A Computerized System, the FANDAS Code, for Design, Flow,
Performance and Noise Predicitions of Industrial Axial FAN
Chan Lee and Hyun Gwon Kil
Department of Mechanical Engineering, University of Suwon, Hwaseong, Korea
Keywords: FANDAS, CFX, Design, Analysis, Optimization, Performance, Noise.
Abstract: FANDAS (FAN Design and Analysis System) code is developed for the design, the performance and the
noise predictions of industrial axial fan. In the FANDAS code, the 3-D geometrical designs for impeller
blades and casing of fan are made through blade angle distribution, camber line determination and blade
airfoil thickness distribution processes along blade span height, and their results are shown in GUI( Graphic
User Interface ) window. Based on the design fan geometry, the FANDAS code automatically predicts the
flow field inside fan and the overall performance map of fan by using flow deviation and pressure loss
models. The noise level and spectrum of designed fan are also evaluated by the FANDAS code which
contains noise analysis models for discrete frequency and broadband noise components of axial fan. All the
performance and the noise prediction results are displayed on GUI windows. The simulation technique of
the FANDAS code is coupled with the CFX code and applied to an actual air-conditioning fan design
practice for optimizing design variables to maximize efficiency and minimize overall noise level of fan. The
optimal fan design obtained from the FANDAS simulation results shows about 10% efficiency
improvement and 11dB noise reduction compared with the commercial market product of a reference
model, and its simulation results are well-agreed with the measurement within a few percent relative error.
1 INTRODUCTION
Axial flow fans are widely used rotating machines in
industrial, ventilation and air-conditioning systems.
However, air-born noise of the axial flow fan is strongly
related with the aerodynamic flow field and the
performance of fan, so the noise control and reduction of
fan must be attempted with the consideration of the
interaction between aerodynamic and acoustic
characteristics. For this reason, the actual fan design
practice in industry calls for reliable analysis method for
predicting both performance and noise level. With the
recent advances in computational fluid dynamics and aero-
acoustic methods, the flow, the performance and the noise
predictions have been being attempted by many previous
researchers (Belamri et al., 2005; Carolus et al., 2007).
However, because their methods are based on
computational fluid dynamics techniques so have still
shortcomings requiring a lot of input data, complicated
modeling work, long computing time and skillful
engineer’s experience for successful iterative computation,
they still remain as analysis tools and can’t be applied to
the actual design stage of axial fan.
Therefore, the present study introduces a simpler and
less time consuming fan design-analysis method,
FANDAS( FAN Design and Analysis System ), for design
and performance/ noise predictions. The fan design
process of FANDAS gives 3-D rotor and stator blade
geometries from fan design requirements and
specification, and can transfer them to CFD code.
FANDAS also analyzes the internal flow field and the
performance of designed fan by combining quasi-3D
inviscid computation scheme, flow deviation and pressure
loss models. With the predicted fan flow field and
performance data, FANDAS predicts the discrete
frequency fan noise at blade passing frequency and its
harmonics due to rotating steady aerodynamic lift and
blade by secondary and tip leakage flows. Broadband
noise of fan is predicted with the use of the correlation
model expressed in terms of the performance parameters.
The present study applies the FANDAS method
coupled with CFX code to optimize actual air conditioning
fan. Through the parametric study by FANDAS, optimal
fan design variables are firstly determined for high
efficiency and low noise, and more optimized fan
geometry is achieved by the 3-D CFD simulation of CFX
code. Furthermore, the FANDAS prediction results are
compared with measured results to verify the reliability
and the prediction accuracy of FANDAS.
331
Lee C. and Gwon Kil H..
A Computerized System, the FANDAS Code, for Design, Flow, Performance and Noise Predicitions of Industrial Axial FAN .
DOI: 10.5220/0004403303310336
In Proceedings of the 3rd International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH-2013),
pages 331-336
ISBN: 978-989-8565-69-3
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
2 DESIGN AND ANALYSIS
METHOD OF FANDAS
2.1 FAN Blade Design Method
Once fan design requirements are given from the GUI of
FANDAS as shown in Fig. 1(a), FANDAS determines fan
rotor and stator blade geometries by using conventional
design processes through spanwise blade angle
distribution, camber line determination, airfoil thickness
distribution to blade section element stacking. Fig. 1(b)
shows the fan blade geometry designed by FANDAS. For
the design versatility of fan, FANDAS contains various
airfoil geometry data base for NACA, DCA, C4 and even
curved plate, and also can export the designed fan blade
geometry to CFD code.
2.2 FAN Flow and Performance
Analysis Models
Based on designed fan blade geometry, FANDAS
conducts internal fan flow field analysis by the through-
flow technique using the assumption of axisymmetric flow
within fan blades, so all the flow variables such as flow
angle, velocity, pressure loss and so on, are computed on
the pitch-averaged flow surface. The computing scheme of
the present through-flow is applied to each streamline
from hub to tip, and uses and combines inviscid pitch-
averaged Navier-Stokes equation, Euler work equation,
flow angle and pressure loss models. The present flow
angle and pressure loss models use already well-verified
correlations available from open literature and previous
related researches (Lieblein, 1959; Lieblein, 1960;
Horlock and Lakshiminarayana, 1973; Koch and Smith,
1976; Lee and Chung, 1991).
It is noted all the flow angle and pressure loss models
are very complicated functions expressed in terms of
various fan blade design parameters and flow variables, so
the through-flow analysis requires iterative computation
process. Once iterative computation on flow filed is
carried out until satisfying overall and local mass
conservations, spanwise flow distributions for all the flow
variables can be achieved and then overall performance
parameters such as pressure rise, power and efficiency are
also calculated by the mass-averaging of computed flow
variables along blade span height( Refer to Fig. 1(c) and
Fig. 1(d) ).
2.3 FAN Noise Analysis Models
After the flow distribution and the fan operating condition
are determined by above-mentioned through-flow method,
fan noise level and spectrum are predicted by the noise
analysis method of FANDAS that is constructed as a
combination of two models for the discrete frequency
noise due to BPF ( blade passing frequency ) and blade
interaction, and for the broadband noise due to turbulent
boundary layer and wake vortex shedding. The discrete
(a) Input data window
(b) 3-D blade geometry design
(c) Spanwise flow velocity and angle distributions
(d) Aeroacoustic performance map
Figure 1: FANDAS input and output results.
frequency noise for rotating steady fan blade thrust is
analyzed by Gutin’s theory(Wright, 1976 ) where fan
blades are assumed as compact moving sources, and its
SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand
Applications
332
sound pressure level at BPF or its harmonics can be
expressed by the following equation:
baemB
eo
T
mB
mBMmBJ
MRa
NL
SP
)cos()
sin
sin(cos
dr
sc
cVBdrBL
m
span span
T
)]tan(tancos
)/(
2
2
1
[
21
2
o
e
e
ee
b
e
a
a
Nr
M
Mr
b
r
c
mBx
x
x
2
,
cos2
,
2
,,
sin
(1)
where SP
mB
is peak sound pressure at mB mode( B: no. of
fan blade, m: 1,2,3 … ), N is fan rotating frequency, L
T
is
total steady lift of fan blades which is determined by
combining cascade theory for section lift(l) and predicted
through-flow field results for flow angle(), and
velocity(V). Here, r
e
is effective radius as the 80% of fan
blade tip radius and c, b and β are fan blade chord, span
and setting angles. In addition, a
o
, R and σ represent the
speed of sound, the measuring distance and the elevation
angle form fan blade tip respectively.
The blade interaction noise due to the secondary flow
and the tip leakage flow within fan blades is produced also
at multiple BPFs ( mB mode ), and its sound pressure is
expressed by
bawwp
e
e
mB
mBELD
Rr
M
SP
sec
2
e
w
w
w
w
r
w
nB
mB
2
,
)sin(
,
o
o
p
if
if
D
90)sin(
0sincos
(2)
where L
sec
is the lift fluctuation due to secondary and tip
leakage flows, w and E are the width and the number of
load excursion, and means the azimuth angle between
blade tip and sound measuring location. In calculating
L
sec
, the present study assumes the lift fluctuation due to
secondary flow as 20% of steady lift, L
T
, and uses the
Sarajona’s correlation for the tip leakage flow (Lee and
Chung, 1991; Sjolander and Amrud, 1987 ).
The broadband noise is modeled by the Mugridge’s
correlation (Mugridge, 1976 ) as shown in eqn.(3). In his
model, fan sound power level and its spectrum are
expressed in terms of fan performance parameters such as
efficiency, flow and pressure rise coefficients, which can
be predicted through the through-flow analysis mentioned
before in the section 2-1 of this paper.
PWL(f)=K
2
+25 log
10
P
s
–2 +10log
10
Z +F
2
(f) [dB]
Z=[(1-
s
)/
s
] [(
2
+1-+
2
/2)/ ]
3/2
(3)
where Ps, , and s are fan static pressure, flow
coefficient, pressure rise coefficient and efficiency. In
addition, K
2
and F
2
(f) are given by Fig. 2.
As shown in Fig. 3, fan noise levels, sound power and
pressure spectra are obtained from the prediction results
by the above-mentioned fan noise models.
Figure 2: Flow (K
2
) and correction (F
2
) factors.
Figure 3: Fan noise spectrum predicted by FANDAS.
3 FAN DESIGN OPTIMIZATION
BY FANDAS
3.1 FAN Design Requirements
and Specifications
For verifying the reliability of FANDAS in actual fan
design practice, FANDAS is applied to the actual air-
conditioning fan design problem under the following
design requirements:
- Static pressure: 30 mmAq
- Flow capacity: 200 m
3
/min
- Efficiency: maximum
- Sound pressure level: lower than 84 dBA @ 1m
Table 1 also summarizes fan design specifications of this
study.
AComputerizedSystem,theFANDASCode,forDesign,Flow,PerformanceandNoisePredicitionsofIndustrialAxial
FAN
333
Table 1: Fan design specifications and methods.
RPM 1170 Stator chord 0.1 m
Tip diameter 0.63 m No. of stators 11
Hub/tip ratio 0.44 Angle distrib. Free vortex
Rotor chord 0.09-0.11 m
Camber design Circular arc
No. of rotors 8-12
Rotor airfoil NACA65-010
Tip clearance 0.0025 m Stator airfoil Cambered plate
3.2 Parametric Study and CFD
Verification
As shown in Table 1, this study is focused to optimize fan
rotor blades design variables such as chord length and
number of rotor blades for maximizing fan efficiency. The
parametric study results on the two design variables by
FANDAS are represented in Figs. 3-5.
From the results of Figs. 4 and 5, it is noted that the
number of rotor blades should be larger than 11 to meet
the design requirements for fan static pressure and noise
level. In addition, Fig. 6 shows the best fan efficiency of
78% is achieved at the rotor chord length of 0.1 m and the
number of rotor blades of 12, which are optimum design
conditions
.
The 3-D CFD simulation by CFX code is carried out
on the firstly optimized fan blades, and uses frozen-rotor
scheme and SST( shear stress transport ) turbulence model
for numerical computation. Fig. 7 shows the computed
results at three spanwise locations by CFX code and they
are compared with FANDAS quasi 3-D analysis results.
The predicted flow angle and velocity distributions are
well agreed between the FANDAS and the CFX.
Figure 4: Parametric studies for fan pressure.
As shown in the CFD results of Fig. 7, the effect of tip
leakage flow is very remarkable at 90% span location
while air flows at hub and mid-span regions are
streamlined along blade section surfaces with no flow
separation. However, in the aspect of pressure loss, the
CFD analysis results show the hub region produces more
pressure loss than the mid-span region. From these CFD
analysis results, it can be judged that the chord length of
rotor blade section at hub can be somewhat reduced to
Figure 5: Parametric studies for fan noise.
Figure 6: Parametric studies for fan efficiency.
decrease the pressure loss within the range to secure
attached and streamlined flow along hub blade surface.
Therefore, final optimal design of fan rotor blade is chosen
with the spanwise chord length distribution from
0.07m(hub) to 0.1m(tip) and the number of blades of 12.
3.3 FAN Performance and Noise Tests
The 3-D shapes of optimized rotor and stator are presented
in Fig. 8, and manufactured fan model is shown also in
Fig. 9. The performance and noise tests for the optimized
fan model are conducted in the chamber test facility of
KTC( Korean Testing Certification ) according to AMCA
and ISO standards. The test results at design point are
compared with the FANDAS prediction results in Table 2,
and they are very well agreed within a few percent relative
error. It is also shown from Table 2 that the optimization
by FANDAS can improve fan efficiency by 10% and
reduce noise level by 10 dB compared with the market
product of Korea, not optimized one.
SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand
Applications
334
Figure 7: Flow analysis results by FANDAS and CFX.
Figure 8: 3-D Fan blade geometries(optimized).
Figure 9: Manufactured fan model(optimized).
Table 2: Performance and noise test results.
M
odel
Pred
Flow
m
3
/min
Pressure
mmAq
Effici.
%
PWL
dB
SPL
dBA
O
ptimal
m
odel
FANDAS 200 32.4 78.2 91.3 78.09
Test 200 36.0 75.0 90.0 79.00
M
arket
p
roduct
Test 210 34.5 65.0 101.0 90.00
Figure 10: Aero-acoustic performance map.
Figure 11: FAN noise spectrum.
Fig. 10 compares the aero-acoustic performance
results by FANDAS with the test, and the FANDAS
analysis method is known to give reliable prediction for
the performance and noise level at off-design condition.
Fig. 11 also shows the good agreement between the
FANDAS prediction and the test results on the noise
spectrum at Q=220 m
3
/min.
4 CONCLUSIONS
The FANDAS code is developed with GUI for design,
flow, performance and noise analyses of axial flow fan.
AComputerizedSystem,theFANDASCode,forDesign,Flow,PerformanceandNoisePredicitionsofIndustrialAxial
FAN
335
The FANDAS code is applied to high-efficiency and low-
noise air conditioning fan development case, and its
design results are verified by the comparison with CFD
and test results within a few percent relative errors. The
fan design optimization by the FANDAS code can
improve fan efficiency by 10% and reduce noise level by
10 dB compared with the conventional market product as
baseline one.
ACKNOWLEDGEMENTS
This research work was conducted under the research
grants of the High Efficiency and Low Noise Fan
Technology Development Corporation, the Ministry of
Knowledge Economy.
REFERENCES
A T. Belamri, S. Kouidri, D. Fedala, and R. Rey, 2005,
“Comparative Study of the Aeroacoustic Behavior of
Two Axial Flow Fans with Different Sweep Angles,”
Proceedings of ASME FEDSM'05.
T. Carolus, M. Schneider and R. Hauke, 2007, “Axial
Flow Fan Broad-band Noise and Prediction,” J. of
Sound & Vibration 300, pp.207.
S. Lieblein, 1959, “Loss and Stall Analysis of Compressor
Cascades,” ASME J. of Basic Eng., 81, pp.387-400.
S. Lieblein, 1960, “Incidence and Deviation Angle
Correlations for Compressor Cascades,” ASME J. of
Basic Eng., 82, pp.575-587.
J. H. Horlock and B. Lakshminarayana, 1973, “Secondary
Flows: Theory, Experiment and Applications in
Turbomachinery Aerodynamics,” Annual Review of
Fluid Mechanics, 5, pp. 247-280.
C. C. Koch and L. H. Smith, Jr., 1976, “Loss Sources and
Magnitudes in Axial-Flow Compressors,” ASME J. of
Eng. for Power, 98, pp. 411-424.
C. Lee and M. K. Chung, 1991, “Secondary Flow Loss
and Deviation Models for Through-Flow Analysis of
Axial Flow Turbomachinery,” Mechanics Research
Communications, pp. 403-408.
S. E. Wright, 1976, “The Acoustic Spectrum of Axial
Flow Machines,” J. of Sound & Vibrations, 45(2),
pp.165-223.
S. I. Lee, 1991, A Study on the Development of Acoustical
Design Software for Rotating Machinery, KAITECH-
91-PNR0140.
S. A. Sjolander and K. K. Amrud, 1987, “Effects of Tip
Clearance on Blade Loading in a Planar Cascade of
Turbine Blade,” ASME J. of Turbomachinery, 109(2),
pp. 237-245.
B. D. Mugridge, 1976, “Noise Characteristics of Axial and
Centrifugal Fans as Used in Industry”, Shock and
Vibration Digest, 45(3).
SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand
Applications
336
