Fuel Spray and Mixture Preparation in Split-Cycle Engine A Nuffield Project with STEM Sussex

Transcription

1 Fuel Spray and Mixture Preparation in Split-Cycle Engine A Nuffield Project with STEM Sussex Larissa R Taylor Park College Sussex Downs Eastbourne Supervisors; Dr Steve Begg Dr Oyuna Rybdylova Dr Elena Sazhina Sir Harry Ricardo Laboratories Centre for Automotive Engineering University of Brighton

2 Fuel Spray and Mixture Preparation in Split-Cycle Engine Abstract A Nuffield Project with STEM Sussex A new type of engine is being designed that is aimed to be more efficient by employing the split-cycle engine concept. The engine operates as a two-stroke by dividing the conventional four-strokes (the induction, compression, expansion and exhaust strokes) between two separate cylinders; air at high pressure is fed into the combustor cylinder from the compressor cylinder via a cross-over passage and valve. Feasibility studies of mixture preparation for a given mass flow rate of air and fuel injection into the combustion chamber of a split-cycle engine have been performed by numerical simulation. Fuel spray evaporation, vapour mixing with air and the onset of combustion are key factors that influence the efficiency of an engine. If the spray is placed in the right position then the gaseous cloud of fuel will be formed near the spark plug when the fuel evaporates in the chamber. It is essential that the fuel evaporates in the chamber in the correct position for complete and stable combustion. The aim of research is to use a numerical CFD modelling program, ANSYS FLUENT, to test where the vapour cloud of fuel will end up for ignition. By setting up the desired spray parameters for realistic engine geometry and operating conditions, it can be determined where the fuel droplets will end up inside the engine. Three injector locations were considered, and three spray breakup cases were compared, namely; no breakup model, TAB and WAVE breakup models. Interaction of spray and in-cylinder flow pattern has revealed results that can be used for further design improvements to the new engine under the EU INTERREG CEREEV project. The fuel droplets were considered as a multicomponent liquid; 95% of iso-octane C 8 H 18 and 5% of n-heptane C 7 H 16 by mass. The study of the evaporation process of multicomponent fuel droplets is relevant to the EU INTERREG E3C3 project, focusing on biofuels. The rate of evaporation as predicted by ANSYS FLUENT, is monitored for both the C 7 H 16 and C 8 H 18 components. It is observed that n-heptane evaporates faster than iso-octane as it can be expected. The results have shown that the optimum location of the injector and ignition for a combustible mixture is close-spaced and central in the chamber. The current project has provided a feasibility study of conventional CFD capabilities for multicomponent fuel droplets. Thus, it lays foundation for CFD implementation of the newly developed spray models at SHRL. As the SHRL experimental and modelling work grows in sophistication and depth, newly developed spray models take into account temperature gradients and species diffusion inside a multicomponent droplet. Further research on validation of the CFD results against available experimental data is needed. 2

3 Acknowledgements I would like to thank the members of the Sir Harry Ricardo Labs at the University of Brighton for their help and support and also the staff members at Park College Sussex Downs for their support. I would also like to give special thanks to STEM Sussex staff, in particular, Daniel Hawkins, Bronagh Liddicoat and Patricia Harwood, and gratefully acknowledge the financial help of the Nuffield Research Placement grant ( 3

4 List of Figures Figure 1 Four stroke spark ignition engine Figure 2 Four Stroke Petrol Engine and Definition of TDC and BDC Figure 3 Four Stroke Diesel Engine Figure 4 Two Stroke Engine Figure 5 Scuderi Split Cycle Engine Figure 6 Principal Elements of Split Cycle Engine Figure 7 Mesh Components for CFDs Figure 8 TAB break up activated in FLUENT. Figure 9 Wave break up activated in FLUENT. Figure 10 Mesh and Initial Conditions Figure 11 Mesh and fuel injector positioned near inlet valve injector Figure 12 Mesh and fuel injector positioned near exhaust duct Figure 13 Mesh and fuel injector positioned near spark plug Figure 14 Mesh and all fuel injector positions Figure 15 Percentage of octane in the fuel vapour by mass for no-break up model Figure 16 Percentage of octane in the fuel vapour by mass for TAB breakup model Figure 17 Percentage of octane in the fuel vapour by mass for Wave breakup model Figure 18 In-cylinder mass of octane vapour for TAB breakup model Figure 19 Mass of octane vapour for TAB and no-breakup models Figure 20 Mass fraction of octane at 2ms for no-breakup model for near-inlet injector Figure 21 Mass fraction of octane at 2ms for TAB model for near-inlet injector Figure 22 Mass fraction of octane at 2ms for Wave model for near-inlet injector Figure 23 Initial velocities of droplets for injector positioned by the exhaust duct Figure 24 Initial velocities of droplets for group injectors positioned by the exhaust duct Figure 25 Velocity magnitude contours for in-cylinder flow for group injectors positioned by the exhaust duct; no-breakup model for spray Figure 26 Mass fraction of octane vapour for group injectors positioned by the exhaust duct; no-breakup spray model Figure 27 Mass fraction of octane vapour of group injectors by the exhaust duct for Wave breakup model Figure 28 Initial velocities of droplets for the single injector positioned by spark plug Figure 29 Mass fraction of octane vapour for no-breakup model for the spark-plug location of the injector Figure 30 Mass fraction of octane vapour for Wave breakup model for the spark-plug location of the injector Figure 31 Ratio of octane in fuel vapour by mass for the spark-plug location of the injector for the injection velocity of 50 m/s; Wave spray breakup model. 4

5 Glossary BDC Bottom dead centre CA BTDC Crank angle before top dead centre CFD Computational fluid dynamics ICE Internal combustion engines MFR Mass flow rate SHRL Sir Harry Ricardo Laboratory SI Spark ignition SSC Scuderi Split Cycle TDC Top dead centre 5

6 Fuel Spray and Mixture Preparation in Split-Cycle Engine 1. Background An engine is a machine designed to convert chemical energy of fuel into useful mechanical energy. Heat engines such as internal combustion engines (ICE) in cars and external combustion engines such as steam engines burn fuel releasing heat in the process, which then creates motion (Engine 2013). The fuel in ICE is normally a fossil fuel (hydrocarbons) and the combustion of the fuel occur using oxygen in air as the oxidizer. An example of a fuel combustion equation for octane is: C 8 H ½ O 2 9H 2 O + 8CO 2 Improvements in engine design address both better fuel efficiency and less pollution simultaneously. Fossil fuel is a non-renewable resource and its cost is rising. The product of combustion is carbon dioxide, a green house gas associated with global warming (Gent 2007). An alternative could be using biofuels, such a biodiesel, which are renewable. At the Sir Harry Ricardo Laboratory (SHRL) at the University of Brighton, a team of theoretical and experimental researchers are investigating different engine designs and fuel injection capabilities for engine efficiency and emissions reduction (SHRL 2013). The University is actively involved in several research programmes in this field. Upon arriving at the University of Brighton I met with some of my project supervisors, Dr Elena Sazhina and Dr Steve Begg. Dr Begg gave me a tour of the SHRL experimental facilities. The SHRL has a variety of engine test beds and they talked me through some of the research being carried out at the laboratory. I also met some of the postgraduate students, in particular K. CK, MSc student. My supervisors were keen that I should get a university experience. I had an opportunity to attend a doctoral conference in July (Research Student Conference 2013). I attended the key note speech on Stem Cell research and even though it was not relevant to my project I found it very interesting. After that I attended the second stream at the doctoral conference which was mostly about engines. It was very interesting to see the projects that the PhD students were doing and how it related not just to my project but to the problems of the current world. Angad Panesar presented his findings on fuel consumption improvement by waste heat recovery for automotive application which was interesting to me as it showed how making engines more efficient is a very important subject at the moment, especially with the regulations regarding emissions This is why the split cycle engine will be very useful as it will be very efficient. Where the current ICE is maybe only about 40% efficient, the split 6

7 cycle engine will be 60% meaning that extra processes to make it more efficient will not be needed. After the conference I was given a tour of the Vetronics lab by Dr Panagiotis Oikonomidis (Vetronics, 2013). I saw the Vetronics laboratory room where they test the control box of a car and the small systems used for it. What really caught my attention was the car they had built. They had designed it and built it from scratch. It had this system where the steering wheel was not directly attached to the wheels but to a control box which was connected to the wheels. The control box sensed the direction the steering wheel had been turned to and so made the wheel turn too. This is useful in that sensors on the control box could be used to make driving safer. The use of the control box also means that the car can be steered using a joy stick too and because it is not connected directly to the wheels it can be steered from places like behind the car. I was also shown a quick tutorial on how to use SolidWorks by Dr Manzanares, a commercial graphics computer program used by students and researchers for design and development. In early August I was shown how to use ANSYS FLUENT, a numerical modelling program that would assist me in calculations for this project. I also attended an Automotive Engineering research workshop in August on modelling of droplet and spray dynamics, heating and evaporation. Everyone participating had some really good ideas and had clearly thought hard about how to solve their problems. Heated discussion on each presentation showed the interest and the level of quality of work from everyone involved; it was a learning experience for me. It highlighted how research collaborations work (Automotive Engineering 2013). My supervisors have given me the background I needed to carry out a project related to ongoing studies at SHRL. The project focused on numerical simulation studies on the feasibility of mixture preparation for given mass flow rate of fuel injection into a split-cycle engine. Fuel spray and mixture preparation for combustion are key factors in the efficiency of an engine. A numerical modelling program, ANSYS FLUENT, has been used to carry out this study. FLUENT is a computational fluid dynamics (CFD) software package. A brief description of CFD is given in Section 2. The outcome of the calculations guides future design considerations or used by researchers at SHRL as experimental parameters.in what 7

8 follows, literature review will be followed by original results of the CFD simulations and analysis of the results. Four stroke engines A four stroke engine has four main stages: intake, compression, power and exhaust (see Figure 1). The most common ICE engine is the piston-type reciprocating engine where the crankshaft is turned by piston moving up and down in a cylinder (Heywood 1988). The stroke refers to the movement of the piston. A four-stroke engine completes a cycle in four strokes and two crankshaft revolutions. Figure 1. Four stroke spark ignition engine (reproduced from Internal Combustion Engine uploaded August 2013) Gasoline direct-injection engines The piston rests at top dead centre (TDC) stroke position 1 (up). As the piston moves down, air is forced into the cylinder due to pressure difference. The piston descends down the cylinder to bottom dead centre (BDC) stroke position 2 (down). The piston is then pushed back up to TDC compressing the air stroke position 3 (up). For direct-injection engines, fuel spray is injected into cylinder well before TDC. It takes time to evaporate and mix with air before the combustible gaseous mixture is ignited with a spark plug. Rapid combustion of 8

9 the fuel and the sudden increase in pressure causes the piston to be pushed down to the bottom of the cylinder again stroke position 4 (down), thus powering the crank shaft. At the next stroke the piston returns to TDC, the exhaust valve opens and the combustion products are expelled via exhaust manifold. This process finishes after two complete revolutions of the crank shaft. TDC BDC Figure 2. Four Stroke Petrol Engine and Definition of TDC and BDC (reproduced from a screenshot of uploaded by codene on Jul 20, 2007) Diesel Engines Diesel engines differ to the gasoline engines in that they use hot compressed air to ignite the fuel rather than a spark plug. The fuel is injected as small droplets. The hot compressed air vaporises fuel from the surface of the droplets and then the heat from the compressed air ignites the fuel vapour. The droplets continue to get smaller as the surface fuel is vaporised and then ignited until all the droplets are used up. This increase in pressure due to the rapid combustion of the gases causes the piston to be pushed down and power the engine. This is illustrated in Figure 3 Figure 3 Four Stroke Diesel Engine (reproduced from seen on 12/08/2013) 9

10 A four-stroke engine completes a power cycle in four strokes, this means that it requires 2 revolutions of the crankshaft to complete the process. Another engine has been designed to only need one crankshaft revolution to complete the power cycle. It is a two-stroke engine. It only requires two strokes, or up and down movement, of the piston to fully complete to process. Two Stroke Engines Two stroke engines complete the power cycle in just one crankshaft revolution. It does this by having the end of the combustion stroke happen simultaneously with the start of the compression stroke, and having exhaust and intake (otherwise known as scavenging) happen at the same time too. Exhaust Intake Figure 4 Two Stroke Engine (reproduced from seen on August 2013)) The disadvantage of two stroke engines is that some fuel is lost to the exhaust duct when it enters the chamber and so can cause pollution when it leaks out of the exhaust. Two stroke engines are also not very fuel efficient. (How Stuff Works 2013) Scuderi Split-cycle Engine The Scuderi Split Cycle (SSC) engine is designed so that it is more efficient. It is similar to the gasoline cycle in that it uses a spark plug not hot compressed air. It is different in that it does intake and compression in one cylinder, known as the compression cylinder and then expansion and exhaust in another, known as the power cylinder. The two cylinders are connected by a crossover port, through which high pressure gas is transferred from the compressor cylinder to the expander cylinder between the compression and power strokes. (Philips et al 2011) This means that it only requires one complete revolution of the crank shaft for the process to be completed. A normal 4 stroke engine would have to do two complete crankshaft revolutions, as shown in Fig. 1, to match the power of the Scuderi engine. The use of a turbocharger (a device with a turbine powered by the kinetic energy of the exhaust to improve volumetric efficiency) with this Scuderi engine also comes in useful as it allows one compression cylinder to provide air flow to multiple power cylinders. 10

11 (Scuderi 2013; Meldolesi 2012) High turbulence in the engine means that the fuel and air mixes quickly and this prevents knock, giving successful combustion. Figure 5a and 5b. Scuderi Split Cycle Engine (reproduced from SCUDERI Group, Inc 2013) EU INTERREG project CEREEV The EU INTERREG project CEREEV (CEREEV, 2012) establishes collaboration between University of Brighton, IRSEEM (Institut de Recherche de l ESIGELEC), and Jules Verne University of Picardie. The research under the CEREEV aims to overcome problems of volumetric efficiency and combustion phasing. A major functional benefit of the split cycle engine is the separation of the compression and power cylinders. This allows optimum conditions for each can be achieved (Meldolesi 2012). Figure 6 shows a schematic representation of the cylinders in split cycle engine. Figure 6 Schematic of Split Cycle Engine (reproduced from Meldolesi and Badain, 2012) The CEREEV project aims to develop a new type of split-cycle engine when intake and compression are done in a separate cylinder. The proposed two-stroke process requires rapid filling in the power cylinder, where combustion will take place, while simultaneously injecting fuel and air from different places. 11

12 2. Computational Fluid Dynamics (CFD) Computational Fluid Dynamics (CFD) is the term used to describe numerical codes that can calculate the properties of a fluid, such as temperature, velocity, chemical composition throughout a region of space (ANSYS 2013). CFD breaks down geometries into cells that make up a mesh, and algorithms based on conservation of energy, mass and momentum are applied to each individual cell to compute the fluid flow, species and temperature (Fig. 7). Boundary conditions are set by the user. Partial differential equations that describe the flow are converted into simultaneous algebraic equations which are set up for each cell. These equations are solved using numerical methods and when the answers found for each cell coincide with the specified tolerance then they converge and the solution is found. (ANSYS 2013) Figure 7. Mesh Components for CFDs (reproduced from ANSYS FLUENT ANSYS FLUENT is a CFD code. It is a numerical simulation program that allows the user to model flow, chemical species and temperature for complex geometries; the calculation starts by reading a mesh from its case and data file or from creating one in the program itself. By setting the properties, boundary conditions and any desired conditions to be tracked, FLUENT can calculate a solution for the given parameters and give a variety of data outputs. FLUENT also allows you to see pictures showing things like fluid flow, giving the user a wide variety of results to see from the calculated solution. (ANSYS 2013). 12

13 Breakup models Within FLUENT, the user can choose to activate atomization models for their run. The term breakup refers to fragmentation and splitting of the droplets injected. If each individual droplet is broken up into smaller droplets then this will create more surface area for the same amount of fuel and so it should evaporate quicker. The idea of breakup is similar to what actually happens in an engine so is a helpful tool for realistic modelling. There are different models of breakup: Taylor analogy breakup (TAB) model. This model uses oscillations of certain amplitude to break up the droplet. External forces acting on the droplet are caused by the motion of it. The droplet has both surface tension, which stops it from falling apart, and a damping force, which stops the oscillations from getting too big (Turner 2012). The droplet will break up if the distortion reaches a critical level. The distortion should to be equal to half the radius of the droplet in order for breakup to succeed. The child droplets are assumed to be neither oscillating nor distorted. (ANSYS 2013). From Fig. 8, you can see that the number of breakup parcels is 2 meaning that the parent droplet will breakup into 2 similar child droplets. Wave break up. Figure 8 TAB breakup model activated in FLUENT. Otherwise known as stripping breakup, it relies on the relative velocity between the liquid and gas phases. The Kelvin Helmholtz instability occurs when there is a velocity difference across the interface between liquid fuel and gas; instability of the fuel jet phase causes child droplets to be stripped from liquid core of the jet, resulting in the gradual decrease in size of the injected fuel. The flow travelling in the opposite direction to the fuel spray can also help with the stripping of the core liquid (ANSYS 2013). The CFD code provides a breakup constant for the value of stable droplet sizes. This constant will depend on the type of injector used (Turner 2012) Figure 9 Wave breakup model activated in FLUENT. 13

14 3. Aims and Objectives It has been decided to focus research on optimization of injector location for a good mixture preparation in this study. The objectives are: To setup CFD simulation for a realistic engine geometry and boundary conditions To explore spray tracking for several injector locations To assess mixture preparation for each case To formulate recommendations for the engine design For my study, a mesh of the engine to work with was taken from 2-ACE project. An EPSRC project, A Fundamental Study of the Novel Poppet Valve 2-stroke Auto-ignition Combustion Engine has been carried out at SHRL. As part of the EPSRC study, a CFD simulation using FLUENT was set and explored for a realistic engine geometry (2-ACE 2012). The focus of my study is to set the injector position and run cases for different breakup models to test where the vapour cloud of fuel will end up for ignition. Three injector positions (Fig. 14) were investigated, namely: fuel injection by the inlet valve; fuel injection by the outlet valve fuel injection near the spark plug. Three injector types: flat fan single injection group injections were modelled. Three breakup models were explored, namely: no breakup, TAB and Wave. An accompanying study by CK, 2013 is focusing on volumetric efficiency for a range of inlet pressures. This facilitates team work for my research. 14

15 4. Methodology Input parameters. Each CFD case had the same base parameters set, whilst varying the position of the fuel injector and its type (single, group, flat-fan). The following base input parameters for each CFD case were used: Inlet Pressure = 10bar at T= 400 K (1 bar = 100 kilopascals) Initial in-cylinder condition Pressure = 1bar at T= 300 K In-cylinder initial air Taken as mixture of nitrogen and to 0.23 of oxygen by mass composition Liquid fuel Multicomponent fuel consisting of 95% octane (C 8 H 18 ) and 5% heptane (C 7 H 16 ) by mass. Injector flowrate kg/s The inlet duct pressure was set to about 9.9 bar as initial condition. Figure 10 shows the mesh and initial conditions. For the first CFD case, the injector used was a flat-fan-atomizer that had 80 streams of multicomponent droplets. The width of the orifice was 0.147mm. The droplets had an initial speed of 200m/s. After evaporation, the in-cylinder charge shall be a combustible mixture of air with fuel, namely iso-octane and n-heptane vapours. The ANSYS FLUENT model was set to Discrete Phase On for spray calculation, and interaction of gas with droplets was activated to make it more realistic. Mass flow rate was kg/s. This is a historical value taken from the 2-ACE project. This means that stoichiometric ratio will be achieved at about 4ms for in-cylinder mass of 0.4g. This corresponds to the volume of around 5.05e-05m 3 at 40 CA BTDC (crank angle before top dead centre) and in-cylinder pressure of 10 bar at T = 400 K. The CFD run was set to monitor pressure, temperature, speed, cylinder mass, cylinder volume, temperature, density, mass fraction of octane, mass fraction of heptane averaged over in-cylinder volume, thus giving a large amount of output data files. In-cylinder air composition 23% O 2 10bar 9.9 bar 1bar Figure 10 Mesh and Initial Conditions (reproduced from 2-ACE EPSRC project, 2012) 15

16 All runs used these same parameters for the base case but the position of the injector changed. Three positions were tested: A) Injector near inlet valve Fuel injector near inlet valve: Direction in which fuel was sprayed Figure 11 Mesh for 2-ACE engine and fuel injector positioned near inlet valve B) Injector near exhaust duct Direction in which fuel was sprayed Fuel Injector near exhaust duct: 95% octane (C 8 H 18 ) and 5% heptane (C 7 H 16 ). Figure 12 The 2-ACE mesh and fuel injector positioned near exhaust duct 16

17 C) Near spark plug Fuel Injector near spark plug: 95% octane (C 8 H 18 ) and % heptane (C 7 H 16 ). Direction in which fuel was sprayed Figure 13 The 2-ACE mesh and fuel injector positioned near spark plug Three different atomization models: no-breakup, TAB breakup and Wave breakup were tested on the three different injector positions; near the inlet valve, near the exhaust and by the spark plug (see Fig. 14). The runs were performed for 800 times steps (8ms). Near spark Near inlet valve Near exhaust duct Figure 14 Mesh and all fuel injector positions (reproduced from 2-ACE Project, School of CEM. University of Brighton, Brighton, 2012) 17

18 As described in the previous section, each case for each injector had been set up first before any calculations were made. This case and data files with the desired parameters then had to be read into FLUENT. The time step size has been set as 0.01ms. On convergence, for post-processing of the results, FLUENT allows the user to create and view velocity vectors and contours of scalar variables. The output data files created from the monitors set. The breakup models for each injector were tested consecutively with the data being gathered and analysed after each individual run. No-breakup atomization model did not require any change to the set up. So after checking that all the parameters were correct, the only thing needed to be done was to initialise, patch and then calculate the solution. No-breakup calculation ran until 300 time steps (3ms) where it was stopped due to length it was taking but still provided sufficient data for analysis. TAB breakup required TAB breakup menu option to be turned on from the Models menu of FLUENT. Figure 8 (in Breakup Models section) shows the set-up of the Discrete Phase On editing menu where breakup can be turned on. Once changed to TAB breakup on, it is initialised by patching initial conditions, and then transient run is launched to calculate the solution. The TAB breakup run was calculated for 800 time steps (8ms). Wave breakup needed the Wave breakup menu option to be turned on. The same process as TAB break up was used to do this; in the Models menu with Discrete Phase On, go to Edit and click on the Physical Models tab. At the bottom of the menu is the breakup options. Figure 9 (in Breakup Models section) shows the activation setup for Wave breakup. Wave breakup calculation also ran to 800 times steps (8ms). After each calculation was finished the data were post-processed. FLUENT allows images of filled contours to be made, showing the output of the calculation on any part of the mesh and can be easily adjusted to the desired needs. Images of mass fraction of octane and heptane vapour, and vectors of velocity were useful in showing where the vapour cloud ended up. There are images that show tracks of fuel droplets making it useful to see what has actually happened with the spray inside the chamber. Due to setting up monitors, output files are generated by FLUENT. This data was transferrable to Microsoft Excel and was manipulated to give lots of graphical representation of the data. This makes all three models easier to compare quantitatively. 18

19 5. Results: Injector near Inlet Valve: The first scenario used was to test the fuel spray without breakup. This meant that the droplet stayed the same size they were when they were injected. The graph below shows the percentage of octane in the fuel vapour for the no-breakup model. The octane vapour mass fraction is density-averaged over in-cylinder charge by ANSYS FLUENT code and it is written to an output text file at each timestep. This gives the incylinder mass of octane vapour. The output files are further processed in Excel. The liquid fuel is injected at time = 0, and in few timesteps the octane percentage in the vapour rises to 80%. Timestep is set to 0.01ms. Percentage of octane in the fuel vapour is defined as the ratio of vapour mass of octane to the fuel vapour mass. The fuel vapour mass is the sum of vapour mass for octane and heptane. Heptane evaporates quicker than octane so there is less percentage of gaseous octane at initial period; it stays at 80% for about 0.2ms (20 time steps). Once reaching this time step the evaporation process seems to be completed, and octane percentage rapidly rises to 95%, where it stays constant for the remaining time. Figure 15 Percentage of octane in the fuel vapour for no-break up model The next graph shows the percentage of octane but with TAB break up on. 19

20 Figure 16 Percentage of octane in the fuel for TAB breakup. Figure 16 shows that initially it rises to 80% due to the liquid fuel evaporation. But instead of staying constant like the breakup model did, it continues to rise unsteadily until about 0.4ms (40 time steps), showing how gradually it evaporates. It reaches the 95% level for mass fraction of octane in fuel vapour, and remains there for the rest of the time. This shows that evaporation started almost instantly upon injection but did not develop as rapidly as without break-up. The next graph shows the octane percentage for the Wave break up model. Figure 17 Percentage of octane in the fuel for Wave breakup 20

21 Figure 17 is similar to the no-break up model. When the liquid fuel is injected the percentage increases to 80% after one time step. It then gradually increases very slightly for about the next 0.1ms (10 time steps) perhaps showing some slight evaporation before rapidly increasing to 95%. The Wave break up model reaches a constant of 95% octane faster than the other models. The plots above have shown percentage of octane in fuel vapour by mass. The next figure shows transient behaviour of in-cylinder octane vapour mass. Figure 18. In-cylinder mass of octane vapour (kg) for all breakup model upto 8ms Figure 19. In-cylinder mass of octane vapour: magnified view In Fig 18, the oscillations in mass of octane vapour are observed that settle down with time. The mass of octane vapour varies with in-cylinder mass. The variations of in-cylinder mass 21

22 are ascribed to reflected pressure waves causing backflow of in-cylinder charge into inlet manifold. As you can see from Figure 19 the evaporation of fuel for TAB model begins almost immediately and slowly increasing. After a certain delay, evaporation for Wave and nobreakup model rises steeply reaching up to just under 1.80E-08 kg. During the first 0.1ms, the octane mass fraction for the TAB model is still increasing. It reaches a maximum of around 1.7E-08kg, then vapour mass start decreasing for all breakup models, as in-cylinder charge mass is decreasing due to the flow up the inlet duct because of reflected pressure waves. These oscillations are believed to be caused by pressure waves inside the cylinder; they are slowly damping with time as can be seen on Figure 18. The following images show fuel vapour distribution as predicted by FLUENT. No-breakup model, mass fraction of octane at 2ms Figure 20 Mass fraction of octane vapour at 2ms for near-inlet valve injector; no-breakup The bulk of the evaporated fuel remains by the injector. Some of it goes up the inlet duct, meaning that fuel is wasted; the ignition will be hard as the bulk of the fuel is far from the spark plug. 22

23 Although the maximum is almost 4 times smaller than the Wave model (see Fig. 22), a larger portion of fuel has ended up spread around the chamber. The mass fraction near spark plug is about 4e-05 with the highest points being around 5e-05. With TAB model, you can see that the fuel particles did not travel very far as the whole section around the injector contains most of the evaporated fuel. Again some escapes up the inlet valve. Lower values of mass fractions of octane get around the chamber but like the Wave model, it is quite close to zero. TAB model mass fraction of Octane shows at 2 ms: Figure 21 Mass fraction of octane vapour at 2ms for TAB model for near-inlet valve injector Wave breakup model mass fraction of Octane for 2ms: 23

24 Figure 22. Mass fraction of octane vapour at 2ms for Wave model for near-inlet valve injector Although there is fuel all around the chamber, the main part of the evaporated fuel is centred around the injection point. This is not good as it s too far away from the spark plug. Part of the fuel also ends up going up the inlet valve, again this is not good as fuel is wasted by settling up there rather than being ignited. A very small proportion of the fuel ends up around the chamber for evaporation. From these results, we conclude that the fuel is evaporating too quickly and because of this we end up with a large portion of fuel vapour right by the injector, not by the spark plug. Fuel is also being lost due to going up the inlet duct. If the inlet valve is to remain open for the duration of the combustion process then relocating the injector could help to overcome this problem. This will be explored below. Injector near exhaust duct The injector was moved to the opposite side of the chamber near the exhaust duct. Spray was aimed towards the spark plug. A new type of injection, single injector was setup. Thus required setting a new location of the injector and initial droplet size. The chemical composition of fuel and mass flowrate reamined the same. 24

25 New position of injector Figure 23 Initial velocities of droplets for single injector positioned by the exhaust duct This test unfortunately diverged so instead a group injection of 10 injectors was tested, without breakup activated. The mass flow rate (MFR) was decreased to 10% of the original for each injector to provide the same overall mass injected. Thus, each injector had a MFR of kg/s. Figure 24 Initial velocities of droplets for group injectors positioned by the exhaust duct 25

26 Figure 25 Velocity magnitude contours for in-cylinder flow for group injectors positioned by the exhaust duct; no-break up model for spray The results from the group injector (no breakup) show that the fuel vapour overshoots the spark plug and ends up going into the inlet duct. The point at which the fuel vapour is fastest is already past the optimum position near the spark plug for successful combustion. Thus indicates that the gas cloud of fuel will continue to travel further and end up quite far away from the spark plug. The Fig 25 shows the fastest point to be quite near the inlet duct. From this point the fuel vapour could either continue to the other side of the chamber or go up the inlet valve like before. Neither of these options is good as both positions will be too far from the spark plug. Next Figure shows that the vapour does in fact go up the inlet valve again. 26

27 Figure 26 Mass fraction of octane at 1.6ms for the group injectors positioned by the exhaust duct; no-break up spray model All the fuel vapour is accumulated near the inlet valve. This causes the same problem as before; not enough fuel vapour by the spark plug and it will mean that there won t be a successful combustion. Figure 27 Mass fraction of octane at 0.5ms for the group injectors by the exhaust duct for WAVE break up model For Wave breakup model, the results for the first steps after fuel injection have shown that the fuel evaporated far too quickly (Fig. 27). Though vapour was not lost to the inlet valve it was still too far away from the spark plug. Another location of the injector was to be tested to try to solve this problem. 27

28 Injector near spark plug: The new single injector was placed at the top, near where the spark plug is. It was set to have interaction with continuous phase. Two cases are explored: The spray is not atomized (no-breakup model is selected). Injection velocity 100 m/s Wave breakup model for spray atomization. Injection velocity 50 m/s New injector location Figure 28 Initial droplets velocities for the case of the injector positioned near spark plug Due to the position of the injector near spark plug, and the fact that liquid fuel spray goes directly down towards the piston, for fast-enough evaporation the vapour cloud could be formed near spark plug, thus facilitating a successful combustion. This is true if spray is not deflected by the crossflow of intake air. 28

29 Figure 29 Mass fraction of octane vapour for no-breakup model. The injector is near spark plug For no-breakup option, the vapour cloud overshoots and is formed on the surface of the piston. This is clearly undesirable as fuel film on the piston does not evaporate quickly and it is still far from the spark plug. Figure 30. Mass fraction of octane vapour for Wave breakup model. The injector is near spark plug For Wave breakup, the fuel evaporates quickly and thus ends up around the spark plug due to the position of the injector. This is the optimum place for the vapour cloud to be and hence this is the best place to position the injector. It shall be observed though that the fuel cloud is significantly deflected towards the wall by the inlet air flow 29

30 6. Discussion For the injectors near the inlet valve and near the exhaust duct, all models showed that whilst the injector was directed towards the spark plug, the flow pattern was not appropriate to create a fuel vapour cloud under the spark plug. Fuel vapour either remains mostly around the injector or is lost up the inlet valve. To overcome this loss of fuel, a solution could be to keep the inlet valve closed during fuel injection, ensuring that all the fuel remains in the chamber. Some fuel might be lost due to impinging on a cylinder wall and piston. The injector near the spark plug was directed towards the piston. The no-break up model shows that the spray overshoots impinging on the piston, and the vapour cloud was too far from the spark plug. Wave breakup models gave the desired outcome for the case under consideration; the fuel evaporated almost immediately and so the vapour cloud was adjacent to the spark plug, though it was deflected to the wall of the cylinder head by strong crossflow of inlet air. The injector near the spark plug with Wave breakup model was losing fuel to the inlet duct after 1.8ms. This shows the complexity of the interaction between fuel spray and air jet. Guiding the flow so that the fuel vapour ends up under the spark plug could help solve this problem and give a successful combustion. Methods on how to guide the flow should be further investigated. Dissemination of knowledge and recommendations for further research CEREEV workshop I gave a presentation of my findings to the CEREEV group on 12 th September 2013 at the University Brighton (CEREEV, 2012). The title of the workshop was: CEREEV Researcher exchange, technical workshop and project review (Appendix 1). The minutes of the meeting relevant to this project, read: CFD: Simulations by E. Sazhina, O. Rybdylova and L. Taylor showed that high-pressure intake conditions produced choked sonic flows conditions in the poppet valve inner seat area. Tests were carried out with three spray options: no breakup model, TAB and WAVE breakup. Comparisons of spray velocity and fuel mass distributions in the cylinder were made between three injector locations; central, inlet and exhaust. The central position was considered the preferred solution however further work will need to be undertaken to investigate the pressure waves generated across the chamber. 30

31 From the feedback I received, I got some insight into the direction of further research for the split-cycle engine. The meeting gave much motivation; it was great to meet the people my project had actually been helping, and talking with the team was a real joy. Contribution to the E3C3 EU INTERREG project The study of evaporation process of multicomponent droplets is relevant for the EU INTERREG E3C3 project (E3C3, 2013). One of the tasks set in the E3C3 project focuses on evaporation of biofuels containing many hydrocarbon components. Hence the investigation of heating and evaporation of multicomponent fuel droplets is relevant for feasibility studies of CFD capabilities in this area. The rate of evaporation as predicted by ANSYS FLUENT, is monitored for both C 7 H 16 and C 8 H 18 components. It is observed that, as expected, at the initial stage n-heptane evaporates faster than iso-octane. This can be seen from Fig. 31 showing the transient behaviour of octane concentration in fuel vapour (on mass basis). In other words, the plot shows the ratio of mass fraction of octane vapour, to the mass fraction of fuel vapour. It is less than the value of 0.95 until 0.5ms. This demonstrates faster evaporation rate of n-heptane, as it can be expected based on physical properties. 1.00E E-01 Ratio of octane in fuel vapour by mass 9.00E E E E Figure 31. Ratio of octane vapour to fuel vapour by mass, for the spark-plug location of the injector with injection velocity of 50 m/s; Wave spray breakup model. 31

32 The results of the CFD simulations are used as a feasibility studies for multicomponent spray modelling under the E3C3 EU INTERREG research. The newly developed CFD spray models will take into account temperature gradients and species diffusion inside a multicomponent droplet. Whilst current project lays foundation for CFD implementation of the newly developed spray models at SHRL, further research on validation of the CFD results against available experimental data is needed. Conclusions The project focused on modelling of spray injection and fuel evaporation for studies of mixture preparation in a split-cycle engine. The modelling has been performed by numerical simulation using CFD code ANSYS FLUENT. The fuel droplets were set as multicomponent liquid, 95% of iso-octane C 8 H 18 and 5% of n-heptane C 7 H 16 by mass. Fuel spray evaporation and vapour mixing with air is calculated by the CFD code, and it can be monitored by post-processing of the results. CFD runs for three injector locations and various spray atomization options: No-Breakup, Wave and TAB breakup models, have been performed for realistic engine conditions. This has a direct relevance to EU INTERREG CEREEV project (CEREEV, 2012). The study of evaporation process of multicomponent droplets is relevant for the EU INTERREG E3C3 project (E3C3, 2013). One of the tasks set under the E3C3 project, focuses on biofuels containing many hydrocarbon components. Hence the investigation of heating and evaporation of multicomponent fuel droplets is relevant for feasibility studies of CFD capabilities in this area. The rate of evaporation as predicted by ANSYS FLUENT, is monitored for both C 7 H 16 and C 8 H 18 components. From my work and all the results I got, my conclusion would be that further research is needed to explore realistic flow patterns and injection strategies. Evaporation process for multicomponent droplets must be further explored. Engine data obtained by experimental methods shall be compared with CFD predictions. As a summary of current project, at this stage of development the recommended position for the injector would be by the spark plug. 32

33 References ANSYS (2013) Theory Guide, Products/ANSYS+Fluent as seen on 01/08/2013. Automotive Engineering (2013) Modelling of droplet and spray dynamics, heating and evaporation, Research Workshop, Centre for Automotive Engineering, Watts Bldg, University of Brighton, 16 th August CEEREV (2012) &Itemid=&lang=en E3C3 (2013), Energy Efficiency and Environment: a Cross-Channel Cluster Engine (2013) Definition of engine, as seen on 01/08/2013. FLUENT (2013) as seen on August Gent (2007) Gent, D. and Ritchie, R. OCR AS Chemistry, Heinemann, Essex UK, 2007 How Stuff Works (2013) as seen on August 2013 Nuffield (2013) Research Student Conference (2013) Science Accessible, Huxley Building, The Faculty of Science and Engineering Doctoral College, The University of Brighton, July Heywood (1988) Heywood, J.B., Internal Combustion Engine Fundamentals, McGraw-Hill Book Company, Singapore,

34 ICE (2013) Internal Combustion Engine, as seen on August Meldolesi (2012), Meldolesi, R., and Badain,N. 2012, Scuderi Split Cycle Engine: Air Hybrid Vehicle, SAE , doi: / ACE (2012), A fundamental study of the novel poppet valve 2-stroke auto ignition combustion engine (2-ACE), and EP/F058276/1 seen on 12/07/2013 Scuderi (2013) SCUDERI Group, Inc as seen in August Philips et al (2013), Phillips, F., Gilbert, I., Pirault, J., and Megel, M., Scuderi Split Cycle Research Engine: Overview, Architecture and Operation, SAE Int. J. Engines 4(1): , 2011, doi: / SHRL (2013) seen on 01/07/2013 Turner (2012), M. R., Sazhin, S.S., Healey, J.J., Crua, C. and Martynov, S.B. A breakup model for transient Diesel fuel sprays Fuel, 97, pp ISSN Vetronics (2013), as seen in August

35 Appendix 1: Attendance of research workshops and laboratory tours Sir Harry Ricardo Laboratories July 2013, Dr Steve Begg The University of Brighton and Ricardo UK jointly opened the Sir Harry Ricardo Laboratories on 14 November The SHRL are one of the largest UK research teams dedicated to internal combustion engines, the development of laser-based measurement techniques, fundamental modelling and computational simulation. It is regarded as one of the foremost centres for automotive engine research in Europe. The group's international esteem is demonstrated by its breadth of collaboration with over 40 academic institutions and industrial partners across the world. Tour of Vetronics laboratories July 2013, Professor E. Stipidis and Dr Panagiotis Oikonomidis as seen on 12/09/2013 The Vetronics Research Centre (VRC) is the only Academic Centre of Excellence in the UK conducting research and training in the subject area of Vetronics (Vehicle Electronics), sponsored by the UK Ministry of Defence (MOD) and supported by Defence Science Technology Laboratory (DSTL) and Defence Equipment and Support (DE&S). We focus on specialised and targeted research to investigate new technologies and methodologies that can be applied in the immediate and near future. Almost all of our research programmes include a functional demonstrator deliverable that can give a practical hands-on experience of the output. These systems are designed in a modular fashion to make them re-usable for an extension of the programme or even in other related applications, allowing us to keep re-development costs down and have access to an ever increasing set of testing environments. 35

36 SHRL Research workshop, 16 August 2013: Modelling of droplet and spray dynamics, heating and evaporation Centre for Automotive Engineering Research Workshop Friday, 16 th August, 2013 Room W623, Watts Building, University of Brighton Modelling of droplet and spray dynamics, heating and evaporation 14.00: Dr Natalia Lebedeva and Prof Alexander Osiptsov Application of the full Lagrangian and viscous-vortex methods to modelling of impulse two-phase jets 14.45: Dr Oyuna Rybdylova Numerical modelling of two-phase vortex ring flow Prof Vlad Gun ko Quantum chemical approach to study evaporation of Diesel fuel droplets 15.55: Dr Rasoul Nasiri Quantum chemical studies on n-alkane droplets 16.15: Mansour al Qubeissi Biodiesel fuel droplets: modelling of heating and evaporation processes 36

37 CEREEV Researcher exchange, technical workshop and project review 12 September 2013 i) Minutes ii) The presentation by the author 37