a cfd-based approach to coaxial rotor hover … - spring 2007... · coefficients were interpolated...
TRANSCRIPT
A CFD-Based Approach to Coaxial Rotor Hover Performance Using Actuator Disks
Jonathan Chiew
AE4699 - Spring 2007
Dr. Lakshmi Sankar
Georgia Institute of Technology
2
Table of Contents
Table of Contents 2
Introduction 3
Methodology 5
Results 8
Conclusions and Recommendations 11
Acknowledgements 11
References 12
3
Introduction
Up until Igor Sikorsky successfully flew the VS-300 in 1939, symmetric rotor
systems were preferred by early designers, but now the single main rotor and tail rotor
design is nearly ubiquitous in the helicopter industry. Boeing has successfully used the
tandem rotor arrangement on the CH-47 Chinook and other helicopters, while Kamov is
the only company to have put a coaxial helicopter in production. However, the coaxial
rotor has been making a recent resurgence, especially with the need for maneuverable,
high-speed helicopters as well as heavy-lift rotorcraft.
A coaxial rotor has several significant advantages over other configurations. First,
the counter-rotating coaxial rotors automatically conserve angular momentum; hence, no
anti-torque system is required and the power that a single-rotor helicopter uses to drive
the tail rotor is “recovered” and used to generate useful lift and thrust in the coaxial
system. The removal of the tail rotor also allows a coaxial helicopter to be designed with
a smaller footprint since the long tail boom is no longer required for rotor separation,
which is especially important for maritime operations. Furthermore, a coaxial helicopter
is inherently directionally stable and thus safer to fly at low speeds in close quarters. The
restrictions on sideward flight in single-rotor helicopters to prevent the tail rotor from
entering vortex-ring state can be relaxed or removed in a coaxial design.
There are advantages of a coaxial rotor configuration beyond those of involving
the removal of the anti-torque device. A coaxial rotor system is aerodynamically
symmetric and therefore immune to the retreating blade stall problems of a single-rotor
system, allowing coaxial helicopters the potential fly at much high advance ratios. In
addition, coaxial rotors can achieve higher thrust coefficients making them much more
4
maneuverable than conventional single-rotor helicopters. The pilot’s workload is also
reduced in because constant pedal input is not required for longitudinal flight and the
cross coupling of controls is removed.
However, coaxial rotor systems are not without their disadvantages. There is a
significant increase in the amount of mechanical linkages and supports necessary for the
rotor control systems. Also, there must be enough separation between the two rotors so
that the blades can flap without hitting each other or the fuselage. Both of these factors
increase the parasitic drag of the aircraft as well as construction and maintenance costs.
Furthermore, there are undesirable interference effects between the coaxial rotors
increasing induced power; finally there are significant vibration and weight issues.
Despite these disadvantages, coaxial rotor systems impart many benefits over
other helicopter rotor designs. This study is a preliminary step in creating a model for
considering the hover performance of a coaxial rotor system. The XV-15 rotor tested by
McAlister, Tung, et al. (Ref. 1) was examined in FLUENT using momentum theory. In
this study a coaxial rotor system was tested at various separation distances to determine if
the model could match the experimental data.
5
Methodology
This study uses a combination of generic momentum theory and a blade element
method to calculate the power required for hover. An unstructured, axisymmetric grid
comprising approximately 240,000 tetrahedral cells was created in GAMBIT, and the
corresponding flow field was solved using FLUENT 6.2.16. The coaxial rotor system
was modeled as two actuator disks acting on an inviscid fluid, using the “fan” boundary
condition in FLUENT. The other outer surfaces of the flow field were designated as the
pressure far field, while the pressure jump applied to the fluid by the disk was set equal to
the rotor’s disk loading and constant over the disk area:
Area
ThrustLoadingDisk =
The thrust data was interpolated from the coaxial rotor plot of thrust versus non-
dimensional separation distance, S/D (Ref. 1) and the rotor area was calculated to be
approximately 1.2 m2.
The flow was given a small, initial velocity in the negative z-direction,
perpendicular to the actuator disks. FLUENT was used to solve the 3D flow field and
compute the average velocity over each disk. The calculated induced power was:
( ) ( )AvgIndind VelThrustPwr ,=
Profile power was calculated using a blade element model. Table 1 describes the
properties of each blade. All
dimensions were scaled down to
match the tip chord tested at
Ames (Ref. 1) resulting in a scale
Table 1: XV-15 Rotor and blade properties (Ref. 2)
6
factor of approximately 1/7. Because the collective pitch was referenced at the blade tip,
the inflow angle, zero lift line angle, and effective angle of attack were each calculated in
the following manner:
)(1
r
VelTanAngleInflow directionz
Ω== −−φ
twistlocaltipcollectiveAngleLineLiftZero θθθ +== ,
φθα −== effAttackofAngleEffective
Using the calculated value for αeff, the local section lift (cl) and drag (cd)
coefficients were interpolated from the airfoil polars. Since Abbott and Doenhoeff report
(Ref. 3) only had a polar for the NACA 64-208 airfoil, XFOIL 6.96 and Javafoil were
used to determine the other airfoil polars. Using the section coefficients, the section
torque was calculated:
( )[ ][ ] drrcccVelrQTorqueSection ldInd )sin()cos(2
1'
22φφρ ++Ω== ∞
where ρ∞ is 1.2 kg/m3 and c is the local chord length at that particular radial position. The
total torque was then calculated as follows:
∑=Tip
Cutout
Total TorqueSectionTorque
Since the model had a 17% root cutout, only nine of the ten points were used in
the computation. The resulting profile and total power equations are
( )( )Ω= Totalofile TorquePwrPr
( )( ) ( )( )Ω+=+= TotalAvgIndofileIndTotal TorqueVelThrustPwrPwrPwr ,Pr
This total power was then compared the actual power required, given by
7
( ) ( )Ω= ActualActual TorquePwr
The actual torque was interpolated from the plot of torque versus S/D. The
predicted and actual power required comparisons were made at various separation
distances between 0.1 and 0.8. The coaxial rotor thrust was computed using the following
formulas.
( )[ ][ ] drcccVelrT dlInd )sin()cos(2
1'
22φφρ −+Ω= ∞
∑=Tip
Cutout
TThrust '
8
Results
The ambient conditions were set to closely match those of McAlister’s
experiments for a coaxial rotor in hover out of ground effect. Specifically, the angular
velocity of the disks was 800RPM while the pressure far field was set to 101,325 Pa and
294.261 K (70°F). For each case of S/D, FLUENT calculated 4000 iterations, which was
sufficient to generate a converged solution. Initial research conducted in 2006 with a
single rotor (Figure 1) shows a significant disturbance in flow properties near the lower
boundary which was approximate 2.5 radii below the disk.
Figure 1: The rotor wake has not reached farfield conditions
Figure 2: New grid size
After several experiments with a variety of grid sizes, the lower boundary was
moved to 100 radii beneath the lower actuator disk, as shown in Figure 2.
The computational model consistently overpredicted the power required for hover
by approximately 50% and 30% for the lower and upper rotors, respectively. Upon
further analysis, the model was found to be producing around 25% more thrust than the
coaxial rotor in the wind tunnel experiments. This brings into question whether the two
rotors are actually at the same test conditions.
9
In order to compensate for this discrepancy in the model, the collective pitch of
the rotor was adjusted in the blade element code to match the thrust output of the rotor in
the experiment. The resulting induced power and total power required are plotted in
figures 3 and 4 for the lower and upper rotor respectively.
Lower Rotor - Power Required
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Rotor Separation Distance (S/D)
Po
wer
Req
uir
ed
(H
P)
Induced Power Total Power Actual Total Power - Calculated
Figure 3: Computed and actual power required for the lower rotor
Upper Rotor - Power Required
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Rotor Separation Distance (S/D)
Po
wer
Req
uir
ed
(H
P)
Induced Power Total Power Actual Total Power - Calculated
Figure 4: Computed and actual powers required for the upper rotor
10
This thrust matching adjustment gives closer correlation between the computed
and actual power required for hover. The average overprediction errors are 35% for the
lower rotor and 15% for the upper rotor. Finally, figure 5 shows a comparison between
theoretical (Ref. 4) and calculated non-dimensional inflow velocities on the upper rotor,
and figure 6 shows the streamlines superimposed on the flow velocity profile (S/D = 0.1).
Upper Rotor Inflow
0
0.02
0.04
0.06
0.08
0.1
0.12
0.0 0.2 0.4 0.6 0.8 1.0
Radial Station (r/R)
Infl
ow
Velo
cit
y ( λλ λλ
)
Theoretical Predictions CFD Computation
Figure 5: Rotor inflow comparison
Figure 6: Streamlines and velocity profile near rotor
11
Conclusions and Recommendations
For a coaxial rotor in hover out of ground effect, the hybrid momentum
theory and blade element method do not closely model actual rotor data. Even with thrust
matching, the model is not accurate enough to do hover performance predictions. It is
likely that the assumption of constant pressure over the actuator disk is the cause of the
computation errors. It is known that pressure varies along the span of the rotor blade.
FLUENT can model this using a boundary profile specification of the pressure increase
across the fan. Preliminary investigation shows a significant increase in accuracy using
this method (only 1-5% error) but more research needs to be done with this model. In
addition, FLUENT can specify a swirl velocity to be added to the fluid as it crosses the
fan. It may be possible to account for wake swirl in this fashion. Finally, an actual rotor
blade could be modeled and imported into FLUENT as a periodic wedge with a mixing
plane model between the upper and lower rotors in order to compare accuracy with actual
wind tunnel tests.
Acknowledgements
Dr. Sankar, Byung-Young Min, and Alan Egolf have been a tremendous help in
giving advice over the duration of this research project.
12
References
1McAlister, K. W., Tung, C., Rand, O., Khromov, V., and Wilson, J.S., “Experimental
and Numerical Study of a Model Coaxial Rotor,” Proceedings of the 62nd
Annual AHS
Forum, Phoenix, Arizona, May, 2006.
2Coffin, C. D., “Tilt Rotor Hover Aeroacoustics,” NASA CR 177598, June, 1992.
3Abbott, I. H.; von Doenhoff, A. E.; and Stivers, L. S., “Summary of Airfoil Data,”
NACA Report 824, 1945
4Leishman, J., Ananthan, S., “Aerodynamic Optimization of a Coaxial Proprotor,”
Proceedings of the 62nd
Annual AHS Forum, Phoenix, Arizona, May, 2006.