**1. Introduction**

The problem of the environmental impact of energy conversion systems is particularly felt in the automotive field as a consequence of the wide diffusion of internal combustion engines within the transportation systems, and of the very high concentration of vehicles in the urban areas. Several actions, therefore, are today being taken by car manufacturers and researchers towards the development of more and more efficient propulsion systems, characterised by lower and lower pollutants emissions. In fact, despite the recent efforts aimed at developing alternative technologies, it is likely that the internal combustion engine will remain dominant for the next 30 years and beyond. This implies that the study and the optimisation of the thermo-fluid-dynamic processes characterising its operation will undoubtedly continue to play a determining role in the forthcoming scenario.

The major difficulties today encountered in the experimental characterization of combustion and pollutants formation in both spark ignition (SI) and compression ignition (CI) engines rely in the low spatial and temporal resolution achievable from measurements, as well as in the possible influence of instruments on the same phenomena to be investigated. The diagnostics capability surely benefits of the development of non-intrusive optical techniques, although constructive and economic problems still hinder their broad use. On the other hand, the introduction of increasingly accurate physical and chemical models and the simultaneous growth of the processors speed have led to a diffuse use of computational fluid dynamics (CFD) techniques, especially in the phase of engine design. A wide variety of geometrical configurations or sets of engine parameters, indeed, are today suitable of being analysed into detail through models of various complexities at relatively low costs, or optimised according to predefined objectives.

As regards SI engines, in particular, the most pursued solution for the improvement of fuel economy relies on engine downsizing, coupled with turbo-charging and direct injection (DI): the engine displacement is reduced, whereas an increase of the low end torque is realised by air boosting, compression ratio rising and gasoline injection directly in the combustion chamber. These measures allow overwhelming the main shortcoming of engines mounting port fuel injection (PFI) systems, with mixture formation occurring within the intake ducts, namely the significant engine pumping losses at part-load operation (where the engine works during most of an urban driving cycle), caused by the throttle load

Numerical Modelling and Optimization of the

combustion processes are discussed.

the ignition location [Stan, 2000].

space.

**injectors**

Mixture Formation Process by Multi-Hole Injectors in a GDI Engine 179

load and a moderate-load. In this last case, a globally lean stratified operation is studied. The influence of important parameters, as the injector orientation, the start of injection (SOI), the time of spark ignition (SI) and the injection pressure on the mixture formation and

Results of a study of split injection under lean-mixture, moderate-speed, moderate-load working conditions are finally presented. Major aim is the application of an optimization technique for the individuation of the mixture formation process realising a combustion the most effective for the maximization of the engine work. The software modeFRONTIERTM is chosen to automatically run the 3D numerical engine model. The Simplex algorithm is adopted to draw the choice of the input parameters in the design of experiments (DOE)

**2. Experimental characterisation of the GDI spray dynamics from multi-hole** 

Either the desired charge stratification around the spark plug in lean-mixture operation, or the greatest homogeneity under stoichiometric conditions, is achievable in GDI engines through different modes of gasoline-air mixture formation. In the so-called *wall-guided* mode the gasoline spray is directed towards the piston, which exhibits a properly shaped "nose" deflecting the mixture cloud in the vicinity of the spark plug. In the *air-guided* mode the mixture richer region is brought towards the ignition location by the tumble motion of the air entering from the intake ducts. Finally, in the *jet-guided* or *spray-guided* mode typically the spacing between the injector and the spark is smaller, with the fuel spray injected close to

Several kinds of injectors for GDI applications are today available. The earliest solution to reduce the rapidly changing fuel concentration gradients as the fuel passes the spark location during the injection period, hence to increase the combustion robustness, relies on the adoption of air-assisted injection systems, such as the one developed by Orbital Engine [Cathcart and Railton, 2001]. This technology is today still applied, because it offers an additional degree of freedom constituted by the direct injection of air, that allows a more effective control of local oxygen concentration, temperature and charge motion through the cycle [Shim *et al.*, 2008]. Alternative solutions, better meeting the requirements for the development of more efficient GDI engines, are the high pressure injectors: the swirl type injector generates an hollow-cone fuel spray by providing a swirl rotational motion to the fuel, that widely disperses and well-atomizes the spray at moderate injection pressures [Brewster *et al.*, 2008]; the multi-hole configuration, on the other hand, exhibits flexible spray patterns that reduce the fuel impingement on the cylinder walls and improve the spray

Three commercial multi-hole injectors suitable to be mounted on high-performances SI engines are tested within the present work. As mentioned in the Introduction, the major aim is the assessment of a complete database for the development of a 3D numerical model for the spray dynamics. Table 1 reports the holes number and diameter, as well as the exact flow rates of the considered injectors. The axes of the single jets coming from the nozzles are configured to depict different spray footprint structures. Two injectors are manufactured by Bosch, type HDEV 5.1, differing for the holes number, six for Injector #1, seven for Injector # 2, distributed regularly on a circumference to form an ellipsoidal-like hollow-cone geometry. The third injector is a six-hole Continental device, with five holes distributed over

stability (cone shape) with respect to the existing backpressure.

control. The more precise gasoline dosage determines mixture formation processes the more effective for the development and stability of combustion. Knock arising benefits of the charge cooling effect offered by directly injecting the fuel into the cylinder. The engine can even be operated with overall lean mixtures through the so-called stratification of the charge, namely by creating a zone with stoichiometric air-to-fuel ratio around the spark plug and leaner conditions close to the cylinder walls. This reduces the wall heat losses, the unburned hydrocarbons (HC) and the carbon monoxide (CO) formation, and, at the same time, strongly increases the engine volumetric efficiency.

A lean engine operation, to the sake of truth, is generally feasible only at low loads and speeds, while at the higher loads, and at all loads and high speeds, the engine better works as homogeneous/stoichiometric. In fact, the in-cylinder turbulence intensity increases with the engine speed, hindering stable stratifications to be achieved at the higher regimes; on the other hand, increasing loads while attempting to maintain a stratified charge leads to mixtures locally excessively rich, that may cause an undesired increase of soot formation.

The realization of these different mixture conditions, namely of what is commonly called "mixed mode" direct injection boosting, encounters several difficulties in the practice. These range between the need of having at disposal increasingly efficient after-treatment systems for lean operation and the will to gain further insight into the in-cylinder mixture formation and combustion processes [Küsell *et al.*, 1999; Çelik and Özdalyan, 2010; Alkidas, 2007]. At present, technical solutions aimed at realising mixture conditions optimal for stable combustion, with low emissions and gasoline consumption over the whole engine working map, mainly rely on the employment of new generation high pressure GDI injectors (especially in the multi-hole configuration), as well as on the possibility to resort to split injections in the low speed regimes.

By controlling the spray orientation and fragmentation, a flexible charge stratification can be achieved, that, case by case, as the engine load and speed are changed, is able to give rise to an optimal combustion processes development.

Present work is focused on two activities performed by authors within the CFD approach to the study of in-cylinder processes in SI engines: the development of a three-dimensional (3D) numerical model for the GDI spray dynamics [Costa *et al.*, 2010], whose prediction capability is improved through a Simplex optimization algorithm, and the assessment of a procedure for the fuel consumption reduction based on the optimal synchronization of injection within the engine working cycle [Costa *et al.*, 2011]. The work is organised as following.

A certain insight into the behaviour of new generation GDI multi-hole injectors is first given, by experimentally characterizing three different devices both at the mass flow rate test bench and in an optically accessible vessel. Single and double injection strategies are considered.

The experimental activity is finalised to the creation of a database to be used for the assessment of a 3D numerical model for the GDI spray dynamics. The model is developed within the AVL FireTM code [www.avl.com] and exploits a log-normal distribution for the initial droplets diameter, whose expected value and variance are properly defined.

Hints for a GDI engine design are then given. One of the three injector is considered as mounted on a single-cylinder engine, four-valve, four-stroke, 638 cc displacement, suitable for motorbike applications. A 3D model able to reproduce the in-cylinder energy conversion process, namely the whole pressure cycle, is build within the same commercial software environment. Two typical engine operating conditions are taken under examination, a high-

control. The more precise gasoline dosage determines mixture formation processes the more effective for the development and stability of combustion. Knock arising benefits of the charge cooling effect offered by directly injecting the fuel into the cylinder. The engine can even be operated with overall lean mixtures through the so-called stratification of the charge, namely by creating a zone with stoichiometric air-to-fuel ratio around the spark plug and leaner conditions close to the cylinder walls. This reduces the wall heat losses, the unburned hydrocarbons (HC) and the carbon monoxide (CO) formation, and, at the same

A lean engine operation, to the sake of truth, is generally feasible only at low loads and speeds, while at the higher loads, and at all loads and high speeds, the engine better works as homogeneous/stoichiometric. In fact, the in-cylinder turbulence intensity increases with the engine speed, hindering stable stratifications to be achieved at the higher regimes; on the other hand, increasing loads while attempting to maintain a stratified charge leads to mixtures locally excessively rich, that may cause an undesired increase of soot formation. The realization of these different mixture conditions, namely of what is commonly called "mixed mode" direct injection boosting, encounters several difficulties in the practice. These range between the need of having at disposal increasingly efficient after-treatment systems for lean operation and the will to gain further insight into the in-cylinder mixture formation and combustion processes [Küsell *et al.*, 1999; Çelik and Özdalyan, 2010; Alkidas, 2007]. At present, technical solutions aimed at realising mixture conditions optimal for stable combustion, with low emissions and gasoline consumption over the whole engine working map, mainly rely on the employment of new generation high pressure GDI injectors (especially in the multi-hole configuration), as well as on the possibility to resort to split

By controlling the spray orientation and fragmentation, a flexible charge stratification can be achieved, that, case by case, as the engine load and speed are changed, is able to give rise to

Present work is focused on two activities performed by authors within the CFD approach to the study of in-cylinder processes in SI engines: the development of a three-dimensional (3D) numerical model for the GDI spray dynamics [Costa *et al.*, 2010], whose prediction capability is improved through a Simplex optimization algorithm, and the assessment of a procedure for the fuel consumption reduction based on the optimal synchronization of injection within the engine working cycle [Costa *et al.*, 2011]. The work is organised as

A certain insight into the behaviour of new generation GDI multi-hole injectors is first given, by experimentally characterizing three different devices both at the mass flow rate test bench and in an optically accessible vessel. Single and double injection strategies are

The experimental activity is finalised to the creation of a database to be used for the assessment of a 3D numerical model for the GDI spray dynamics. The model is developed within the AVL FireTM code [www.avl.com] and exploits a log-normal distribution for the

Hints for a GDI engine design are then given. One of the three injector is considered as mounted on a single-cylinder engine, four-valve, four-stroke, 638 cc displacement, suitable for motorbike applications. A 3D model able to reproduce the in-cylinder energy conversion process, namely the whole pressure cycle, is build within the same commercial software environment. Two typical engine operating conditions are taken under examination, a high-

initial droplets diameter, whose expected value and variance are properly defined.

time, strongly increases the engine volumetric efficiency.

injections in the low speed regimes.

following.

considered.

an optimal combustion processes development.

load and a moderate-load. In this last case, a globally lean stratified operation is studied. The influence of important parameters, as the injector orientation, the start of injection (SOI), the time of spark ignition (SI) and the injection pressure on the mixture formation and combustion processes are discussed.

Results of a study of split injection under lean-mixture, moderate-speed, moderate-load working conditions are finally presented. Major aim is the application of an optimization technique for the individuation of the mixture formation process realising a combustion the most effective for the maximization of the engine work. The software modeFRONTIERTM is chosen to automatically run the 3D numerical engine model. The Simplex algorithm is adopted to draw the choice of the input parameters in the design of experiments (DOE) space.
