Академический Документы
Профессиональный Документы
Культура Документы
Abstract
A gearless one-motor concept for contra-rotating fans is presented in this article. The rotors are mounted to an electric
motor using only one shaft. The coupling between both rotors is realised by utilising the conservation of angular
momentum. The contra-rotating fans has a diameter of 200 mm at a design speed of 2100 min1 for the first stage
and 1200 min1 for the second stage. It has been designed and investigated through a series of experiments by the
Institute of Air Handling and Refrigeration in Dresden. The performance map and 2D particle image velocimetry
measurements have been conducted. Numerical models for 3D quasi-steady state and transient simulations have
been implemented and carried out by the Institute of Mechanics and Fluid Dynamics. The results show a good agreement
between the quasi-steady, the transient simulations and the experiment. However, when close to stall, the time-resolved
simulations show a superior performance compared with steady-state computations.
Keywords
Contra-rotating fan, gearless, computational fluid dynamics, simulation, validation
Date received: 26 August 2015; accepted: 15 January 2016
Introduction
Contra-rotating fans (CRF) reduce the swirl in the
wake ow as well as achieve higher power densities
compared with single-rotor fans (SRF) with discharge
guide vanes. In practice, this approach is already
being used to expand the application range of axial
fans due to the higher total pressure rise and increased
eciency. This is due to the improved recovery of
kinetic energy from the rst rotor. Conventional
CRF utilise one motor for each rotor or a gear for
power transmission. The compact design makes the
CRF favourable for various applications, where due
to lack of space, large industrial fans cannot be used,
e.g. tunnel and mine ventilation, as well as highloaded fans for high-speed propulsion systems.
However, using two motors or one motor with a
gear is more complicated, expensive and requires
more maintenance compared with SRF.
The CRF system has been around for many years.
The very rst experimental investigations of open
contra-rotating propellers were already being conducted in the 1930s1,2 showing the increase of eciency
compared with conventional propellers. In recent
years, more research work focusing on CRFs has
been published. Advantages of CRF has been shown
by Wang et al.3 He compared a SRF, a conventional
rotorstator stage and a CRF. The results showed
n1 P1
:
n2 P2
Heinrich et al.
Max. static
pressure rise
Rotational speed
Shroud diameter
Hub-to-tip ratio
Mass
Material
Rotor 1
Rotor 2
prel
100 Pa
40 Pa
n
d
h
m
2100 min1
0.2 m
0.45
2.13 kg
Steel sheet
1200 min1
0.2 m
0.45
0.45 kg
Polyamide
PA2200
Experimental setup
The measurements are performed on a test rig based
on DIN EN ISO 5801:2011 (see Figure 2). It has a
total length of 2.4 m and a pipe diameter of 200 mm. It
consists of three parts: (a) the air inow channel with
a curved inlet nozzle, (b) the CRF and (c) the outlet
with a ow control valve to adjust the ow rate. The
inlet nozzle is used to achieve a homogeneous velocity
prole and to measure the mass ow rate using four
pressure tabs (Ashcroft XLDP2 TSI VELOCICALC
Model 9565) at the tube walls. Four pressure tabs
at the outlet pipe are used for the calculation of the
static pressure rise of the fan. The rotational speed for
both fans is measured using a laser tachometer
(Compac Visible Optical Sensor).
The housing of the fan is made of acrylic glass to
enable optical access for stereo PIV measurements.
The Nd:YAG laser (NewWave Research Solo PIV
Laser, 50 mJ @ 200 Hz) creates a thin light sheet
30 mm behind the second fan normal to the axis of
rotation.
Two cameras (Dantec Dynamics HiSense 610 CCD
camera and Dantec Dynamics FlowSense CCD, both
1600 1200 pix2) record the particle movement for
the 2D PIV measurements. These are arranged at an
angle of 45 towards the laser sheet. A timerbox which
is connected to the tachometer of the second rotor
synchronises all cameras and the laser.
The ow is visualised using DEHS (di-ethylhexyl-sebacate) tracer particles of (0.51.5) mm size
created by a seeding generator in front of the inlet
Numerical setup
The three-dimensional, incompressible and Reynoldsaveraged NavierStokes equations are solved using
the open source computational uid dynamics
(CFD) library OpenFOAM Version 2.3.x.19 These
equations are given as follows:
r u 0,
@u
p
r uu r
r RS
@t
second interpolation for u and rst-order upwind interpolation for the turbulent quantities. Gradients are discretised with the central dierencing scheme. In case of
transient sliding mesh simulations, a rst-order explicit
Euler scheme is used for temporal discretisation. A
maximum Courant number of 2.0 is used to adjust
the time step size. The nal residuals for steady-state
simulations are set to 105. Furthermore, the static
pressure at the inlet is monitored to judge convergence.
Steady-state simulations employ the MultipleReference-Frame (MRF) approach to model the rotation of the fans. Here, all parts of the mesh are xed in
space and additional terms are added to the momentum equation to account for the rotation of the rotor.
This way, no blade passage or time-dependent eects
can be resolved. Transient simulations are performed
using the sliding mesh approach. In this case, the
mesh in the rotor is moving around the axis of rotation. The connection between the rotating and xed
mesh parts is achieved using mesh interfaces. Six revolutions with respect to the second rotor are computed
for the unsteady simulations and the results are averaged of three rotations.
The numerical model of the test bench is shown in
Figure 3. It consists of the air inow channel, the
CRF and the outlet channel. At the inlet, a mapped
boundary condition is used for the velocity and turbulent quantities to achieve a fully developed ow
prole. The average inlet velocity is varied to achieve
dierent ow rates. At the outlet, the relative static
pressure with respect to the ambient conditions is set
to zero and the other variables are treated as a zero
gradient. The uid is air with a kinematic viscosity of
1.85 105 m2/s and a density of 1.190 kg/m3.
This resembles from ambient conditions of 24 C at
101500 Pa.
The domain is meshed using an unstructured tetrahedral mesh. It is rened in the wake of the blades and
around the tips. Furthermore, an extrusion mesh has
been used for the inlet and outlet channel, where the
inlet pipe is twice the diameter and the outlet pipe four
Heinrich et al.
0.44
Experiment
Simulation
[-]
0.42
0.4
0.38
0.36
4
10
12
Results
Experiment
The experimental measurements have been conducted
by Czarnecki21 at the ILK. Two dierent setups are
investigated. The rst one is a standard SRF using the
rst rotor only. In this case, the second rotor is
replaced by a straight pipe. The other aspects of the
setup are the same as for the CRF. In the second case,
the CRF is analysed. The ow and pressure rise coefcients as well as the power coecient are dened as
follows:
V_
2
3
D n1
4
ptot
2
2 2
D n1
2
, l
PF
4
5 3
D n1
8
based on the total pressure rise ptot . The experimental measurements of the fan performance are shown in
Figure 6. The general shapes of the pressure rise coefcient curves are similar and the maxima are nearly at
the same ow rate. However, the pressure rise of the
CRF is considerably higher compared with the SRF.
The reason for this is the second rotor, which
reduces the swirl in the wake of the fan and therefore
increases the static pressure rise. Furthermore, the
maximum eciency of the CRF is increased by
approximately 5% and shifted to lower ow rates.
Additionally, the range of higher eciencies is wider
compared with the eciency peak of the SRF. This
improves the performance at o-design conditions.
Figure 7 shows the angular velocities of the rst
and second rotor for the CRF. These vary by up to
20% with respect to the design speed. This indicates
that the momentum distribution between the rst and
second stage is uneven in case of xed rotational
speeds. It further shows the advantage of the new
approach to vary the rotational speed based on the
loading of the blades. Because of this variation, the
design speed cannot be used for the numerical
Simulations
Figure 8 compares the performance map of the CRF
for the MRF and sliding mesh simulations with the
experiment. The MRF simulations show a reasonable
agreement at high and low ow rates. However, the
characteristic maximum and drop in pressure rise
are not captured. This indicates that the unstable
region of the performance curve is not modelled
1.2
0.6
Pressure rise coefficient
0.5
Efficiency
0.8
0.4
0.6
0.3
CRF
0.4
0.2
0.2
Efficiency [-]
[-]
0.1
SRF
0
0
0.1
0.2
0.3
0.4
[-]
Figure 6. Experimental measurements of the pressure rise coefficient and efficiency for the SRF and the CRF.
SRF: single-rotor fan; CRF: contra-rotating fan.
2400
2000
-1
n [min ]
1600
1200
800
-1
400
-1
0.1
0.2
0.3
0.4
[-]
Figure 7. Rotational speed of the first and second stage of the contra-rotating fan with design speed in brackets.
Heinrich et al.
0.6
0.5
[-]
0.4
0.3
0.2
Experiment
0.1
MRF
SlidingMesh
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.25
0.3
0.35
[-]
Figure 8. Pressure rise coefficient for the experiment and the simulation.
0.12
Experiment
0.11
[-]
0.1
MRF
SlidingMesh
0.09
0.08
0.07
0.06
0.05
0.05
0.1
0.15
0.2
[-]
Figure 9. Power coefficient for the experiment and the simulation. V_ 373 m3 (Off-design point).
Conclusion
This article presents a gearless one-motor concept for
CRF. For comparison, the single-rotor and the CRF
are investigated through several experiments, which
include a test rig for performance measurements and
stereo PIV for ow eld measurements. The relative
speed of the CRF is 3300 min1. As expected, the
CRF shows a signicantly higher pressure rise coecient and power coecient compared with the SRF.
The rotational speeds of the rst and second rotor
vary by up to 20% with respect to the design speed.
This shows the uneven momentum distribution
between both stages in case of xed rotational
speeds. For this reason, the proposed approach with
varying speeds based on the loading of the blades may
be better for CRF.
Figure 10. Lambda2 isosurface downstream the CRF to illustrate large turbulent structures at design point (top) and off-design
point (bottom).
CRF: contra-rotating fan.
12
Experiment
Axial Velocity [m/s]
10
MRF
SlidingMesh
8
6
4
2
0
45
55
65
75
85
95
Radius [mm]
Figure 11. Circumferential averaged axial velocity of the simulation and experiment 30mm behind the second rotor at design point.
Heinrich et al.
Figure 12. Velocity magnitude at a radius 75 mm for the steady-state (left) and transient simulations (right) at design point.
Funding
The author(s) received no nancial support for the research,
authorship, and/or publication of this article.
References
1. Lesley E. Experiments with a counter-propeller. Technical
Report 453, National Advisory Committee for
Aeronautics, 1933.
2. Lesley E. Tandem air propellers. Technical Report
689, National Advisory Committee for Aeronautics,
1939.
3. Wang J, Ravelet F and Bakir F. Experimental comparison
between a counter-rotating axial-flow fan and a conventional rotor-stator stage. In: 10th European turbomachinery conference, Lappeenranta, Finland, 2013.
4. Cho L, Choi H, Lee S, et al. Numerical and experimental
analysis for the aerodynamic design of high performance
counter-rotating axial flow fans. In: ASME 2009 fluids
engineering division summer conference, Vail, Colorado,
USA, 2009.
5. Wang W, Zhichi Z, Li S, et al. Shaft power matching and
performance prediction for contra-rotating axial flow
fans. HVAC&R Research 2007; 13: 141162.
6. Shin H-W, Whitfield C and Wisler D. Rotorrotor interaction for counter-rotating fans, part 1:
Three-dimensional flowfield measurements. AIAA
Journal 1994; 32(11): 22242233.
7. Li Q, Li Z and Lu Y. Experimental study on the
unsteady flow between rotors of counter-rotating fans.
Progress in Natural Science 2002; 12(4): 289293.
8. Grasso G, Christophe J and Schram C. Numerical
performance and accuracy of wake interaction noise
prediction models. In: 26th international conference on
noise and vibration engineering, ISMA 2014, Leuven,
Belgium, 2014.
9. Grasso G, Christophe J, Schram C, et al. Aerodynamic,
aeroacoustic and structural optimization of contrarotating fan. In: 15th international symposium on transport phenomena and dynamics of rotating machinery,
ISROMAC 2014, Honolulu, USA, 2014.
10. Nouri H, Ravelet F, Bakir F, et al. Design and experimental validation of a ducted counter-rotating axialflow fans system. Journal of Fluids Engineering 2012;
134(10): 16.
11. Sharma P, Jain Y and Pundhir D. A study of some
factors affecting the performance of a contra-rotating
axial compressor stage. Proc IMechE, Part A: Power
and Process Engineering 1988; 202: 1521.
12. Mistry C and Pradeep A. Effect of variation in
axial spacing and rotor speed combinations on the performance of high aspect ratio contra-rotating axial fan
stage. Proc IMechE, Part A: J Power and Energy 2012;
227(2): 138146.
13. Wilcox W and Wright L. Investigation of two stage
counter-rotating compressor Idesign and over-all performance of transonic first compressor stage. Cleveland,
Ohio, USA: NACA RM E56C15, 1956.
14. Wilcox W and Wright L. Investigation of two
stage counter-rotating compressor IVover-all performance of compressor with modified second-stage
rotor. Cleveland, Ohio, USA: NACA RM E58A27,
1958.
10
15. Nagano S, Takata H and Macmida Y. Dynamic performance of stalled blade flow. Bull JSME 1971; 11:
112.
16. Sharma P, Jain Y and Pundhir D. A study of some
factors affecting the performance of a contra-rotating
axial compressor stage. Proc IMechE, Part A: Power
and Process Engineering 1998; 201: 1521.
17. Roy B, Rao P, Basu S, et al. Flow studies in ducted
twin-rotor contra-rotating axial flow fans. In:
International gas turbine and aero engine congress and
exposition, Paper no. 92-GT-390, Germany, 1992.
18. Chen Y, Liu B, Xuan Y, et al. A study of speed ratio
affecting the performance of a contra-rotating axial
compressor. Proc IMechE, Part G: J Aerospace
Engineering 2008; 222: 985991.
19. OpenFOAM, Version 2.3.x, http://www.openfoam.org
(2015, accessed 1 February 2016).
20. Menter F and Esch T. Elements of industrial heat transfer prediction. In: 16th Brazilian congress of mechanical
engineering (COBEM), Uberlandia, Minas Gerais,
Brazil, 2001.
21. Czarnecki T. Aufbau und Messung eines gegenlaufigen
Axial-Kompaktlufters. Bachelor Thesis, RWTH Aachen
University, E.ON Energieforschungszentrum, 2014.
22. Joly M, Verstraete T and Paniagua G. Full design
of a highly loaded and compact contra-rotating fan
using multidisciplinary evolutionary optimization. In:
ASME turbo expo 2013, San Antonio, Texas, USA,
2013.
Appendix 1
Notation
d
h
k
m
M
n
P
p
t
u
V_
y
Diameter
Hub-to-tip ratio
Turbulent kinetic energy
Mass
Momentum
Rotational speed
Shaft power
Pressure
Time
Velocity
Flow rate
Wall distance
l
Power coefficient
Kinematic viscosity
Density
Shear stress tensor
Flow coefficient
Pressure rise coefficient
Specific dissipation