SPECIAL INVESTIGATION : DESIGN OF AN INTAKE BELLMOUTH

Gordon P. Blair, Senior Associate, Prof Blair & Associates and

W. Melvin Cahoon, Senior Engineer-Specialist, Volvo Penta of the Americas,
discuss how to optimise the design of an engine air intake bellmouth and the
use of advanced software as an aid to the process

Best
bell
T

Fig.1 Measurement of Discharge Coefficients

of the pressure in all of the CFD computation cells across the duct at
the Ps measuring station and is not simply the atmospheric pressure

tmdot (g/s), through the pipe area Ap, at the experimental pressure

divided by the plenum pressure. This average computation pressure Ps

ratio PR, i.e., Po/Ps, using some such theory as the St Venant

corresponds precisely with that obtained in an actual experiment by

equation, or other subsonic nozzle theory. That researcher then

the manometer pressure gauge shown in Fig.1.

conventionally quoted the discharge coefficient CD as the ratio of the

measured to the theoretical mass flow rate, i.e., mdot/tmdot.
A useful number, maybe, but one that is well-nigh useless for

Further evidence is given in Fig.3 by the computed temperature

contours for the same process with a drop of 8 degC to the vena
contracta and a recovery of 3 degC by the end of the pipe. The most

application into an unsteady flow simulation of the flow in a real

significant evidence is provided by Fig.4, showing nozzle-like flow

engine, assuming that one expects accuracy from that computation.

from zero velocity at the entry to a high particle velocity (as Mach

Some readers may well be alarmed to read that you can still pay

number) of 0.3 at the vena contracta, a still region surrounding it, and

megabucks for a theoretical engine simulation that precisely uses that

a reduction of velocity with diffusion to the pipe exit. That being the

particular approach. We will not belabour you with a full description

case, then that is how it must be theoretically analysed in order to

of why there is a correct/incorrect way to derive CD values, which

derive a realistic discharge coefficient.

he design of a bellmouth at the end of the intake tract of

of air is sucked through it into the tank by a vacuum pump. Typically,

are accurately applicable within an engine simulation for it has been

a reciprocating internal combustion engine is not a topic

most production/commercial rigs like this will induce a tank pressure

thoroughly covered already [2,3].

that has ever occupied much space within the pages of the

some 28 inches of water below atmospheric pressure, which is a

technical literature. One could come to the not unnatural

pressure ratio, PR, of some 1.07. As you will find, the numerical value

conjunction with Fig.1. In Fig.1, the illustrated theoretical contention

contracta of area Ac, to be followed by a non-isentropic diffusing flow

conclusion that it cannot be a topic of any real significance. That

of the discharge coefficient is a considerable function of pressure ratio

is that the flow will form a vena contracta of area Ac inside the

with entropy gain and pressure recovery to a pressure Ps at the full

viewpoint, right or wrong, is very much at odds with the efforts made

and, as many pipe end boundaries are exposed to pressure ratios up

pipe somewhat less than the full pipe area Ap. The actual discharge

pipe area Ap, the upshot of which is a computed value of mass flow

by the designers of the nacelles for aircraft gas-turbine engines who

to the sonic flow condition (where PR is virtually 2), a rig which can

coefficient CD is then defined as Ac/Ap and a theoretical analysis must

rate tmdot (g/s) which must correspond precisely with the measured

put much experimental and theoretical effort into the design shape of

generate a maximum pressure ratio of just 1.07 is just not adequate.

be created to compute that Ac value at any given pressure ratio [2,3].

value of mass flow rate mdot (g/s). The actual [2,3] discharge

the leading edge of their engine pods.

However, as an intake bellmouth is normally exposed to pressure

Speaking personally (writes Blair), I have always been curious

The correct way to derive CD values is exemplified in Figs.2-4, in

To illustrate the reality of this flow regime, Fig.2 shows the outcome

The theoretical situation is sketched in Figs.1 and 2. The theory, see

the temperature-entropy diagram as an inset to Fig.2, must prescribe
isentropic nozzle flow from the opening pressure Po through a vena

coefficient CD is then calculated as Ac/Ap.

ratios of 1.1 or less, this type of commercially-available experimental

of a computation by the FLUENT CFD code for the case of inflow into

about the proper design method for intake bellmouths, indeed I have

rig would suffice for that purpose. At The Queens University of Belfast,

the sharp-edged plain pipe from the atmosphere (at 1.0 atm and 25

Firstly, the theoretical equations are all non-linear polynomial

been known to dangerously pontificate about it, dangerously in

at an earlier point in history, we had quite superb experimental

degC or 298 K). The Fig.2 shows the computed contours of pressure

functions of pressure, temperature, density, and particle velocity.

the sense that my real experimental or theoretical knowledge of that

facilities and could measure most pipe end boundary conditions up to

for the flow process. That there is indeed a vena contracta is evident

No single solution is a direct solution, but the theoretician must

design process is dangerously inadequate. Behind the writing of this

the sonic threshold PR values approaching 2.0 [2,3].

by the pressure drop from the entry at Po to Pc followed by pressure

This is neither a simple nor a straightforward computation process.

continuously vary the value of Ac in the computation until a unique

Traditionally, in the literature, one measured the mass flow rate,

recovery to Ps at the pipe exit. It is worth mentioning that the pressure

value of Ac produces precisely the measured values of Ps and mdot.

(CFD) and with the expert efforts of my co-author using the FLUENT

as seen in Fig.1, and noted that as mdot (g/s). Some rather carelessly

ratio PR is indeed Po/Ps and that Ps value is determined as an average

The iterative process to get there is not for the mathematically faint-

code [1], not only can real design information be provided on the

computed it merely as a volume flow rate. In the particular case of

topic but also our mutual curiosity has been satisfied. We present this

the bellmouth, the researcher then computed a theoretical mass flow,

design systems.

DERIVATION OF DISCHARGE COEFFICIENTS

The effectiveness of the flow regime at any boundary at the end of a
pipe in an engine is expressed numerically as a discharge coefficient,
i.e., a Coefficient of Discharge, or CD. In history, and even today, they
were/are measured experimentally using a steady flow rig, much as
shown in Fig.1.
The pipe-end boundary under examination, in this case an intake
bellmouth, is placed before a settling tank/plenum and a steady flow

I have been
known to dangerously
pontificate about
intake bellmouths

Fig.2 Thermodynamics of flow into a plain pipe end

t
t

paper, with the modern availability of computational fluid dynamics

here both for your interest and for numerical assimilation into your

34

This is neither a
simple nor a
straightforward
computation process

Fig.3 Temperature flow profiles into a plain pipe end

35

SPECIAL INVESTIGATION : DESIGN OF AN INTAKE BELLMOUTH

Fig.7 Velocity flow profiles into a bellmouth pipe end

Fig.4 Velocity flow profiles into a plain pipe end

hearted and it is little wonder that many a major engine simulation

difficult to separate them.

package supplier has shied away from this, the only approach which

Fig.6 Velocity flow profiles into a radiused pipe end

will produce accuracy of engine simulation when the attained

in CD for all profiled bellmouths is expressed as a percentage over

CD values are re-employed to help compute pipe end boundary

The elliptical
profile comes out
as the winner over
the aerofoil profile

that for the simple radius bellmouth. It becomes apparent that the

conditions [2].

THE BELLMOUTH DESIGNS TO BE ANALYSED

In Fig.5 is sketched, to scale, the bellmouths which will be analysed by
the FLUENT CFD software. The first is a simple semi-ball wrap-round

Fig.5 Nomenclature and shape of various bellmouths

radius installed at the end of the pipe. The second is a bellmouth with
an aerofoil profile (NACA type) and the third is a bellmouth with an

more sophisticated bellmouth case with an elliptical profile, the CD is

elliptical profile [5].

0.743 and the measured mass flow rate is 36.15 g/s. The fundamental

All bellmouths are characterised by their basic data for Type,

message is that there is a considerable benefit in either CD (27%) or

improvement in CD is very much a function of the entry diameter Di

and is less of a function of either the profile or the length L, which
rationale is echoed by the colour and symbol coding of the several
graphs. This is but a small selection of all of the bellmouths studied
but showing them all would merely provide further confusion, not
enhanced clarity.
While there is not much in it, the elliptical profile comes out as the
winner over the aerofoil profile. In the all-important pressure ratio
PR range up to 1.1 one can conclude that the best bellmouth has an

length L, exit diameter De, entry diameter Di, and entry corner radius

mass flow rate (16%) by the addition of even a simple radius at a pipe

Rc. The Type can be a sharp edged plain pipe (PP), a simple radius

end to make a bellmouth, but the gain in CD above that simplicity to

that is not really a major issue as the instantaneous pressure ratio at

In design terms, one can usefully conclude that short and fat is

(RAD), an aerofoil profile bellmouth (AER), or an elliptical profile

an optimum may be only 4% more.

an actual engine bellmouth during the peak of reflection of the intake

best with an optimum length criterion L of one diameter De, and

pulse will rarely exceed 1.1. However, even at low pressure ratios

an optimum entry diameter Di of some 2.13 times the exit diameter

bellmouth (ELL). A wide range of dimensions for all such bellmouths

advantage in CD terms of some 3.5% over the simplest bellmouth.

were tested and most of the more significant ones are reported upon

THE MEASURED MASS FLOW RATE

the CD values will vary by 20% over a PR range from 1.04 to 1.1 and

De, and with an elliptical profile. Although the investigations are not

below. Before that point in the discussion, look at Figs.6 and 7, which

In the discussion thus far, there is reference to a measured mass flow

therefore cannot be considered as a constant. On this evidence, the

presented here, the corner radius Rc can be usefully designed as 0.08

show the computed Mach number (particle velocity) plots for the

rate (mdot) into the bellmouth when actually it is really referring to

FLUENT code can be trusted to provide us with an accurate prediction

times the entry diameter Di.

simple radius (as RAD-46-23-35-6 of Fig.5) and the ellipse profile (as

a mass flow rate as computed by the computational code FLUENT

of air flow rates into intake bellmouths.

ELL_23-23-49-3 of Fig.5). The simple radius in Fig.6 shows less of the

modelling the bellmouth attached to a settling tank as shown in

pronounced vena contracta so evident in Fig.4 for the plain pipe, but

Fig.1. In short, the CFD code is modelling the intake bellmouth and

for the same plain pipe and the same simple radius bellmouth. Both

the elliptical profile in Fig.7 has almost no vena contracta so smooth is

the entire apparatus as a replica for an actual experiment with real

are also a function of pressure ratio and here the difference between

the flow entry. A more fundamental message, reflecting the increasing

hardware instead. Is this justified?

them remains at a near constant 16% at any given PR value. Lau [3]

At The Queens University of Belfast (QUB), much experimental

also quotes measured mass flow rates for the plain pipe and the simple

case, the CD is 0.5672 and the measured mass flow rate is 30.023 g/s.

work was conducted in this area (2,3) and one series of experiments

6 mm radius and those calculated here by CFD agree very closely with

In Fig.6, where there is a simple radius as the bellmouth, the CD is

did measure the inflow of air into a plain ended pipe and a simple

those measured, as seen in Fig.9.

now 0.719 and the measured mass flow rate is 34.83 g/s. In Fig.7, the

radiused bellmouth, the physical dimensions of which were identical

to those described here as PP-46-23-23-0 and RAD-46-23-35-6. It was

DISCHARGE COEFFICIENTS FOR BELLMOUTHS

conducted as a final-year project by a most capable student, H.B. Lau

We will now discuss the results of the FLUENT CFD analysis of the

[3]. In short, we can now directly compare the CD values as measured

fluid mechanics of the 3D flow and the analysis of its output data to

by Mr Lau and as computed by FLUENT. They are shown in Fig.8.

acquire the actual discharge coefficients. In Fig.10 is shown the CD

The correspondence between the measured and CFD-computed CD

To illustrate the potential effect on engine performance, as air mass

flow breathed is potentially engine torque produced, in Fig.12 is

t
t

In Fig.9 are shown the (computed by FLUENT) mass flow rates mdot

area Ac, is also given on those diagrams. In Fig.4, the sharp edged pipe

There is a considerable
benefit in mass flow rate
by the addition of even
a simple radius
36

The visualisation problem is rectified in Fig.11, where the change

values for a range of elliptical and aerofoil profile bellmouths and

values are very close both numerically and as a trend with pressure

also for the simple radius bellmouth RAD-46-23-35-6. All profiled

ratio. You will observe that the CD values are indeed a considerable

bellmouths have the same exit diameter De of 23 mm, some have

function of pressure ratio. You will also note that at QUB we could not

lengths L of 23 or 46 mm, some have entry diameters Di of 40, 46 or

exceed an experimental pressure ratio of about 1.3 even though the

49 mm, and all have a corner radiis Rc of 3 mm. It can be seen that

apparatus at QUB had a flow capacity at least 5 times more than most

all profiled bellmouths exhibit a step increase in CD over the simple

commercially available flow rigs. However, for an intake bellmouth

radius bellmouth but all lie rather closely together so that it is visually

Fig.8 Measured and computed CD data at pipe ends

37

SPECIAL INVESTIGATION : DESIGN OF AN INTAKE BELLMOUTH

Fig.9 Measured and computed airflow rates at pipe ends

Fig.10 CD data for intake pipe bellmouths

shown the change of mass flow rate for all of the profiled bellmouths

radius bellmouth and the elliptical profile bellmouth are curve-fitted

over that of the simple radius bellmouth. The change is expressed as a

with a third order polynomial and the equations are printed at the top

percentage. In the relevant pressure ratio PR band up to 1.1, the best

of Fig.13. The y value is the CD and the x value is the pressure

bellmouths are those that are short and fat with the elliptical profiled

ratio PR. The quality of the curve fit is visibly good.

bellmouth ELL-23-23-49-3 hailed as the winner. But the winning

However, if the CD line is required within an engine simulation

margin is very much a short head as its advantage over simplicity is

for a bellmouth fitted at an intake pipe where there is located a

a mere 1.5%. However, as that might just be an extra 1.5 hp per 100

restrictor diameter, to be followed by a pipe, or more usually a

hp, we would sooner have it as not!

diffuser pipe, as described in our previous articles [6,7], then a CD-PR

equation only fitted over a pressure ratio up to 1.4 is not very helpful.

CD DATA APPLIED WITH AN ENGINE SIMULATION

Hence, the CD line for the best elliptical profile bellmouth, ELL-23-23-

While the graphs and discussion above may be useful as an aid to

49-3, as graphed in Fig.10, is curve-fitted over the full pressure ratio

understanding of the air flow behaviour at intake bellmouths, it is

range up to 1.8 and that trend line is expressed below as,

numbers that engineers need to be able to generally employ in design

or, perhaps more specifically, within an engine simulation during the

Fig.11 CD variations at intake bellmouths

Whereas a 1D simulation
will take but minutes to
complete, a 1D-3D cosimulation can take days

computation of pipe end boundary conditions. For the specific case

mm diameter intake duct, then new bellmouths for both the simple
radius type and the elliptical profile type were designed [5] and their
dimensions are shown labelled on Fig.14. In Fig.14 is presented
snapshots of their particular particle velocity characteristics (as Mach
number contours) at an engine speed of 7000 rpm. A plain-ended
intake pipe is also included in this co-simulation computation series
for its curiosity value.
In Fig.14a for each of the pipe end conditions tested is a snapshot
at a particular crankangle of the outflow process from the
atmosphere (to normal folk it is obviously an inflow process and

during the cycle [2]. That, of course, is what an accurate engine

only thermodynamic pedants such as your authors would consider

of it as an inflow process. Actually, this is an outflow case as the

would wish to use the best bellmouth design possible at the entry to

simulation computes, crankangle by crankangle, but normally using

it otherwise) at the point where maximum particle velocity is taking

air is outflowing from a plenum, i.e., the atmosphere, through a

the restrictor/diffuser and then simulate it accordingly. This same CD

what is theoretically described as a 1D (one-dimensional) procedure.

place at the bellmouth. The similarity of velocity contour (as Mach

restriction, i.e., the bellmouth, to a pipe, i.e., the intake pipe leading

trend line is also applicable to another higher pressure ratio situation

to the engine. For more theoretical information on the reverse flow

where a bellmouth is used and that is at the entry to the compressor

3D (three-dimensional or CFD) codes, it is possible to co-simulate

apparent. These similarities lend some credence to the oft-used phrase

case of inflow, i.e., spitback, you should read a textbook [2] but this

of a turbocharger or a supercharger, particularly if that entry is

where elements of an engine ducting are segregated and computed by

of quasi-steady flow as used to describe the conventional theoretical

common phenomenon you have seen elsewhere in the engine as the

geometrically restricted by the mandates of some racing engine

FLUENT (say) and the remainder of the engine ducting and cylinders

approach in unsteady gas dynamics, which approach is founded in

exit of particles of exhaust gas at the end of an exhaust pipe!

classification.

are computed by the 1D engine simulation. The 1D engine simulation

In Fig.13 is shown some of the previous CD graph data replotted

This information is presented on the assumption that the designer

In recent times, and with the advent of ever more sophisticated

This information is also presented on the assumption that these

then feeds the instantaneous thermodynamic state and gas dynamic

up to a pressure ratio just below 1.4. This range of pressure ratio

CD-PR equations will be applied into an accurate engine simulation

conditions to the CFD computation at either end of the segregated

covers the pressure wave reflection behaviour at the end of an intake

where the theory used therein for its mathematical, gas dynamic and

region and similarly receives updated instantaneous data in return

pipe for naturally aspirated engines where that end either meets the

thermodynamic replay is that described briefly above and thoroughly

with which to continue its 1D calculations. This is referred to as co-

atmosphere itself or the conditions of an intake airbox. The individual

in a textbook [2]. If that is not the case then the trend-line data in

simulation and is a most powerful tool to examine regions of an

graphs of discharge coefficient for the plain ended pipe, the simple

Fig.13, or the equation above, is meaningless and should not be used,

engine ducting where the 1D simulation is theoretically weak, such as

i.e., garbage in is garbage out.

at branches in pipes or at an exhaust collector junction where the flow

The best bellmouths

are those that are short
and fat with the
elliptical the winner
38

which are fully described elsewhere [2]. As the G50 engine had a 38

number) with the steady flow pictures of Figs.4, 6 and 7 is very

t
t

in question, at the intake bellmouth, the reader will naturally think

Fig.12 Airflow rate variations at intake bellmouths

is decidedly three-dimensional.

DYNAMIC FLOW AT THE BELLMOUTH

These computations are best conducted on high performance

The CFD analysis of the flow by FLUENT [1] and the subsequent

computer workstations. Whereas a 1D engine simulation will take but

analysis of that flow computation to determine the CD coefficient

minutes to complete on a modern fast PC, a 1D-3D co-simulation can

[2,3] is conducted under steady flow conditions, just as if it was

take many hours, even days, to conclude.

experimentally executed on a flow bench. However, the actual

In this case, we have prepared a co-simulation by FLUENT of the

bellmouth is placed on an engine which breathes most unsteadily

entire bellmouth including a short segment of the intake pipe beyond

and so the pressure ratio across the bellmouth varies with crankangle

which the 1D engine simulation takes over. The engine used within

and the air particles will not only enter that intake pipe from the

the 1D simulation is the Seeley-Matchless G50 racing motorcycle

atmosphere but will also reverse (spitback) during various periods

engine of yesteryear, the geometry and performance characteristics of

Fig.13 Curve fitted CD data for intake bellmouths

39

SPECIAL INVESTIGATION : DESIGN OF AN INTAKE BELLMOUTH

The computed CD
results for the simple
rectangular bellmouth
are worse
computational difficulties for any CFD code.
Quite irrespective of the above caveats, the computed CD results for
the simple rectangular bellmouth are worse than that for the simple
simple radius bellmouth was a numerical step below all of the profiled
bellmouths. As it is rather difficult to design a rectangular profiled

There is a message
here for those who install
fuel injectors pointing
into intake bellmouths

experiments. The bellmouth is created by a simple 6 mm radius

bellmouth, it will inevitably have a rectangular entry, the general

around the perimeter, making it the rectangular equivalent of the

conclusion must be that rectangular intake ducts and rectangular

round pipe RAD-46-23-35-6. This rectangular bellmouth is labelled as

intake bellmouths should be avoided by design if at all possible.

REC-29.878_14.939-6.

That the flow regime has the asymmetric fluid mechanic difficulties

process to a large plenum called the atmosphere. The particle flow

entering the atmosphere is more pronouncedly strong for the weakest

Fig.14a Dynamic particle

inflow during co-simulation

Fig.14b Dynamic particle

spitback during co-simulation

the notion that unsteady flow is but a sequence of differing steady flow
processes conducted over very short time intervals.

The FLUENT CFD computations are run at varying pressure ratios

described in the literature [8,9] is confirmed in Fig.18, where the

up to 1.7 and the CD values are analysed at each PR from the mass

computed (particle velocity) Mach number profiles across each axis

flow data determined by CFD [2,3]. The results of these calculations

and at a section 6 mm inside the entry are illustrated.

are seen in Fig.17, but they might have been anticipated by reading
almost any text on fluid mechanics [8] or studying the experimental

CONCLUSIONS

observations on the loss-creating vortices at the entry corners to

While the specific conclusions have already been highlighted at

rectangular pipes [9]. Any such texts will show that the hydraulic

each stage of the discussion above, the general conclusion might be

are somewhat inconclusive, although up to the very relevant pressure

radius, conventionally calculated as area/wetted perimeter, for the 23

glibly stated that the design of an intake bellmouth is not as difficult

ratio of 1.1 it can be stated that a sharp-edged bellmouth is marginally

mm round pipe is D/4 or 5.75 mm but that for our selected rectangular

nor as vital to good engine breathing as might have been imagined.

bellmouth, i.e., the plain-ended pipe, and vice-versa for the elliptical

inferior to a bellmouth with a Rc corner radius and that the use of a

pipe is 5.24, a loss of some 9%. However, in Fig.17, the loss of CD for

On the other hand, in racing, where the last few hp per 100 hp is

profile bellmouth. Indeed, it is so strong at the plain pipe end that

full ball radius is unnecessary.

the rectangular pipe is computed through FLUENT as no worse than

the difference between winning and losing, the design exemplars

0.83% at the lowest pressure ratio. The reality of an actual CD steady

discussed above are not to be lightly ignored.

In Fig.14b are the movie snapshots at a particular crankangle at

the peak of the reverse flow process, commonly called spitback,
and again your pedantic authors will tell you that this is an inflow

it has formed a toroidal vortex (smoke ring!) at the pipe end. Such a

The second obvious set of questions will doubtless relate to the oft-

phenomenon has been seen and photographed before in high speed

used rectangular intake duct shape as seen in Fig.16. In a four-valve

flow measurement might show greater losses in CD for the rectangular

Schlieren image experiments conducted at QUB more than thirty years

head design it is somewhat difficult to smoothly connect the twin

bellmouth as the corner vortices seen by Schlicting [9], with their

REFERENCES

ago [2, pp 154-157]. There is a message here for those who install

intake passages at each of the intake valves into a single round intake

inherent rotational turbulence characteristics, always provides

[1] FLUENT Inc., Lebanon, New Hampshire, www.fluent.com

fuel injectors pointing into intake bellmouths; use a short and fat

duct and often a rectangular

[2] G.P. Blair, Design and Simulation of Four-Stroke Engines, Society

bellmouth to reduce the spitback of fuel for it is the spitback of air

intake duct is considered the

of Automotive Engineers, 1998, SAE reference R-186, p813.

which propels it.

effective compromise. But

[3] G.P. Blair, H-B. Lau, A. Cartwright, B.D. Ragunathan, D.O.

is it, especially if it leads to

Mackey, Coefficients of Discharge at the Apertures of Engines, SAE

ANSWERING THE OBVIOUS QUESTIONS

a rectangular bellmouth? In

International Off-Highway Meeting, Milwaukee, September 1995, SAE

At this point many a reader will be saying, is that it? and

Fig.16 is seen a photo of just

paper no. 952138, pp 71-85.

forming a question beginning with what if ? . To forestall many

such a bellmouth and above it

[4] G.P. Blair, F.M. Drouin, The Relationship between Discharge

an e-mail, we have examined a couple of such cases. The first obvious

the CFD geometric model to

Coefficients and the Accuracy of Engine Simulation, SAE Motorsports

questions may well relate to the wrap-round radius Rc. Is it necessary?

assess its CD characteristics by

Engineering Conference and Exposition, Dearborn Michigan,

How big should it be? Is it OK as a half-radius as we show it here, or

FLUENT [1].

December 8-10, 1996, SAE paper no. 962527

should it be a complete ball? The answers are contained in Fig.15.

40

radius RAD-46-23-35-6 and, as a glance at Fig.10 will confirm, this

Fig.15 CD variations for bellmouth edge geometries

Fig.17 Loss of CD by a rectangular bellmouth

The rectangular duct

[5] 4stHEAD design software, Prof. Blair and Associates, Belfast,

The steady flow CFD analysis is conducted with three similar

geometry used has an aspect

Northern Ireland, www.profblairandassociates.com.

elliptical profile bellmouths but one has our common half-radius

ratio of 2:1 with four corner

[6] G.P. Blair, W.M. Cahoon, Airbox Design Part 1, Race Engine

as seen in Fig.5 (Rc is 3 mm), another has a full ball radius of 3 mm

radii each of 6 mm and the

Technology, Vol.1, No.2, 2003.

that folds right back to the outside of the bellmouth. Yet another has

width and height are 29.878

[7] G.P. Blair, W.M. Cahoon, Airbox Design Part 2, Race Engine

no radius at all but has a sharp-edged pipe end at the (common to

and 14.939 mm, respectively,

Technology, Vol.1, No.3, 2003.

all three) 46 mm entry diameter Di. The Fig.15 shows the variation

giving the same area as the

[8] V.L. Streeter, Fluid Mechanics, McGraw Hill, New York, 1961.

from the worst case, which is the zero radius case to the other two

round 23 mm pipe used

[9] H. Schlichting, Boundary Layer Theory, McGraw Hill, New York,

cases, as a change of CD expressed as a percentage (%). The results

for all previous CFD flow

Fig.16 Rectangular bellmouth design

Fig.18 Velocity flow profiles at a rectangular bellmouth

1961.
41