Analysis of In-Cylinder Air Motion with LDV
Measurement and Multi-Dimensional Modeling
Teruyuki Itoh, Yasuo Takagi, Tokuhei Ishida,
‘Shigwo Ishikawa and Tsuyoshi Ishikawa
Nissan Motor Co., Ltd
Central Engineering Labi
Natsushima-cho, Yokosuka 237
ABSTRACT
‘The tangential and axial velocity in
a hemispherical combustion chamber using a
conventional gascline engine is measured
by means of ‘laser doppler velocimetry
(L.D.V.). At the same. time, an
axi-synmetric two-dimensional model for
Simulating the gas flow in combustion
chambers is applied. The svirl ratio as
initial conditions for the numerical
calculation was derived from LDV
Measurement. The shift of the swirl
center and the incline of the avirl plane
is observed through the results of LBV
measurement. Good agreement between
measured and ‘calculated results is
obtained under the driving condition where
the measured swirl center is close to the
cylinder axis. Moreover, modifying the
calculated velocities in giving an
excentricity of swirl center obtained from
experiment results in good quantitative
agreement with the measured results even
Af the swirl center is not coincident with
the cylinder axis. Numerical calculation
of gas motion combined with LDV
Measurement’ improves understanding of the
gas flow in a combustion chamber under
realistic driving conditions,
INTRODUCTION
Making use of optimized air motion
has long been a goal for obtaining stable,
smooth and economical operation in spark
ignition engines. Several engines that
emphasize intake Jair motion to improve
fuel economy (1)(2} and meet the severe
emission standards in the U.S.A. and
Japan, have been put on the market. “In
addition, a lean burn engine system (3)
has been developed with a swirl. control
valve at the herical intake port that
generates high swirl ratio during partial
load operation. However, details of
quantitative air motion, if a combustion
chamber, such as the spatial distribution
of velocity and turbulence and its
variation in time, are not. fully
understood. A long-used hot wire
anemonetry (4) that could not detect “the
18s
direction of flow has been replaced by
laser doppler velocimetry (L.D.V.) because
LDV can ‘determine the direction of air
flow in the combustion chanber,
Results of measuring the direction of
mean flow and turbulence intensity of
squish flow in a side valve spark ignition
engine with LDV are reported by Asanuma et
al, (5) and those in a piston bowl by
Vatidis (6) and Murakami et al. (7)~
However, LDV is not always the best and
only way to understand complicated three
dimensional air motion in a cylinder
because measuring points are restricted to
a limited space by the presence of valves
and the motion of a piston in actual
engines.
On’ the other hand, comprehensive
nunerical multi-dinensional models to
Simulate the in-cylinder dynamics of air
motion and combustion are being developed
(81(9)(10). One smerit. of the
multi-dimensional model is that it can
present three-dimensional distribution of
Fluid flow, gas temperature and chemical
species. One of the problens to be solved
in multi-dimensional modeling is
verification of calculated results with
experimental data
Comparison of calculations for gas
flow with a two-dimensional mode! and
experimental results with LDV was reported
by Gosman et al. (11), Matsuoka et als
(42) among others. Most of the works were
carried out with diesel engines having a
bowl in a piston (aIP) with a high one way
swirl. Few reports on’ gasoline engines
have “been published. This is because Gas
dynamics in the combustion chamber of a,
conventional gasoline engine are a little
more ‘complicated than those of a. BIP
diesel engine. In general, most of the
chambers have a flat configuration and the
swirl is not so high.
The main objectives of this study are
to examine the three-dimensional
in-cylinder gas dynamics of a gasoline
engine with a flat hemispherical
combustion chamber by LDV measurement and
calculation with an __axi-symmatric
fhunbers in parenthese Gesignate references
at end of papermodel. In addition, a
proposal is presented for extending
applications of the two-dimensional
axi-synnetric model.
two-dimensional
EXPERIMENTAL APPARATUS
vest Engine
A test engine used in this experiment
is a water-cooled 4 stroke single cylinder
engine that have flat hemispherical
combustion chamber configuration. The
main specifications of the test engine are
shown in Table 1, Lubricant-free piston
rings were used to prevent the windows
from staining with oil and seeding
particles.
TABLE 1. ENGINE SPECIFICATIONS
Bore x stroke (mm x mm) 76 x 78
Displacement volume (cc) 352.8
Chamber configuration hemispherical
Compression ratio gil
optical set Up for upv
‘The forward scattering LOV was used
throughout this experiment in order to
obtain better signal to noise ratio of the
Scattered light than that of the backward
scattering LDV. Bragg cell shifter was
The optical access for_— tov
measurement was provided by two quartz
windows installed in the cylinder head.
The diameter of guarte windows was 10
millimeters. Laser and LDV. optics were
set on the bed which enabled them to nove
three dimensionally. Seeding particles
used in this experiment is a SAE-fine dust
which has mean diameter of 0.6 microneter.
The seeding particles were supplied
to an intake port by a seeding generator
which could keep the density of seedin:
particles steady according to the esr
amount supplied to the cylinder. A detail
of the optical components and data. are
shown in Appendix A.
Experimental Condition
The engine vas motored at an engine
speed of 1400 rotation per minute (r.p.m.)
and a charging efficiency of 408, the
equivalent load’ conditions for 40° kn/n
road-load operation. Velocity measurement
at an engine speed of 700 rpm end
charging efficiency of 40 % as also
performed.
The statistical treatment for
acquiring the mean velocity was the moving
time average method, Sampling intervalsa
@
Consisting of arbitrarily _ shaped
guadrilaterals, was employed for
finite-difference “calculation. The
velocity at a point corresponding to the
LDV measuring was obtained through
interpolation by using the velocity of
several grids which surround the measuring
points. The boundary condition throughout
this study was free slip.
g0°BTDC = 6OBTOC30°BTDC 10° BT
INE : 700 rpm)
®
E
a !
i= t
Fig. ¢
1 Hee Be
OC
87
RESULTS
Tangential Velocity
R tine history of tangential
component in measured mean velocity at
each measuring point under engine speed of
3400 rpm are shown in the upper part o:
Fig. “3, distributions of the tangential
component at the selected crank angle are
shown in the upper of Fig. 4. The
following characteristics can be observed
in these figures.
Between around 130 deg TOC and 90
deg BTDC, absolute velocity values at each
Measuring point increase slightly
(indicated by an arrow "a"). The piston
motion that pushed up the main flow toward
the measuring point aight cause the change
in velocities.
e
Pi
e < ‘
# sf
2 ce:
z X Te =40 te
5 ‘
# i
a \
2. Fe OO cea SO este TOC ers 0 e000.
CRANK ANGLE (deg ATOC )
Tangential and axial mean
velocity - measured
Toc wrAToc §=—-S0"ATDC. BOAO
assis
Tangential mean velocities - measuredNE: 1400rpm
B so°BTDe YP BTOC , 30°BTDC 10° BTC
=
8 ty
4 me
vig. 5 ania nean
Between around 90 deg BTDC and 10 deg
arpc, all velocities seen to change their
values almost evenly. (indicated by "pb"
and "e") It can be explained by following
Understanding obtained from Fig. 4.
fough tangential air flow in the chanber
behave almost like a solid body in all
crank angles, a center of the swirl flow
does not coincide with a center of the
cylinder and the center changes its
Yocation time by time. From the
observation it can be said that tangential
velocity measured at a fixed point’ might
be strongly affected by the eccentricity
and the movement of the swirl center. As
the eccentricity of the swirl center under
700 rpm operation is small, especially
after 30 deg BIDC, as is shown in the
lower part of Pig. 4, the effects of
gwizl ‘center movement’ on measuring
velocity should depend on engine speed.
Moreover, it can be observed fron
Fig. 4 that entire tangential velocities
tend to decay and show more complicated
velocty distribution after around 60 deg
aTpc.
Axial Velocity
RES’ Eine history of axial component in
measured mean velocity at each measuring
point under engine speed of 1400 rpm are
Ehown in the lower of Fig. 3,
distributions of the axial component at
the selected crank angie are shown in Fig.
5. It can be observed fron the figure
188
Toe
i woraToC ; SOrATOC one
ie
locities - measured
that large twin vortices are generated
between 90 deg. and 60 deg, BTDC as are
shown with arrows. On the other hand,
single counter-clockwise _ voltex is
observed at an crank angle after 10 deg
RTDC. Between these periods axial
component of mean velocity consists only
upward flow. These results also suggest
that flow pattern in the
complicated behavior and
configuration tine by time.
Taree-dimensional Gas Flow
The tangential component and the
axial component of measured mean velocity
under 1400 rpm operation shown in Fig. 4
and Fig. $, respectively, are illustrated
in a same coordinate as are shown in Fig.
6. From this figure, it is clear that air
motion in the combustion chamber is very
plicated and it is not so easy to
capture the direction and velocity value
of mainflow plane on the measuring line.
In order to understand these complicate
flow field more deeply, data of a
three-dinensional velocity’ distribution
are required, such as by means of
three-dimensional LDV measurement or by
nulti-dimensional numerical calculation.
Understanding of Flow Field with
‘Axi -Synmetric Two-Dinent ional vodelin
valve_for
‘to adopE
Swirl ratio as an initial
punerical calculation. In order
WW'BTOC60°BTOC 30°BTDC 1e°BTOC
Ne Ua0rpm
‘SWIRL RATIO: 2.15,
FREE SLIP.
MEAN VELOCITY IN ms!
Toc
60°ATDC 30°ATOC TOPATOC
aa
Fig. 7 Contours of tangential velocities - calculated
the results of numerical calculation for It is seen that the calculated result is
Standing of flow field in the quantitatively and qualitatively far from
experiment, calculaton was performed with measured one under his operation. The
the “same Combustion chamber configuration existance in the eccentricity of swirl
a8 was used in the experiment and with an center, however, was observed in’ the
inital value of swirl ratio which is experiment. As 'the profile of tangential
erimental result. velocity has a steep inclication, a small
equivalent to the @:
perimental swirl ratio was calculated shift of swirl. center might” cause
iron an inclination of velocity differences "in _measuring” reales,
Zistribution of tangential conponent as is Inaginning that” calculated value ‘aigh¢
geen in Fig. 4. The swirl ratios given have the’ eccentricity, the center of
zron the initiel Value are 2.15 at. crane calevalated tangential’ volocity ficlas,
fagle of 130 deg. STDC both under 700 and such is shown in Fig. 7, ‘were thifeed ¢o
1160 ren operation. the swirl center that vas decided by the
experiment. “A dashed line in the Fig, 8
Gonpacison of calculation and Shows velocity at point D, which renaine
eS _—___E_||___E
contours of Taigential” velocity obtained compensated velocity shave quantitatively
fon ‘numerical calculation at selected and qualitatively, good "ageeenent with
rank angles under 1400 cpm. Tine curves measured one. it “can be sald fron the
of the velocity both calculated and result thats flow field caleulates with
Measured at point D are shom in fig. 8. two-dimensional axi-syanceric sumaseal
nodeling aight indicate che equivaicnt
mesmo flow field in the combustion chanber of
8 ui tow realistic engine expect for the
z easing it = eccentricity of the swicl center. Using
ea the calculated flow field. of tangential
E velocity. shown in Fig. 3, “following
= deeper conprenensions than those with LOv
Ea measurement are obtained. During
8 Gonpression stroke up to around 10” deg.
a BTbc tangential velocity. increases
g according to upward piston’ motions A
S velocity” boundary layer appears around at
Zo} 30 deg. aTDc and it develops mainly on a
3 Surface of cylinder and cylinder heads
2 The decay in’ swirl velocity that was
a o a observed in the measured results around a
FM Yoo in Fig, 4 is seen to be caused by the
See eee (aed AIDE) boundary layer on the surface of cylinder
Fig. 8 Comparison between calculated Read. A conplex behavior in tine history
and measured tangential velocities of measured velocity that was seen afteTDC in Fig. 3 might be understood because
neasuring points are arranged in the
complicated boundary layer in Fig. 7.
Moreover, the existance in the
eccentricity of swirl center makes the
Feasured velocity more complex. On the
contrary, calculated tangential velocity
shows good agreenent with measured one
under “700 rpm operation shown in Fig. 9.
It is because the eccentricity of swirl
center under 700 rpm operation is small
and swirl pattern keep solid-body like
velocity distribution through the stroke
a5 is shown in the lover of Fig. 4.
[Re = 70D rem
Me: 20%
8 7 [Measuring Point :0)
MEAN VELOCITY (rn/s)
4 = 0 oO
CRANK ANGLE (deg ATOC)
Fig. 9 Comparison between calculated
and measured tangential
velocities
concLUsroNs
Following comprehensions to air flow
field in such an flat hemispherical
combustion chanber as is used for spark
ignition engines were derived from this
study.
(7) Although distribution of tangential
velocity. in the chamber is basically
solid-body-like, a center of the swirl
does not coincide with the center of the
cylinder and the swirl center should be
wondering around tine by time.
(2) & decay in swirl in TDC and vicinity
should be affected by the boundary layer
Geveloped on the surface of cylinder head.
(3) Axial flow has twin vortices in the
compression stroke and it turns to single
counter-clockwise voltex after TDC.
(4) An axi-symmetric two dimensional gas
flow model if ane of methods to give muita
dimensional informations to the data which
were obtained with non-molti dimentional
measurement.
REFERENCES
+ Harada, M., Kadota, T., and Sugiyama,
Yu, Nissan NAPS“Z Engine Realizes
Better Fuel Economy and Low NOx
Emission," SAE Trans, Vol. 90, Paper
No, 810010, 1981.
: Nakamura, H., Ohinouye, T., Mori, x.
Kyota, | Y., and Nakagami, | T.,
"Development of a New Conbustion
system in Gasoline Engine," SAE Paper
780007, 1978.
190
3. Inoue, T., et al., “Development of
Toyota Lean Combustion System,” S8=
Paper 850044, 1985,
4, Witze, P.0., "A Critical Comparison of
Wire'Angnomentzy and Laser Doppler
cimetry | for T.c. Engine
Applications," SAE Paper 600132, 1860.
5. Asanuma, T., and Obokata, 7., "Gas
Velocity Measurenents of a Motored and
Firing Engine by Laser Anenonetry,”
SAE Paper 790096, 1973.
Vafidis, C., "Influence of Induction
irl and Biston Configuration on Air
Flow ina Four-Stroke Model Engine"
Proc. of T.M.E., Vol. 198C No. 8,
1964.
+ Murakami, K., Arai, M., Hiroyasu, #.,
and Magiyana, ke, || "Flow ina
Combustion Chamber of the D.
Diesel Engine," JSAE pre-Print 6
p.135-140, 1984. “(in Japanese)
Borgnakke,’ C., Davis, G. C., and
Tabacaynski, "predictions of
In-Cylinder’ Swirl Velocity and
Turbulence Intensity for an
hanber Cup in Piston Engine,"
Paper 810224, 1981.
Kisaka, T., Shimamoto, Y., Isshiki,
Ye, and shibata, ?., "#wumerical
Simulation Method in cylinder Gas Flow
of 4 Cycle Engine" (Part 1: The
Analysis of Gas Flow in Induction
Period), 4th Joint symposium on
Internal Combustion Engines. (in
Japanese), 1984.
10. Gesman, A. D., Teui, Y. ¥., and
Watkins, A. P., "Calculation of Three
Dinensional Air Motion in Model
Engines," SAE Paper 840223, 1984,
V1. Gosnan,' A. De, aud, ¥. Ye, and
vafidis, ‘c, “Plow ih a Hodel Engine
with a Shrouded Valve - A Combined
Experimental and Computational study,”
SAE Paper 850496, 1985.
12, Matsuoka, —S.,'_-Kamimoto, :
Mochinaru, 'y., and
Urushihaza, 7.
Morita, #., “LDA Measurement and a
Theoretical’ ‘Analysis of the
Cylinder Air Motion in a Diesel
Engine," SAE Paper 850106, 1985.
13. Kondoh, T., Fukumoto, a., Ohsawa, K.,
and Ohkubo, Y., "An Assessment of a
Malti-DinenSional Numerical Method to
Predict the Flow in Internal Combus-
tion Engines," SAE Paper 850500, 1985.
14. Cloutman, L. D., Dukowiez, J. Key
Ranshaw, J. D.," "and Amsden, 2. Avy
"CONCHAS-SPRaY," Los Alamos’ National
Lab. Report, 1A-9294-Ns, 1962.
ACKNOWLEDGEMENT
We wish to particularly thanks mr.
Yoshitaka Suzuki in the Conputer Service
Section and Mr. Takao Itoh for. his
remarkable skill throughout. the LDV.
measurement,APPENDIX A
Laser Doppler velocimetry
a argon ion laser with an output of
4 watts was used in this study. The
optical arrangement of the LDV system is
Shown in Fig. Al, ‘The two-dimensional
backward scattering measurenent can be
employed with this LDV. However, the
forvard scattering LDV was used throughout
the experiment because of a good signal to
noise ratio, The measuring direction in
velocity vector of notion in a
bastion chamber was ched in turn
fron tangential component to axial
conponent. The total angle of the bean
intersection was chosen point by point to
avoid disturbance by the oraque cylinder
wall, The optical specifications are
Shows in Table Als
Ar ion laser
Bragg cell
Color separator prism
Beam expander
Front lense
Measuring volume
Pin hole
Photo multiplier tube
Fig. Al.
@2980800
optical arrangements
of LDV systen
Table Al. OPTICAL OXTA
Laser wavelength 514.5 mm
Lager source power 50 aw
Focal length of 310 mm
a focusing lens
Total angle of 6.4%, 3.2%
intersection
Intersection volume 1.29, 2.57 mm
length
Intersection volume 0.072 mm
diameter
Range of frequency 40 Mie
shifter
Frequency responce 120 whe
processor
calibration factor 9.4, 4.8 msl/tne
APPENDIX B
Application of Drop out Signa:
When the mean velocity was calculated
by using an analogue velocity signal from
19
Doppler burst signal
a tracker processor, the drop out signal
was simultaneously’ employed. "Drop Out"
Means worse conditions such as an absence
in a particle and existance of plural
particles in a measuring volume. as the
velocity signal from the tracker doesn't
Show the correct velocity under drop ow
condition, it was used to select correct
data digitalized by the crank angle pulse
fron a rotary encoder attached to a
camshaft. An example of the drop out
Signal and doppler burst signal are shown
in Fig. BI. The mean velocities
calculated by the moving average method to.
be described later in detail were obtained
under three threshold levels of 40, 70 and
percentages. The results are shown in
g.. 82, Since no difference among these
results could be seen, the threshold level
50 8 was used throughout this study.
90
[Dep Out ta Sn
Re 700 rom
Me 8a
[Sampling Inia "SEA
T ie
carer ity
Schober.
MEAN VELOCITY ms
CRANK ANGLE (deg ATOC)
E£fect of slice level
about the drop out signal
Fig. B2.
Investigation in Statistical Calculation
Tn general, there are sone methods
for statistical calculation of random
signals. According to the investigation
so far, the ensemble average method cannot
separate the turbulence fron cyclic
variations of mean velocity and. the
stationary time averaging method do not
give an instantaneous mean velocity. On
the contrary, since the moving average
method is seemed to be superior to then an
Account of its available flexibility such
as obtaining the variation of mean
Velocity and instantaneous mean velocity,
this method was used throughout this
study. The averaging cycle number forstatistical calculation was 400. As the
analogue out-put sign: 5 digitalized
with 2 0.25 deg. crank angle through each
the sampling frequency under the
of 1400 rpm operation
33.6 kHz. Measurenen
cessor was used throughout
are turbulent intensity
both processors. The data
ate shown in
As the mean velocities obtained
gnals from and track
fo have no difference during
compression and expansion stroke, only the
output signal from tracker was’ used in
this study.
Counter signal
Tracker signal
Fig. 83. Signals from tracker and counter
APPENDIX ¢
NOMENCLATURE
two-dimensional vector operator
two-dimensional fluid velocity
Wo = swirl velocity
R coordinate parameter
cylindrical coordinate
rectangular coordinate
partial mass density of
total mass density
species diffusivity
rate of change of 2 due to
evaporation or condensation
= pressure
viscous stress tensor
cylindrical viscous stress
ronentun per unit volume per
unit tine transferred from
the spray droplets to the
fluid
external force per unit mass
swirl stress vector
angular momentum per unit volune
per unit tine transferred from
spray drops to the fluid
specific internal energ:
the £1)
heat flux vector
the rate of chemical heat
release
Qs = source term associated with the
interaction between the spray
droplets and fluid
= the dimensionless stoichiometric
fficients for the rth
faction =
Xe = One mole of species x
a
of
Is
192