Graham Pitcher1,* and Graham Wigley2 - ULisboaltces.dem.ist.utl.pt/lxlaser/lxlaser2012/upload/189_paper_mjeojf.pdf · the steady state measurements that was not present in ... this

16th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 09-12 July, 2012

- 1 -

A Comparison between In-cylinder Steady Flow and

Motored Engine Air Velocities Using LDA

Graham Pitcher1,* and Graham Wigley2

1: Advanced Concepts, Lotus Engineering, Norwich, UK

2: Department of Aero and Auto Engineering, Loughborough University, Loughborough, UK * correspondent author: [email protected]

Abstract Steady state measurements on engine cylinder heads have been performed for many decades as a simple and early design tool in an effort to predict the performance of an engine. These integral techniques have proved very successful for swirl based engines but have failed to show the same success for tumble flow, where the absence of the piston in steady state work is believed to have a big impact. Modern spark ignition engine design, particularly with direct injection, is finding the tumble flow in the cylinder to be a critical part, for both guiding of the fuel for stratified operation and as an aid to the combustion speed and stability under all operating conditions. This had led to laser diagnostics being applied in place of the simpler integral methods, but with the assumption of a close correspondence between the steady state flow and that in a running engine. This paper looks at a comparison between in-cylinder flows under steady state and motored engine conditions to test the above assumption. LDA measurements were made in a horizontal plane for the steady state measurements and vertical planes for the motored engine flows. Calculated tumble ratios and correlation factors are used to give a quantitative comparison between the flows generated for the two cases. The results from these comparisons showed the tumble ratio to be higher in the engine flows, but that an unexpected cross tumble was measured for the steady state measurements that was not present in the engine. The correlation coefficients were approximately 0.5, which although not high, does indicate some similarity between the flows. A final qualitative comparison is given between PIV vector fields, now in a vertical plane, for the steady state measurements and the LDA vector fields for the motored engine. 1. Introduction For many years the steady flow bench has been the standard method to characterise cylinder heads for discharge and flow coefficients in the automotive industry. The techniques and subsequent data processing that have been developed and utilized have proved to be very successful in guiding port optimisation for air flow into the cylinder. The introduction of a swirl vane meter or torque momentum meter into the flow, to measure the swirl ratio allowed a correlation of the flow structure to engine performance. These techniques performed very well with a swirl based engine geometry but could not deliver the same accuracy with tumble based engines. Here it was much more difficult to measure the rotation of the flow in the axis of the cylinder, and additionally it was believed that not being able to allow for the piston, which would be expected to have a direct influence on tumble flow, caused greater problems in the predictive capabilities of these bulk flow techniques. With the present drive to reduce both emissions and fuel consumption, while maintaining performance and the now quite common use of direct injection for gasoline engines, the tumble flow structures developed in the engine cylinder are becoming critical for both fuel guiding in stratified operation and aiding the combustion speed and stability in all modes of operation, Middendorf et al., 2005, Kessler et al., 2006, Boccadoro et al., 2007, Szengel et al., 2007 and Boccadoro et al., 2009. One of the effects of a high tumble induced flow is the detrimental effect this has on the efficiency of inducting the air into the cylinder, Tsuji et al., 2006, Kessler et al., 2006 and Andriesse et al, 2008, but the flow structure has been deemed to be the more important criteria in these cases. These last two points has seen a change to the way in which these measurements are conducted, with laser diagnostics, both LDA and PIV, being applied to understand the flows being developed, Glanz, 2000 and Bensler et al., 2002. The laser techniques have allowed the tumble flow to be measured directly, rather than forcing the flow


- 2 -

through ninety degrees to generate a pseudo-swirl flow using either an ‘L’ or a ‘T’ junction. Additionally, they allow information on the flow structures to be obtained locally throughout the cylinder, aiding with some of the design considerations for direct injection engines. However, most of the work is still performed under steady flow conditions for several understandable reasons. Firstly, this work is usually required to be completed at an early stage of the design, before a complete engine may have been constructed. Secondly, these steady state measurements are much simpler to set up and obtain and thirdly, they are much cheaper and faster to perform. The main assumption from this is that there is a close correspondence between steady state measurements and those that would be obtained in the cylinder of a running engine. This paper looks at a comparison between the in-cylinder air velocity measured under steady flow conditions and that in a motored engine to observe the correlation between the flows and hence test the above assumption. 2. Experimental Setup The experimental setup can be split into three basic components: the steady flow rig, the optical engine, and the LDA system. 2.1 The Steady Flow Rig The steady flow rig was a propriety item specifically designed to perform the standard port flow measurements, a Superflow SF600. It consisted of the pump system, manometers to measure rig and valve pressure drops, an inclined manometer to measure the flow rate through an orifice plate and a thermocouple to measure ambient air temperature. The rig actually had a series of user selectable orifice plates for the different flow rates generated by changes in pressure drop and valve lift. The air is pulled through the ports and valve gap, via a dummy engine cylinder, which in this instance, was a fused silica liner as it was intended to perform LDA measurements on the rig as well as the conventional flow measurements. A hemispherical shaped piece of plasticine was placed around the entry to the ports as an aid to air entry, with this work being done without an inlet manifold. The procedure for this piece of work was to open the inlet valves to the required valve lift, switch on the machine and adjust the flow rate until the required pressure drop was obtained between atmospheric pressure and the pressure in the engine cylinder, so the pressure drop was fixed across the valve gap. The standard measurement then consisted of measuring the flow rate using the orifice plate and inclined manometer. Additionally, LDA measurements of the axial air-flow velocities were performed in radial scans for many radii around the engine cylinder. These measurements were made in a horizontal plane half bore down from the flame face and for different pressure drops and valve lifts. 2.2 The Optical Engine The Lotus optical engine has been designed specifically for the use of optical diagnostics, Allen et al, 1999 and is shown in Figure 1. The cylinder bore is 88 mm and the stroke is 82.1mm. The compression ratio is 10:1, producing a theoretical peak adiabatic compression pressure of 25.1bar. Optical access is provided through a full length fused silica liner and a sapphire window in the piston crown. A carbon fibre piston ring runs in the optical liner to maintain correct compression pressures. An extended bifurcated piston allows a 45° mirror to be located between the upper and lower piston crown allowing illumination or viewing into the cylinder from below. A maximum engine speed of 5000rpm has been achieved through the use of light weight materials for the piston components and the use of primary and secondary balance shafts in the crankcase. The bifurcated piston is manufactured from aluminium while the piston crown is made from titanium, chosen due to the similarity in thermal expansivity between titanium and the sapphire of the piston window. Engine timing data were provided by two optical encoders: One with 1.0° resolution mounted on the end of the crankshaft and one with 0.2° resolution mounted on a 2:1 drive representing the camshaft. The


- 3 -

overhead poppet valves are actuated by means of a fully variable electro-hydraulic valve system, ‘Lotus Active Valve Train’, which allows complete control of the individual valve profiles. The engine specifications used for this work are given in Table 1, and the valve timings and profiles are shown in Table 2 and Figure 2 respectively. The valve timing data is given relative to TDC valve overlap.

Figure 1: Single Cylinder Optical Research Engine (SCORE)

Table 1: Engine Specification

Bore 88.0 mm Peak Otto Cycle Compression 25.1 bar Stroke 82.1 mm Peak Compression (750 rpm) 19.2 bar Capacity 500 cc Peak Compression (2000 rpm) 22.1 bar Maximum Speed 5000 rpm Peak Compression (3500 rpm) 22.0 bar Compression Ratio 10:1

Table 2: Valve Timing Data IVO 675° IVC 264° EVO 462° EVC 45°

Max. Inlet Lift 9.35 mm Max. Exhaust Lift 9.35 mm


- 4 -

Figure 2: Valve timing profiles

2.3 The LDA System Two distinct LDA systems were employed in this work. A diffraction grating based, discrete optics system was used for the steady state work and a Bragg cell based, fibre optics system for the motored engine work. This change of optical system was implemented to ease traversing of the measurement volume in the cylinder of the engine. Both systems were operated in single component mode and utilized an Argon ion laser as the light source, operating in single line mode at 514.5 nm. The scattered light was collected in the direct backscatter direction, aiding traversing for all measurement sets. The main parameters for the two systems are given in table 3.

Table 3: LDA Parameters Discrete Optics Fibre Optics

Parameter Value Parameter Value Wavelength 514.5 nm Wavelength 514.5 nm

Beam separation 50 mm Beam separation 60 mm Beam diameter 5 mm Beam diameter 3 mm

Focal length 300 mm Focal length 310 mm Crossing angle 9.53° Crossing angle 11.06° Fringe spacing 3.10µm Fringe spacing 2.66µm

Length δz 473µm Length δz 702µm Height δy 39.30µm Height δy 67.69µm Width δx 39.44µm Width δx 68.01µm

Number of fringes 12 Number of fringes 25 Frequency Shift 10.0 MHz Frequency Shift 40.0 MHz

It can be seen from table 3 that the major differences between the two systems are a change in the frequency shift from 10 MHz to 40 MHz and the size of the measurement volume. Neither of these had any undue effects on the data collection. 3. Results


- 5 -

The steady flow measurements were made at half bore distance down from the flame face, 44mm, and were performed as radial scans for multiple orientations of the cylinder head. The radial positions were on a non-linear scale to give equal segment areas in the plane for a separate piece of work looking at mass flow calculations. Measurements were made for two pressure drops across the valves, 250mm H2O and 635mm H2O and for valve lifts of 2mm, 5mm and 10mm. The scan orientations are shown in figure 3 with the radial scan positions given in table 4. One thousand samples were collected at each measurement location and processed to give mean and RMS velocities.

Figure 3: Radial scan locations

Table 4: Radial positions

Position Number Radial Distance mm 1 42 2 40.6 3 38.1 4 35.5 5 32.6 6 29.5 7 26.0 8 22.0 9 17.0

10 9.8 11 0.0

The motored engine measurements were made in four vertical planes, the main tumble diameter, the cross tumble diameter and two chords across the centre of the inlet and the exhaust valves, figure 4. Seven locations were used in this vertical plane from 10mm below the flame face to 70mm below the flame face in 10mm steps. The measurements were performed as horizontal scans at each vertical location, with a step size of 5mm, beginning 4mm in from the cylinder wall. For this transient case all three components of velocity were measured independently. Data were collected throughout the inlet and compression stroke, with the data collection window being adjusted with vertical location to reflect


- 6 -

the position of the piston. The number of samples collected was also adjusted to match the size of the total window with the number of samples set to equate to 100 samples per degree. The data were then ensemble averaged into 4 degree crank-angle windows to produce mean and RMS values of velocity.

Figure 4: Position of measurement planes for motored engine study

The results from the steady flow work can best be observed as mean velocity profile plots across the main tumble and symmetry, cross tumble diameters. An example for the 5mm valve lift and for both pressure drops of 250mm and 635mm of H2O is shown in figure 5.

Mean Velocities (5mm lift)

-20

-10

0

10

20

30

40

50

60

-50 -40 -30 -20 -10 0 10 20 30 40 50

Radial Distance [mm]

Velo

city

[m/s

]

Main 250Symmetry 250Main 635Symmetry 635

Figure 5: Mean velocities for the 5mm valve lift across the main and symmetry

diameters Several comments can be made from the plots in figure 5. The flow on the symmetry diameter does


- 7 -

show a reasonably degree of symmetry, but with one valve apparently flowing faster than the other. This is in agreement with the standard bulk flow measurements of mass flow rate where the same feature was observed. More significantly, there is no real increase in the mean velocity with the change in pressure drop, unlike on the main diameter where the expected increase in velocity is observed. The flow on the main diameter is far from symmetrical about the cylinder centre line, and it is this asymmetry that is responsible for the tumble motion being generated in the cylinder. The measurements in the motored engine, where a full plane of data was collected, are best viewed as vector plots at different time steps and so allowing the transient nature of the flow to be observed. Examples of these plots for crank-angles of 90 and 180 degrees and for the same diameters as in figure 5 are shown in figure 6.

Main diameter 90° crank-angle Symmetry diameter 90° crank-angle

Main diameter 180° crank-angle Symmetry diameter 180° crank-angle

Figure 6: Mean velocity vectors for the main and symmetry diameters With the transient results, again a high degree of symmetry can be seen in the symmetry plane, but now there appears to be a much closer correspondence between the flows from the two inlet valves. This is somewhat surprising as these measurements were performed on the same head as the steady flow study. The generated tumble flow is clearly observable in the main tumble plane, as an anti-clockwise rotation, and at 180 degrees crank-angle, a reverse tumble flow can be seen just above the piston on the inlet valve side of the cylinder. This reverse tumble is generated from the flow exiting between the two inlet valves and so moving down the cylinder on the inlet valve side of the cylinder. Two parameters have been selected to give a quantitative comparison between the steady flow and the


- 8 -

transient, motored engine results; namely the calculated tumble ratio and a correlation coefficient giving an indication of the similarity between the two sets of data. Due to the measurement technique employed for the steady state results, the comparison will be made between these results on the two main diameters and the profiles obtained from the motoring engine on the appropriate diameters. Examples of this are shown in figure 7 for the 5mm and 10mm valve lifts and for a pressure drop of 635mm H2O. To better match the results, the pressure drops from the steady flow work have been taken to be representative of fixed engine speeds, with 2500rpm and 4000rpm being representative of 250mm and 635mm of H2O pressure drop respectively. The measured motored engine velocities were then scaled with the ratio of the engine speed measured to the representative engine speed from the pressure drop. This is shown in figure 7, where both the velocity and a corrected velocity are plotted from the motored engine results.

Tumble Flow

-10

0

10

20

30

40

50

60

-50 -40 -30 -20 -10 0 10 20 30 40 50

Distance [mm]

Velo

city

[m/s

]

SteadyMotoredCorrected

Cross Tumble Flow

-15

-10

-5

0

5

10

15

20

25

-50 -40 -30 -20 -10 0 10 20 30 40 50

Distance [mm]Ve

loci

ty [m

/s]


Tumble flow – 5mm lift Cross tumble flow – 5mm lift

Tumble FLow

-20

-10

0

10

20

30

40

50

60

70

-50 -40 -30 -20 -10 0 10 20 30 40 50

Distance [mm]

Velo

city

[m/s

]


Cross Tumble Flow

-20

-15

-10

-5

0

5

10

15

20

-50 -40 -30 -20 -10 0 10 20 30 40 50

Distance [mm]

Velo

city

[m/s

]


Tumble flow – 10mm lift Cross tumble flow – 10 mm lift Figure 7: Comparisons between steady flow and transient velocities

There are many ways to calculate the tumble ratio, particularly when the full field is available as is the case for the transient measurements, but as was shown by Pitcher and Wigley, 2003, the variation in the final result is not that great. For this work, the tumble ratio will be based on the angular velocity as that is the most relevant for the steady flow work where only one diameter is available. With this algorithm, the average angular velocity of all the points is normalized by the engine speed in radians per second, to give the tumble ratio, equation 1.

Tumble Ratio πω2

11∑==

N

n n

n

RU

nTr

1

Where: U = Air velocity ms-1 R = Radial position m ω = Engine speed Hz The correlation coefficient for the comparison is defined in equation 2, and has been chosen for the


- 9 -

feature of giving a negative correlation when the velocities from the steady flow and the transient measurements are in different directions, and so the sum is biased towards a low number when this occurs. ∑ +

=n

smms

nR

122

21 2

Where: m = motored velocity ms-1 S = Steady flow velocity ms-1 N = number of positions correlated The results from the above analyses are shown in table 5, with the steady flow tumble results for the main tumble plane and the cross tumble plane given first, followed by the same quantities for the motored engine results and finally the correlation between the profiles for the two sets of measurements. For the motored engine studies, the analysis has been performed at the time in the engine inlet stroke that corresponds to the same valve lift as that used in the steady flow analysis. There is only one value of the tumble ratio with the motored engine results, as the engine speed is used in the calculation and the measurements were only performed for one engine speed.

Table 5: Tumble ratios and correlation factors Steady Flow Data

Valve Lift Tumble Cross tumble 2mm 0.517 0.391

5mm Low -0.083 -0.312 5mm High 0.08 -0.213 10mm Low 0.372 -0.288 10mm High 0.185 -0.429

Engine Data Valve Lift Tumble Cross tumble

2mm 1.07 -1.07 5mm 1.27 0.17

10mm 0.37 -0.01 Correlation Data

Valve Lift Tumble Cross tumble 2mm 0.50 -0.08

5mm Low 0.50 0.41 5mm High 0.45 0.31 10mm Low 0.66 0.46 10mm High 0.54 0.42

The cross tumble in the engine is essentially zero apart from the 2mm valve lift. The steady flow analysis however, gave a non-zero value for the cross tumble at all valve lifts and although small it is still a significant value. The principal difference that could be responsible for this effect is the inlet manifold on the engine that was not present on the steady flow measurements. For the main tumble


- 10 -

diameter, the tumble value is always greater or equal for the engine measurements. This is the result that would be expected given that the presence of the piston is believed to be largely responsible for the generation of the tumble flow. The expected increase in the tumble value with increasing valve lift was not realized. Also, the difference in the value for different pressure drops was unexpected, as the values are normalized to a representative engine speed and would be expected to generate similar values. The correlation factors are not particularly high, but with values around 0.5 do show some degree of correlation between the steady state flows and those in the engine. The graphs in figure 7 support this showing similarity between the two flow types and even for the 5mm valve lift in the tumble plane, where there is a lateral shift in the position of the minimum value of velocity, the curves still retain the same shape. The correlation factors are consistently higher in the tumble plane where the higher velocities are leading to a greater stability in the flow. The results presented so far do not lend much confidence in using steady flow data to estimate engine performance. Although there are similarities in the flow profiles, there are significant differences in the calculated tumble ratios and the correlation factors are only in the region of 0.5. However, some later work, which has not yet been fully analysed, does appear to indicate that, with the correct tools, a closer correspondence may be found between steady flow and engine measurements. The plots in figures 8 and 9 show PIV measurements on the steady flow rig compared to the LDA vector results from the engine, for the 2mm valve lift. The plots in figure 8 are for the main tumble diameter, while those in figure 9 are for the cross tumble diameter.

Figure 8: Main tumble diameter with 2mm valve lift

The two images in figure 8 are not to exactly the same scale, as to reduce the PIV image more would remove the possibility to see any of the vectors. The first comment to make is that the steady flow structure is more complex than could have been imagined from looking at a single line of data as observed from the LDA measurements. There does seem to be some correspondence between the flow structures from the two sets of measurements. There is a definite flow movement across the cylinder from the exhaust to the inlet side in both cases. This was assumed to be due to the presence of the piston in the engine, but the fact that this feature exists in the steady flow measurements suggests that the fast moving flow from the back of the valves causes a pressure drop across the cylinder and hence pulls the


- 11 -

air over from the exhaust side. There are flow reversals visible in the steady flow fields, but they do not coincide with those seen in the motored engine. The flows in the engine are of course modified by the piston whereas those in the steady flow case can only be modified by local areas of higher or lower pressures.

Figure 9: Cross tumble diameter with 2mm valve lift

There are also similarities in the flow fields for the cross tumble plane with the 2mm valve lift, figure 9. Both sets of data show flow moving downwards in the centre of the cylinder and back up at the walls. The main difference between the flows comes from the magnitude of the vectors, and this was observed when studying the LDA results from the two sets of measurements. The next stage in this further piece of work will be to calculate the tumble ratios for the PIV data and the full field LDA vector data to ascertain whether there is closer correspondence in this case. Study of the two sets of data to date have suggested that the flows show remarkable similarities away from the piston crown but this has yet to be quantitatively verified. 4. Conclusions In-cylinder flow measurements have been made under steady state and motored engine conditions using LDA. The steady state measurements were performed in a plane half bore diameter below the flame face while the motored engine measurements were performed in vertical planes. Angular velocity was used to calculate tumble ratios across the main tumble diameter and the cross tumble diameter for both measurement series, and a correlation coefficient was defined to quantify the similarity of the flows from the two cases. The tumble ratio on the main tumble diameter was greater for the engine flow but as the presence of the piston is believed to be one of the generators for tumble motion, this was not unexpected. The cross tumble ratio did show an anomaly for the steady flow measurements in having a non-zero value, whereas in the engine, except for the 2mm valve lift, this was neutral. The correlation factors gave values of approximately 0.5, which although not high does indicate a degree of similarity between these flows. Finally, vertical planes of PIV data for the steady flow showed visually a lot of similarity to the vector fields in the engine. This suggests that local pressure fields are responsible for some of the tumble motion, away from the vicinity of the piston. This data has yet to be analysed for tumble ratios but shows some promise that with the correct selection of measurement tools and strategy, there is a possibility of using steady flow measurements to gain insight in the flows likely to occur in the engine.


- 12 -

5. References Allen J., Simms A., Williams P., Hergrave G. and Wigley G., An Advanced Optically Accessed Single Cylinder Research and Development Engine for Fuel/Air Mixing and Combustion Diagnostics, JSAE 9935086, Seoul, Korea, 1999. Andriesse D., Comignaghi E., Lucignano G., Oreggioni A., Quinto S. and Sacco D., The new 1.8 l DI Turbo-Jet Gasoline Engine from Fiat Powertrain Technologies, 17th Aachener Kolloquium Fahrzeug- und Motorentechnik, 2008 Bensler H., Kapitza L., Raposo J. and Reisch, A New Experimental Method for Determining Port Generated Swirl Flow, SAE 2002-01-2846, 2002 Boccadoro Y., Kermanc’h L., Siauve L. and Vincent J., The new Renault TCe 130 1.4l turbocharged gasoline engine, 30th Internationales Wiener Motorensymposium, 2009 Boccadoro Y., Tranchant O., Pionnier R. and Engelhardt H., The new Renault TCe 1.2l turbocharged gasoline engine, 28th Internationales Wiener Motorensymposium, 2007 Glanz R., Differentielle Erfassung von Tumble-Strőmungsfelden, MTZ, 2000

Kessler F., Sonntag E., Schopp J., Simionesco L., Keribin P. And Bordes F., The new family of small 4 cylinder engines from BMW/PSA co-operation, 15th Aachener Kolloquium Fahrzeug- und Motorentechnik, 2006 Middendorf H., Krebs R., Szengel R., Pott E., Fleiss M. And Hagelstein D., Volkswagen introduces the worlds first double charge direct injection petrol engine, 14th Aachener Kolloquium Fahrzeug- und Motorentechnik, 2005 Pitcher G. and Wigley G., LDA Analysis of the Tumble Flow Generated in a Motored 4 Valve Engine, Proceedings Laser Anemometry Advances and Applications, 7th International Conference, Limerick, Ireland, 2001 Szengel R., Middendorf H., Pott E., Theobald J., Etzrodt T. And Krebs R., The TSI with 88kW – the expansion of the Volkswagen fanily of fuel-efficient gasoline engines, 28th Internationales Wiener Motorensymposium, 2007 Tsuji N., Sugiyama M. and Abe s., The new 3.5l V^ gasoline engine adopting the innovative stoichiometric direct injection system D4S, 27th Internationales Wiener Motorensymposium, 2006

Documents

Graham Pitcher1,* and Graham Wigley2 - ULisboaltces.dem.ist.utl.pt/lxlaser/lxlaser2012/upload/189_paper_mjeojf.pdf · the steady state measurements that was not present in ... this