• Tidak ada hasil yang ditemukan

Mixing homogeneity improvement

7.3 Homogeneous-charge direct injection (DI) system design and optimization

7.3.2 Mixing homogeneity improvement

The fuel–air mixing process for homogeneous-charge operation starts with fuel injection into the cylinder at a prescribed fuel pressure. The fuel spray is atomized into fuel droplets, and the fuel droplets are transported throughout the combustion chamber. The droplets subsequently vaporize and mix with fresh air through turbulence mixing and molecular level mixing, then form a well-mixed fuel–air mixture for combustion. Even though all these processes are important to the mixture homogeneity, even distribution of fuel droplets and sufficiently rapid droplet evaporation are the overriding characteristics essential for good mixture quality. The evaporation process is determined by the droplet size, the droplet temperature, and the surrounding gas thermodynamic environment. The droplet size depends on the injector type, the nozzle opening size, and the operating fuel pressure. The fuel dispersion is dominated by the interaction between the fuel spray and the in-cylinder flow motion. This interaction depends on the spray droplet characteristics such as droplet size and penetration, injector plume orientation, and also the in-cylinder flow structure. This interaction can be optimized through adjustment of the injector spray pattern, through controlling the injector

Advanced direct injection CET and development

7.12 Schematic of a CFD-based port optimization process (Yi et al., 2002).

Experimental data

Calibrated numerical results

Automated processes with in-house software

CFD analysis

Cd

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 L/D

Hard points defined by design engineer

operating conditions, such as fuel pressure or injection timing, and through intake port or chamber design. This interaction is highly three-dimensional and time dependent, and is also a strong function of engine operating conditions such as speed and load.

There are two major factors that make fuel–air mixing homogeneity very challenging at high engine speeds and high-load operation. One factor is that it is very hard to distribute the fuel spray evenly in the cylinder, because

(a)

30°

(b)

7.13 Audi FSI combustion system with a valve in the intake port to improve tumble motion, (Wurms et al., 2002).

the fuel spray trajectory is highly influenced by the intake flow with high momentum. The other factor is that there is very limited flexibility in the fuel injection timing due to the fact that the injector flow rate is generally designed such that the injection window is about the same as the duration of the intake process at high-speed, high-load conditions. Increasing the injector flow rate helps mixing homogeneity at high-speed, high-load conditions. However, it may cause fuel metering difficulty at low loads such as idle conditions. At a moderate engine speed and high-load conditions, such as 3000 rpm WOT (wide open throttle), the unique challenge to mixing homogeneity is that the injection duration is not short enough to allow much flexibility for the injection strategy, such as split injection, while at the same time the engine speed is not high enough to promote strong turbulence mixing. At low- speed, full-load operation, some flexibility exists for advanced fuel injection strategies. However, low turbulence mixing, due to the low engine speed, creates a challenge for good mixing. Shown in Fig. 7.14 are engine smoke emission measurements from a research DiSi engine at full-load operation over a wide engine speed range. it can be seen that the smoke emissions are higher at both low and high speed operations. The smoke emissions can be used as an indicator of the mixture homogeneity, because the smoke emission is believed to be closely associated with fuel-rich mixtures at these operating conditions.

The work of Yi et al. (2000a, 2002b, 2002) illustrated how three-dimensional modeling can help to optimize combustion chamber design with improved

0 1000 2000 3000 4000 5000 6000 7000 Engine speed (rpm)

Soot (filter smoke number)

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

7.14 Engine soot emission measurements vs. engine speed for a typical DISI engine with a high tumble intake port. The engine is equipped with a SCV (swirl control valve) in one of its two intake ports. The corresponding operation mode is full-load operation with the SCV open (Yi et al., 2002).

fuel–air mixing homogeneity at high-speed and high-load conditions by alleviating the intake flow impact on the fuel spray trajectory. Shown in Plate II is the simulated air–fuel ratio distribution at 6000 rpm and full- load condition with the initial design of this combustion system (Yi et al., 2002). The combustion system is based on a typical tumble port concept. As shown in Plate II (between pages 172 and 173), even at 20° bTDC (before TDC) during the compression stroke, about the time of ignition, there is still very strong stratification in the mixture. A rich mixture pocket is located on the intake side, while a lean mixture exists on the exhaust side. Yi et al.

identified the root cause of the mixing inhomogeneity as uneven spray droplet distribution inside the cylinder. The fuel spray trajectory was strongly affected by the intake flow motion, which pushed the fuel spray away from its axial penetration direction towards the intake port. The bowl-in-piston geometry in this DiSi engine further complicated the interaction between the intake flow and the fuel spray by drawing the airflow into the bowl and forming a vortex structure. The direction of the left branch of the vortex structure was opposite to that of the fuel droplet penetration, which further prevented the fuel droplets from penetrating across the cylinder, as shown in Fig. 7.15.

7.15 Corresponding velocity and fuel spray distribution at BDC. The intake flow pushes the fuel spray to the intake side, which causes fuel–air mixing inhomogeneity as shown in Plate II (Yi et al., 2002).

It was found that such an interaction between the intake flow and the fuel spray could be alleviated by designing a mask around the injector pocket to deflect the intake flow from impinging on the injected fuel spray. As a result of this local redirection of the intake flow motion, the fuel spray was more evenly distributed and the fuel–air mixing homogeneity was improved, as shown in Plate III (between pages 172 and 173).

Lippert et al. (2004a) demonstrated that three-dimensional modeling can be used to reveal the physical insight of the injection timing effect on the engine torque and volumetric efficiency as observed in dynamometer testing at 3000 rpm full-load condition in a four-valve pent-roof DISI engine. The dynamometer results indicated that as the injection timing advanced, the volumetric efficiency decreased between EOI (end of injection) of 180° to 320° bTDC. Their modeling showed that at early injection timings with an EOI of 280° bTDC, as much as 33% of the injected fuel impinges on the cylinder liner. However, the liner wetting with EOI of 180° bTDC is only about 1.5%, as shown in Fig. 7.16. The high liner wetting with early injection timings results in the fuel spray absorbing significant heat from the cylinder wall, which reduces the charge cooling effect that is characteristic of DiSi engines. Lippert et al. also noted that the predicted fuel evaporation due to impingement may be greater than in the actual engine due to the Leidenfrost effect, whereby a vapor cushion insulates the liquid droplet from the hot wall. The wide spectrum of the fuel composition may also contribute to a discrepancy in the predicted evaporation compared with that observed. Another factor that can contribute to the discrepancy is the existence of an oil film on the cylinder liner. However, the proper trend of the fuel impingement over the injection timing range is valid. Excessive liner impingement could also impact engine durability, due to oil dilution, and fuel impingement on the piston surface could be a source of soot emissions. in engine combustion system optimization, it is important to minimize both liner wetting and piston wetting.

ikoma et al. (2006) applied CFD modeling to optimize their slit injector spray pattern in the new Toyota 3.5-liter DISI gasoline engine (2GR-FSE) development. This engine is equipped with a dual fuel system with both PFI and DI fuel injection systems as shown in Fig. 7.17. The combustion system is an evolution of Toyota’s 3GR-FSE engine with a pent-roof-type chamber, tapered squish, and an SCV (swirl control valve) in one of its intake ports. The DI fuel system in the dual fuel system employs a slit injector that generates a dual-fan-shaped spray with wide dispersion. CFD modeling was utilized to assist the selection of the DI fuel spray pattern. The CFD modeling results shown in Plate IV (between pages 172 and 173) suggest that the slit injector emits fuel perpendicularly to the piston with wide dispersion in the cylinder, while the conventional spray causes the fuel to be trapped by the piston cavity and prevents penetration toward the exhaust side of the combustion

P1-MH3 P1-MH4 P1-Fan1 CFD P1-Fan1 CFD P1-MH4

180 200 220 240 260 280 300 320

End of injection (BTDC) (a)

5%

Volumetric efficiency (%)

–360 –320 –280 –240 –200 –160 –120 –80 –40 CA (°aTDC)

(b)

Vaporized mass (% injected)

100 90 80 70 60 50 40 30 20 10 0

Wall vaporized

EOI = 280 EOI = 250 EOI = 181

7.16 (a) Injection timing effect on volumetric efficiency, and (b) vaporized fuel from cylinder wall, at 3000 rpm full-load conditions with multi-hole (MH) and slit (fan) injector sprays (Lippert et al., 2004a).

33%

15%

1.5%

chamber. The combination of the new fuel system and elimination of SCV valve improves the low speed torque by about 7%.

As mentioned earlier, excessive fuel impingement on the cylinder and chamber surfaces could cause damage to a DiSi engine. As in PFi engines, liner wetting may cause oil dilution that affects engine durability. Surface impingement can also cause soot emissions in a DiSi engine, whereas that is seldom the case in PFI engines. Soot formation is a very complex process that includes many stages of soot particle generation and growth (Heywood, 1988). The capability to accurately predict soot formation in CFD modeling is limited by the understanding of the fundamental physical and chemical processes during soot formation. Even if the steps to soot formation were well understood, the computational resources required for three-dimensional simulation of practical engine geometry could still be beyond current computer power. Certain correlations between predicted wall wetting and oil dilution or soot emissions have been developed to provide an engineering guideline for combustion system optimization. Significant progress has been made in the last two decades in the understanding of the physics of the fuel impingement and fuel film dynamics on a solid surface for engine operating conditions (Naber and Reitz, 1988; Bai and Gosman, 1996; Han et al., 2000).

Yi et al. (2004a) identified that excessive amounts of fuel impingement on the intake valve surfaces could cause high soot emissions in their DiSi engine equipped with a swirl injector mounted underneath the intake ports.

They further established a correlation between the amount of fuel impinging on the intake valves and the engine-out soot emissions. Shown in Fig. 7.18 is the correlation between CFD predicted intake valve fuel wetting and the engine-out soot emissions for two swirl injectors with 70° and 60° spray

Spark plug

PFI injector

High flow efficiency intake port High-pressure fuel injector Shell-shaped piston

Fuel spray

Front view Side view

Needle Seat

Sac Slit

7.17 Toyota D-4S combustion system equipped with both PFI and DI fuel injection systems (left); and DI fuel system with dual-fan-shaped spray (right) (Ikoma et al., 2006).