Analysis of In-Cylinder Flow for a Boosted GDI Engine ...

17th International Symposium on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 07-10 July, 2014

- 1 -

Analysis of In-Cylinder Flow for a Boosted GDI Engine using High Speed

Particle Image Velocimetry

Yasuhiro Okura1,*, Makoto Segawa1, Hiromi Onimaru1, Yasuhiro Urata1, Mamoru Tanahashi2

1: Honda R&D Co., Ltd., Tochigi, Japan

2: Department of Mechanical and Aerospace Engineering, Tokyo Institute of Technology, Tokyo, Japan * correspondent author: [email protected]

Abstract In this study, in-cylinder flow for a boosted GDI engine is investigated on an optical engine with a transparent sleeve, a high tumble port and a combustion chamber using high speed PIV. For measurements of overall flow patterns, measurement region is set to transparent cylinder entire and PIV is conducted at 3 kHz. As results of analysis of overall flow characteristics, the generated tumble flow in the intake stroke is kept until the end of the compression stroke and the tumble vortex center moves in a clockwise direction. For an investigation of turbulence, effects of time resolution of PIV are investigated and adequate time resolution is shown. In these cases, measurements are conducted in a narrow region at upper cylinder and time resolution of PIV is varied up to 10 kHz. It is clarified that appropriate time resolution for turbulence evaluation from the intake through the compression stroke should be higher than 5 kHz for this engine. A specific frequency where slope of energy spectrum of velocity changes is used as a cut-off frequency which decomposes mean and turbulent components. This specific frequency approximately corresponds to the inverse of the integral time scale of flow from the middle of the intake stroke to the end of the compression stroke. This cut-off frequency also represents a relatively large time scale of flow field generated by the piston behavior in the intake stroke. To investigate effects of piston top shape on the flow characteristics, a cavity piston and a flat piston are tested. It is found that an optimization of piston top shape is effective for enhancement of in-cylinder flow. 1. Introduction In recent years, downsized boosted engines have been developed for the purpose of improving both fuel consumption and fun driving. This engine concept is intended to expand low engine speed and high load region against natural aspirated engines by using turbo charger etc., which contributes an improvement of the fuel economy (Szengel et al 2007; Heiduc et al 2011). The improvement of thermal efficiency is limited by occurring of knocking at low engine speed and high load. This phenomenon has been a big issue in the development of a boosted GDI (Gasoline Direct Injection) engine. To solve this issue, optimizations of an intake port or a combustion chamber shape (Kojima et al 1990; Sasaki et al 2010) and introduction of a switching device for flow enhancement in the cylinder such as a tumble control valve (Nogawa et al 2010; Mito et al 2011; Adomeit et al 2010) have been investigated and developed for the improvement of degree of constant volume and knocking reduction. In research and development of these engines, CFD (Computational Fluid Dynamics) has been used actively. However, it is hardly said that detailed design of an engine is performed only by CFD since detailed reference data of turbulence or flow characteristics in a combustion chamber of an engine was very few and verification of CFD was insufficient. As turbulence measurement method in an engine combustion chamber, a hot-wire anemometer or LDV (Laser Doppler Velocimetry) has been performed for a long time. Evaluation of whole combustion chamber field is not always easy because they are a point measurement technique. In recent years, PIV (Particle Image Velocimetry) has been applied for in-cylinder flow measurements. Since the results obtained by PIV give flow velocity distribution, larger amount of information is obtained by PIV compared to LDV, and more detailed evaluation of in-cylinder flow can be performed.

Decomposition methods of turbulence component from the measured velocity have been proposed by many researchers. Kido et al (1984) measured velocity in the compression stroke by


- 2 -

using a hot-wire anemometer and decomposed turbulence component by using an ensemble-average method, where ensemble-averaged velocity is defined as mean flow component. Here, it has been pointed out that turbulence component obtained by the cycle-average method includes cycle variation of the mean flow component. Wakisaka et al (1982) and Hamamoto et al (1987) measured velocity around the compression TDC by a hot-wire anemometer and tried to decompose mean flow component and turbulence component from instantaneous velocity by using a stationary time average method. In this method, flow is assumed to be steady in short period of each cycle around the compression TDC and mean flow component is defined independently in the compression stroke and in the expansion stroke, which is divided by the compression TDC, by setting a time interval, 𝛥𝑡 (or a crank angle interval, 𝛥𝜃). However a clear physical definition about the time interval has not been given. Kamimoto et al (1987) and Ohtani et al (1990) have proposed a cut-off frequency for the decomposition. The cut-off frequency determined to be a frequency which includes 90% energy by integrating of power spectrum obtained from cycle-averaged velocity measured by LDV. However it has been suggested that selection of the cut-off frequency may includes some arbitrariness. Furthermore, in PIV measurements, ensemble-averaged velocity has been treated as mean flow component and the fluctuation from this ensemble-averaged velocity defined as turbulence component (Jeong et al 2011; Muller et al 2010). Although a cut-off length was also introduced in other decomposition method using a spatial filter were reported, determination method of the cut-off length was not clear. Thus, although various methods have been proposed for the decomposition of mean flow and turbulence component, the applicability of those methods to different engine specifications or operating conditions is still unclear. In our previous study (Okura et al 2013), a time filtering method for turbulence decomposition in-cylinder flow has been proposed. In the time filtering method, based on the advantage of the high speed PIV, velocity field is split into large-scale main flow component and turbulence component from one cycle velocity field by setting an appropriate cut-off frequency, fc. This method may have potentials to exclude cycle variation of mean flow component from the turbulence component and to give turbulence characteristics which contribute to combustion promotion. However, further investigations on effects of time resolution of PIV and the cut-off frequency selection have been required.

In this study, in-cylinder flow in a boosted GDI engine is investigated by using high speed PIV on an optical engine with a transparent sleeve, a high tumble intake port and a combustion chamber shape. The overall flow characteristics flow such as temporal developments of tumble vortex center and its relation with turbulence characteristics at each crank angle are clarified. Furthermore, the high speed PIV is performed with different time resolution to show the adequate time resolution for the investigation of turbulence in in-cylinder flow in a boosted GDI engine. In addition, validity of the cut-off frequency is investigated by considering an integral time scale of flow and piston behavior. 2. Experimental technique 2.1 Engine specification Specification and appearance of the engine which is investigated in the present study are shown in Table 1 and Fig. 1(a), respectively. This optical engine (Koyama garage) is equipped with a transparent sleeve made by quartz. Bore and stroke are 73 mm and 89.4 mm, respectively. 2 intake and exhaust valves are equipped with the engine head. The quartz sleeve is set between the head and the block by a sleeve holder. This quartz sleeve allows one optical access of in-cylinder flow from the intake to the compression stroke. The intake port has a shape which is easy to generate a tumble flow. Here, the average tumble ratio measured in steady flow state is 1.93 and is larger than that for a natural aspiration engine port, 0.59 (Okura et al 2013). Measurement region is shown in Fig. 1(b). Available measurement area of the bore and the stroke direction is limited by a sleeve holder. For measurements of overall flow patterns, measurement region is set to be 65 mm in the bore direction and 75 mm in the stroke direction. For high time resolution measurements up to 10 kHz, high speed PIV is performed in a narrow region at upper cylinder and the size of measurement region is 36 mm in the bore direction and 6 mm in the stroke direction. 2.2 Experimental setup


- 3 -

The experimental setup is shown in Fig. 2. High speed PIV system is used for in-cylinder flow measurement. A high repetition double-pulsed Nd:YAG laser (Lee, LDP-50MQG) is used for light source and its wavelength is 532 nm. A high speed CMOS camera (Photron, APX-RS) is used for particle image capture. For large domain measurements with 65 mm × 74 mm, the high speed CMOS camera is operated with 6 kHz with 768 × 656 pixel resolution. For the small domain

Table 1 Engine specification

Engine An optical single cylinder engine with a quartz sleeve

Bore × Stroke 73 mm × 89.4 mm

Displacement 374 cm3

Valve train DOHC 4 valves

Fig. 1 (a) Appearance of engine and (b) measurement region.

Fig. 2 Experimental setup (a) side view and (b) upper view.

measurements with 36 mm × 6mm, the frame rate of the camera is varied up to 20 kHz with 512 × 256 pixel resolution. The laser beam is stretched into a sheet with 0.5 mm thickness by cylindrical lenses and the laser sheet is led into the cylinder through the quartz sleeve from the engine side. The measurement plane is set to include a cylinder bore center. As tracer particles, SiO2 (Suzuki Yushi Industrial, God Ball B-6C) of which diameter is 3 µμm is used. The tracer particles are supplied into the cylinder at the throttle chamber through a custom-made tracer particle supply device. It has been shown that this tracer particle is available for turbulence measurement since traceability is 94 % against 10 kHz velocity fluctuation (Hand book of particle image velocimetry 2002). Crank angle range measured in the present study is shown in Fig. 3. Here, intake valve lift, in-cylinder pressure, 𝑃!"# and intake manifold pressure, 𝑃!"#$%& are also shown additionally as a reference. The measurements are performed from crank angle, 𝜃=360 deg BTDC to 0 deg BTDC


- 4 -

which corresponds to from the intake TDC (Top Death Center) to the compression TDC. Multiple cycles are measured for each condition. Velocity vectors on the measurement plane are evaluated by commercial PIV analysis software (Seika, Koncert Ver1.0). In this commercial software, a recursive FFT cross-correlation method is used. In this study, the interrogation window is changed from 64 × 64 pixels to 16 × 16 pixels by a 3-step hierarchical method with 50 % overlaps. Since the final interrogation window is 16 × 16 pixels, the spatial resolution of the present study is 1.1 mm for the large domain measurement and 0.6 mm for the small one. Time resolution of the PIV is 3 kHz since the camera records particle images with 6 kHz for the large domain measurement. For the small domain measurement, that is up to 10 kHz. In this study, velocity component in the bore direction is denoted by u and that in the stroke direction is done by v. It should be noted that, in the present study, a physical quantity which is averaged conditionally with a certain crank angle over multiple cycles is called as a cycle-averaged variable.

Fig. 3 Crank angle range measured in the present study

2.3 Test conditions Test conditions are shown in Table 2. Engine speed, Ne is 700 rpm, 1500 rpm and 2000 rpm. Throttle opening is WOT (Wide Open Throttle) under natural aspiration condition in each engine speed. Number of measured cycles is 28 for 700 rpm, 50 for 1500 rpm and 2000 rpm since this number is limited by memory size mounted the camera. In each engine speed, interval time between successive laser pulses, 𝛥t is adjusted in a range from 10 to 25 µμs. Change of piston top shape and style of expected flow pattern by them are schematically shown in Fig. 4. 2 type of piston top shape are tested. One is cavity piston with cylindrical shape [Fig. 4(a)]. The purpose of this shape is efficient conversion of the intake flow to upward flow by the piston top and generation of a strong tumble flow. Another one is flat piston [Fig. 4(b)] which is compared with the cavity one to show cavity effects on the flow field. Hereafter, unless otherwise stated, test is performed using cavity piston.

Table 2 Test conditions Engine speed Throttle opening Cycle number 𝛥t

700 rpm WOT 28 25 µμs 1500 rpm WOT 50 15 µμs 2000 rpm WOT 50 10 µμs


- 5 -

Fig. 4 Piston top shape and expected flow pattern.

3. Experimental results 3.1 Flow distribution analysis Cycle-averaged velocity distribution obtained by the large domain measurement for Ne=1500 rpm WOT condition is shown in Fig. 5(a). The obtained results well represent overall feature of flow pattern in the cylinder. At 𝜃=299.5 deg BTDC which is in the early stage of the intake stroke, the maximum velocity of the intake flow is about 30 m/s. At 𝜃=269.5 deg BTDC which is in the middle of the intake stroke, a tumble flow pattern which represented by a vertical vortex around cylinder center is generated clearly. The high speed flow initiated at the intake stroke moves downward with descent of a piston. The flow impinges on the piston top and comes back to the upper cylinder at 𝜃=179.5 deg BTDC which is around BDC (Bottom Death Center). At 𝜃=59.5 deg BTDC which is in the end of the compression stroke, the center of the tumble vortex is observed around the upper left side of the cylinder, which suggests that the tumble flow is kept until the end of the compression stroke. A tumble vortex center is calculated from cycle-averaged velocity distribution and a trajectory of that is shown in Fig. 5(b). Notes that the coordinate in the bore direction is denoted by x and that in the stroke direction is done by y. The origin of this coordinate system is set at the bore center (x=0 mm) and the head gasket section (y=0 mm). The method for determination of a tumble vortex center is following. A tumble ratio is calculated at all positions and then positions where the tumble ratio is more than 75 % of the maximum one are detected. From the detected positions, the location with lowest velocity magnitude is defined as the tumble vortex center. Here, other method which is determined from distribution of velocity and vorticity in spatial low-passed flow field has been proposed (Muller et al 2010). Since the cut-off length used for low-pass filter was not discussed in the present study, this method was not applied. Although the method used in this study cannot be applied for investigation on cycle variation of the vortex center, it has an advantage that the tumble vortex center is determined relatively easily from cycle-averaged flow distribution. The obtained result shows that the vortex center is generated at the upper position of the exhaust side and it moves downward direction with drastic sliding to the intake side in the descent phase of the piston. After that, it moves to the upper direction closing to the exhaust side in the rising phase of the piston. In this process, the tumble vortex center moves in a clockwise direction. 3.2 Decomposition of turbulence component As is mentioned above, we have proposed a time filtering method for turbulence decomposition. In this paper, effects of time resolution of PIV and cut-off frequency are investigated. First, time resolution of PIV is investigated by the small domain measurement shown in Fig. 1(b). The time resolution of PIV is varied up to 10 kHz. Cycle-averaged velocity at each crank angle is shown in Fig. 6 at the position which is denoted by a red symbol in Fig. 1(b). Test condition is Ne=1500 rpm WOT. It should be noted that velocity data from 𝜃 =360 to 330 deg BTDC and from 30 to 0 deg BTDC cannot be obtained and is excluded from analysis because a piston moves into the measurement region. Figure 6 shows that both velocity components have the maximum absolute values in the intake stroke and the maximum velocity of u component is about 22 m/s. Although magnitudes of both velocity components decreases at 𝜃=180 deg BTDC, they increase in the end of


- 6 -

Fig. 5 (a) Cycle-averaged velocity distribution and (b) a trajectory of the tumble vortex center calculated from cycle averaged flow field for Ne=1500 rpm WOT condition.

Fig. 6 Cycle-averaged velocity at the analysis point for Ne=1500 rpm WOT condition.

the compression stroke again. The reason of this behavior is that the intake flow generated in the intake stroke comes back into the analysis area.

Effects of time resolution on energy spectrum for u and v components are shown in Fig. 7. Notes that instantaneous velocity data from 𝜃=330 to 30 deg BTDC in each cycle is analyzed by Fourier transform and the obtained spectrum is cycle-averaged. Time resolution of PIV is changed by thinning 10 kHz data. Fig. 7(a) shows that energy of u component in high frequency regime (turbulence component) decreases with the decrease of the time resolution, which is caused by truncation errors. On the other hand, the energies in low frequency regime (mean flow component) less than 200 Hz scarcely affect by the time resolution. Here, it should be noted that energies of u component more than 200 Hz (turbulence component) increase with the decrease of the time resolution due to alias errors. A similar tendency can be seen in v component [Fig. 7(b)].

Figure 8 shows that effects of time resolution on cycle-averaged kinetic energy of mean flow component, 𝐾! and turbulent kinetic energy, 𝑘! obtained by the time filtering method. Here, 𝑓! is selected to 360Hz. 𝐾! and 𝑘! are calculated by following equations from one cycle instantaneous velocity data and then they are cycle-averaged.


- 7 -

𝐾! =

!!𝑢! + 𝑣! (1)

𝑘! =

!!𝑢!′! + 𝑣!′! (2)

Where 𝑢 and 𝑣 denote mean flow component decomposed by the time filtering method. 𝑢′! and 𝑣′! are turbulence components. It can be seen that Kf and kf are scarcely affected if the time resolution is decreased to 5 kHz from 10 kHz. When the time resolution is decreased to 2.5 kHz, significant

Fig. 7 Effects of time resolution on cycle-averaged energy spectrum of (a) u component and (b) v component at analysis point for Ne=1500rpm WOT condition.

Fig. 8 Effects of time resolution on cycle-averaged kinetic energy of mean flow component, Kf (a) and cycle- averaged turbulent kinetic energy, kf (b) at the analysis point for Ne=1500rpm WOT condition. change both in Kf and kf cannot be observed and they are slightly decreased in the intake stroke. Furthermore, when the time resolution is decreased until 1.0 kHz, Kf in the intake stroke increases and kf decreases in all phases. The above results suggest that the turbulent kinetic energy tends to decrease with the decrease of the time resolution. Especially, in the 1.0 kHz case, the turbulence kinetic energy is significantly underestimated in the compression stroke. These facts show that an appropriate time resolution for evaluation from the intake stroke through the compression stroke should be higher than 5 kHz for in-cylinder flow field of this engine.

In this study, the specific frequency where slope of spectrum changes is used as a cut-off frequency. This specific frequency approximately corresponds to the inverse of the integral time scale of flow from the middle of the intake stroke to the end of the compression stroke (Okura et al 2013). Validity of the cut-off frequency is investigated by using an integral time scale of flow and piston behavior. A spatial-averaged integral time scale of flow in the analysis area and a characteristic time scale of piston motion, 𝜏! are presented in Fig. 9. The integral time scale of flow is determined by a following method. First, auto-correlation coefficient is calculated from instantaneous velocity data. The integral time scale of flow is determined by integrating the auto-correlation coefficient until 0. Here, the integral time scale of flow of u component is denoted by 𝜏! and that of v component is done by 𝜏!. In addition, 𝜏! is obtained by dividing the combustion


- 8 -

chamber height, hp by the instantaneous piston velocity, up. Here, hp is distance from top of the combustion chamber to the piston top at a certain crank angle and is 15 mm at TDC. 𝜏! tends to decrease toward to the compression stroke from the intake stroke, whereas the decreasing rate becomes once gradual around 𝜃=180 deg BTDC. 𝜏! first decreases until 𝜃=240 deg BTDC. After that it increases again toward 𝜃=150 deg BTDC and then decreases toward the compression TDC. As a reason for changes of 𝜏! and 𝜏! around 𝜃=180 deg BTDC, one can considered that the intake flow generated in the intake stroke comes back to the analysis area. Since the tendency of 𝜏! with respect to the crank angle is similar that of 𝜏!, it is considered that the change of piston speed affects a relative large time scale in fluctuating flows. The cut-off frequency proposed in this study is close to the integral time scale of flow in the middle of the intake stroke, and therefore this cut-off frequency is equivalent to the relatively large time scale of flow generated by the piston behavior in the intake stroke. Due to these results, fc is determined from the inverse of averaged value of 𝜏! and 𝜏! at 𝜃=270 deg BTDC in the middle of the intake stroke.

In order to investigate the validity of an integral time scale of flow further to determine the cut-off frequency, comparison with an integral length scale and time scale of in-cylinder flow is shown in Fig. 10(a). The length scale in the bore direction and in the stroke direction is denoted by Lx and Ly, respectively. In addition, an integral length scale Lτ , which is calculated by multiplying the inverse of fc and turbulent intensity, is shown. Here, the turbulent intensity is square root of turbulent kinetic energy, kf. This result shows that Lx and Ly are smaller than Lτ in the intake stroke. The reason for this tendency is considered that analysis area is small with respect the length scale

Fig. 9 Comparison of a spatial-averaged integral time scale of flow, 𝜏𝑢 and 𝜏𝑣 in analysis area with a time scale of piston speed, 𝜏𝑃 for Ne=1500rpm WOT condition. 𝜏𝑢 and 𝜏𝑣 are cycle-average. of in-cylinder flow. In the compression stroke, Lx is close to Lτ . In Fig. 10(b), distribution of mean flow and turbulent intensity obtained by the time filtering method and spatial filtering method at 𝜃=40.0 deg BTDC is shown. fc is set to be 360Hz and Lc does 8.1mm which is an average of Lx and Ly. From this result, both of the mean flow and the turbulent intensity calculated by the both filtering methods shows similar distribution and coincide very well. Therefore, it is considered that the cut-off frequency determined by the above method is appropriate in the compression stroke. However further investigations might be required since Lx or Ly is different form Lτ in the intake stroke. 3.3 Comparison of turbulence characteristics Using the method investigated in the previous section, turbulence characteristics of this engine are analyzed in various operating conditions. Effects of engine speed on mean flow kinetic energy, Kf and turbulent kinetic energy, kf are shown in Fig. 11(a) and (b), respectively. Engine speed, Ne is 700 rpm, 1500rpm and 2000 rpm. The cut-off frequency, fc at each engine speed is 222 Hz, 360 Hz, 640 Hz. Furthermore, Kf and kf are spatially averaged in the analysis area (see [Fig. 1]), and then cycle-averaged. The result shows that Kf and kf show higher value for higher engine speed. It can be seen that turbulence becomes stronger with higher engine speed, which may general tendency. With respect to the crank angle 𝜃, Kf shows the maximum value around 𝜃=300-270 deg BTDC and decreases after. Furthermore, it shows a second local maxima around 𝜃=40 deg BTDC and decreases again. Although overall tendency of kf is similar with that of Kf, kf increases after the


- 9 -

decrease of Kf at 𝜃=40 deg BTDC. These results suggest that energy is transferred from mean flow

Fig. 10 (a) A cycle-averaged integral length scale of flow, (b) comparison of instantaneous flow distribution obtained by time filter and spatial filter at 𝜃=40.0 deg BTDC for Ne=1500 rpm WOT condition. A cut-off frequency for time filter is 360Hz and a cut-off length for spatial filter is 8.1mm.

Fig. 11 Engine-speed effects on (a) cycle-averaged mean flow kinetic energy, Kf and (b) cycle-averaged turbulent kinetic energy, kf. Kf and kf are spatial average. to turbulence component and turbulence is produced in the end of the compression stroke. Since turbulence characteristics can be obtained according to the operating condition, the time filtering method will be applicable to strong tumble flow field like in a boosted engine

Since optimization of piston top shape is one promising method for enhancing in-cylinder flow to promote combustion (Nogawa et al 2010), effects of piston top shape is investigated. Flow characteristics for different piston top shapes are shown in Fig. 12. Here, velocity vector maps are presented at upper part of the figure, and mean flow velocity distribution and turbulent kinetic energy, kf is shown at middle and at lower of the figure, respectively and all variables are cycle-averaged. The comparison shows that flow which seems to be turned to upward direction on the piston top is observed at the intake side at 𝜃=180.4 deg BTDC in case of the cavity piston [Fig. 12(a)]. In case of the flat piston [Fig. 12(b)], upward flow similar to that for cavity piston is not observed and the intake flow from intake valves is only observed. Mean flow distributions for the cavity and flat pistons are quite similar and velocity magnitude for cavity piston is larger than that for flat piston. At 𝜃=40.0 deg BTDC, faster velocity flow is observed toward the intake side from the exhaust side in case of the cavity piston [Fig. 12(c)]. In case of the flat piston, although similar flow pattern is created, velocity magnitude is smaller [Fig. 12(d)]. Turbulent kinetic energy, kf for the cavity piston is about 2 times larger than that for the flat piston at 𝜃=40.0 deg BTDC. Since turbulent kinetic energy for the cavity piston in combustion chamber is expected to be higher than that of flat piston, it is implied that optimization of the piston top shape is effective for enhancing in-cylinder flow to promote combustion.


- 10 -

Fig. 12 Effects of piston top shape on cycle-averaged in-cylinder flow, mean flow velocity and turbulent kinetic energy, 𝑘! for Ne=1500rpm WOT condition. (a)(c) Cavity piston and (b)(d) flat piston. 4. Conclusion In this study, in-cylinder flow in a boosted GDI engine is investigated by using high speed PIV on an optical engine with a transparent sleeve and following conclusions are obtained. (1) Tumble flow generated in the intake stroke is kept until the end of the compression stroke and

the tumble vortex center of this engine moves in a clockwise direction. (2) It is clarified that the appropriate time resolution of PIV for the evaluation of flow field by using

the time filtering method from the intake through the compression stroke should be higher than 5 kHz for in-cylinder flow field of the engine treated in the present study.

(3) In this study, the specific frequency where slope of spectrum changes is used as a cut-off frequency. This specific frequency corresponds to the inverse of the integral time scale of flow from the middle of the intake stroke to the end of the compression stroke, approximately. Therefore, this cut-off frequency is equivalent to the relatively large time scale of in-cylinder flow generated the piston behavior in the intake stroke.

(4) It is implied that optimization of the piston top shape is effective for enhancing in-cylinder flow to promote combustion speed.

Reference Adomeit P, Weinowski R, Ewald J, Brunn A, Kleeberg H, Tomazic D, Pischinger S, Jakob M (2010) A new approach for optimization of mixture formation on gasoline DI engines. SAE International 2010-01-0509:1-17. Hamamoto Y, Tomita E, Miwa H (1987) Measurement of turbulence of air swirl in a 4-stroke cycle engine cylinder. Transaction of the Japan Society of Mechanical Engineers Series B Vol.53 No.491:2226-1205. Heiduc T, Dornhofer R, Eiser A, Grigo M, Pelzer A, Wurms R (2011) The new generation of the R4 TFSI engine from Audi. 32th International Wiener Motor Symposium. Jeong H, Furui T, Nishiyama A, Ikeda Y (2011) Turbulence characteristics measurement near spark plug with PIV and LDV. 22th Internal Combustion Engine Symposium 24:139-144. Kido H, Wakuri Y, Murase E, Fujimoto K, Wang Z, Tomita O (1984) Variation of turbulence spatial scale in an engine cylinder in compression stroke. Transaction of the Japan Society of Mechanical Engineers Series B Vol.50 No.452:1114-1121. Kamimoto T, Yagita M, Moriyoshi Y, Kobayashi H, Morita H (1987) An experimental study of in-cylinder air flow with a transparent cylinder engine. Transaction of the Japan Society of Mechanical Engineers Series B Vol.53 No.492:2686-2693. Kojima S, Katsumi N, Miyagawa H, Okumura T, Ueda T (1990) Mechanism of knock-suppression by the slant-shaped squish combustion chamber : knock-suppression by flame acceleration in the latter half of combustion. Transaction of the Japan Society of Mechanical Engineers Series B Vol. 65 No. 638:287-293. Mito Y, Tanzawa K, Watanabe M, Eiyama Y (2011) Advanced combustion performance for high efficiency in new I3 1.3L supercharged gasoline engine by effective use of 3D engine simulation.


- 11 -

JSAE Annual Congress (Autumn) 2011562:23-26. Muller H R S, Bohm B, Gleiβner M, Grzeszik R, Arndt S, Dreizler A (2010) Flow field measurements in an optically accessible direct-injection spray-guided internal combustion engine using high-speed PIV. Experiments in Fluids Vol. 48 No. 2:281-290. Nogawa S, Nakata K, Kanda M (2010) Improvement of thermal efficiency by lean boosted engine. 21th Internal Combustion Engine Symposium A7-4:533-538. Sasaki J, Shishime K, Fujikawa T, Sato K, Wada Y, Omori H, Oda Y (2010) Engine torque improvement with high compression ratio. 21th Internal Combustion Engine Symposium B1-2:99-104. Ohtani H, Moriyoshi Y, Yagita M, Kamimoto T (1990) The effect of swirl on the decay and generation of in-cylinder turbulence during the compression stroke (LDV measurement of turbulence length scales using the spatial correlation method). Transaction of the Japan Society of Mechanical Engineers Series B Vol.56 No.530:3137-3180. Okura Y, Higuchi K, Urata Y, Someya S, Tanahashi M (2013) Measurement of in-cylinder turbulence in an internal combustion engine using high speed particle image velocimetry. Transaction of the Japan Society of Mechanical Engineers Series B Vol.79 No.809:2193-2206. Szengel R, Middendorf H, Pott E, Theobald J, Etzrodt T, Krebs R (2007) The TSI with 90 kW – the expansion of the volkswagen family of fuel efficient gasoline engines. 28th International Wiener Motor Symposium. The Visualization Society of Japan (2002) Visualization of flow by tracer particle. Hand book of particle image velocimetry. Morikita publishing, Tokyo, pp 26-38 Wakisaka T, Hamamoto Y, Kinoshita S (1982) Turbulence characteristics in combustion chamber of internal combustion engine. Transaction of the Japan Society of Mechanical Engineers Series B Vol.48 No.430:1198-1205.

Analysis of In-Cylinder Flow for a Boosted GDI Engine ...

Documents