Professional Documents
Culture Documents
i
=arctan()
)
`
1
32
1
16
r
a
+
a
=
(3.1)
(3.2)
R
r
= r (3.3)
Figure 7 - Attack angles and induced angle along span at 6000 rpm for a
24x10 static MenZpropS propeller
R
bc
=
ITA Aeronautical Institute of Tchnology
June 30/2011, So Jose Dos Campos, SP, Brazil
rad
d
dc
= a
l
/ 5.73 ~ (3.4)
In equations above, R is the blade radius, r is the element radius position, b is the number of blades and c is the
element chord. The value of a is fixed, in this model, and its value is based on experimental results for profiles. The
model considers only axial induced velocity, disregarding tangential one.
Figure 8 - Blade schematic representation of two-dimensional flow. Source: Cavcar (2004)
Figure 7 shows a graph of induced angle along blade span. These angles may be significant. Its main consequence
is an increase in drag. An incompatibility appears, however: induced velocity at tip is not sufficient to reduce angle of
attack to zero. Actually, the angles predicted are trustful until 97% of radius (MacCormick, 1967). After this station, lift
decreases and reaches zero at tip.
Contribution of each element to thrust and torque are calculated by:
( ) ( )
i i
+ dD + dL = dT | | sin cos (3.5)
( ) ( ) | |
i i
+ dD + + dL r = dQ | | cos sin (3.6)
( ) dr c c V
= dL
i l e
|
2
2
(3.7)
( ) dr c c V
= dD
i d e
|
2
2
(3.8)
As an example, it is valuable to analyze thrust and torque distribution. The geometry is the same of Fig. 7,, a 24 x
10 MenZpropS.
ITA Aeronautical Institute of Tchnology
June 30/2011, So Jose Dos Campos, SP, Brazil
Figure 9, and Fig. 10, show the expected trends. Towards center of blade, angle of attack increases beyond that
specified by experimental data. As discussed, values of 12 are maintained. In this way, estimates of thrust will be
higher than real values. Torque analysis reveals the opposite problem. If Cd is maintained in 12 level, it under-predicts
drag and torque, consequently.
Working with fitted surfaces to represent aerodynamics coefficients maybe created difficulties which caused the
corners noted in Fig. 9 and Fig. 10.
4. RESULTS AND COMPARISONS
Based on the blade element model described, a code was written in LabVIEW
TM
platform in order to predict
different blades behaviors. Summing up the contributions of the elements, thrust T and torque Q obtained were used to
calculate dimensionless coefficients C
T
and C
Q
given by the following formulas:
Figure 9 - Example of element thrust contribution along span.
Figure 10-Example of element torque contribution along span
ITA Aeronautical Institute of Tchnology
June 30/2011, So Jose Dos Campos, SP, Brazil
4 2
D N
T
= C
T
(4.1)
5 2
D N
Q
= C
Q
(4.2)
N represent rotational speed of the propeller, and D, its diameter, while stands for air density. In this study
simulation, three different geometrical size of propeller were studied. As a common practice, a propeller is defined by
its diameter and its pitch at 75% station, measured in inches and placed in this order. Adopting this convention, the
simulated propellers were 24x10, 24x12 and 22x12. Rotation was normalized by its maximum value, taken as 10000
rpm. This number was chosen because it's the maximum experimental value of engine Zenoah GT-80 data. The series
of simulations were conducted with airfoils RAF-6 and CLARK-Y, for which the experimental coefficients were
adjusted.
Figure 11 Theoretical model curves of C
t
vs. N/N
max
.
Figure 12 Experimental results of C
t
vs. N/N
max
.
Using propellers of same dimensions, experimental results were obtained and are shown in Fig. 11 and Fig. 12,.
Propeller test stand is the same as described in Martins & Venson (2010). Data of thrust coefficient were neglected for
propeller 22x12, since the deviation was very high when compared to other measures.
Care must be taken when comparing these experimental results and blade element models. First, the propellers
MenZpropS tested have airfoils different from those simulated. Generally, the geometry varies between different
sections, in real props. So, just trends may be compared between data of the two theoretical figures and the
experimental one.
The data obtained through the program was satisfactory when compared to the experimental data (see Fig. 11 and
Fig. 12). Ct (thrust coefficient) curves as a function of the non-dimensional rotation for 24x10 and 24x12 propellers,
using a range between 0.4 and 0.7 of N/Nmax, showed similar trends in both the theoretical model and the experimental
data.
ITA Aeronautical Institute of Tchnology
June 30/2011, So Jose Dos Campos, SP, Brazil
Figure 13 Theoretical model curves of C
q
vs. N/N
max
.
Figure14 - Experimental results of C
q
vs. N/N
max
.
The Cq (torque coefficient) curves as a function of the fraction of rotation for the same range (between 0.4 and 0.7
of N/Nmax), also shows similar trends for both the model and for the experimental analysis done for the propellers
22x12, 24x10 and 24x12.
These results show that the model is worth in predicting trends while some propeller parameters are varied. On the
other hand, there is no guarantee that the numeric values are valid results for design purposes. However, it may be
useful when choosing the adequate geometry of the propeller.
After this experimental validation, the model results were used to estimate witch propeller would provide greater
thrust at take-off conditions (V = 0). Power requirements of the propeller at each rotational speed were compared with
the power characteristics of Zenoah GT-80 driver as shown in Fig. 15 and Fig. 11. The matching point between
driver and propellers curves gives the rotational speed for which the maximum power can be delivered, and therefore
the maximum thrust for each propeller.
ITA Aeronautical Institute of Tchnology
June 30/2011, So Jose Dos Campos, SP, Brazil
Figure 15 Power curves for Clark-Y propeller. Observe the matching points with Zenoah driver power curve.
Figure 16 Power curves for RAF6 propeller. Observe the matching points with Zenoah driver power curve.
Table 1 shows the summary of maximum thrust and related rotational speed for the propellers.
Table 1 Theoretical maximum thrust and related rotational speed.
ClarkY Speed (RPM) T (N) RAF6 Speed (RPM) T (N)
22x12 7200 199 22x12 6950 181
24x10 6900 239 24x10 6700 214
24x12 6200 206 24x12 6000 184
For comparison, the experimental maximum thrust results are shown in
Table 2:
Table 2 Experimental maximum thrust and related rotational speed.
Experimental Speed (RPM) T (N)
22x12 6600 126
24x10 6330 140
24x12 5970 136
Again, it can be seen the same trends between model and experimental results. The maximum thrust was
achieved by 24x10 propellers at intermediate speeds. Absolute values should not be compared since experimental work
was done with unknown blade airfoil distribution. However, propeller 24x10 also revealed to the best one in
experimental results, according to blade element model prediction.
ITA Aeronautical Institute of Tchnology
June 30/2011, So Jose Dos Campos, SP, Brazil
4. CONCLUSIONS
This work proposed a simple method for propeller performance calculation, based on blade element model, as
shown. For its simplicity, it has some limitations, but preserves its utility. Moreover, simplicity does not mean the
method implementation was straightforward. Its useful to comment the implementation tasks with its challenges as
well as the usefulness and shortcomings of this model.
Firstly, wind tunnel testing results for propeller airfoils are not easy to found. Actually, the team was able to found
data only for RAF-6 and Clark-Y airfoils. Moreover, real propellers have variable airfoils along wingspan, and this
distribution information is not easy to obtain. Therefore, the calculations were performed for propellers with constant
airfoil.
The c
l
and c
d
of an airfoil is function of Mach, Reynolds and attack angle. Mach, attack angle and Reynolds varies
with radius along propeller blade. The way chosen to deal with this problem was to create a surface fitting of C
L
=
C
L
(M,) and C
D
= C
D
(M,) from experimental data, since the available curves implicitly considers Reynolds variation
as function of Mach (because experimental results were obtained for constant chord models, varying velocity). The
program xyExtract was very useful for collecting the experimental points, while MATLAB