Modelling Combustion in DI-SI using the G-equation Method and Detailed Chemistry: Emissions and knock Realize innovation. M.Zellat, D.Abouri, Y.Liang, C.Kralj
Main topics of the presentation 1. Context and combustion modelling problematic 2. Presentation and description of the proposed approach 3. Application SI-GDI low load and low speed : EGR sweep SI-GDI full load : Emissions SI-GDI full load : knock 4. Conclusion Page 2
Introduction : Schematic of Premixed and partially premixed Regimes Flame brush is thickened Flame structure changes Thermal and molecular diffusion Velocity Turbulence Flame speed Karlowitz > 1 Laminar flame thickness > smallest turbulent scale Broken Reaction Zones - Kolmogorov is now smaller than reaction sheet thickness - Flame now lacks local structure (extinguishes) Length Turbulence Flame thickness Karlowitz < 1 Laminar flame thickness < smallest turbulent scale Corrugated & Wrinkled Flamelets - Flame still retains laminar flame structure Page 3 - Wrinkled or corrugated by turbulent eddies
Introduction : Schematic of Premixed and partially premixed Regimes Flame brush is thickened Flame structure changes Thermal and molecular diffusion ENGINE MAP Page 4 Part load : Low speed, EGR Fuel consumption, Emission (chemistry) Mid load (EGR?) Driving comfort, Emission (mixed mode ) Full load Selling point, knock (propagation and chemistry)
Model need to cover whole flame regime : Turbulence and detailed chemistry Detailed chemistry for post and front flame COUPLED G-equation model for flame propagation ELEMENTS h c o n END SPECIES c7h16 o2 n2 co2 h2o co oh h2o2 ho2... REACTIONS MOLES CAL/MOLE c7h16<=>h+c7h15-2 1.300e+88-21.01 1.395e+05 rev/ 2.263e+83-20.31 4.083e+04 / c7h16+h<=>c7h15-2+h2 2.600e+06 2.40 4.471e+03 rev/ 1.807e+01 3.38 9.318e+03 / c7h16+o<=>c7h15-2+oh 9.540e+04 2.71 2.106e+03 rev/ 3.481e-01 3.67 5.541e+03 / c7h16+oh<=>c7h15-2+h2o 1.900e+06 2.00-5.960e+02... Averaged C profile Instantaneous C profile Fuel A transport equation for the G to track the front flame A transport equation for the variance from which we can obtain flame thickness An averaged profile for the progress variable Additional equations for mixture inhomogeneity Page 5 Background : AIR+EGR
The Detailed chemistry solver: CPU time Reduce the expense to obtain the required results Detailed chemistry solver : Dynamic Reduction Direct Relation Graph Detailed chemistry solver : Dynamic multizone method (DMZ) Create a groups of cells with similar thermochemical state Number of species history The number of groups will be much smaller than the number of cells Very useful for knock analysis Diesel combustion : in-cylinder pressure Consider thermal and species diffusion (Lewis number 1) Temperature history in homogeneous reactor Page 6 Applications are all with the reduced ERC PFR mechanism (73 species and 296 reactions)
Application I : Central Hollow cone injector Spray Guided Operating condition : 2000 rpm Low Load Increasing trapped Residual Burn Gases : Valve overlapping Low Moderate Medium High Page 7 Same injected quantity Load is maintained by Spark Timing
Application I : Central Hollow cone injector Spray Guided Operating condition : 2000 rpm Low Load Mean In-cylinder Pressure Low Moderate Experiments Prediction Medium High Page 8
Application I : Central Hollow cone injector Spray Guided Operating condition : 2000 rpm Low Load Apparent Rate Of Heat Release Low Moderate Experiments Prediction Medium High Page 9
Application I : Central Hollow cone injector Spray Guided Operating condition : 2000 rpm Low Load Integral of Apparent Rate Of Heat Release Low Moderate Experiments Prediction Medium High Page 10
Application I : Central Hollow cone injector Spray Guided Operating condition : 2000 rpm Low Load Mean angular momentum Low Max=0.4 Turbulent viscosity H2O2 field intermediate temperature branching High Compensate lower reactivity due to EGR In addition to larger spark timing (Low~20 CA BTDC High~ 40 CA BTDC) Max=0.1 Page 11
Application II : GDI Spray Guide 6holes Nozzle type Full load Page 12
Application II : OP3 : 2000 rpm full load (maximum torque) Pressure history Apparent rate of Heat Release Integral of Apparent rate of Heat Release Experiments Prediction Page 13
Application II : OP4 : 3000 rpm full load (maximum torque) Pressure history Apparent rate of Heat Release Integral of Apparent rate of Heat Release Experiments Prediction Page 14
Application II : OP5 : 5800 rpm Full load (maximum power) Pressure history Apparent rate of Heat Release Integral of Apparent rate of Heat Release Experiments Prediction Page 15
Application II : Full load : Engine cases : OP3, OP4 and OP5 NOx (NORA model), CO, CO2 and I.M.E.P Levels 6 5 4 3 2 1 0 3000 2500 2000 1500 1000 500 0 EXPERIMENTAL STAR-CD OP3 OP4 OP5 EXPERIMENTAL STAR-CD OP3 OP4 OP5 CO emissions NOx emissions 14 12 10 8 6 4 2 0 25 24 23 22 21 20 19 18 EXPERIMENTAL STAR-CD OP3 OP4 OP5 CO2 emissions EXPERIMENTAL RUN OP3 OP4 OP5 IMEP on closed cycle [bar] Page 16
Application II : GDI Spray Guided GDI Engine cases : Case, 3 4 and 5 Full load Soot Distribution SOOT SECTIONAL METHOD Diameter 3 4 5 Page 17
Application II : GDI Spray guided Spark timing +3 degrees CA knocking condition@ 5200 RPM Full load Iso-G Iso-Temperature fuel TDC 5CA ATDC As expected the thin reaction zone is in the thinner layer Page 18
Application II : GDI Spray guided 2000 RPM Full load Spark timing +3 degrees CA knocking condition Mean In-cylinder pressure Monitoring local pressure Experiments Prediction + 3 CA advance ST Severe knock (expected) Monitoring local pressure Monitoring local pressure Propyl radical H2O2 Intermediate Temperature Chain Branching Page 19
Application II : GDI Spray guided 2000 RPM Full load Spark timing +3 degrees CA knocking condition : Chain Branching radical indicating knock onset in space and time Monitoring local pressure Progress variable of combustion Page 20 Propyl radical H2O2 Intermediate Temperature Chain Branching
SUMMARY G-equation and detailed chemistry in STAR-CD V4.28 * has been investigated and validated for different engine configurations Low and full load combustion have been investigated The combination of G-equation/detailed chemistry and tables demonstrate good capabilities in a large range of the engine map (speed, load, EGR, knock) Good match for global thermodynamic quantities Good match for combustion history Absolute for emissions (Soot and NOx) are well predicted The soot sectional method is very promising to control soot particle size at source (source terms tabulated or computed on the fly from the detailed mechanism) Currently investigating Diesel combustion and detailed chemistry, SI Large eddy simulation and detailed chemistry Other details IMEM@SAE 2017 (IMEM@SAE2016) Date release STAR-CD V4.28 : May 2017 Page 21 *Actually STAR-CD V4.27 STABLE development version