A measuring stand for a ducted fan aircraft propulsion unit

Abstract

The UL-39 ultra-light aircraft which is being developed by the Department of Aerospace Engineering, Faculty of Mechanical Engineering, Czech Technical University in Prague, is equipped with an unconventional ducted fan propulsion unit. The unit consists of an axial fan driven by a piston engine and placed inside a duct ended with a nozzle. This article describes the arrangement of a modernised measuring stand for this highly specific propulsion unit which will be able to measure the fan pressure ratio and velocity field in front of and behind the fan and its characteristic curve.

Full text

a
Corresponding author: David.Hlavacek@fs.cvut.cz
A measuring stand for a ducted fan aircraft propulsion unit
David Hlaváþek
1,a
1
Department of Aerospace Engineering, Faculty of Mechanical Engineering, Czech Technical University in Prague,
Karlovo námČstí 13, 121 35 Praha 2, Czech Republic
Abstract. The UL-39 ultra-light aircraft which is being developed by the Department of Aerospace
Engineering, Faculty of Mechanical Engineering, Czech Technical University in Prague, is equipped with an
unconventional ducted fan propulsion unit. The unit consists of an axial fan driven by a piston engine and
placed inside a duct ended with a nozzle. This article describes the arrangement of a modernised measuring
stand for this highly specific propulsion unit which will be able to measure the fan pressure ratio and velocity
field in front of and behind the fan and its characteristic curve.
1 Introduction
The UL-39 aircraft is being developed by a team of
students and employees of the Institute of Aerospace
Engineering, Faculty of Mechanical Engineering, Czech
Technical University in Prague. Its propulsion unit
consists of a single-stage axial fan in stator-rotor
arrangement placed in a duct inside the fuselage of the
aircraft. The duct ends with a nozzle. The fan is driven by
a piston engine. In the past, a Yamaha YZF-R1 engine
was used to drive the measuring stand placed in the
laboratories of the Institute. For this engine, the present
fan was designed. A new engine, the BMW S1000RR
was bought in 2011 for use in the flying prototype of the
aircraft. This engine has the advantage of a lighter weight
along with a higher maximum horsepower.
Figure 1. Model of the UL-39 propulsion unit.
In order to optimize the operating conditions of the
propulsion unit, a new fan will be designed. The main
objective of its design was a higher propulsion unit thrust,
a higher efficiency while lowering the noise level at the
same time.
The air passes through a pair of inlet channel to a
stator guide vane. The very specific flow conditions
inside the short and strongly curved channel led to an
unconventional inlet channel design. Its splitting plane is
twisted by an angle of 45 deg around the longitudinal axis
of the aircraft (see figure 3). Along with the stator guide
vane, this arrangement ensures an even distribution of
rotor angles of attack in the design point (figure 4).
Figure 2. Model of the UL-39 aircraft.
The flow field inside the ducted fan propulsion unit,
especially the inlet channel, has already been investigated
using CFD calculations and the method of materialized
streamlines. An experimental investigation is being
prepared now which should lead to the knowledge of
velocity magnitudes (and an idea of its directions as well)
downstream of the fan rotor and of the fan characteristic
curve.
The measurement will be a part of the development of
a new fan. At this time, it should validate the results of
the previous design calculation.
DOI: 10.1051/
C

Owned by the authors, published by EDP Sciences, 2014
,
/
02036 (2014)
201
67
epjconf
EPJ Web of Conferences
46702036
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 2.0, which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Article available at http://www.epj-conferences.org or http://dx.doi.org/10.1051/epjconf/20146702036

Figure 3. The twisted inlet channel of the propulsion unit.
The aim of this paper is to propose the method and
instrumentation for measuring the velocity distribution
downstream of the rotor and the fan characteristic curve.
2 The measuring stand arrangement
For measuring the fan's characteristic curve and its
parameters in the design point, it is necessary to know the
total pressure ratio (or difference) and mass or volume
flow rate of air passing through the fan. Therefore,
pressure and velocity sensing tubes will be used during
the measurement. The instrumentation for measuring the
velocity profile behind the fan rotor will be identical.
First, we need to evaluate the suitable places and
conditions of the measurement. We will do so with the
Figure 5. The velocity magnitudes in the outlet channel as
computed by CFD. [3]
help of calculations performed in [1] and relationships
describing the velocity field downstream of compressor
blade vanes.
2.1 Goal of the measurement
The main goal of the measurement is determining the
characteristic curve of the fan.
The characteristic curve describes the dependence of
the fan pressure ratio on its flow rate and is one of the ¨
Figure 4. The contours of rotor blade angles of attack as
computed by CFD
. [2]
fundamental means of indicating the fan's behaviour
during operation. It determines the operating point of the
fan and its reaction to any changes of the operating
conditions. The fan characteristic curve is depicted in
figure 16. Each branch of the characteristic curve
corresponds to a respective shaft speed.
For measuring one branch of the characteristic curve,
one needs to measure the total pressure upstream and
downstream of the fan, and the air flow rate. Using a
throttling device, the flow rate is then changed which
leads to a change of the fan total pressure ratio.
2.2 The points of measurement
The choice of the suitable places for measuring the
pressures and velocities is influenced by the
unconventional arrangement of the propulsion unit
EPJ Web of Conferences
02036-p.2

(especially of its inlet channel) and the accessibility
requirements. Therefore, the measurement will be made
in both inlet channels and a plane in a certain distance
downstream of the fan. A measuring plane upstream of
the fan is not suitable for this type of measurement since
the flow field is highly distorted there while our goal is to
find the mean values of velocity in both space in time.
Furthermore, this place is not easily accessible.
Figure 6. The present measuring stand of the propulsion unit.
There is another reason for this arrangement of measuring
places. The operational parameters of the fan are affected
by the design of the inlet channel to a great extent.
Therefore, it is better to consider the fan and its inlet
channel a compact unit rather than separate components.
The plane for placing the velocity sensing tubes in the
inlet channel can be chosen considering only an easy
access for mounting the tubes and their piping. Therefore,
the measuring plane downstream of the fan rotor will be
the chief subject of our interest. Its distance from the
rotor blade vane depends on the presence of wakes
behind the blades and hub, and, of course, on the
arrangement of the outlet channel.
Since, for the reasons of weight, the hub of the fan is
not equipped by any cone, the flow area downstream of
the rotor increases suddenly. A cone-shaped wake forms
in an area close to the rotor axis (see figure 5). According
to the CFD calculations performed in [3], its length is
approx. 350 mm. The distance between the rotor exit and
the inlet of the by-pass channel with a radiator in it (see
also figure 5) is 410 mm. The measuring plane should
best be placed between these two distances. If it was
placed further downstream, it would be necessary to
account for the division of the channel in two branches.
Moreover, the section of the channel is not circular
anymore.
Now, we need to investigate the effect of the rotor
blade wakes on the flow field in the area downstream of
the fan. In the following paragraphs, the dimensions of
the affected area will be calculated based on relatively
simple models. The results will lead to a decision,
whether it is possible to place the measuring plane in the
above- mentioned distance downstream of the fan.
For determining the distance in which the influence of
the rotor wakes will not be significant, a model published
in [1] will be used.
Figure 7. Model of a wake behind an airfoil. [1]
According to [1], the backflow velocity in a wake
behind a rotor blade is a function of the blade drag
coefficient and the axial distance behind it.
R
RD,
0
c
x
c
v
v
∝
∞
(1)
Particularly, the author of [1] claims that, according to
experiments performed, the relationship
025.0
6.1
R
RD,
0
+
=
∞
c
x
c
v
v
(2)
should best be used.
The ratio of velocities in the previous equation should
be as close to zero as possible. The unknown quantity in
this relationship is the distance denoted by x.
That rotor blade drag coefficient c
D,R
is not very often
used to account for losses in a blade vane. More often,
the total pressure loss coefficient ȗ
R
according to Ainley's
definition is used.
2
1
c
R
2
1
ǻ
w
p
ρ
ζ
=
(3)
The following relationship between ȗ
R
and c
D,R
can be
found in [4]:
∞
=
αζ
cos
R
R
RRD,
c
s
c
(4)
The pressure loss coefficient can be taken from a
diagram [5] which describes the dependence of this
coefficient (denoted by Ȧ) and of deviation angle on the
angle of attack (denoted by i). It contains values
measured using modified NACA 65A10 airfoils which
are also used in the UL-39 propulsion unit fan.
The angle of attack results from the CFD calculation
made in [2]. Therefore, Į

= 1.1° (see figure 4). From the
diagram in figure 9, the value of ȗ
R
= 0.03 can be read.
The pitch-to-chord ratios s
R
/ c
R
of the UL-39 fan
rotor are different for each radius and can be found in [6].
For example, s
R
/ c
R
= 0,704 at the hub. The rotor blade
drag coefficient is then:
0211.01.1cos704.003.0cos
R
R
RRD,
=°××==
∞
αζ
c
s
c
This calculation is to be repeated in the other sections
of the rotor vane. The results of this calculation are
EFM 2013
02036-p.3

summarized in table 1. Since the angle of attack is nearly
constant along the rotor blades, the pressure loss
coefficient is considered to be constant as well.
Table 1. Rotor blade drag coefficients in eight
circumferential sections of the rotor vane.
Section No. r (mm) s/c c
D
,
R
1
126
0.704
0,021
2
148
0.897
0,027
3
170
0.975
0,029
4
192
1.324
0,040
5
214
1.549
0,046
6
236
1.775
0,053
7
258
1.999
0,060
8
280
2.216
0,066
Therefore, the ratio of the backflow velocity to the
mean undisturbed flow velocity will be a function of two
variables:
025.0
)(
)(6.1
),(
R
RD,
0
+
=
∞
rc
x
rc
xr
v
v
(5)
This function is presented as a diagram in figure 8. It
is clear that the difference between the velocity ratios
along the blades is not significant. The curve describing
this dependence is very steep for the first 50 mm and,
conversely, very flat further downstream.
Figure 8. Wake velocity defect as a function of distance.
At the outer radius of the rotor, where the backflow
velocity is the highest, a value of v
0
/ v

= 0.2 is achieved
in a distance of 260 mm. A value of v
0
/ v

= 0.15 and
0.10 corresponds to x = 460 and 1035 mm, respectively.
In longer distances behind the rotor, the calculation is
clearly not valid anymore since the shape and area of the
duct are different.
This means that if the measuring tubes are to be
placed in a distance between 350 and 410 mm, it is
necessary to count on a backflow velocity of
v
0
= (0.16 ÷ 0.17) v

(at the outer diameter), not taking
into account the widening of the wake.
Figure 9. Deviation angles and total pressure loss coefficients
of a NACA 65A10 airfoil
. [5]
However, from the fundamentals of aerodynamics, it
is known that the wake behind airfoils becomes wider
when passing downstream (as sketched in figure 7). So
now we need to focus our attention on the width of the
wake.
Theoretically, an even velocity distribution would be
achieved in a distance in which the width of the wake
(denoted by į) equals one half of the vane blade pitch.
For roughly estimating the coordinate of this place, we
will used relationships described in [7].
The equations found in [7] describe the wake width as
a function of the longitudinal coordinate and rotor blade
drag coefficient:
),(
RD,
c
c
x
f
s
į
=
and
),c
c
x
f(
c
į
RD,
=
(6)
According to [7], these relationships can be used for
compressor blade vanes in a wide range of pitch-to-chord
ratios (or solidities).
For high-solidity vanes with a value of s/c  1 it is
recommended to normalize the width of the wake with
the blade pitch. The following equation is used:
1268125.0
048.031875.0
8
RD,
8
RD,
+
+
=
c
c
x
c
c
x
s
δ
(7)
EPJ Web of Conferences
02036-p.4

For low-solidity vanes (s/c > 1), on the other hand, the
width of the wake is normalized with the blade chord.
Therefore:
1357.0
034125.02375.0
8
RD,
8
RD,
+
+
=
c
c
x
c
c
x
c
δ
(8)
Since the rotor blades of the UL-39 fan are relatively
long and their chord is not constant over their length,
both these relationships will be used at different radii. As
can be seen from table 1, in sections 1 to 3 Eq. (7) will be
used while in sections 4 to 8 Eq. (8) is valid.
The results of these equations, expressed by į/s, are
presented in figure 10. It can be observed that, according
to these calculations, the even velocity distribution (į/s =
1) is theoretically achieved in a distance of x = 210 mm.
Figure 10. Wake thickness as a function of distance.
This leads to a conclusion that the velocity
distribution in distances between 350 and 410 mm
downstream of the rotor should not be extensively
affected.
Placing the sensing tubes in these distances is
therefore possible. The real distance will be determined
during their assembly.
2.3 Distribution of the sensing tubes in the
channels
Now that the possible distance of the sensing tubes
downstream of the fan is known, the distribution of
measuring points in the sections of the inlet and outlet
channel should be determined
Since, apart from total pressure, we also intend to
measure static pressure downstream of the fan, it is
necessary to bear in mind the fact that a circumferential
velocity component will always be present in the flow
field downstream of the fan (except for, theoretically, the
design point).
This means that a swirling motion of the air will take
place inside the channel and that the static pressure
distribution over the respective channel section will be
uneven. Around the axis, the static pressure will be lower
than near the channel walls. It is therefore not possible to
measure its value using bores in the channel walls. The
static pressure sensors should be placed directly in the
sensing tubes.
Figure 11. A scheme of vortex generation behind a fan [8]
For measuring the total and static pressure inside the
outlet channel, three radial lines should be used and the
mean value in each circumferential section will then be
computed (at least two radial lines should be used
according to [8]).
In places where the sensing tubes will enter the
channel, it is necessary to strengthen the composite
channel walls with inserts.
In the inlet channel, a few ribs equipped with total
pressure sensing bores will be placed (in the left part of
figure 13, these places are marked by yellow lines). The
connecting tubes will go through the inside of the channel
leading edge, where enough space is available for their
placement. The resulting "measuring inlets" will be
available for quickly putting on during measurement in a
wind tunnel or in the laboratory. Another advantage of
this arrangement is that no traversing is needed which
would require a traversing mechanism to be designed
while also extensively consuming time and fuel.
There are also two minor disadvantages – a slight
blockage of the inlet section and the influence on the flow
field around it. For similar reasons, ladder probes will be
used downstream of the fan.
3 The instrumentation
3.1 Instrumentation for measuring the velocity
distribution
In this experiment, measuring the flow field pulsations
caused by turbulence is not required. Its goal is just the
opposite - the mean values of total and static pressure and
velocity need to be measured. Therefore, using
anemometric probes is not necessary. Conventional Pitot
and Prandtl tubes will be used instead. For determining
the direction of the velocity vectors downstream of the
fan, multi-hole probes will be used.
3.2 Instrumentation for controlling the flow rate
The air flow rate will be controlled by a suitable
throttling device. According to the arrangement of the
whole propulsion unit, this device should be placed in its
aft part (inside the outlet nozzle or in a chamber
downstream of it).
A throttling valve is one of the devices that may be
used for controlling the flow rate. Conventional valves
used in air conditioning fans are not designed for the
range of velocities encountered in the UL-39 propulsion
unit (a mass flow rate of up to 20 kg/s). Throttling valves
EFM 2013
02036-p.5

for heating systems or industrial compressors could be
used but they can be rather expensive.
Another major disadvantage of using throttling valves
is their considerable influence on the flow field to a long
distance upstream of the valve. This leads to a decrease
of the measurement accuracy as the measuring conditions
start to vary with the real case.
Using a variable-geometry outlet nozzle is also an
option. The laboratory if the Institute of Aerospace
Engineering is equipped with converging nozzles of
various exit diameters. However, they are unable to cover
the whole range of flow rates and unavailable in a
sufficient quantity to ensure a satisfying stepwise flow
rate control.
Figure 12. One of the exit nozzles used by the Department of
Aerospace Engineering [9]
Seemingly, the most suitable controlling element for
this application will be a strainer with a variable density
which means that several strainers will be used placed
behind each other. The control will also be stepwise in
this case but with a sufficient range and quantity of steps
for measuring the whole extent of the fan characteristic
curve.
The measurement of the UL-39 fan characteristic
curve is highly specific because the fan together with its
inlet and outlet channel forms an integrated system.
Figure 13. Position of probes in the two measuring planes.
If the fan was separated from the influences of its
channels, its operational behaviour would be completely
different.
When air conditioning fans are measured, special
elements are placed inside the piping to make the velocity
distribution more even and thus help the fan to operate
independently from the piping. Taking these measures in
the case of the UL-39 fan would cause deviations from
the real operating conditions.
4 The measurement procedure
4.1 Visualising the flow in the inlet channel
The first step of the experiment will be flow visualisation
in the measuring section of the inlet channel. The flow
will be visualised using smoke or cotton. One of its goals
is checking whether the flow field is qualitatively the
same in both branches of the inlet channel. Furthermore,
the presence of transversal velocity components or
vortices can be detected.
Since the measuring stand is made if composites and
is not equipped with windows, visualising the flow in
other sections is not possible.
4.2 Determining the velocity directions
downstream of the fan
For determining the directions of the absolute velocity
vectors downstream of the fan rotor, cylindrical of multi-
hole probes will be used.
Measuring the velocity directions is important for two
reasons. First, the angle of the Prandtl tubes downstream
of the fan will be determined, and second, the knowledge
of the circumferential velocity components at various off-
design conditions can be gained.
EPJ Web of Conferences
02036-p.6

4.3 Measuring the velocity distribution
The velocity idstribution will be measured dowstream of
the fan. The Prandtl tubes should be directed against the
flow. The tubes will be distributed over the section of the
piping using the equal area rule.
The velocity will not be measured near the axis since
an extreme value would be captured there.
The radii for placing the Prandtl probes can be
computed using the following equation:
n
i
R
n
i
Rr
i
2
12
2
1
−
=
−
=
(9)
At each radius, a velocity value will be measured
which will be considered mean velocity in its annulus.
The mean value over the whole section which determines
the volumetric flow rate will then be computed using
graphical or numerical integration (see [8] or [10] for
details). The flow rate is computed using the following
equation:
SwSwV
S
==
³³
d

(10)
According to the calculations made in [6], the axial
velocity component inside the fan should be
c
a
= 90 m/s. A Mach number of 0,264 corresponds to this
velocity considering the conditions in a height of h = 0 m
(ISA), similar to these expected in the laboratory during
the experiment. Therefore, the flow in the absolute
coordinate system is assumed to be incompressible. The
mass flow rate is then computed using this simple
equation:
SwSwm
S
∞∞
==
³³
ρρ
d

(11)
Figure 14. Velocity profile graphical integration [10].
The mean values of total pressure behind the fan will
be determined by averaging the measured values in each
measuring plane. From these values, the total pressure
ratio ʌ
V
of the fan can be computed and then plotted in
the characteristic curve.
4.4 Conditions at the design point
The measurement described above will, first of all, be
carried out at the design conditions of the fan (according
to [6]). It means using an unthrottled exit nozzle with an
exit diameter of 450 mm and reducing the ambient
conditions during measurement to the conditions
corresponding to h = 0 m (ISA).
Figure 15. Division of the outlet channel cross-section using
the equal area rule. [10]
4.5 Plotting the characteristic curve
Afterwards, the whole characteristic curve of the fan will
be measured. Each curve, corresponding to a respective
shaft speed, can be plotted according to the known values
of pressure ratio and mass flow
Figure 16. The fan characteristic curve (above) and the
alternative means of measuring it (below) [11]
When measuring each of these curves, the flow rate is
controlled by a throttling device while the shaft speed is
kept constant. This leads to a change of the fan total
pressure ratio. Each curve should be constructed using at
least about ten points (ten settings of the throttling
EFM 2013
02036-p.7

element). This technique, although often used in practice,
is not suitable for the UL-39 fan propulsion unit.
Considering the difficulty of keeping the combustion
engine speed constant as well as the necessity of
switching the strainers several times at each speed, the
measurement will be carried out using an alternative
technique, described in [11].
A set of curves in the pressure ratio-flow rate
coordinates will be measured, each of them
corresponding to a constant nozzle exit area (a constant
throttling device setting, see the bottom part of figure 16).
The curves corresponding to the respective shaft
speeds may then be constructed by connecting the
appropriate sets of points in these coordinates. Slight
deviations of the shaft speed, which may occur during the
experiment, may be corrected using equations presented
in [11].
This way, for each throttling element setting, the fan
will be accelerated from the lowest speed to the highest.
This makes the measurement more practical and faster
while not negatively affecting its accuracy.
Figure 17. A scheme of the future stand.
The characteristic curve can then be reduced to the
standard conditions (see [11]) to make it comparable with
future measurements or other types of axial
turbomachines.
4.6 Evaluation and further steps
The characteristic curve of the fan, as described above, is
an important indicative factor of its operational
behaviour. The operating point should lie in a sufficient
distance from the surge limit and in the area of the
greatest fan efficiency. The efficiency at various points
can be determined by using total temperature sensing
devices together with pressure sensors.
Another important result of the experiment will be the
knowledge of the velocity distribution in the inlet channel
and downstream of the fan. So far, only CFD calculations
were used to determine the velocities in these areas. The
experiment should help to validate the calculations made.
Last but not least, the values of the pressure ratio and
flow rate in the design point will be checked.
Furthermore, the velocity direction should be measured in
this point and any deviation from the axial direction
determined. These calculations should validate the one-
dimensional design calculations of the present fan.
5 Conclusions
In this paper, the redesign of a measuring stand for a
ducted fan aircraft propulsion unit was described,
together with the procedures of future measurements.
The results of the experiments will be used during the
development of a new fan for this propulsion unit which
should produce less noise or increase the cruise speed of
the aircraft.
Since any measurements of the fan characteristic
curve or the design point conditions have not yet been
made, the experiment will also be used for validating the
design calculations of the present fan and the location of
the design point on the characteristic curve.
References
1. J. H. Dittmar. Methods for Reducing Blade Passing
Frequency Noise Generated by Rotor-wake-Stator
Interaction. Technical Memorandum. (Washington:
NASA, 1972). Report No. NASA TM-X 2669.
2. E. Ritschl. Analýza tvaru vstupních ploch sacího
kanálu dmychadla. Doctroal Thesis. (CTU in Prague,
Faculty of Mech. Eng., 2009)
3. Z. Ilich. Rekonstrukce výstupního kanálu
ventilátorového pohonu malého sportovního letounu.
Master’s thesis. (CTU in Prague, Faculty of Mech.
Eng., 2012)
4. I. R. Schwartz, H. T. Nagamatsu, W. C. Strahle.
Aeroacoustics: Fan Noise and Control; Duct
Acoustics; Rotor Noise. Technical Papers from
AIAA 2nd Aero-Acoustics Conference. (New York:
AIAA, Cambridge: MIT Press, 1976)
ISBN 0-915928-09-4
5. S. Lieblein. Incidence and Deviation Angle
Correlations for Compressor Cascades. (Cleveland:
NASA, 1959)
6. R. Poul. Termodynamický a aerodynamický návrh
axiálního ventilátoru v uspoĜádání rotor-stator a
pĜedstator-rotor. Research report. (CTU in Prague,
Faculty of Mech. Eng., 2009)
7. R. K. Majjigi, P. R. Gliebe.. Development of a Rotor
Wake/Vortex Model. Volume I – Final Technical
Report. Final Contractor Report. (Washington:
NASA, 1984) Report No. NASA CR 174849
8. B. Eck. Ventilatoren. 6. Aufl. (Berlin, Heidelberg,
New York: Springer, 2003) ISBN 3-540-44058-5
9. M. Helmich. Tests of the Propulsion Unit of an
Unconventional Ultralight Aircraft. In: Studentská
tvĤrþí þinnost. Sborník 2013. (Prague: CTU in
Prague, 2013) ISBN 978-80-01-05232-7
10. T. Matuška. Experimentální metody v technice
prostĜedí. 1st Ed. (Prague: CTU in Prague, 2005)
ISBN 80-01-03291-4
11. P. Šoch, J. Vrátný, J. MaĜík. Mechanika tekutin –
experimentální metody. 1st Ed. (Prague: CTU in
Prague, 1989)
EPJ Web of Conferences
02036-p.8