Experimental and numerical analysis of a high-pressure outwardly opening hollow cone spray injector for automotive engines

Fuel 196 (2017) 508–519

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Experimental and numerical analysis of a high-pressure outwardly opening hollow cone spray injector for automotive engines Marianna Migliaccio ⇑, Alessandro Montanaro, Carlo Beatrice, Pierpaolo Napolitano, Luigi Allocca, Valentina Fraioli Istituto Motori – Consiglio Nazionale delle Ricerche, Italy

a r t i c l e

i n f o

Article history: Received 13 May 2016 Received in revised form 1 December 2016 Accepted 5 January 2017 Available online 16 February 2017 Keywords: Pintle-type injector Hollow cone nozzle Diesel engine Gasoline direct-injection compression ignition engine

a b s t r a c t The manuscript describes the characterization of a novel outwardly opening pintle-type injector, characterized by a Hollow Cone Nozzle (HCN), as a possible solution for ultra-low emission high-efficiency DI diesel and GDICI engines. In its prototypal version, the injector is capable of generating the hollow cone spray at relatively high injection pressure with a very precise fuel metering. The study concerns some experimental and numerical activities aimed at the analysis of the spray generated by the HCN. A commercial supplier provided the prototypal version of the injector with a dedicated piezoelectric actuation system and a proper choice of geometrical design parameters. The experimental characteristics of the HCN concept (in terms of spray pattern and spatial penetration) were analyzed in a constant volume vessel. The OpenFOAMÒ libraries, in the lib-ICE version of the code, were employed for the simulation of the spray dynamics after a first validation phase based on the experimental data. Results show a typical spray structure and the overall fluid-dynamics of the outwardly nozzle with a finely atomized spray, circumferentially well distributed, but displaying a reduced tip penetration. These characteristics appear very interesting, being well suited for the application of strategic spray concepts, to be coupled with dedicated combustion chamber designs both for CI and GDICI engines development. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Combustion quality in modern diesel engines strictly depends on the efficiency of the air-fuel mixing, and in turn, on the quality of spray atomization process. As well known, air-fuel mixing is strongly influenced by injection pressure, geometry of the nozzle and the hydraulic characteristics of the injector. Some characteristics are essential for a modern diesel Common Rail (CR) Fuel Injection System (FIS) in order to assure a satisfying spray quality. In particular, some of the fundamental requirements

Abbreviations: ALMR, Adaptive Local Mesh Refinement; Cd, discharge coefficient; CI, Compression Ignition; CR, Common Rail; DDM, Discrete Droplet Model; ET, Energizing Time; FIS, Fuel Injection System; GDI, Gasoline Direct Injection; GDICI, Gasoline Direct Injection Compression Ignition; HCN, Hollow Cone Nozzle; ICE, Internal Combustion Engine; KH, Kelvin Helmholtz; MHN, Multi-Hole Nozzle; PECU, Programmable Electronic Control Unit; PM, Particulate Matter; RT, Rayleigh Taylor; SMD, Sauter Mean Diameter; TP, tip penetration. ⇑ Corresponding author at: Istituto Motori, Consiglio Nazionale delle Ricerche, Via Marconi, 4, 80125 Napoli, Italy. E-mail address: [email protected] (M. Migliaccio). http://dx.doi.org/10.1016/j.fuel.2017.01.020 0016-2361/Ó 2017 Elsevier Ltd. All rights reserved.

are the possibility of using very high injection pressures (Pinj), multiple and very near injections and high rate of the opening and closure nozzle lift. The current CR FIS reaches a maximum value of 200–220 MPa of Pinj, while systems up to 300 MPa are coming on the market [1,2]. Spray concepts alternative to the conventional Multi-Hole Nozzles (MHN) could be considered as solutions to the extremely high Pinj increase to assure a high and faster fuel-air mixing in the piston bowl, with the final target of increasing the fuel efficiency and reducing pollutant emissions at engines’ exhaust. At the same time, swirl supported combustion systems usually employed in diesel engines configurations do not help reducing heat transfer while, in principle, quiescent combustion systems could reduce the wall heat transfer through the piston, bowl and liner walls [3,4]. However, the hypothesis of reducing heat transfer losses via the reduction of swirl motion could be theoretically considered only leaving the well-assessed MHN configuration that needs swirl motion for an appropriate spray atomization level. Only proper solutions for nozzle configurations, intrinsically capable of producing a finely atomized spray homogeneously

509

M. Migliaccio et al. / Fuel 196 (2017) 508–519 Table 1 Energizing timings for the investigated injection strategies. Qinj/Pinj

30 MPa

80 MPa

120 MPa

1 mm3/shot 10 mm3/shot 30 mm3/shot 60 mm3/shot

280 ls 500 ls 920 ls 1560 ls

200 ls 390 ls 660 ls 1050 ls

140 ls 340 ls 560 ls 900 ls

Tension [V] -Current [A]

140

Inj. Rate MHN

120

Inj. Rate HCN

100

Needle Lift HCN

45 40 35 inj. rate

30 25

80

needle lift

tension

60

20 15

40

10

current

20

5

0 0

Injection rate [mm³/ms] lift [um]

50

160

0 200 400 600 800 1000 1200 1400 1600 1800 2000

ET [μs] Fig. 1. Spray pattern comparison between the HCN (right) and MHN (left). Fig. 3. Main functional characteristics of the HCN injector in terms of voltage command, needle lift and injection rate.

Fig. 2. Schematic view of geometrical features of HCN (left) compared with a conventional MHN (right).

distributed in the whole engine combustion chamber, could be coupled with quiescent combustion systems. In this sense, a diesel HCN concept could be an alternative solution, as shown in Fig. 1. Another potential application of a high-pressure HCN could be relative to the Gasoline Direct Injection Compression Ignition (GDICI) engines. Research interest is growing up on this kind of engine due to its capability of producing very low PM and NOx emissions with a fuel efficiency comparable to diesel’s one [5]. Literature studies on the GDICI technology indicate that complex injection strategies coupled with an accurate control of the fuel metering are crucial features for the development of this type of engines [6,7].

Fig. 4. Liquid spray sequence at different Pinj for Qinj 30 mm3/shot.

510

M. Migliaccio et al. / Fuel 196 (2017) 508–519

By the way, results from GDICI engine development activities indicate that low-swirl or quiescent combustion chamber with reduced heat losses represent the optimal solution for the combustion system design [6]. In this sense, a HCN injector capable of precise fuel metering and high-pressure multi-pulse injections could be strategic for the GDICI engine development. In this context, a prototypal high-pressure HCN was tested from an experimental and numerical point of view to characterize its whole spray behaviour. In the following paragraphs the main results on the operating features of the HCN, tested preliminarily for diesel fuel applications

only, are described, together with the adopted experimental facilities and numerical tools.

2. The HCN prototype injector The proposed solution is an outwardly opening pintle-type nozzle able to generate a hollow cone spray similar to that produced in some GDI injectors [8]. A very simple scheme of the geometrical configuration of the HCN, compared with conventional MHN, is reported in Fig. 2. The reference MHN injector is a 7 hole,

Fig. 5. Frontal and lateral Liquid spray sequence for different Qinj at Pinj of 120 MPa.

511

M. Migliaccio et al. / Fuel 196 (2017) 508–519

Fig. 6. Experimental methodology for the radial (left) and axial (right) TP measurements.

141 lm in diameter and an inner spray-cone angle of 148°. The peculiar characteristics of the prototype HCN consist in:
direct actuation of the needle employing piezo-electric actuators;
high quality of the dynamic behaviour with a rapid and precise control of the injection event;

40

3

40

3

Qinj: 30 mm

radial penetration [mm]

radial penetration [mm]

The first research activities were devoted to preliminarily characterize the spray pattern generated by HCN. For the HCN

Pinj 80 MPa

Qinj: 60 mm

35


maximum Pinj of about 160 MPa;
ability to operate at high flow rate and very low needle lift (some dozens of microns).

3

Qinj: 10 mm

30

3

Qinj: 1 mm

25 20 15 10 5

3

Pinj 120 MPa

30 25 20 15 10

35

Qinj: 30 mm

30

Qinj: 10 mm

0 40

Pinj 120 MPa

Qinj: 60 mm

3

radial penetration [mm]

radial penetration [mm]

Qinj 30mm /str

Pinj 80 MPa

5

0 40

3

3

Qinj: 1 mm

25 20 15 10 5 0

3

Pinj 30 MPa

35

3

Pinj 30 MPa

35

Qinj 60mm /str

Pinj 80 MPa Pinj 120 MPa

30 25 20 15 10 5

0

200

400

600

800

1000

1200

time [ s] Fig. 7. Radial liquid TP profiles for different Qinj.

1400

0

0

200

400

600

800

1000

1200

time [ s] Fig. 8. Radial liquid TP profiles for different Pinj.

1400

512

M. Migliaccio et al. / Fuel 196 (2017) 508–519

Table 2 Setup of spray sub-models for spray simulation.

3. Experimental layout and test methodology

Models

Setup

Injector Atomization Breakup Evaporation Heat transfer

Hollow cone Off KHRT Frossling Ranz-Marshall

The experimental characterization of the injection process was carried out under non-evaporative conditions, injecting the fuel in a high-pressure constant-volume cylindrical vessel in order to measure the spatial and temporal spray pattern at engine-like gas densities. Therefore, SF6 gas (density 6.2 kg/m3) was used in the vessel at room temperature, thus reaching the desired gas densities at lower pressures than air. Pressure of 0.25 MPa was set in order to achieve a gas density of 14.8 kg/m3, correspondent to incylinder engine conditions of 4.2 MPa at 1000 K. A Bosch CP1H high-pressure injection pump was coupled with a Programmable Electronic Control Unit (PECU) for managing the whole injection apparatus. Fast electronic drivers have allowed setting precise and stable injections like the one involving small fuel quantities in multiple injections. The Bosch CP1H was driven by a variable speed electric motor while a heat exchange system, allocated on the hydraulic circuit, was adopted to keep the fuel temperature constant in the tank (21 ± 1 °C). The spray images were collected by a Photron (FASTCAM SA4) high-speed C-Mos camera synchronized with both the injection system and a high-intensity flash. The images were acquired both parallel and orthogonal configuration with respect to the spray propagation. In this way, frontal and lateral spray evolutions were acquired for each investigated condition in non-simultaneous mode. They refer to different injection cycles. For both acquisition

Table 3 Set of injection conditions for numerical spray characterization. Injection strategy

Injection pressure [MPa]

Fuel delivery [mm3/shot]

Cone angle [°]

1 2 3

60 80 120

10 30 60

116 100 100

prototype, the needle seat diameter (d) is 4 mm and the cone angle 100°. Its circumferential area depends linearly with the needle lift (see Fig. 2) with a maximum value of 30 lm in the operating conditions tested. Referred to the needle lift of dozens of microns, the discharge area is well larger than that the classical MHN one. As an example, for a needle disk diameter of 4.0 mm, when the needle lift reaches 30 lm the discharge area is of 0.377 mm2, while for the reference MHN injector the discharge area (Ad) is 0.109 mm2.

50 μs

50 μs

150 μs

150 μs

250 μs

250 μs

350 μs

350 μs

500 μs

500 μs

650 μs

Fig. 9. Frontal and lateral spray evolution for injection strategy Pinj = 80 MPa and Qinj = 30 mm3/shot. Comparison between experimental and numerical spray images.

513

M. Migliaccio et al. / Fuel 196 (2017) 508–519

The spray sequences in Fig. 4 show a partial sample of the spray structure, frontally-acquired, with the fuel freeze at different time from the start of injection. The test conditions were: Qinj 30 mm3/ shot and three Pinj: 120 (top), 80 (middle), and 30 MPa (bottom). The correspondent ETs are reported in Table 1. To better understand the injection system stability, two spray sequences are reported in Fig. 5 at Pinj (120 MPa) for different Qinj, 1.0 (top) and 60 mm3/shot (bottom). Both frontal and lateral views were acquired. It is worthwhile to highlight that the frontal images (sequences on the top) show a uniform fuel distribution for both the low and high injected amounts since the first instants from the start of injection. The circular shape of the fluid grows in regular mode during the entire injection duration. Furthermore, the spray looks denser and more penetrating when moving from the 1.0 to 60 mm3/shot condition. Images collected orthogonally to the injector axis show a significant presence of fuel cloud both in forward and backward directions, meaning the last toward the rear of the constant volume vessel. The superimposed arrows in Fig. 5 highlight this recirculation zone. The reverse flow results in the experimental, typical of HCN sprays, derives from several contemporary effects. First, the intense air flow from the outside region enters into the conical spray and produces a recirculation zone with toroidal vortices [12]. In addition, the hydraulic dynamic behaviour of the needle lift itself, coupled with the air entrainment into the spray, forms air

50

Pinj=80MPa Qinj=30mm3/shot

45

axial penetration [mm]

4. Experimental results

bubbles and cavitation pockets in the gap between the pintle and the seat [13]. Their position depends on the fluid radial or tangent preferred direction. The phenomenon of cavitation could be responsible for vapor sacs rather than a continuous vapor film around the circumference of the needle sealing area, as evidenced by experimental visualization of HCN sprays [8,12]. The analysis of spray images was carried out by a professional software (Image-Pro Plus) to extract the measures of the spray tip penetrations (TP) at the different injection strategies. A spatial calibration of 3.7 pixels/mm was realized by using the 25 mm focal length lens on the camera. The top sequences of Fig. 5 show that the annular liquid jet develops in a rather regular mode with equal penetrations along the circular crown at all the time-steps. The sprays developing at the highest Pinj appear more compact compared to the lowest values cases. Due to its greater compactness, the spray is much brighter giving off more scattered light. The sequence at Pinj 120 MPa (Fig. 5) demonstrates the capability of the HCN injector to generate a high-pressure hollow cone. The TPs of the sprays were acquired measuring the radial distance between the outer fuel and the centre of the injector nozzle. A set of five images was captured per each injection condition to take into account the cycle-to-cycle dispersion. Fig. 6 shows a sketch of the radial and axial TP acquisition. tFig. 7 reports the radial TP profiles versus the time from the start of injection for two injection pressures, 80 (top), and 120 MPa (bottom) and for different amount of injected fuel. Each data is averaged on five measurements at the same time and the bars indicate the standard deviation. As known, the TP is a func-

40

Experiments

35 30

Numerical simulation

25 20 15 10 5 0 0

100

200

300

400

500

600

700

800

700

800

time [ μs] 50

radial penetration [mm]

sides, a 25 mm focal lens for the camera was used with a field of view covering the entire evolution of the spray. The frontal spray images were acquired on a 320  320 pixels window size with a time resolution of 33 ls (30,000 fps). The spatial resolution resulted 3.7 pixel/mm. Vice versa for the lateral configuration, a size of 320  672 pixels was set for the acquisition window with a time step of 66 ls (15,000 fps) realizing a spatial resolution of 12.3 pixel/mm. The shutter times of the camera were 3.7 and 12.3 ls for frontal and lateral configuration, respectively. The analysis of the liquid fuel spray images was carried out by the following image processing procedure: image acquisition, background subtraction and filtering, edges determination, tip penetration and cone angle measures. A sketch of experimental apparatus for tests under non-evaporative conditions and further details were described in [9]. Table 1 reports the complete set of the tested injection strategies in order to simulate different real engine conditions. They have been defined in terms of Energizing Timing (ET) for a single pulse, varying Pinj and Qinj between 30 and 120 MPa and 1.0 and 60 mm3/shot, respectively. The injector supplier provided a complete flow rate map for the HCN prototype injector as function of Pinj and ET. It was acquired by the Bosch EMI2 device, and the injection rate profile measured by means of an injection-rate indicator (Bosch EVI) for the test conditions of Table 1 [10,11]. As an example, Fig. 3 shows the comparison of main functional characteristics measured for the MHN and HCN injectors in terms of current and voltage command, respectively, and in terms of injection rate profiles. The figure corresponds to an injected quantity of 10 mm3/shot at 600 MPa of injection pressure and shows that the HCN injector can reproduce the same injection rate profile of MHN adequately modulating its voltage profile. This is possible due to the piezo-direct actuation of the HCN needle. In Fig. 3, the needle lift was reported as the result of simulation of the HCN injector model realized with AMESim code, provided by the supplier.

Pinj =80MPa Qinj=30mm3/shot

45 40

Experiments

35

Numerical simulation

30 25 20 15 10 5 0 0

100

200

300

400

500

600

time [ μs] Fig. 10. Comparison between experiments and numerical simulation for the HCN prototype in terms of axial (top) and radial (bottom) spray penetration for the injection strategy Pinj = 80 MPa and Qinj = 30 mm3/shot.

514

M. Migliaccio et al. / Fuel 196 (2017) 508–519

tion of Pinj, ambient gas density and discharge area for a fixed fluid. The trends of the TPs are in agreement with literature data for HCN sprays operating at injection pressure values typical of GDI applications [13]. The analysis of the TP curve confirms that different phases in the spray propagation can be identified for the highest Pinj values here employed. The initial linear penetration, corresponding to a constant velocity, is followed by a square root shaped phase, indicating a deceleration of the droplets then involved in circular vortices [13]. Looking at Fig. 7, similarly to conventional MHN, no differences concerning the ETs were registered meaning that the longest penetrations are an extension of the shortest ones, overlapping in the common time. Further analysis were carried out for evaluating the Pinj effects on the TPs (Fig. 8) for Qinj of 30 (top) and 60 mm3/shot (bottom). The trend of the curves depicts a strong effect of the injection pressure on the TP for both the conditions showing a linear and wellscaled behaviour with respect to the different Pinj.

50 μs

50 μs

150μs

150 μs

5. The CFD code The OpenFOAMÒ numerical code [14] was employed to simulate diesel spray experiments in the constant volume vessel. The numerical characterization of spray pattern in non-evaporating conditions was performed with the Lib-ICE set of libraries [15,16]. Injection rate from EVI measurements and needle lift from AMESim simulation were assumed as input data together with the ambient conditions. The needle lift curve of HCN injector was used to calculate the instantaneous geometrical discharge area (Aid) during the injection event as in the following:

Aid ¼ p  d  h where Aid is the instantaneous discharge area from HCN injector, d is the needle seat diameter and h is the instantaneous needle lift that reaches its maximum value equal to 30 lm. The correspondent variation of injection velocity (u) in the simulation is calculated in the following:

250 μs

250 μs

350 μs

350 μs

500 μs

500 μs

650 μs

Fig. 11. Spray evolution for the injection strategy Pinj = 120 MPa and Qinj = 60 mm3/shot. Comparison between experimental and numerical spray images.

515

M. Migliaccio et al. / Fuel 196 (2017) 508–519

_ m

50 45 40

Experiments

35

Numerical simulation

30 25 20 15 10 5 0 0

100

200

300

400

500

600

700

800

700

800

time [μs] 50

radial penetration [mm]

where Cd is the discharge coefficient, q is the fuel density and Aid is the instantaneous discharge area from HCN injector. As previously evidenced, the influence of inner-nozzle flow on the spray behaviour is very strong in reducing the effective discharge area at high Pinj values for HCN injectors, due to the contemporary effects of cavitation pockets and/or air bubbles produced by the air entrainment from the outside regions into the nozzle [13]. In the present simulation, such phenomena are neglected because information from inner-nozzle flow analysis was not yet available. The evolution of the spray was simulated through the classical models for high pressure spray simulation. The Discrete Droplet Model (DDM) based on a Lagrangian approach is used to describe the liquid phase while an Eulerian description is employed for the gas phase [17]. The spray droplets are described by stochastic particles that are usually referred to as parcels [18]. The initial condition of SMD is the most uncertain parameter for HCN spray simulation, related to the unknown shape and morphology of fuel structures at the nozzle exit [19] from high-pressure HCN spray (above 30.0 MPa). No experiments were available for HCN nozzle regarding the droplet diameter size neither at injection nor during the breakup process, therefore the present simulation assumed a constant initial droplet diameter of 30 lm corresponding to the highest value of needle lift in the operating conditions. Different tests, where the diameter is fixed at either a constant value or according to a Rosin-Rammler distribution function [14], show that no significant effects on spray are evident due to the very small initial diameter values [19]. Once injected, droplets were subjected to primary atomization, secondary breakup and droplet collisions. The setup of spray sub-models is reported in Table 2. Some information are available in literature about physical mechanisms involved in the breakup process for an outwardly HCN injector operating at low Pinj typical of GDI applications [8,12,13,19] while a few studies are relative to higher Pinj typical of diesel fuel (like CR FIS) [20]. In the hypothesis of physical breakup mechanisms similar to those typical of MHN, the Kelvin-Helmholtz (KH) and Rayleigh-Taylor (RT) mechanisms by Reitz [21] were employed. In the following simulations, the standard drag model [12] and a so called ‘random walk procedure’ for turbulent dispersion [22] were adopted. All simulations were performed adopting the standard k-e model for turbulence description [14]. Drop to drop collisions, generally neglected due to their high dependency from the quality of the mesh, were here considered thanks to the adoption of the Adaptive Local Mesh Refinement (ALMR) technique [23]. Starting from an initial mesh quite coarse, the ALMR was employed for a correct prediction of the spray behaviour, keeping an acceptable computational time. The mesh refinement is performed only in the cells where the total fuel mass fraction is within a certain range. In this way, the simulation started with a cell size of 2.2 mm and 19,683 cells, which was progressively refined during the simulation up to a cell size of 0.25 mm in the spray region and about 252,000 cells. The predictive capabilities of the code were assessed in previous work [23] when the ALMR technique is used. The set of injection conditions chosen for the numerical simulations include two experimental injection conditions of previous Table 1: Pinj = 80 MPa and Qinj = 30 mm3/shot and Pinj = 120 MPa and Qinj = 60 mm3/shot. In these conditions, the predictive capabilities of the code in capturing spray characteristics for HCN injector were evaluated comparing numerical spray penetration and diffusion with experimental results.

axial penetration [mm]

qAid C d

Pinj=120MPa Qinj=60mm3/shot

45 40

Experiments

35

Numerical simulation

30 25 20 15 10 5 0 0

100

200

300

400

500

600

time [μs] Fig. 12. Comparison between experiments and numerical simulation for the HCN prototype in terms of axial (top) and radial (bottom) spray tip evolution for the injection strategy Pinj = 120 MPa and Qinj = 60 mm3/shot.

35 30

Pinj=80MPa 25

SMD [μm]

u¼

Pinj=120MPa 20 15 10 5 0 0

100

200

300

400

500

600

700

800

time [μs] Fig. 13. Comparison between SMD variation from numerical simulation for the HCN prototype for the injection strategies Pinj = 80 MPa, Qinj = 30 mm3/shot, and Pinj = 120 MPa, Qinj = 60 mm3/shot.

Once validated the spray model setup with the experimental data, a numerical comparison between the spray performance HCN and MHN injector was carried out, considering the same injection rate shape in order to specifically evaluate the effect of different injector configurations. In this case, a different injection condition was considered (Pinj = 60 MPa and Qinj = 10 mm3/shot), due to the availability of the common injection profile.

516

M. Migliaccio et al. / Fuel 196 (2017) 508–519

Fig. 14. Frontal view of spray evolution for the injection strategy Pinj = 60 MPa and Qinj = 10 mm3/shot; coloured drops for SMD and injection velocity respectively. Comparison between MHN (first row) and HCN (second row). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

In summary, Table 3 describes the set of injection conditions for numerical spray characterization. 6. Numerical results The comparison between experiments and simulations in terms of spray penetration is reported in Fig. 9 for the injection pressure of 80 MPa and Qinj = 30 mm3/shot. The figure reports the frontal and lateral images of the spray obtained from simulation, in the first and fourth row, while corresponding images collected by the C-Mos camera are shown in the second and third row. The spray evolution is analyzed from 50 to 500 ls after the start of injection. The comparison of frontal images, in the first and second rows, evidences a quite good agreement in terms of radial shape and radial TP of the spray. In the images obtained from simulations, parcels are coloured according to the droplets diameter, ranging from about 3 to 30 lm. Droplets having the minimum diameter (about 10 lm) are located in the peripheral area of the annulus. Lateral views of the spray are also reported in Fig. 9: in this view calculated drops are coloured1 according to the droplets velocities. It can be noted that calculated TP along the axis of the jet seems to be in agreement with experiments up to 300 ls while, at subsequent times, TP is slightly overestimated. This can be ascribed to the reverse flow of the spray toward the direction of the rear wall 1 For interpretation of color in Fig. 9, the reader is referred to the web version of this article.

of the constant volume vessel as previously described and not predicted by simulation, even though a weak toroidal vortex is captured (as indicated in the superimposed arrow in the last frame of Fig. 9) [24]. Probably, the role of cavitation effects previously mentioned has a greater importance with respect to the final spray shape directly affecting the formation of vapor sacs, air bubbles and pockets in the gap between the pintle and the seat of HCN injectors [8,12]. More efforts should be necessary to take into account cavitation models in the simulation to improve numerical prevision of the toroidal vortices [25]. Concerning the spray TP experiments allow capturing the radial and axial TP as previously outlined. Fig. 10 shows the comparison between measured and calculated data in terms of radial (top) and axial (bottom) TP for injection strategy at Pinj = 80 MPa and Qinj = 30 mm3/shot. The comparison highlights a quite good agreement for radial TP while the axial one is slightly overestimated after 300 ls, confirming previous results of Fig. 9. These results show that the adopted setup of spray sub-models, well assessed in capturing the whole spray behaviour for high pressure MHN injectors [26,27], also provides a quite good representation of the spray behaviour for the new HCN injector. Fig. 11 shows the comparison between experimental and numerical results for the frontal and lateral views, respectively, for the injection strategy of Pinj = 120 MPa and Qinj = 60 mm3/shot. The high value of Pinj 120 MPa produces very low values of SMD at spray completely developed and injection velocities up to 200 m/s are reached. The comparison of the images confirms for this injec-

517

M. Migliaccio et al. / Fuel 196 (2017) 508–519

tion strategy a quite good agreement between numerical and experimental results in terms of radial TP, while the axial one appears overestimated in the simulation with respect to the experiments. In Fig. 12 the comparison between experimental and numerical simulations of the TP is shown both for radial (top) and axial (bottom) fuel propagation. This figure evidences results similar to those obtained in Fig. 10 with a quite good agreement in terms of radial TP and a slight overestimation of the axial one over the entire injection event, confirming the analysis based on image processing of the spray. In Fig. 13 the SMD evolution is reported for both the injection strategies of 80 and 120 MPa injection pressures. Here, the high breakup rate exhibited by the HCN prototype is evident in the first phase of the injection. Due to the breakup process that starts at the beginning of the injection, the drop diameter for HCN injector is halved in the first 100 ls and reaches very low values around 8 and 6 lm for 80 and 120 MPa injection pressure, respectively. As previously discussed, the initial condition of SMD is the most uncertain parameter for HCN high-pressure spray simulation (above 30.0 MPa). Its value was considered equal to the reference nozzle hole size and equal to the maximum needle lift of 30 lm. Despite no SMD data are available from literature for HCN validation neither at the injection start nor during the breakup process, the final estimated droplet diameters seem coherent with droplet size measurements by PDPA method for low-pressure (10.0– 20.0 MPa) gasoline systems and diesel hollow-cone sprays with SMD values in the range of 10–30 lm [8,13,19,28,29].

160 140

MHN

SMD [μm]

120

HCN

100 80 60 40 20 0 0

200

400

600

800

Simulations also indicate that the phenomenon of coalescence is more evident with lower injection pressure, causing a lowering of the breakup rate between 100 and 300 ls. Looking at the overall evolution of the SMD traces, it could be outlined that starting from low initial SMD values, the tested HCN allows to reach very small droplets in short time (within 300 ls). In principle, such peculiarity could improve the rate of fuel-air mixing in the neighbour of the spray location area. Such considerations are supported by the comparison between a MHN spray behaviour and the HCN one at the same injection rate profile (see Fig. 3). As already reported before, the MHN is a 7 holes-141 lm with an inner spray angle of 148°. This comparison was here done only from a numerical point of view and on the basis of the well assessed predictivity of the code for MHN injector [23,26,27]. Fig. 14 shows the spatial and temporal spray behaviour for both the MHN (first row) and HCN (second row) with an injection setting of Pinj = 60 MPa and Qinj = 10 mm3/shot, in terms of drop diameter (top) and velocity distributions (bottom). In order to better observe the differences between the two nozzles, both the frontal and lateral views are displayed in the figure for sprays delivering the same quantities. The most evident difference consists in the lower radial TP for the HCN spray. In fact, the HCN spray is significantly bounded in the central part of the constant volume vessel. The reduced TP of HCN can be ascribed to the reduced spray momentum deriving from the simultaneous reduction of the droplet diameter and the injection velocity. The figure also shows the reduction of injection velocity due to the increased discharge area in HCN, three times more than in the case of MHN. Concerning the reduction of the drop diameter, Fig. 15 shows the evolution of SMD calculated for the MHN and HCN injectors. It must be noticed the different hydraulic conditions of the injectors: the injection simulation of MHN is based on the initial value of SMD equivalent to the nozzle hole diameter of 141 lm [23,26,27] whereas, as discussed in the previous paragraph, the initial condition of SMD for HCN injector was fixed equal to the reference nozzle hole diameter of 30 lm. Fig. 15 evidences that the code predicts a quite similar breakup rate for both nozzles within the first 80 ls. Final values predicted in non-evaporating conditions for HCN injector are about 10 lm, lower than the final values for MHN injector around 18 lm and in line with experimental data reported in literature for classical MHN and HCN injectors [8,23,24]. Fig. 16 reports the spray TP evolution for both injectors, in terms of radial and axial components. As evidenced, the spray

time [μs] 40 MHN radial

35

5.E-05

MHN

tip penetration [mm]

gas momentum [kg m/s]

6.E-05

HCN

4.E-05 3.E-05 2.E-05 1.E-05

HCN radial

30

MHN axial

25

HCN axial

20 15 10 5

0.E+00 0

200

400

600

800

time [μs]

0 0

100

200

300

400

500

600

700

800

time [μs] Fig. 15. Comparison between MHN and HCN injectors in terms of SMD evolution (top) and gas momentum induced by the spray (bottom) for the injection pressure of 60 MPa.

Fig. 16. Comparison between MHN and HCN injectors in terms of radial and axial spray TP (bottom) for the injection strategy Pinj = 60 MPa and Qinj = 10 mm3/shot.

518

M. Migliaccio et al. / Fuel 196 (2017) 508–519

behaviour in terms of axial TP is very similar for both injectors, while the main difference concerns the radial TP of HCN that is lower of about 50% with respect to MHN, confirming the images of previous Fig. 14. As evidenced both by the experimental and numerical analysis, the tested prototype HCN provides finely-atomized high-pressure spray homogenously distributed over 360°. In line of principle, such feature should permit the use of the prototypal HCN injector in a low-swirl or quiescent combustion chamber in diesel engines with a positive effect on heat exchange losses and thermodynamic efficiency. The moderate TP of the HCN injector with respect to the MHN, the circumferential quasi-uniform and fine atomized spray close to the nozzle, as the capability of injection modulation due to the direct piezo actuation, represent important specific features that could be exploited for future low-swirl/quiescent diesel combustion systems, providing high thermal efficiency and low pollutant emissions. Dedicated experimental campaigns could be devoted to the possibility of using adequate geometrical designs of the combustion chambers or higher Pinj, for a good control of the fuel-air mixing process and the air exploitation in the combustion chamber of CI engines. 7. Conclusions The present paper describes some experimental and numerical activities on the behaviour of an innovative high-pressure HCN injector for automotive application. The prototypal HCN with a dedicated direct piezoelectric actuation system reproduces the characteristics of an hollow cone spray concept from an outwardly opening pintle-type nozzle. The analysis of the spray behaviour in terms of spray penetration and diffusion was accomplished in a constant volume vessel and validated by numerical simulations. The direct actuation piezo system permits a fine control of the needle lift. The small needle lift of the HCN, of dozens of microns, allows to realize very little initial droplet dimensions, providing finely atomized spray along the whole circumference. Numerical simulation describes a rapid breakup process immediately at the fuel discharge. Future SMD measurements, not yet available, should aim at confirming the numerical results. Experimental results show that a critical aspect of the prototypal version of the injector lies in the spray tip penetration. The global spray TP appears reduced with respect to a classical MHN one operating at the same injection conditions, due to an evident reduction of radial TP measurements. The reduced spray TP of HCN derives from the increased discharge area responsible for reduced injection velocity. This effect, combined with very little initial droplet dimensions, produces a reduction of spray momentum. Further studies are needed to analyze these effects, in order to assure an appropriate control of spray penetration and atomization levels. Further research steps will consist in optimizing the spray pattern exploiting the peculiar characteristics of the outwardly HCN prototype driven by a direct piezo actuation. Future steps will be the analysis on the injector behaviour and performance in research engines to evaluate potentialities and critical aspects for real application in diesel CI or GDICI combustion systems. Declaration of conflicting interests The Authors declare that no conflict of interest are in this paper.

Acknowledgments The activities reported in the present work were performed in the framework of the MIUR project PON01_02238 and the authors are grateful to Dr. Sergio Stucchi and his team for the support relatively to the prototypal injector.

References [1] Johnson J, Naber J, Lee S, Hunter G, et al. Correlations of non-vaporizing spray penetration for 3000 bar diesel spray injection. SAE Technical Paper 2013-240033; 2013. http://dx.doi.org/10.4271/2013-24-0033. [2] Palanisamy M, Lorch J, Truemner R, Baldwin B. Combustion characteristics of a 3000 bar diesel fuel system on a single cylinder research engine. SAE Int J Commer Veh 2015;8(2):479–90. http://dx.doi.org/10.4271/2015-01-2798. [3] Whitehouse ND. Effect of compression and expansion on instantaneous heat transfer in reciprocating internal combustion engines. Proc Inst Mech Eng, Part A: J Power Energy 1987;201(August):175–86. [4] Fridriksson H, Tuner M, Andersson O, Sunden B, et al. Effect of piston bowl shape and swirl ratio on engine heat transfer in a light-duty diesel engine. SAE Technical Paper 2014-01-1141; 2014. http://dx.doi.org/10.4271/2014-011141f. [5] Ciatti S, Subramanian SN. An experimental investigation of low-octane gasoline in diesel engines. J Eng Gas Turbines Power 2011;133:092802. [6] Sellnau M, Foster M, Hoyer K, Moore W, et al. Development of a gasoline direct injection compression ignition (GDCI) engine. SAE Int J Engines 2014;7(2). http://dx.doi.org/10.4271/2014-01-130. [7] Dempsey AB, Curran SJ, Wagner RM. A perspective on the range of gasoline compression ignition combustion strategies for high engine efficiency and low NOx and soot emissions: effects of in-cylinder fuel stratification. Int J Engine Res 2016. http://dx.doi.org/10.1177/1468087415621805. [8] Nouri JM, Hamid MA, Yan Y, Arcoumanis A. Spray characterization of a piezo pintle-type injector for gasoline direct injection engines. 2007 Publishing Ltd, Journal of Physics: Conference Series, Volume 85, Volume 85. [9] Montanaro A, Allocca L. Impact of the nozzle coking on spray formation for diesel injectors. SAE Technical Paper 2013-01-2546; 2013. http://dx.doi.org/ 10.4271/2013-01-2546. [10] Bosch R. GmbH, Diesel-Engine Management, Bosch Technical Books; 1999. ISBN-13: 978-0470026892. [11] Bosch W. The fuel rate indicator: a new measuring instrument for display of the characteristics of individual injection. SAE Paper No. 660749; 1966. [12] Dong Q, Ishima T, Kawashima H, Long W. A study of the spray characteristics of a piezo pintle-type injector for DI gasoline engines. J Mech Sci Technol 2013;27(7):1981–93. [13] Schmid A. Dissertation Thesis ETH No. 20852. Experimental characterization of the two-phase flow of a modern, piezo-activated hollow cone injector. . [14] OpenFOAMÒ web site. The Open-source CFD toolbox. OpenCFD Limited 2007. . [15] Lucchini T, D’Errico G, Ettorre D, Ferrari G. Numerical investigation on nonreacting and reacting diesel spray in a constant-volume vessel. SAE Int J Fuels Lubr 2009;2(1):966–75. http://dx.doi.org/10.4271/2009-01-1971. [16] Lucchini T, Fiocco M, Onorati A, Montanaro A, et al. Full-cycle CFD modeling of air/fuel mixing process in an optically accessible GDI engine. SAE Int J Engine 2013;6(3). http://dx.doi.org/10.4271/2013-24-002. SAE Technical Paper 201324-0024 (ISSN 1946-3944). [17] Dukovitz JK. A particle-fluid numerical model for liquid sprays. J Comput Phys 1980;35:229–35. [18] Bracco FV. Modeling of engine sprays. SAE Paper 850394; 1985. http://dx.doi. org/10.4271/850394. [19] Sim J, Badra JA, Elwardany A. Spray modeling for outwardly-opening hollowcone injector. SAE Paper 2016-01-0844; 2016. http://dx.doi.org/10.4271/201601-0844. [20] Obokata T, Long W, Ishima T. PDA and LDA measurements of large angle hollow cone spray proposed for hot-premixed combustion type diesel engines. SAE Paper 960772; 1992. http://dx.doi.org/10.4271/960772. [21] Reitz RD. Modeling atomization processes in high pressure vaporizing sprays. Atom. Spray Technol. 1987;3:309–37. [22] O’Rourke PJ. Statistical properties and numerical implementation of a model for droplet dispersion in a turbulent gas. J Comput Phys 1989;83:345–60. [23] Migliaccio Mna, Ettorre D, Aprovitola A, Alfuso A. Effects of turbulence modulation addition in OpenFOAMÒ toolkit on high pressure fuel sprays. SAE Paper 2011-01-0820; 2011. http://dx.doi.org/10.4271/2011-01-0820. [24] Sudhakar D, VanBrocklin PG. Effect of design and operating parameters on the spray characteristics of an outward opening injector. In: Thirteenth international multidimensional engine modeling user’s group meeting, March 2, 2003 Detroit, MI. [25] Jia M, Xie M, Liu H, Lam WH, Wang T. Numerical simulation of cavitation in the conical-spray nozzle for diesel premixed charge compression ignition engines. Fuel 2011;90:2652–61.

M. Migliaccio et al. / Fuel 196 (2017) 508–519 [26] Briani M, Fraioli V, Migliaccio Mna, Di Blasio G, et al. Multi-dimensional modeling of combustion in compression ignition engines operating with variable charge premixing levels. SAE Paper 2011-24-0027; 2011. http:// dx.doi.org/10.4271/2011-24-0027. [27] Beatrice C, Guido C, Fraioli V, Migliaccio Mna. Experimental and numerical analysis of nozzle flow number impact on full load performance of an Euro5 automotive diesel engine. SAE Paper 2012-01-0679; 2012. http://dx.doi.org/ 10.4271/2012-01-0679.

519

[28] Huang C, Yasari E, Lipatnikov A. Simulation of mixture formation and combustion for gasoline direct injection engine using OpenFOAM. CFD simulation of I.C. engines by OpenFOAM, Milano, Italy (2011/11/07). [29] Kawaharada N, Sakaguchi D, Ueki H, Ishida M. Effect of injection pressure on droplet behaviour inside diesel fuel sprays. SAE Technical Paper 2015-011841; 2015. http://dx.doi.org/10.4271/2015-01-1841.