You are on page 1of 10

CFD Analysis of Compressible

Diego A. Arias Flow Across a Complex Geometry

Solar Energy Laboratory,
Department of Mechanical Engineering,
University of Wisconsin-Madison,
Madison, Wisconsin, 53706
e-mail: A commercial computational fluid dynamics (CFD) package was used to develop a three-
dimensional, fully turbulent model of the compressible flow across a complex-geometry
Timothy A. Shedd1 venturi, such as those typically found in small engine carburetors. The results of the CFD
simulations were used to understand the effect of the different obstacles in the flow on the
Multiphase Flow Visualization and Analysis
Laboratory, overall discharge coefficient and the static pressure at the tip of the fuel tube. It was
Department of Mechanical Engineering, found that the obstacles located at the converging nozzle of the venturi do not cause
University of Wisconsin-Madison, significant pressure losses, while those obstacles that create wakes in the flow, such as the
Madison, Wisconsin, 53706 fuel tube and throttle plate, are responsible for most of the pressure losses. This result
e-mail: indicated that an overall discharge coefficient can be used to correct the mass flow rate,
while a localized correction factor can be determined from three-dimensional CFD simu-
lations in order to estimate the static pressure at locations of interest within complex
venturis. 关DOI: 10.1115/1.2754321兴

1 Introduction for compressible flow of an ideal gas through a converging isen-

At the time of publication, over 35 million small engines are tropic nozzle. The air mass flow rate ṁa is given by
sold every year in the United States and their emissions comprise ṁa = Cd,tAt冑2␳a0共Pv,0 − Pv,t兲⌽ 共1兲
a significant percentage of total pollutants in the U.S. and world-

冉 冊
wide. As demonstrated by the automotive industry, significant re-
ductions in emissions are technologically possible, particularly 关␥/共␥ − 1兲兴关共Pv,0/Pv,t兲2/␥ − 共Pv,0/Pv,t兲共␥+1兲/␥兴 1/2
⌽= 共2兲
with the use of electronic fuel injection. However, due primarily 1 − Pv,0/Pv,t
to cost constraints, small engine manufacturers rely on small, in-
expensive carburetors to generate the fuel mixture for their en- where CD is the discharge coefficient based on the throat area At,
gines. Thus, a better understanding of carburetor performance and ␳a0 is the air density at total inlet conditions, Pv,0 is the isentropic
modeling could lead to better fuel mixture control and lower emis- stagnation pressure at the inlet of the venturi, and Pv,t is the static
sions from small engines. pressure at the venturi throat 关1兴. ⌽ accounts for the compressibil-
The equations used to represent the flow across a carburetor ity effects, where ␥ is the ratio of specific heats. These expressions
venturi are typically based on isentropic compressible flow rela- can be used for real gases by using the compressibility factor Z in
tions. The deviation from this ideal flow is corrected with a dis- the denominator of Eq. 共1兲, as used by Cornelius and Srinivas 关3兴.
charge coefficient. This discharge coefficient is influenced by Equations 共1兲 and 共2兲 may be regarded as a steady-state one-
many factors, including geometry, mass flow rate, and fluid prop- dimensional model of compressible flow across a variable area
erties 关1兴. A real carburetor venturi has details in its geometry that duct. For a given flow rate, they can be used to predict the static
create disturbances in the flow and may cause pressure losses that pressure as a function of the local duct area, assuming that all the
cause deviations from an ideal isentropic flow. Examples of these properties of the flow are uniform across the cross-sectional area.
carburetor parts are the choke plate, throttle plate, fuel tube, side As it will be shown later, this may not be true for the flow in the
passages to secondary systems, and, sometimes, an additional carburetor throat due to acceleration, as well as the presence of
concentric fuel tube in the venturi throat. Some details of typical different obstacles, such as the fuel tube.
carburetors used in small engines are shown in Fig. 1. The pres- Aside from the intake valves, the throttle plate is the largest
sure losses created by these elements reduce the mass flow rate restriction that the air encounters in its way through the engine
that could be driven through the venturi for a given pressure dif- intake system. Harrington and Bolt 关4兴 applied Eqs. 共1兲 and 共2兲 to
ference between the inlet of the venturi and the intake manifold. the throttle plate and calculated the discharge coefficient based on
In the present study, the inlet obstacles, fuel tube, and throttle the open cross-sectional area for the actual throttle plate angle.
plate were modeled with FLUENT 关2兴 in order to gain a better Based on the analysis of the different elements in the throttle plate
understanding of the flow in complex venturis. The results of this 共shaft and plate兲, Harrington and Bolt 关4兴 derived an expression
study can be used to aid in the design of devices employing ven- for the projected cross-sectional area available for the air to flow
turis, such as jet pumps, ejectors, venturi scrubbers, and industrial through

冉 冊 冋
4Ath cos共␺兲 2 a
2 Background on Prediction of Airflow in Carburetor = 1− + 共cos2 ␺ − a2 cos2 ␺0兲1/2
␲D2 cos ␺0 ␲ cos ␺

冉 冊 册
cos ␺ a cos ␺0
2.1 Zero- and One-Dimensional Studies. The simplest − sin−1 − a共1 − a2兲1/2 + sin−1 a 共3兲
cos ␺0 cos ␺
model of airflow in a carburetor venturi is based on the equations
where a = d / D, d is the throttle shaft diameter, D is the throttle
bore diameter, ␺ is the throttle plate angle, and ␺0 is the minimum
Corresponding author.
Contributed by the Fluids Engineering Division of ASME for publication in the
angle when the throttle is closed. An example of this function is
JOURNAL OF FLUIDS ENGINEERING. Manuscript received July 22, 2006; final manuscript shown in Fig. 2. A minimum area is available due to small leaks
received April 10, 2007. Review conducted by Ye Zhou. and reaches a maximum due to the blockage created by the

Journal of Fluids Engineering Copyright © 2007 by ASME SEPTEMBER 2007, Vol. 129 / 1193

Downloaded From: on 05/30/2013 Terms of Use:

engine testing and simulation: the discharge coefficient is deter-
mined experimentally on a steady flow bench at a given pressure
difference; the results are then used for all other pressure
The natural extension to the theoretical model of airflow rate in
the intake system is to use a quasi-steady-state approximation.
With this assumption, the compressible nozzle model may be ap-
plied to the pulsating flow in a single- or two-cylinder engine.
Woods and Goh 关6兴 studied the compressible flow across a throttle
plate under steady and unsteady conditions. They measured the
airflow in the intake system of a single-cylinder engine and found
that the prediction of the flow by using a quasi-steady-state as-
sumption for the discharge coefficients produced good results for
the engine operating at 530 rpm and 1000 rpm.
However, only the simultaneous solution of instantaneous mass,
momentum, and energy equations for compressible flows can cap-
ture the dynamic effects of the flow inside of intake manifolds 关7兴.
The solution of the instantaneous one-dimensional flow equations
Fig. 1 Details of carburetor parts inside the venturi
has been generally performed with two methods: the method of
characteristics 共e.g., 关8,9兴兲 and the finite difference method 共e.g.,
throttle plate shaft. 关10兴兲. Both these methods can predict the pressure and velocity
Although the steady flow assumption may seem too restrictive fields for motored engines with good agreement with experiments.
for the application of this analysis in small utility engines, it was The solution to the one-dimensional flow equations has been
used by Harrington and Bolt 关4兴 to predict the airflow through the implemented in several commercial packages, such as GT-Power
intake system of an eight-cylinder engine. In this case, the airflow 关11兴, Ricardo-Wave 关12兴, and AVL-Boost 关13兴. These packages
was nearly constant; thus, the steady flow assumption was valid. are based on the assumptions of one-dimensional flow, ideal gas
From these results 关4兴, it is possible to deduce the following: behavior, quasi-steady boundary conditions, no interaction be-
tween air and fuel, and that heat transfer, friction, and discharge
• For a given throttle plate angle, the mass flow rate increases coefficients are valid for both steady and unsteady flow 关1兴.
for a decreasing manifold pressure. The flow behaves lin-
early when the pressure difference is small 共high manifold 2.2 CFD Studies. Although the one-dimensional methods ad-
pressure兲, but the lines curve at higher pressure difference dress many important questions in relation to the transient gas
indicating the compressibility effects. dynamics, they do not provide detailed information about the flow
• The flow chokes when the intake manifold is lowered to field in specific parts of the system 关14兴. This kind of information
around the critical pressure P / P0 = 0.528. may be obtained by using computational fluid dynamics 共CFD兲.
• At large throttle angles, the flow does not reach choked However, the use of CFD for the analysis of the flow across the
conditions, even at the highest engine speeds. complex geometries in the carburetor venturi has been very
• For small throttle plate angles, the flow will be choked most limited.
of the time. Tekriwal 关15兴 performed a CFD analysis of venturis used as
flow measurement devices. FLUENT was used to model the flow
As indicated in Eq. 共1兲, the actual mass flow rate is corrected with an axisymmetric geometry; two turbulence models were
with a discharge coefficient, which must be determined experi- compared in the simulations: Renormalized Group 共RNG兲 k-⑀
mentally. Pursifull et al. 关5兴 found that the discharge coefficient is model and Reynolds stress model 共RSM兲. It was found that
a strong function of throttle plate angle and a weak function of
intake manifold pressure. The discharge coefficient varies at small • Both models predicted the pressure drop at the venturi throat
pressure differences but reaches an almost constant value for with very good agreement with experimental results, but the
lower intake manifold pressures. This result is widely used in overall pressure drop 共from the inlet to the end of the dif-
fuser兲 was better predicted by the RSM model. It was ar-
gued that this difference was caused by the assumption used
in the k-⑀ model of isotropic turbulence in all of the Rey-
nolds stresses. The use of the RSM model provided a better
accuracy at the cost of more computations.
• Inlet turbulence intensity up to 10% had no effect on the
pressure change through the venturi.
• The effect of air viscosity on the pressure drop is negligible
for the conditions tested.

Examples of other studies that have used CFD for the charac-
terization of airflow across venturi nozzles without obstacles and
under subsonic conditions are those performed by Guessous 关16兴
and Sera et al. 关17兴. CFD has also been used for choked condi-
tions, such as the study by von Lavante et al. 关18兴.
Sanatian and Moss 关14兴 used a standard k-⑀ turbulence model to
study the steady three-dimensional flow across a throttle valve in
the intake system of a two-cylinder engine. The studies were con-
ducted for two throttle-plate angles 共30 deg and 50 deg兲. The re-
sults were given in terms of the mean velocity and turbulence
Fig. 2 Projected open area for airflow across throttle plate: intensity profiles along the intake pipe that houses the throttle
closed= 0 deg, open= 90 deg valve. Comparisons to visualization experiments of stream lines

1194 / Vol. 129, SEPTEMBER 2007 Transactions of the ASME

Downloaded From: on 05/30/2013 Terms of Use:

Fig. 3 Mesh size sensitivity of the mass flow rate
Fig. 4 Comparison of discharge coefficient for a Briggs &
Stratton carburetor as function of throttle plate angle
and limited hot-wire anemometer measurements showed that the
simulations could give an good indication of the flow fields in this
A more complete three-dimensional analysis of the flow across side of a blower. Once the throttle plate is set to a known position,
a throttle valve was performed by Alsemgeest et al. 关19兴. They the pressure in the outlet plenum is adjusted until a recommended
used a standard k-⑀ model to study the unsteady compressible static vacuum pressure is achieved. The volumetric airflow rate at
flow across a throttle valve under constant boundary conditions. these conditions is recorded. Figure 4 shows the measured dis-
They found that the obstruction created by the throttle plate pro- charge coefficient of a Briggs & Stratton carburetor as function of
duced vortex shedding behind it at a frequency of 200– 600 Hz. throttle plate angle.
The only known work that has used CFD for the characteriza- In order to validate the simulations with experimental data, a
tion of the flow across the carburetor was done by Wu et al. 关20兴. Briggs & Stratton carburetor venturi was modeled in FLUENT. The
However, the carburetor was represented as a two-dimensional carburetor venturi had an inlet diameter of 25 mm, a throat diam-
channel, where the fuel tube was a large obstacle in the flow field. eter of 12 mm and exit diameter of 20 mm. This venturi had inlet
The only results shown in this work were the static pressure drop obstacles, a fuel tube, and a throttle plate. The inlet boundary
along the axis of the carburetor. conditions in FLUENT were set to the laboratory conditions 共T0
= 293 K and P0 = 1 atm兲 and the outlet boundary condition to the
3 Three-Dimensional CFD Study outlet pressure in the low-pressure plenum in the flow bench
共Pout = 94.5 KPa, or a static pressure difference of −28 in H2O兲.
3.1 General Characteristics of the Numerical Model. A Four throttle plate angles were simulated: 90 deg 共wide open
three-dimensional model of a carburetor venturi was generated in throttle兲, 75 deg, 60 deg, and 45 deg. Figure 4 compares the dis-
GAMBIT and used in FLUENT to study the effect of different venturi charge coefficient calculated from the simulations to the experi-
parts on the flow field. The geometry was discretized with an mental results. It was found that the FLUENT-calculated discharge
unstructured tetrahedral mesh, with a refined mesh near the ven- coefficient was within 10% of the experimental value and closely
turi throat. The RNG k-⑀ turbulence model was used, with stan- followed the trends of the experimental results. The smaller values
dard wall functions for near-wall treatment. The discretization of CD predicted by FLUENT indicated that the numerical model
scheme used was second order in space. The pressure-velocity overpredicted the pressure losses.
coupling was performed with the SIMPLE algorithm. The good agreement with experimental results over the range of
The convergence criteria were set to a maximum residual equal throttle plate angle indicates that the characteristics of the simu-
to 1 ⫻ 10−6 for the energy equation and to 1 ⫻ 10−4 for the other lations are appropriate to capture the flow conditions in the car-
equations. The convergence of the solution with respect to mesh buretor venturi with all of the inner parts. This was an encourag-
size was assessed by running the simulations with three meshes of ing result, as the range of throttle plate angles created complex
different size. The result for an infinitely small mesh was approxi- details in the geometry, such as small passages when it is in a
mated using the Richardson extrapolation 关21,22兴. Figure 3 shows more closed position.
the effect of mesh size on the mass flow rate, at a representative The CFD results were used for assessing the details of the flow,
flow condition in the venturi without obstacles. According to Fig. the values of the discharge coefficients, and localized values of the
3, the intermediate mesh size was selected for the following simu- flow variables; specifically, the static pressure at the tip of the fuel
lations because it ensured a mesh independent solution. tube. The following sections present a systematic study of the
The inlet boundary condition was defined with the isentropic effect of different carburetor parts. First, the Briggs & Stratton
stagnation pressure and temperature, and the outlet boundary con- carburetor venturi was modeled without obstacles. Second, the
dition was defined with the outlet static pressure. An ideal gas inlet obstacles were added, and then the fuel tube was added to the
model was used in order to take into account the compressibility geometry. Finally, the effect of throttle plate angle was studied.
of the airflow.
3.3 Venturi Without Obstacles. Figures 5共a兲 and 5共b兲 show
3.2 Comparison to Experimental Results. Carburetor ven- the static pressure and Mach number for the compressible airflow
turis are typically tested on steady-state flow benches, and the across the carburetor venturi without obstacles in the flow. Figure
experimental discharge coefficient is found as a function of the 5 shows that the static pressure is almost uniform in the radial
throttle plate angle. On the steady-state flow bench, the inlet of the direction, with an exception occurring at the venturi throat, where
carburetor is open to the laboratory conditions and the outlet is the static pressure changes next to the wall. The velocity increases
connected to a low-pressure plenum, created by the low-pressure at the converging nozzle and then separates from the wall at the

Journal of Fluids Engineering SEPTEMBER 2007, Vol. 129 / 1195

Downloaded From: on 05/30/2013 Terms of Use:

Fig. 5 Steady airflow across carburetor venturi without obstacles: „a… static pressure „in Pascal…, „b… Mach number, „c…
turbulent kinetic energy „in meters squared per seconds squared…, and „d… gage total pressure „in Pascal…

diffuser in the region of adverse pressure gradient. The velocity and effective area for the flow downstream of the venturi throat.
field then resembles a free jet entering a constant pressure reser- The static pressure shows a similar behavior to the previous
voir, although in this case the jet interacts with the surrounding cases: uniform pressure in the cross section everywhere but in the
wall. venturi throat. At this location, there is a different static pressure
The turbulent kinetic energy is shown in Fig. 5共c兲. It shows that in the radial direction. In addition, the sharp leading edge of the
the regions of high turbulence are those next to the walls of the fuel tube creates a separation region, which results in a lower
diffuser and that they extend downstream of the carburetor ven- pressure at the tip of the fuel tube. Downstream of the fuel tube,
turi. This result is similar to the turbulence intensity in a free jet, the static pressure is almost uniform in the radial and axial
where the highest turbulence region occurs in the region of veloc- directions.
ity transition from the high-velocity zone to the quiescent air 关23兴. The velocity field shows the wake region created by the fuel
The final effect is the reduction of isentropic stagnation pressure tube. The final effect of the fuel tube on the airflow is to reduce
at the outlet of the venturi, shown in Fig. 5共d兲. This result indi- the effective area used by the flow behind the venturi. The size of
cates that, in a venturi without inlet obstacles, the converging the wake region is increased with the length of the fuel tube. The
nozzle does not cause noticeable losses; it is the separation at the wake region increased when the fuel tube was modeled with a
diffuser and the turbulence at the free shear zone of the jet that length equal to 1 / 2 of the throat diameter. This wake zone may be
causes the pressure losses. responsible for fuel puddling after the carburetor: once a fuel
3.4 Effect of Inlet Obstacles. It was found that the presence droplet is captured in this region, there is no momentum to drive
of the inlet obstacles did not create a noticeable effect on the flow it downstream of the carburetor.
field, as shown in Fig. 6. As these obstacles are located in the The turbulent kinetic energy field indicates that the nature of
converging zone of the venturi, the favorable pressure gradient the pressure losses are quite different when the fuel tube is present
keeps the velocity profile attached to the walls. The inlet obstacles in the carburetor. The wake zone has the region of highest turbu-
affect the convergent flow but do not cause any wake or free shear lence intensity; the turbulence intensity next to the walls of the
region. In the same manner as in the venturi without obstacles, the diffuser is almost negligible in comparison. The effect of turbu-
high turbulence zones and pressure losses are located in the sepa- lence is seen in the total pressure: the wake zone is also the region
ration regions of the diffuser at the free shear region downstream where the isentropic stagnation pressure is reduced most
of the venturi throat. significantly.
Because the analysis was performed with the same pressure
3.5 Effect of Fuel Tube. The presence of the fuel tube re- difference for all of the venturi geometries, the effect of the fuel
sulted in a strong change in the flow field in the carburetor ven- tube is a reduced mass flow rate of air. If the analysis were per-
turi. Figure 7 shows the effect of a fuel tube with length equal to formed at constant mass flow rate, the mass conservation in the
3 mm and diameter equal to 3 mm. The length was equivalent to reduced area would have produced a higher velocity and lower
1 / 4 of the throat diameter. The presence of the fuel tube produced static pressure.
a reduced cross-sectional throat area and a large wake zone behind These images allow for a better understanding of the complex
it. This wake completely changes the nature of the pressure losses interaction between the fuel tube and the airflow: in current car-

1196 / Vol. 129, SEPTEMBER 2007 Transactions of the ASME

Downloaded From: on 05/30/2013 Terms of Use:

Fig. 6 Steady airflow across carburetor venturi with inlet obstacles: „a… static pressure „in Pascal…, „b… Mach number, „c…
turbulent kinetic energy „in meters squared per seconds squared…, and „d… Gage total pressure „in Pascal…

Fig. 7 Steady airflow across carburetor venturi with inlet obstacles and fuel tube: „a… static pressure „in Pascal…, „b… Mach
number, „c… turbulent kinetic energy „in meters squared per seconds squared…, and „d… gage total pressure „in Pascal…

Journal of Fluids Engineering SEPTEMBER 2007, Vol. 129 / 1197

Downloaded From: on 05/30/2013 Terms of Use:

Fig. 8 Steady airflow across carburetor venturi with fuel tube, inlet obstacles, and throttle plate at 90 deg: „a… static pressure
„in Pascal…, „b… Mach number, „c… turbulent kinetic energy „in meters squared per seconds squared…, and „d… gage total
pressure „in Pascal…

buretor designs, a fuel tube extending into the venturi throat, be- shaft, and screw are responsible for the largest turbulence regions
yond the throat wall, is necessary. It brings the fuel flow near the in the flow. This turbulence results in a reduced isentropic stagna-
centerline of the venturi, which is intended to help generate an tion pressure.
even distribution of the fuel droplets in the flow field. A fuel tube
that does not extend beyond the wall would not prevent the fuel 3.7 Effect of Throttle Plate at Different Angles. The effect
flow from staying next to the wall. But the fuel tube itself also of throttle plate angle on the flow field in the carburetor venturi is
completely disturbs the airflow, increasing the pressure losses and, a strong function of the angle at which it operates. As the throttle
therefore, decreasing the mass flow rate at a given pressure drop. plate closes, the mass flow rate is reduced for the given pressure
difference across the entire carburetor. Figures 9 and 10 show the
3.6 Effect of Throttle Plate at Wide Open Angle. Besides throttle plate at 75 deg and 60 deg, respectively. In these two
the intake valves, the throttle plate is the largest restriction to the cases, there are increased pressure losses and increased asymme-
airflow in the intake manifold. The carburetor venturi was next try in the flow. A conclusion from this asymmetry would be an
simulated with the inlet obstacles, fuel tube, and throttle plate. The increased tendency for droplets to impact the throttle plate and
throttle plate was modeled as close as possible to the physical one side of the intake manifold.
model: it was composed of the axis rod, plate, and screw.
When the throttle plate angle is further closed to 45 deg, the
Figure 8 shows that, even when it is aligned with the flow, the
flow is different from the previous cases 共Fig. 11兲. The mass flow
presence of the throttle plate causes a large effect on the flow
field, increasing the wake zone and producing asymmetric fea- rate has decreased significantly, and the flow fields resembled
tures in the flow. those in an orifice: the static pressure is uniform everywhere up-
The static pressure field does not change significantly from the stream of the throttle plate 共even at the venturi throat兲, it decreases
previous cases; additional stagnation points are created by the suddenly next to the reduced area created by the plate, and then is
leading edge of the throttle plate, shaft, and screw. However, the uniform downstream of the plate. The velocity field shows the
velocity field is greatly influenced by the throttle plate: the high- same characteristics of an orifice: small magnitude everywhere
speed stream created by the fuel tube now encounters a large but in the region next to the reduced area opening.
obstacle just downstream. The wakes created by the fuel tube and This is a case where it is very likely that the main fuel system
the throttle plate interact between them, producing a vortex shed- of the carburetor is not active; the airflow does not create a pres-
ding seen in both planes of Fig. 8. sure at the venturi throat that is low enough as to drive fuel flow
The flow field shows that, even at wide-open conditions, the from the fuel tube. Under these conditions, the idle system could
screw creates an asymmetry in the flow. The highest velocity on be activated. An interesting feature of these plots is that the asym-
the right side of the throttle plate 共seen on the top view兲 shows metry of the flow would make it different if the idle ports are
that this side of the flow would experience a higher airflow and, it located in one side of the carburetor or the other. This asymmetry
may be inferred, more fuel droplets. could also be a large player in the uneven fuel distribution in the
The turbulent kinetic energy indicates that the throttle plate, intake manifold.

1198 / Vol. 129, SEPTEMBER 2007 Transactions of the ASME

Downloaded From: on 05/30/2013 Terms of Use:

Fig. 9 Steady airflow across carburetor venturi with fuel tube, inlet obstacles, and throttle plate at 75 deg: „a… static pressure
„in Pascal…, „b… Mach number, „c… turbulent kinetic energy „in meters squared per seconds squared…, and „d… gage total
pressure „in Pascal…

Fig. 10 Steady airflow across carburetor venturi with fuel tube, inlet obstacles, and throttle plate at 60 deg: „a… static pressure
„in Pascal…, „b… Mach number, „c… turbulent kinetic energy „d… gage total pressure „in Pascal…

Journal of Fluids Engineering SEPTEMBER 2007, Vol. 129 / 1199

Downloaded From: on 05/30/2013 Terms of Use:

Fig. 11 Steady airflow across carburetor venturi with fuel tube, inlet obstacles, and throttle plate at 45 deg: „a… static pressure
„in Pascal…, „b… Mach number, „c… turbulent kinetic energy „d… gage total pressure „in Pascal…

4 Discussion dimensional representation of the total and static pressure behav-

ior in the carburetor venturi with inlet obstacles, fuel tube, and
4.1 Calculation of the Overall Discharge Coefficient. In ad- throttle plate. The total pressure shows a larger decrease after the
dition to the qualitative analysis of the flow fields in the different fuel tube and throttle plate, and it shows a segregation of the flow,
carburetor venturi configurations, a quantitative comparison was i.e., there are regions of different values of total pressure. Finally,
performed by calculating the discharge coefficient from Eqs. 共1兲 the flow reconverges to a one-dimensional model. The same
and 共2兲. These equations may be regarded as a one-dimensional trends are seen in the static pressure. Table 1 shows a summary of
model of the steady-state compressible flow across a variable-area the different geometries used in this study and presents the FLU-
duct. In an ideal isentropic flow, for a given mass flow rate and ENT results in terms of mass flow rate ṁ, the overall discharge
inlet conditions, T0 and P0, the information about the local area A coefficient CD, and the localized discharge coefficient calculated
is enough to solve for the local static pressure. This is the classical at the fuel tube CD,ft.
behavior of an isentropic converging-diverging nozzle, shown in The results of the overall discharge coefficient as a function of
Fig. 12共a兲 as dotted lines; the total pressure remains constant and
the static pressure depends on the local velocity. As the venturi
studied has an outlet of different diameter than the inlet, the static
Table 1 Effect of carburetor parts on the discharge coefficient
pressures are different. The real flow in this venturi is shown as
solid lines: friction losses reduce the total pressure as well as the Fuel tube Throttle plate ṁ
static pressure. Case 共mm兲 Inlet obstacle 共deg兲 共kg/s兲 CD CD,ft
When the different obstacles are considered in the flow, the
one-dimensional model may seem too simplistic for all the infor- 1 - - - 0.0222 0.630 1.14
mation seen with the CFD analysis. Figure 12共b兲 shows a one- 2 3 共1/4Dthroat兲 - - 0.0177 0.503 1.03
3 6 共1 / 2Dthroat兲 - - 0.0169 0.481 1.02
4 - ✓ - 0.0217 0.617 1.08
5 3 共1 / 4Dthroat兲 ✓ - 0.0177 0.503 1.05
6 6 共1 / 2Dthroat兲 ✓ - 0.0164 0.468 1.03
7 - - ✓共90兲 0.0200 0.569 1.09
8 3 共1 / 4Dthroat兲 - ✓共90兲 0.0179 0.509 0.99
9 6 共1 / 2Dthroat兲 - ✓共90兲 0.0167 0.474 1.06
10 - ✓ ✓共90兲 0.0192 0.547 1.09
11 3 共1 / 4Dthroat兲 ✓ ✓共90兲 0.0178 0.505 1.00
12 6 共1 / 2Dthroat兲 ✓ ✓共90兲 0.0166 0.473 1.00
13 3 共1 / 4Dthroat兲 ✓ ✓共75兲 0.0163 0.463 0.98
14 3 共1 / 4Dthroat兲 ✓ ✓共60兲 0.0119 0.339 0.97
Fig. 12 One-dimensional model of carburetor venturi: „a… clear 15 6 共1 / 2Dthroat兲 ✓ ✓共45兲 0.0066 0.189 0.96
venturi and „b… carburetor venturi

1200 / Vol. 129, SEPTEMBER 2007 Transactions of the ASME

Downloaded From: on 05/30/2013 Terms of Use:

Fig. 13 Effect of carburetor parts on the discharge coefficient
of the carburetor venturi

fuel tube length are plotted in Fig. 13. The results of those cases in Fig. 14 Discharge coefficients
which the inlet obstacles are present, but there is no fuel tube 共fuel
tube length equal to zero兲 show that the inlet obstacles reduce the
discharge coefficient a small amount. The throttle plate, when it is CD, Eqs. 共1兲 and 共2兲 can be used with the actual throat area and
wide open, is responsible for a larger decrease in the discharge CD,ft ⬇ 1 to calculate the static pressure on top of the fuel tube.
coefficient. The largest effect on the discharge coefficient is cre-
ated by the fuel tube. In addition, when the fuel tube is 3 mm 5 Conclusions
long, the discharge coefficient is the same for all of the different
geometries. Furthermore, when the length of the fuel tube length The key findings of this study can be summarized as follows:
is doubled, the discharge coefficient is further reduced and gives • Obstacles located in the converging zone of the venturi do
the same value among the different geometries. not create a noticeable effect on the flow field because the
This quantitative assessment of the discharge coefficient is in favorable pressure gradient prevents separation and wall
agreement with the qualitative description of the flow inside the shear stress does not significantly determine the flow in the
carburetor venturi. The pressure losses in the venturi without ob- nozzle.
stacles are due to the free shear generated at the jet leaving the • The pressure loss in a venturi without objects in the throat or
venturi throat. The presence of the intake obstacles create a very expansion zone is due primarily to the separation of flow
small effect on the discharge coefficient because they affect the from the wall during the expansion. The dissipation of flow
flow in the region of favorable pressure gradient. The fuel tube energy in the free shear zone of the resulting turbulent jet
and throttle plate are the most important elements affecting the causes the reduction in total pressure.
flow: the wake created by these elements is responsible for the • Obstacles in the throat and diffuser, such as the fuel tube and
generated turbulence and the corresponding pressure losses. throttle plate, create wakes and are responsible for most of
4.2 Calculation of a Localized Discharge Coefficient. In ad- the pressure losses in a venturi with complex geometries.
dition to getting the information about the overall discharge coef- • Significantly, once the mass flow rate is corrected using an
ficient to correct the mass flow rate across the carburetor venturi overall discharge coefficient, the knowledge of the actual
given a pressure drop, it is possible to calculate a local discharge cross-sectional area at the venturi throat is enough to calcu-
coefficient that may be used to get the static pressure at a particu- late the static pressure at the tip of the fuel tube.
lar location in the carburetor venturi. For example, it is of great
interest to use the information from the CFD simulations to set the The results of these simulations indicate that CFD simulations
appropriate boundary condition at the tip of the fuel tube in a fuel can be used to understand the nature of the flow field in venturis
flow network. The discharge coefficient at the tip of the fuel tube of complex geometry and to find quantitative information that can
CD,ft was calculated using the average static pressure at the tip of be used as boundary conditions for additional systems coupled to
the fuel tube and the mass flow rate. This discharge coefficient the venturi. Future work should focus on the analysis of the static
was calculated based on the actual cross-sectional area at the ven- pressure at different inlet obstacles, as well as next to the throttle
turi throat Avt. The last column in Table 1 shows the results for all plate, in order to improve on the design of flow systems incorpo-
of the different geometries studied. It can be seen that CD,ft ⬇ 1. rating complex venturis.
This result indicates that the assumption of isentropic flow is valid
for the converging side of the carburetor venturi. Acknowledgment
Figure 14 shows the overall discharge coefficient for the differ- The authors would like to acknowledge the support provided by
ent geometries that had a fuel tube 3 mm long. It shows the strong the Wisconsin Small Engine Consortium.
relationship between the overall discharge coefficient and the
throttle plate angle. However, the discharge coefficient calculated
for correcting the static pressure at the tip of the fuel tube remains Nomenclature
almost constant and equal to 1, even in these cases. Therefore, the A ⫽ cross sectional area 共m2兲
information required to model the actual static pressure in the tip CD ⫽ discharge coefficient
of the fuel tube is only the actual cross-sectional area at the ven- D ⫽ diameter 共m兲
turi throat. Once the mass flow rate is corrected using the overall f ⫽ friction factor

Journal of Fluids Engineering SEPTEMBER 2007, Vol. 129 / 1201

Downloaded From: on 05/30/2013 Terms of Use:

g ⫽ gravitational constant 共m / s2兲 Fluid Flow in Single Cylinder Engine Carburetors,” SAE Technical Report No.
h ⫽ height 共m兲 关11兴 Gamma Technologies, 2001, GT-Power: User’s Manual and Tutorial, Gamma
k ⫽ parameter for sensitivity analysis Technologies, Westmont, IL.
km ⫽ pressure loss coefficient 关12兴 Ricardo Software, 2006, Wave,
L ⫽ pipe length 共m兲 关13兴 AVL, 2006, Boost,
关14兴 Sanatian, R., and Moss, J. B., 1989, “Computer Simulation of Induction Flows
ṁ ⫽ mass flow rate 共kg/s兲 in Spark Ignition Engines,” Proc. of the International Centre for Heat and
P ⫽ pressure 共N / m2兲 Mass Transfer, Hemisphere Publ. Corp, Dubrovnik, Croatia, pp. 275–289.
v ⫽ average velocity 共m/s兲 关15兴 Tekriwal, P. K., 1996, “Pressure Drop Calculations and Measurements in
Converging-Diverging Nozzles,” Technical Information Series 96CRD157, GE
Research & Development Center.
References 关16兴 Guessous, L., 2003, “Theaching CFD: Combining Experiments and Practical
Projects With Numerical Simulations,” Proceedings of the ASME FEDSM’03,
关1兴 Heywood, J., 1988, Internal Combustion Engine Fundamentals, McGraw-Hill,
ASME, New York, pp. 999–1004.
New York.
关2兴 Fluent, 1999, FLUENT 6.2 User’s Guide, Fluent, Inc., Lebanon, NH. 关17兴 Sera, M. A., Bakar, R. A., and Leong, S. K., 2003, “CNG Engine Performance
关3兴 Cornelius, K. C., and Srinivas, K., 2004, “Isentropic Compressible Flow for Improvement Strategy Through Advanced Intake System,” SAE Technical Re-
Non-Ideal Gas Models for a Venturi,” ASME J. Fluids Eng., 126共2兲, pp. port No. 2003-01-1937.
238–244. 关18兴 von Lavante, E., Zachcial, A., Nath, B., and Dietrich, H., 2000, “Numerical
关4兴 Harrington, D. L., and Bolt, J. A., 1970, “Analysis and Digital Simulation of and Experimental Investigation of Unsteady Effects in Critical Venturi
Carburetor Metering,” SAE Technical Report No. 700082. Nozzles,” Flow Meas. Instrum., 11共4兲, pp. 257–264.
关5兴 Pursifull, R., Kotwicki, A. J., and Hong, S., 2000, “Throttle Flow Character- 关19兴 Alsemgeest, R., Shaw, C. T., Richardson, S. H., and S., P., 2000, “Modeling
ization,” SAE Technical Report No. 2000-01-0571. the Time-Dependent Flow Through a Throttle Valve,” SAE Technical Report
关6兴 Woods, W. A., and Goh, G. K., 1979, “Compressible Flow Through a Butterfly No. 2000-01-0659.
Throttle Valve in a Pipe,” Proc. Inst. Mech. Eng., 193, pp. 237–244. 关20兴 Wu, B., Feng, Y., and Liu, J., 1997, “Numerical Simulation of Gas-Liquid
关7兴 Winterbone, D. E., and Pearson, R. J., 1999, Design Techniques for Engine Two-Phase Flow in Motorcycle Carburetor,” Proceedings of the International
Manifolds, SAE International, Warrendale, PA. Symposium on Multiphase Flow, ISMF, Beijing, China, pp. 271–275.
关8兴 Benson, R. S., Baruah, P. C., and Sierens, R., 1974, “Steady and Non-Steady 关21兴 Roache, P. J., 1994, “Perspective: A Method for Uniform Reporting of Grid
Flow in a Simple Carburettor,” Proc. Inst. Mech. Eng., 188共53兲, pp. 537–548. Refinement Studies,” ASME J. Fluids Eng., 116共3兲, pp. 405–413.
关9兴 Zhu, Y., and Reitz, R. D., 1999, “A 1-D Gas Dynamics Code for Subsonic and 关22兴 Celik, I., and Karatekin, O., 1997, “Numerical Experiments on Application of
Supersonic Flows Applied to Predict EGR Levels in a Heavy-Duty Diesel Richardson Extrapolation With Nonuniform Grids,” ASME J. Fluids Eng.,
Engine,” Int. J. Veh. Des., 22共3兲, pp. 227–252. 119共3兲, pp. 584–590.
关10兴 Bajema, D. L., and Gatecliff, G. L., 1978, “Prediction and Measurement of 关23兴 Hinze, J. O., 1975, Turbulence, McGraw-Hill, New York.

1202 / Vol. 129, SEPTEMBER 2007 Transactions of the ASME

Downloaded From: on 05/30/2013 Terms of Use: