Dual Fuel Engine Charge Motion & Combustion Study STAR-Global-Conference March 06-08, 2017 Berlin Kamlesh Ghael, Prof. Dr. Sebastian Kaiser (IVG-RF), M. Sc. Felix Rosenthal (IFKM-KIT)
Introduction: Operation of Dual-fuel Engine Primary fuel (natural gas) Premixed with air either in intake port or injected in the cylinder Secondary pilot fuel (with lower autoignition temperature - Diesel) non-premixed - injected to ignite primary fuel. Figure: Sample high-speed shadowgraph imaging sequence ( t 45 μs); SOI = -20 CAD & Φ = 0.8; D visu = 45 mm. [Used from: Salaun, E., Apeloig, J., Grisch, F., Yvonnet, C.-E., Nicolas, B. and Dionnet, F., Optical Investigation of Ignition Timing and Equivalence Ratio in Dual-Fuel CNG/Diesel Combustion, SAE Technical Paper, 2016-01- 0772, 2016, doi:10.4271/2016-01-0772] www.uni-due.de 06.03.2017 2
Introduction: Challenge(s) Simulations essential part of engine research and development process, as experiments are time and cost consuming. At leaner fuel-air ratio (Φ ~ 0.5), laminar flame speed of CH 4 is reduced. Short time available for combustion (ms) + slow flame speed incomplete combustion releases more pollutants like CO, formaldehyde, and unburnt CH 4. In this study Φ = 0.53 Figure: Burning velocity S L of methane vs. fuel-air-equivalence ratio Φ for various pressures with a fixed unburnt temperature T u of 298 K (solid lines). [Used from: Göttgens, J., Mauss, F. and Peters, N., Analytic approximations of burning velocities and flame thicknesses of lean hydrogen, methane, ethylene, ethane, acetylene, and propane flames, Symposium (International) on Combustion, 24(1): 129-135, 1992.] www.uni-due.de 06.03.2017 3 Laminar flame Turbulent flame
Introduction: Methods to Optimize In-cylinder Flow Pilot fuel injection Intake port geometry Valve lift, timing, and number Compression ratio Squish height Piston bowl geometry Swirl Squish Tumble Turbulence www.uni-due.de 06.03.2017 4
Engine Parameters & Numerical Setup Parameter Type Displacement Valves Fuel Squish height Speed Dimension/Configuration Single-cylinder dual-fuel heavy-duty engine 2 litres 2 intake, 2 exhaust Natural gas (premixed), Diesel (injected) 2 mm 1400 RPM Parameter Model Simulation Star CD es-ice 4.26 Turbulence Model Wall function k- RNG turbulence model Han-Reitz Figure: 1.4 Million cells (1.25 mm cell size) mesh generated by Automatic meshing tool www.uni-due.de 06.03.2017 5
Assumptions and Constants Assumptions: For cold flow (without combustion), only air is used as fluid. Ideal gas. For cold flow, walls of cylinder are adiabatic. Constants: Compression ratio kept constant for all variations i.e. ԑ = 12.45. For piston geometry variation, squish height is constant. www.uni-due.de 06.03.2017 6
Part I: Optimization of Turbulent Kinetic Energy (TKE) by Varying Piston Bowl Geometry and Squish Height Target: Optimize swirl and TKE by changing piston bowl geometry and squish height. Bowl throat radius Original Geometry 40% squish area 1 st Geometry 63% squish area Squish area Flank angle Pip height Toroid radius 2 nd Geometry 67% squish area 3 rd Geometry 79% squish area www.uni-due.de 06.03.2017 7
Swirl (mass) [-] Swirl (mass) [-] Valve Lift [mm] Intensity (u'/u) [-] Intensity (u'/u) [-] Valve Lift [mm] Pressure [bar] Valve Lift [mm] TKE [m²/s²] TKE [m²/s²] Valve Lift [mm] Part I: Results - Piston Bowl Geometries Results until 703 CA (SOI) from simulation are compared with experiment results with combustion Original_geometry 1st_geometry 2nd_geometry 3rd_geometry Experiment_Original Inlet valve lift Exhaust valve lift 360 400 440 480 520 560 600 640 680 720 760 Crank Angle [ CA] 360 400 440 680 480 690 520 700 560 710600 720640730680740720750 760 760 Crank Angle Crank [ CA] Angle [ CA] Increase in TKE, intensity & swirl 360 400 440 680 480 688 520696 560 704 712 600 720 640728 680 736 744 720 752 760 760 Crank Angle Crank [ CA] Angle [ CA] 360 400 440 680 480 690 520 700 560 710600 720640730680740 720750 760 760 Crank Angle Crank [ CA] Angle [ CA] www.uni-due.de 06.03.2017 8
Part I: Piston Bowl Geometries (Velocity)- (through half depth of squish height) Original geometry 1st geometry 2nd geometry 3rd geometry intake exhaust 709 CA High squeezed flow Velocity m/s2 24.50 720 CA = TDC 0.00 High energy back flow 735 CA www.uni-due.de 06.03.2017 9
Part I: Piston Bowl Geometries (TKE and Velocity)- (vertical plane through cylinder axis) Original geometry 1 st geometry 2 nd geometry 3 rd geometry 709 CA (Generates more TKE as well as swirl required for fast flame propagation) More turbulence due to elongated toroid 720 CA High swirl due to more volume above pip TKE (m 2 /s 2 ) 42.00 Velocity (m/s 2 ) 24.50 735 CA 0.00 0.00 www.uni-due.de 06.03.2017 10
Swirl (mass) [-] Valve Lift [mm] Intensity (u'/u) [-] Intensity (u'/u) [-] Valve Lift [mm] Pressure [bar] Valve Lift [mm] TKE [m²/s²] TKE [m²/s²] Valve Lift [mm] Part I: Influence of Squish Height on TKE and Swirl (at constant ε) Original_geometry 2nd_geometry 2nd_geometry_1.5mm 2nd_geometry_0.5mm Inlet valve lift Exhaust valve lift 360 400 440 480 520 560 600 640 680 720 760 Crank Angle [ CA] 360 400 440 680 480 690 520 700560 710600 720640730680740 720750 760 760 Crank Angle Crank [ CA] Angle [ CA] (more TKE as well as swirl) Increases initial swirl 360 400 440 480 520 560 600 640 680 720 760 Crank Angle [ CA] 360 400 440 680 480 690 520 700 560 710600 720640730680740 720750 760 760 Crank Angle Crank [ CA] Angle [ CA] www.uni-due.de 06.03.2017 11
Part II: Investigation of flame Propagation Target: Investigate flame propagation in original as well as selected geometry and show influence of charge motion on it. Progress Variable Model - Multi Fuel (PVM-MF) combustion model is used Reaction mechanism (from PVM - library): Fuel 1: CH 4 (natural gas primary fuel) Fuel 2: n-c 7 H 16 (n-heptane pilot fuel), reduced mechanism, 160 species and 1540 elementary reaction steps For non-premixed fuel: flamelet concept For premixed fuel: G equation concept Φ = 0.53 www.uni-due.de 06.03.2017 12
Part II: PVM-MF Combustion Model Multi-Fuel Coupling, Simulation Boundary Conditions PVM-MF coupling for dual fuel combustion Assumptions: Uniform mixture of CH 4 and air inside cylinder. Parameter p (690 CA) T (690 CA) Cylinder head temperature Piston temperature Cylinder wall temperature Total Equivalence ratio (Φ) SOI Values 23.74 bar 697 K 485.85 K (constant) 505.85 K (constant) 465.85 K (constant) 0.53 703 CA www.uni-due.de 06.03.2017 13
Part II: Sector Model Suitable for simulation of fuel injection and combustion in axis symmetric engine cylinders. Reduces simulation time considerably as number of cells decreases in comparison to full cylinder model. However, it is unable to model gas exchange and assumes uniform charge flow throughout the cylinder. Original geometry sector mesh Modified geometry (3 rd with 0.5 mm squish height) sector mesh www.uni-due.de 06.03.2017 14
Injection mass flow rate [kg/s] AHRR [J/deg] Pressure [bar] Part II: Investigation of Flame Propagation Combustion Model Mass flow rate AHRR_Sim_default AHRR_Exp Pressure_Sim_default Pressure_Exp 700 701 702 703 704 705 706 707 708 709 710 Crank Angle [ CA] 690 695 700 705 710 715 720 725 730 735 740 Crank Angle [ CA] Default combustion model settings shows longer ignition delay than experimental delay. Longer ignition delay is probably due loss of most reactive mixture fraction which is available only for shorter time. This is due to variation in atomization of pilot fuel and dilution of n-heptane fuel in CH 4 -air mixture. Need more experimental data for validation Thus, reaction rate is scaled in progress variable transport equation. www.uni-due.de 06.03.2017 15
Pressure [bar] AHRR [J/deg] Part II: Results (Pressure and AHRR Comparison) Modified geometry Original geometry Exp original geometry 680 690 700 710 720 730 740 750 760 Crank Angle [ CA] 680 690 700 710 720 730 740 750 760 Crank Angle [ CA] More heat release at first peak due to more CH 4 trapped in pilot fuel cloud Heat release rate (HRR) after 720 ºCA decreasing in simulation of original geometry probably due to different flow field in sector model HRR is higher in modified geometry as a result of faster flame propagation www.uni-due.de 06.03.2017 16
Part III: Results 709 CA 720 CA 735 CA Full cylinder model m/s 2 24.50 Sector Model 0.00 Probable reason for variation in flame propagation after TDC is due to difference in flow field between full model and sector model of original geometry www.uni-due.de 06.03.2017 17
Part III: Flame Front, Progress Variable, Temperature CA 709.25 709.50 710.00 712.00 718.00 730.00 750.00 Flame front - Original geometry burnt Flame front - Modified geometry Progress variable - original geometry Progress variable - modified geometry Earlier ignition Faster flame propagation burning unburnt Flame front 1.00 (PV), 2800 K (T) Temperature - original geometry Temperature - modified geometry 0.00 www.uni-due.de 06.03.2017 18
Summary Increase in squish area influences swirl up to an optimal value (0.73 for this case) and after an optimal value, it influences turbulence. Elongated toroidal area decays large tumble flow into turbulence. More area above pip generates more swirl, while more pip height decreases swirl and replaces charge flow in toroid region Re-entrant shape of bowl enhance turbulence and allows faster flame propagation in piston clearance area. Decreasing squish height, increases swirl and TKE. Numerical investigation shows faster flame propagation in modified geometry with high TKE. www.uni-due.de 06.03.2017 19
Thank you for your kind attention! www.uni-due.de 06.03.2017 20