KMB (KIVA Multi-Blocks) based on KIVA-II. â Turbulence model : k-ε. â Atomization : Wave-FIPA (SAE 970881). â Spray/wall interaction, liquid film transport ...
Improved Modelling of DI Diesel Engines Using Sub-grid Descriptions of Spray and Combustion P. BEARD, O. COLIN and M. MICHE IFP, Rueil-Malmaison, France
Work supported by GSM: IFP, Peugeot, Renault
Presentation Sub-grid
models
hSpray hCombustion Results hDiesel simulation cell hDI Diesel engine Conclusion 2
2003-01-0008
Spray model (CLE) Coupled Lagrangian-Eulerian model (SAE 2000-01-1893) Goals : To reduce the mesh sensitivity of HP DI spray modelling, To improve the simulation of combustion in DI Diesel (DID) engines. Similar approaches (1D) : - Wan & Peters (SAE 972866) : Cross-section Averaged Spray - Abraham & Magi (SAE 1999-01-0911) : Virtual Liquid Source 3
2003-01-0008
Description of the CLE model KMB
4
2003-01-0008
Description of the CLE model KMB
KMB + CLE model
⇒ Over-estimated diffusion ! 5
2003-01-0008
Description of the CLE model KMB
KMB + CLE model
⇒ Over-estimated diffusion !
⇒ Reduced diffusion 6
2003-01-0008
Combustion model (ECFM3Z) Extended Coherent Flame Model 3 Zones = ECFM (Oil & Gas Science and Tech., 2003) + CFM3Z (to be published in Comb. Science and Tech.) Goals : To account for the local fuel stratification, To describe the mixing process between fuel and air. Similar model : Bensler, Bühren, Samson & Vervisch (SAE 200001-0662) : Ricardo Two Zone Flamelet
7
2003-01-0008
Description of the ECFM3Z model Sub-grid description of the local stratification : 3 zones Air (+ EGR)
Equivalence ratio
Turbulent mixing
Mixed zone
Fuel
8
2003-01-0008
Description of the ECFM3Z model Sub-grid description of the local stratification
Homogeneous reactor
Air (+ EGR)
Equivalence ratio
Auto-ignition
Mixed zone
Turbulent mixing
Fuel
9
2003-01-0008
Description of the ECFM3Z model Sub-grid description of the local stratification
Homogeneous reactors
Air (+ EGR) Mixed zone Equivalence ratio
Fresh
Burnt
gases
gases
Turbulent mixing
Auto-ignition
Fuel Premixed flame (oxidation) 10
2003-01-0008
Description of the ECFM3Z model Sub-grid description of the local stratification
Homogeneous reactors
Air (+ EGR) Mixed zone Equivalence ratio
Fresh
Burnt
gases
gases
Turbulent mixing
Auto-ignition
Fuel Premixed flame (oxidation) 11
Diffusion flame (oxidation+pollutant formation) 2003-01-0008
Computer code KMB (KIVA Multi-Blocks) based on KIVA-II Turbulence model : k-ε Open structured blocks (with refinement)
Atomization : Wave-FIPA (SAE 970881) Spray/wall interaction, liquid film transport and evaporation (Oil & Gas Science and Tech., 1999)
Pollutant formation : q NO
Moving walls
: extended Zeldovitch q Soot : 10 reaction mechanism (SAE 2001-01-3684)
Same constants for all runs 12
2003-01-0008
Presentation Sub-grid models hSpray hCombustion Results
hDiesel simulation cell hDI Diesel engine Conclusion 13
2003-01-0008
Diesel simulation cell Fuel : C12H26 Pinj ≤ 200 MPa, Dinj ≤ 200 µm
Injector
Window
Pch ≤ 15 MPa, Tch ≤ 1500 K
At SOI (Start Of Injection) : Vgas ≈ 0 m/s, k ≈ 0.06 m2/s2
Corner access port
14
2003-01-0008
Mesh sensitivity of the spray model Pinj = 80 MPa, Pch = 6 MPa, Tch = 800 K KMB + CLE model
Fuel vapor mass fraction 0 ≤ Yv ≤ 0.2
⇒ Reduced mesh sensitivity
∆x = 1 mm ∆z = 2.7 mm 15
∆x = 1 mm ∆z = 1.3 mm
∆x = 0.5 mm ∆z = 1.3 mm
2003-01-0008
Mesh sensitivity of the spray model Pinj = 80 MPa, Pch = 6 MPa, Tch = 800 K KMB KMB + CLE model
∆x = 2 mm ∆z = 2.7 mm
∆x = 1 mm ∆z = 2.7 mm 16
∆x = 1 mm ∆z = 1.3 mm
∆x = 0.5 mm ∆z = 1.3 mm
2003-01-0008
DI Diesel engine Peugeot passenger car engine
3D grid : ≈ 250 000 cells at BDC Simulation from IVO (intake TDC) to EVO (140 deg. ATDC)
17
2003-01-0008
DID engine piston geometry Piston bowl shapes
Constant compression ratio = 18 Cell size inside the bowl ≈ 1 mm 18
2003-01-0008
DID engine operating conditions
21 cases were computed
Engine speed (rpm) : Load : Global equivalence ratio : SOI (deg. BTDC) : EGR (%) :
1250, 1640, 4000 full or partial [0.46;0.83] 3, 5, 15, 29 [0;30]
Jet Wall Impact parameter (mm) : defined at TDC (SAE 2002-01-0495)
[3.3;8.9]
19
2003-01-0008
Influence of sub-grid models Case A : 4000 rpm, full load (Φ = 0.71), bowl 2
20
2003-01-0008
Influence of piston bowl shape Case B : 4000 rpm, full load (Φ = 0.71), bowl 3
Relative variations between cases A and B : - indicated power : measured -3 %, computed -6 % - soot emissions : measured +11 %, computed +18 % 21
2003-01-0008
Influence of JWI parameter (1) Case C8 (JWI = 8.1 mm) : 4000 rpm, full load (Φ = 0.71), bowl 1 KMB
KMB + CLE + ECFM3Z
Fuel vapor mass fraction 22
0 ≤ Yv ≤ 0.2
2003-01-0008
Influence of JWI parameter (2) Case C : 4000 rpm, full load (Φ = 0.71), bowl 1
-7 %
∆zinj = 0.9 mm
-10 %
23
2003-01-0008
Influence of EGR (1) Case F : 1640 rpm, partial load, bowl 2 Experiment
KMB + CLE + ECFM3Z
24
2003-01-0008
Influence of EGR (2) Case F : 1640 rpm, partial load, bowl 2
25
Computed
Soot emissions (mg/s)
Measured
NO emissions (mg/s)
2003-01-0008
Influence of pilot injection Case H : 1640 rpm, partial load (Φ = 0.64), 20% EGR, bowl 2
26
2003-01-0008
Conclusions (1) New sub-grid models CLE and ECFM3Z are applicable for DID engine simulations with multi-hole injectors and multi-injections. ⇒ Reduced mesh sensitivity of the spray model. ⇒ Without any tuning, improved prediction of hthe influence of operating conditions (engine speed and load, bowl shape, nozzle tip protrusion, EGR, pilot injection) on DID engine performance. hthe sensitivity of pollutant (NO, CO and soot) emissions. 27
2003-01-0008
Conclusions (2) The new sub-grid models were successfully used to optimize the design of HCCI engines. Nevertheless, relative variations between cases are sometimes under or over-estimated. ⇒ Additional developments are still required to become fully predictive : hCalculation of an auto-ignition delay accounting properly for the ambient conditions, in particular high EGR rates. hDescription of the mixing process between fuel and air.
28
2003-01-0008