlet Combustor Exit Fuel Inlet Pressure (Pa) 398,515 395,100 398,711 Mass Flow Rate (kg/s) 0.04325 0.0434 8.480E-4 The simulation resulted in a flow field represented by the streamlines shown in Figure 25. They are color coded to show temperature variations as they pass through the combustion region. Figure 25: 3D streamlines showing flow from the compressor outlet to combustor outlet, no water injection The temperature profile shows heating of the combustion region, with hot spots forming where flow velocities are reduced. The temperature profile is shown in Figure 26. 38 Figure 26: Temperature profile of flow through the domain, no water injection The combustion process resulted in a peak temperature of 1748 K in localized areas, while the average combustor exit temperature was 1036 K. This is close to the value given by Benini. [10] 39 2.2.7. 3D Simulation of Water Injection The 3D geometry used to simulate the injection of atomized water into the engine is similar to the 3d model without water injection, except that a heat absorbing region was added in the space between the combustion chamber wall and the inner wall of the problem domain. The water injection location in the 3D model corresponds to the location selected for the 2D axisymmetric model. The geometry used to simulate water injection into the engine is shown in Figure 27: Water Inlet Evaporation Region Figure 27: 3D combustion chamber model with heat absorbing region The mesh generated for this geometry was similar to the model without water injection. The meshing algorithm was set to a "normal" density setting and the number of boundary layers was reduced from 5 to 4. The final mesh consisted of 410,101 volumetric elements, as shown in Figure 28: 40 Figure 28: 3D problem domain mesh, with water injection The thermal power of combustion and evaporation, and the target mass flow rates of air, fuel, and water were all 1/12 of the values given in Table 3. The thermal powers and target mass flow rates for this simulation are shown in Table 9: Table 9: 3D model target flow rates and boundary conditions, with water injection Property Heat of Combustion Target Compressor Air Flow Rate Target Fuel Flow Rate Target Water Flow Rate Target Equivalent Air Flow Rate Latent Heat of Evaporation for Injected Water Value 25837.1 watts 0.044 kg/s 6.051E-4 kg/s 6.051E-4 kg/s 1.128E-3 kg/s 1365.68 watts Initially, the pressure boundary conditions developed during completion of the 3D model without water injection and the 2D axisymmetric model were used as the boundary conditions for this model. The air inlet, the fuel inlet, and the exit backpressure were taken from the 3D model without water injection while the water inlet pressure and heat absorption in the injection region were taken from the 2D axisymmetric model results. This resulted in flow rates that were close to the target values given in Table 9. Further refinements were undertaken to reduce error from the target values. 41 The final boundary conditions and mass flow rates for the 3D model with water injection are as shown in Table 10: Table 10: Operating characteristics for the 3D combustor model, without water injection Variable Air Inlet Pressure (Pa) 398,515 Mass Flow Rate (kg/s) 0.04325 Combustor Exit 395,100 0.0434 Fuel Inlet 398,711 8.480E-4 Water Inlet 395,100 0.0434 Equivalent Water Flow Rate 398,711 8.480E-4 The flow field produced by this simulation is represented as streamlines color coded to represent temperature in Figure 29: Figure 29: 3D streamlines showing flow from the compressor outlet and water inlet to combustor outlet The temperature profile shows heating of the combustion region, with hot spots forming where flow velocities are reduced. The temperature profile is shown in Figure 30: 42 Figure 30: Temperature profile of flow through the domain, with water injection The combustion process resulted in a peak temperature of 1653 K in localized areas, while the average combustor exit temperature was 919 K. 3. RESULTS AND DISCUSSION Two distinct models have been used to simulate the operation of the combustor. In the first case, a simplified model was constructed based on a two dimensional profile which was then revolved around the axis of the engine. This resulted in a reasonable approximation of the flow field, while the temperature effects diverged considerably from available test data. The second model used a three dimensional, 30° section of the engine which allowed for greater detail in the geometry. The second model produced also produced a reasonable approximation of the flow field, and the temperature effect closely matched available test data. This section will report the outputs from these simulations. 3.1. 2D Axisymmetric Models The 2D axisymmetric simulations produced flow field results consistent with available test data. The pressure fields are shown in Figure 31: 43 Figure 31: 2D absolute pressure results without water injection (left) and with (right) Color coding in each figure is identical. The maximum and minimum pressures are indicated at the top of bottom of each figures color scale. The water injection caused a decrease in the minimum pressure at the combustor exit. This is due to the increased overall mass flow. A greater pressure differential is required to achieve the same flow rates for the fuel and compressor air when water is added. The average pressure at the combustor exit decreased from 395,553 Pa without water injection to 394,105 Pa with water injection, a reduction of 0.37%. The velocity magnitude of the fluid in each simulation is shown in Figure 32: Figure 32: 2D Velocity magnitude results without water injection (left) and with (right) 44 The injection of atomized water into the engine had a small effect on the velocity of the fluid traveling through interior of the engine. A decrease in the peak velocity near the combustor exit was observed from 51.632 m/s without water injection to 50.176 m/s with water injection, a reduction of 2.82%. Average exit velocity decreased from 47.11 m/s without water injection to 46.80 m/s with water injection, a reduction of 0.66%. The temperature effects are shown in Figure 33: Figure 33: 2D Temperature results without water injection (left) and with (right) The addition of atomized water into engine significantly decreased the temperature of the combustion products exiting the chamber. Peak temperature decreased from 741.06 K without water injection to 728.7 K with water injection, a reduction of 1.7%. More importantly, average combustor exit temperature decreased from 708.73 K without water injection to 686.97 K with water injection, a reduction of 3.07%. Overall performance of the engine was similar with and without water injection. The operating characteristics in each scenario are compared in Table 11: 45 Table 11: Operating characteristics - 2D axisymmetric models Characteristic Air Inlet Pressure Air Inlet Mass Flow Rate Combustor Exit Pressure Combustor Exit Mass Flow Rate Average Combustor Exit Temperature Fuel Inlet Pressure Fuel Mass Flow Rate Water Inlet Pressure Water Mass Flow Rate Without Water Injection 398,515 Pa 0.52962 kg/s 394,294 Pa 0.5394 kg/s With Water Injection 398,515 Pa 0.53081 kg/s 394,148 Pa 0.55285 kg/s 708.73 K 686.97 K 395,688 Pa 0.00882 kg/s N/A N/A 395,200 Pa 0.00726 401,386 Pa 0.00726 kg/s* **This value represents the equivalent mass flow rate of water from the flow rate of fluid in the simulation The average combustor exit temperatures were used to predict NOx emissions based on Equation 20. The results for NOx emissions are shown in Table 12: Table 12: 2D NOx production with and without water injection Water Flow Rate 0.00000 kg/s 0.00726 kg/s Average Exit Temperature 708.73 K 686.97 K NOx Mass Fraction 2.814E-6 1.954E-6 Exit Flow Rate 0.5394 kg/s 0.5528 kg/s NOx Production Rate 1.518E-6 kg/s 1.080E-6 kg/s NOx emissions are predicted by the 2D axisymmetric models to decrease by 28.84% as a result of water injection equal to approximately 100% of the fuel flow. In addition to the effect on NOx emissions, water injection was expected to significantly decrease component wall temperature close to the injection site. The effect is visible in Figure 33. The wall temperature along the inner wall of the problem domain at r = -37.5 mm is plotted relative to the z coordinate, (the water injection site is at z = 162.75 mm), in Figure 34: 46 Figure 34: 2D wall temperature without water injection (left) and with (right) The figure shows that injecting water in this location maintains the inner wall at about 373 K for the first 70mm as it flows into the engine. This occurs because all of the heat flux through the combustion chamber wall is evaporating approximately 6% of the injected water in this region. 3.2. 3D Models The 3D combustion chamber models produced flow field results consistent with the available test data. The pressure results are shown in Figure 35: 47 Figure 35: 3D absolute pressure results without water injection (left) and with (right) Color coding in each figure is identical. The maximum and minimum pressures are indicated at the top of bottom of each figures color scale. The water injection caused a slight decrease in the minimum pressure at the combustor exit. This is due to the increased overall mass flow. A greater pressure differential is required to achieve the same flow rates for the fuel and compressor air when water is added. The average pressure at the combustor exit decreased from 395,103 Pa without water injection to 395,097 Pa with water injection. The velocity magnitude in each simulation is represented in Figure 36: Figure 36: 3D Velocity magnitude results without water injection (left) and with (right) 48 The injection of atomized water into the engine had a small effect on the velocity of the fluid traveling through interior of the engine. A decrease in the peak velocity near the combustor exit was observed from 51.632 m/s without water injection to 50.176 m/s with water injection, a reduction of 2.82%. Average exit velocity decreased from 47.11 m/s without water injection to 46.80 m/s with water injection, a reduction of 0.66%. The effect of water injection on temperature within the fluid is shown in Figure 37: Figure 37: 3D Temperature results without water injection (left) and with (right) The addition of atomized water into engine significantly decreased the temperature of the combustion products exiting the chamber. Peak temperature decreased from 1778.3 K without water injection to 1653.2 K with water injection, a reduction of 7%. More importantly, average combustor exit temperature decreased from 1037.4 K without water injection to 919.05 K with water injection, a reduction of 11.4%. Overall performance of the engine was similar with and without water injection, with the most significant difference being observed for the combustor exit temperature. The operating characteristics in each scenario are compared in Table 13: 49 Table 13: Operating characteristics - 3D models Characteristic Air Inlet Pressure Air Inlet Mass Flow Rate Combustor Exit Pressure Combustor Exit Mass Flow Rate Average Combustor Exit Temperature Fuel Inlet Pressure Fuel Mass Flow Rate Water Inlet Pressure Water Mass Flow Rate Without Water Injection 398,515 Pa 0.04231 kg/s 372,477 Pa 0.04338 kg/s With Water Injection 398,515 Pa 0.0417 kg/s 395,100 Pa 0.04774 kg/s 1037.4 K 919.05 K 398,711 Pa 0.000848 kg/s N/A N/A 398,711 Pa 0.000840 kg/s 397,000 Pa 0.002592 kg/s* **This value represents the equivalent mass flow rate of water from the flow rate of fluid in the simulation The average combustor exit temperatures were used to predict NOx emissions based on Equation 20. The results for NOx emissions are shown in Table 14: Table 14: 3D NOx production with and without water injection Water Flow Rate 0.00000 kg/s 0.002592 kg/s Average Exit Temperature 1037.4 K 919.05 K NOx Mass Fraction 6.950E-4 9.560E-5 Exit Flow Rate 0.04338 kg/s 0.04774 kg/s NOx Production Rate 3.015E-5 kg/s 4.564E-6 kg/s NOx emissions are predicted by the 3D models to decrease by 84.86% as a result of water injection at a rate of 0.002592 kg/s. Simulation results also indicate significant cooling of the inner wall of the engine resulting from water injection. This effect is visible in Figure 37. The wall temperatures along the edges of this region of the problem domain are plotted in Figure 38: 50 Figure 38: 3D wall temperature without water injection (left) and with (right) Figure 38 shows that injecting water into the inner region of the engine can maintain the inner wall at nearly 373 K. This is because all of the heat flux through the combustion chamber wall is evaporating approximately 1.5% of the atomized water being sprayed into the engine. 51 4. CONCLUSIONS This thesis has demonstrated that the injection of atomized liquid water into a gas turbine engine reduces NOx emissions while reducing engine component surface temperatures. The results of both the 2D axisymmetric models and the 3D models support the hypothesis that water injection can simultaneously reduce engine component wear due to high temperatures while simultaneously reducing pollutant emissions. The results show that for an engine operating at low temperatures, injecting water into the interior of the engine at a rate equivalent to the fuel flow rate reduces NOx emissions by about 30% while successfully maintaining a temperature around 373.15 K along the inner wall of the interior of the engine. This has the effect of protecting the rotating machinery typically housed within the center of a gas turbine engine from the heat of combustion taking place in the combustion chamber. This can lead to significant advantages for aircraft operators because the highest engine power output is experienced during the takeoff and climbout phases of flight and also corresponds to the highest engine temperature and NOx emissions, making it an ideal time to utilize water injection. [6] The engine modeled in this thesis is a small research engine designed to operate at low (~900 K) temperatures. [9] Typical commercial gas turbine engines operate at much higher temperatures, where engine component wear and NOx emissions are much greater. Additionally, these simulations only considered the heat added as a result of complete fuel combustion, and modeled all fluids as air. A more accurate model would consider species transport of the fuel and water being injected into the flow stream, as well as the chemical reactions taking place. Such a model was not possible with the computer resources available. Further research into the use of water injection for component cooling and NOx reduction should be undertaken using commercial aircraft engine designs operating at high temperatures. More accurate combustion models can be used to simulate the species transport and chemical reactions taking place inside the combustion chamber if sufficient processing resources are available. 52 5. REFERENCES Mattingly, J. with foreword by Ohain, H. “Elements of Propulsion: Gas Turbines and Rockets”, AIAA Education Series, (2006) [2] Acharya, S. " J85 ge 17a turbojet engine.jpg" Used unaltered and to develop an illustration under Creative Commons license 3.0. http://commons.wikimedia.org/wiki/File:J85_ge_17a_turbojet_engine.jpg [3] “Alternative Control Techniques Document – NOx Emissions from Stationary Gas Turbines”, U.S. Environmental Protection Agency, (1993), EPA-453/R-93-007 [4] Lefebvre, A., Ballal, D., "Gas Turbine Combustion: Alternative Fuels and Emissions" Third Edition, CRC Press (2010) [5] Kuo, K. "Principles of Combustion" Second Edition, John Wiley & Sons, Inc. (2005) [6] Daggett, D., Fucke, L., Hendricks, R., Eames, D., “Water Injection on Commercial Aircraft to Reduce Airport Nitrogen Oxides”, NASA (2010) NASA/TM-2010-213179 [7] Pletcher, R., Tannehill, J., Anderson, D., "Computational Fluid Mechanics and Heat Transfer" Third Edition, CRC Press (2013) [8] O.C. Zienkiewicz, CBE, FRS, "The Finite Element Method: Its Basis and Fundamentals" Sixth Edition, Elsevier Butterworth-Heinemann (2005) [9] Benini E, Giacometti S, Design, manufacturing and operation of a small turbojet-engine for research purposes, University of Padova, Applied Energy Article 84 (2007) 11021116, dated 27 July, 2007 [10] Benini E, Pandolfo S, Zoppellari S, Reduction in NO emissions in a turbojet combustor by direct water/steam injection: numerical and experimental assessment, University of Padova, Applied Thermal Engineering 29 (2009) 3506-3510, dated 2 June, 2009 [1] 53 APPENDIX A: Determination of Boundary Conditions Appendix A-1: 2D axisymmetric model, no water injection Table 15: 2D model iterative convergence toward target mass flow rates, no water injection Simulation Number Target 1 2 3 4 5 6 0.535 Backpressure (Pa) 395585 395000 394294.7 394294.7 394294.7 394294.7 Simulation Values Inlet Air Flow Rate (kg/s) 0.53 0.50618 0.51698 0.52992 0.53061 0.52962 0.53075 Fuel Pressure (Pa) Target Value 395479 395479 395479 395000 395688 395560 Fuel Flow Rate (kg/s) 0.007261 0.00031 0.00327 0.0075 0.00449 0.00882 0.00766 Poly. (Simulation Values) Mass Flow Rate (kg/s) 0.53 0.525 0.52 0.515 0.51 y = -1.84615385E-05x + 7.80928769E+00 R² = 1.00000000E+00 0.505 0.5 394200 394400 394600 394800 395000 395200 395400 395600 395800 Backpressure (Pa) Figure 39: 2D model curve fit and predicted backpressure value to achieve the target inlet air mass flow rate, no water injection 54 Simulation Values Target Value Linear (Simulation Values) 0.01 0.009 Mass Flow Rate 0.008 0.007 0.006 0.005 y = 0.0000062919x - 2.4808222804 R² = 0.9999986161 0.004 0.003 0.002 0.001 0 394900 395000 395100 395200 395300 395400 395500 395600 395700 395800 Fuel Pressure (Pa) Figure 40: 2D model curve fit and predicted fuel line pressure value to achieve the target fuel mass flow rate, no water injection Appendix A-2: 2D axisymmetric model, with water injection Table 16: 2D model iterative convergence toward target mass flow rates, with water injection Simulation Number Target 1 2 3 4 5 6 7 Backpressure (Pa) 395500 395000 394148 394148 394148 394148 394148 Air Flow (kg/s) 0.53 0.50572 0.5147 0.52982 0.53027 0.53053 0.53071 0.53081 Fuel Pressure (Pa) 395736 395736 395736 395400 395200 395200 395200 55 Fuel Flow Rate (kg/s) 0.007261 0.00247 0.0055 0.01065 0.00852 0.00725 0.00726 0.00726 Water Pressure (Pa) 401700 401700 401700 401700 401700 401500 401386 Water Flow (kg/s) 0.013533 0.01147 0.01259 0.01448 0.01453 0.01457 0.01391 0.01353 Simulation Values Target Value Linear (Simulation Values) Compressor Mass Flow Rate (kg/s) 0.535 0.53 0.525 0.52 0.515 y = -0.00001796x + 7.60890000 R² = 1.00000000 0.51 0.505 0.5 394000 394200 394400 394600 394800 395000 395200 395400 395600 Backpressure (Pa) Figure 41: 2D model curve fit and predicted backpressure value to achieve the target inlet air mass flow rate, with water injection 56 Simulation Values Target Value Linear (Simulation Values) Fuel Mass Flow Rate (kg/s) 0.012 0.01 y = 0.0000063393x - 2.4980335714 R² = 1.0000000000 0.008 0.006 0.004 0.002 0 395100 395200 395300 395400 395500 395600 395700 395800 Fuel Inlet Pressure (Pa) Figure 42: 2D model curve fit and predicted fuel line pressure value to achieve the target inlet air mass flow rate, with water injection 57 Simulation Values Target Value Linear (Simulation Values) 0.0148 Water Flow Rate (kg/s) 0.0146 0.0144 0.0142 y = 0.0000033000x - 1.3110400000 R² = 1.0000000000 0.014 0.0138 0.0136 0.0134 401350 401400 401450 401500 401550 401600 401650 401700 401750 Water Inlet Pressure (Pa) Figure 43: 2D model curve fit and predicted water injection pressure value to achieve the target inlet air mass flow rate Appendix A-3: 3D model, no water injection Table 17: 3D model iterative convergence toward target mass flow rates, no water injection Simulation Number Target 1 2 3 4 5 6 Backpressure (Pa) 395585 395000 394294.7 394294.7 394294.7 394294.7 Inlet Air Flow Rate (kg/s) 0.53 0.50618 0.51698 0.52992 0.53061 0.52962 0.53075 58 Fuel Pressure (Pa) 395479 395479 395479 395000 395688 395560 Fuel Flow Rate (kg/s) 0.007261 0.00031 0.00327 0.0075 0.00449 0.00882 0.00766 0.535 Simulation Values Target Value Poly. (Simulation Values) Mass Flow Rate (kg/s) 0.53 0.525 0.52 0.515 0.51 y = -1.84615385E-05x + 7.80928769E+00 R² = 1.00000000E+00 0.505 0.5 394200 394400 394600 394800 395000 395200 395400 395600 395800 Backpressure (Pa) Figure 44: 3D model curve fit and predicted backpressure value to achieve the target inlet air mass flow rate, no water injection Simulation Values Target Value Linear (Simulation Values) 0.01 0.009 Mass Flow Rate 0.008 0.007 0.006 0.005 y = 0.0000062919x - 2.4808222804 R² = 0.9999986161 0.004 0.003 0.002 0.001 0 394900 395000 395100 395200 395300 395400 395500 395600 395700 395800 Fuel Pressure (Pa) Figure 45: 3D model curve fit and predicted fuel line pressure value to achieve the target fuel mass flow rate, no water injection 59 Appendix A-4: 3D model, with water injection Table 18: 3D model iterative convergence toward target mass flow rates, with water injection Simulation Number Target 1 2 3 4 5 6 7 Backpressure (Pa) 395500 395000 394148 394148 394148 394148 394148 Air Flow (kg/s) 0.53 0.50572 0.5147 0.52982 0.53027 0.53053 0.53071 0.53081 Simulation Values Fuel Pressure (Pa) 395736 395736 395736 395400 395200 395200 395200 Fuel Flow Rate (kg/s) 0.007261 0.00247 0.0055 0.01065 0.00852 0.00725 0.00726 0.00726 Target Value Water Pressure (Pa) 401700 401700 401700 401700 401700 401500 401386 Water Flow (kg/s) 0.013533 0.01147 0.01259 0.01448 0.01453 0.01457 0.01391 0.01353 Linear (Simulation Values) Compressor Mass Flow Rate (kg/s) 0.535 0.53 0.525 0.52 0.515 y = -0.00001796x + 7.60890000 R² = 1.00000000 0.51 0.505 0.5 394000 394200 394400 394600 394800 395000 395200 395400 395600 Backpressure (Pa) Figure 46: 3D model curve fit and predicted backpressure value to achieve the target inlet air mass flow rate, with water injection 60 Simulation Values Target Value Linear (Simulation Values) Fuel Mass Flow Rate (kg/s) 0.012 0.01 y = 0.0000063393x - 2.4980335714 R² = 1.0000000000 0.008 0.006 0.004 0.002 0 395100 395200 395300 395400 395500 395600 395700 395800 Fuel Inlet Pressure (Pa) Figure 47: 3D model curve fit and predicted fuel line pressure value to achieve the target inlet air mass flow rate, with water injection 61 Simulation Values Target Value Linear (Simulation Values) 0.0148 Water Flow Rate (kg/s) 0.0146 0.0144 0.0142 y = 0.0000033000x - 1.3110400000 R² = 1.0000000000 0.014 0.0138 0.0136 0.0134 401350 401400 401450 401500 401550 401600 401650 401700 401750 Water Inlet Pressure (Pa) Figure 48: 3D model curve fit and predicted water injection pressure value to achieve the target inlet air mass flow rate 62