Journal of JSEM, Vol.

10, Special Issue (2010) 7-12

Copyright Ⓒ 2010 JSEM

Relationship between Internal Flow and Fan Noise of Cross Flow Fan

Kazuyoshi NISHIHARA1, Yusuke NAKAHATA2, Charles W. KNISELY3 and Manabu IGUCHI2

1
Graduate School of Mechanical Engineering, Osaka Electro-Communication University, Osaka, Japan
2
Graduate School of Engineering, Hokkaido University, Hokkaido, Japan
3
Mechanical Engineering Department, Bucknell University, Lewisburg, PA, USA

(Received 10 December 2009; received in revised form 28 April 2010; accepted 12 June 2010)

Abstract vortex and the accompanying flow around the vortex. The
The cross flow fan has an eccentrically located vortex inside effect of the tongue geometry on rotation noise level was
impeller. The vortex behavior has a large effect on the fan ascertained by flow measurements and numerical
performance and fan noise. Although investigations on the simulation of the internal flow.
internal flow of cross flow fan have been performed by
many researchers, quantitative relationship between the 2. Experimental Apparatus and Procedure
eccentrically located vortex and fan noise is not sufficiently 2.1 Measurements of fan performance and fan noise
made clear. In our previous study, we developed a noise Fig.1 shows the cross flow fan used in this study. The fan
reduction method of cross flow fan by using a step tongue had 27 blades with an outside pitch of 10 mm and 0.8 mm
and a skew tongue. Unfortunately, a detailed mechanism of blade thickness. The impeller had a 90 mm outside diameter
fan noise reduction is not known yet. In this paper the flow and was 127 mm in axial length. A 66.5 mm diameter hole
pattern and the fan noise of cross flow fan are was bored in one of the end plates to permit the insertion of
experimentally and numerically examined. the one-hole yaw meter. The tongue was flat and the casing
was composed of an arc joined to a flat duct. The
Key words experimental apparatus conforms to JIS B 8330 for the
Cross Flow Fan, Internal Flow, 3-D Numerical Simulation, configuration of air blowers, see Refs.[12, 13]. The
Fan Noise, Rotation Noise Level clearance between impeller and tongue was 3 mm in the
radial direction, and the clearance between leading edge of
1. Introduction casing and impeller was fixed at 11 mm. The flow rate was
In previous numerical simulation studies [1-3], cross flow controlled by changing the aperture ratio of the duct outlet.
fan performance and noise level were examined. In a very The pressure transducers, fan tachometer and the
recent analysis, Jeon and Cho [1] have shown through microphone were used to determine the fan performance
unsteady CFD and unsteady aeroacoustic pressure parameters and the noise level. The rotational speed of fan
calculations that “the interaction between the stabilizer and was maintained at 1400 min-1. A digital manometer was
rotating impeller generates tonal sound. Also, it is found used to record the static and total pressures in the duct. The
that the trailing edge of blade generates more acoustic noise measuring method was based on the noise level
pressure than the leading edge.” Meanwhile, Fukano et al. measurement standard JIS B 8346 for air blowers. The A-
[4-6] examined the effects of impeller, casing and tongue weighting fan noise was measured by a sound level meter,
configurations on the noise performance of cross flow fan. the data from which were input into a PC and processed via
Hayashi et al. [7] revealed that the fan noise could be the FFT contained in the LabVIEW8.0 software.
reduced by changing the blade shape. Lee et al. and Hyoung
[8, 9] conducted experiments on the effects of blade and
tongue configuration on noise reduction.
In our previous study, we developed a noise reduction
method of cross flow fan by using a step tongue and a skew
tongue [10, 11]. Unfortunately, a detailed mechanism of fan
noise reduction is not known yet.
In this study, the relationship between internal flow and
fan noise reduction of cross flow fan was experimentally
and numerically examined. The center of the eccentrically
located vortex was inferred from flow measurements. A
one-hole yaw meter [12, 13] was used to obtain the
pressure distributions along the inner and outer peripheries
of impeller. The behavior of the eccentrically located
vortex was estimated from the results of pressure
distributions.
A 3-D numerical analysis was carried out and was
compared with the experimental result. The position of
eccentrically located vortex was mainly influenced by air
flow rate. The rotation noise level was inferred to result
from the change in the position of the eccentrically located Fig.1 Experimental apparatus

--
 K. NISHIHARA, Y. NAKAHATA, C. W. KNISELY and M. IGUCHI

Fan performance was characterized in terms of the tanβ2=vr2/(vθ2－u2) (7)

total pressure coefficient ψt,

following three parameters [12, 13]:
where, vr1, and vr2 are the radial velocity component on the

ψt=2pt/ρu22
inner and outer peripheries, respectively, vθ1 and vθ2 are the
(1) circumferential velocities on the inner and outer peripheries,

flow coefficient φ,
respectively. In addition, u1 and u2 are the rotational speeds
of impeller based on the inner and outer diameters,

φ=Q/DLu2
respectively. The above-mentioned one-hole yaw meter was
(2) traversed to measure vr1, vr2, vθ1 and vθ2.

specific noise level K, 3. 3-D Numerical Simulation Method

The flow analysis software SCRYU/Tetra V7 (CRADLE
K=SPL－10log10Qpt2 (3) Co., Ltd.) was used to analyze the internal impeller flow.
Figure 4 shows the 3-D mesh used for the numerical
simulation of the internal flow of the cross-flow fan. The
where pt is the total pressure, ρ is the density of air, u2 is the
number of mesh points in the whole flow field was
rotational speed of impeller based on the outer diameter, Q
5,650,000.
is the flow rate, D is the diameter of impeller, L is the width
In the 3-D numerical simulation, three types of boundary
of impeller, SPL is the sound pressure level.
layer elements were used to precisely estimate the flows in
the close vicinity of the blades, the end plates, the casing
2.2 Measurement of internal flow
and the tongue. The working fluid was air at a temperature
To measure the internal flow, the one-hole yaw meter with a
of 20oC. The atmospheric pressure was used as the inlet
1.4 mm outside diameter, 1.0 mm inside diameter and 0.5
boundary condition of the system, and mass flow rate was
mm pressure tap diameter was used. Peripheral distributions
used as the outlet boundary condition. The rotational speed
of total pressure pt, dynamic pressure pv and static pressure
of fan was kept at a constant value of 1400 min-1. The
ps along the inner and outer peripheries of the impeller were
obtained to determine the internal flow pattern. The inner standard k-ε model was used as the turbulence model.
and outer periphery radii were Rin=30.0mm and
Rout=51.0mm, respectively. The yaw probe was positioned
at five axial locations and then rotated in 10o increments
around the axis of the impeller as shown Fig.2. The axial
measurement locations were z1=6.4, z2=31.8, z3=63.5,
z4=95.3, z5=120.7mm from the end of the impeller (opposite
from the motor).
To investigate the effects of inflow and outflow velocities,
w1 and w2, shown in Fig.3 on the fan noise, these velocities
were determined from the following equations [13].

w1=vr1/sinβ1 (4)

w2=vr2/sinβ2 (5)

tanβ1=vr1/(vθ1－u1)
Fig.3 Relative velocities on the inner and outer peripheries
(6)

Inlet

Side plate
Impeller Outlet

Measurement duct
y

z x

Fig.4 The meshes set in calculation

Fig.2 Measurement position by a one-hole yaw meter

--
Journal of JSEM, Vol.10, Special Issue (2010)
4. Results and Discussion of Experiments and 3-D
Numerical Simulation
Fig.5 shows the total pressure coefficient ψt and total
4.1 Fan performance and fan noise
pressure efficiency ηt versus the flow coefficient φ. The
chain line drawn between φ= 0.3 and 0.7 is the results of 3-
D numerical simulation. The values of total pressure
agree with each other for φ= 0.3 to 0.5, while the two
coefficient,ψt , of experiment and 3-D numerical simulation
coefficient, φ. This difference is associated with the fact
results become to depart with a further increase in the flow
simulation becomes greater than the measured value for φ
that the static pressure obtained from the 3-D numerical
≥0.7. The reason for the increase in the numerical value of
the static pressure is not clear at present and must be left for
a future study. The maximum value of the total pressure
efficiency is 39% and it appears forφ= 0.5. The noise level,
SPL, and the specific noise level, K, shown in Fig.6
noise becomes significantly low for φ≥0.7. Fig.7 shows the
gradually decrease as the flow rate increases. The rotation
frequency analysis data for φ=φ max andφ= 0.3, where
φmax denotes the maximum value of φ.
Fig.5 Relation between total pressure coefficient ψt and
total pressure efficiency ηt
Fig.6 SPL, specific noise level K and rotation noise level
--
As the rotation noise level is extremely lower than overall
noise level, as can be seen in Fig.7 (a), the turbulence noise
is dominant at this flow rate. On the other hand, Fig.7 (b)
shows that the rotation noise level and the overall noise
an essential role for φ =0.3. These results collectively
level are comparable. Accordingly, the rotation noise plays
indicate that the rotation noise is significant in the low flow
rate regime, while the turbulence noise is dominant in the
high flow rate regime.
4.2 Position of eccentrically located vortex center
The vortex center is located on the angular location of θ =
θVC [10, 13] where the total pressure, pt, exhibits its
minimum value. Fig.8 shows the distribution of vortex
center along the impeller axis obtained from 3-D numerical
simulation in the high flow rate regime. A good agreement
can be seen between the numerical and measured values of
distribution for φ=0.7 is almost flat as shown in Fig.8,
angular displacement of vortex center. The vortex center
whereas that for φ=0.3 is not flat as shown in Fig.9. In the
case that the turbulence noise is the main noise source, the
vortex center is located near the edge of the tongue.
Meanwhile, when the rotation noise is the main noise source,
the vortex center departed from the edge of the tongue.
(a) φ=Max (high flow rate) (b) φ=0.3 (low flow rate)
Fig.7 Spectral SPL
Fig.8 Vortex center angle θVC for φ=0.7

Fig.9 Vortex center angle θVC forφ=0.3
4.3 Relative velocity vector along the impeller
Figs.10 and 11 show the relative velocity vectors, w1 and w2,
for φ = 0.7 and 0.3, respectively. The simulated results
obtained from experiments and 3-D numerical simulation
based on the 3-D numerical simulation are in good
agreement with the experimental ones in the two cases.
Definite difference cannot be seen between the results
shown in Figs.10 and 11. Consequently, it is difficult at
present to discuss the effects of the relative velocity vectors
on noise generation.
φ=0.7 and z3=63.5mm (high flow rate)
Fig.10 Relative velocity vector of internal flow for
φ=0.3 and z3=63.5mm (low flow rate)
Fig.11 Relative velocity vector of internal flow for
-10-
K. NISHIHARA, Y. NAKAHATA, C. W. KNISELY and M. IGUCHI
4.4 Recirculating streamline of internal flow
obtained from 3-D numerical simulation for φ= 0.7 and 0.3,
Figs.12 and 13 show the streamlines at z3 = 63.5mm
respectively. The center of the eccentrically located vortex
is near the edge of the tongue, as shown in Fig.12. The
velocity gradient near the flow field between the vortex
center and the edge of the tongue is very sharp, suggesting
that the flow is highly turbulent there. The turbulence noise
the vortex center for φ= 0.3 separated far from the edge of
therefore would be the main noise source. On the other hand,
sharp compared to that for φ=0.7. The turbulence intensity
the tongue and, as a result, the velocity gradient is not so
also would not be so high. Under this condition the flow
passing through the clearance between the blade and the
edge of the tongue would play a main role on the noise
generation. In other words, the rotation noise would be the
main noise source.
Vortex center
Edge of tongue
Fig.12 Streamline of internal flow for φ=0.7 and
z3=63.5mm (high flow rate)
Fig.13 Streamline of internal flow for φ=0.3 and
z3=63.5mm (low flow rate)
5. Development of Noise Reduction Method
5.1 Nine configurations of tongues
In our previous study [11], the fan noise could be reduced
by alteration of the tongue shape. Unfortunately, a detailed
mechanism of the fan noise reduction is not known yet. In
this section, we mention about noise reduction caused by
the alteration of tongue shapes. The nine different dividing
tongue geometries investigated in this study are shown in
Fig.14. These geometries can be summarized as follows:
Tongue A affords the reference condition (default tongue
position, taken as the reference datum). The shaded regions
in Fig.14 show those portions of the dividing tongue
geometry that are at the default reference position. Tongues
B and C are both still flat tongues, but positioned one-half
Journal of JSEM, Vol.10, Special Issue (2010)
blade pitch above or below the datum. Tongues D, E, F and
G have a half pitch offset over a half of the tongue width.
The tongue D is a mirror image of the tongue E. The
tongues F and G are rotation of the tongues E and D about
the tongue centerline, respectively.
5.2 Experimental results and discussion
5.2.1 Fan performance
The fan performance is essentially unaffected, or very
slightly improved, by modifications to the dividing tongue
by comparison of the performance of the modified tongues
(step tongue D, skew tongue I) with that of flat tongue A in
Fig.15. Meanwhile, Fig.16 shows a relatively significant
decrease in the SPL and K, especially for tongues D and I.
The SPL are reduced except for high flow rate (φ≥0.7).
Fig.14 Nine tongues of configuration
Fig.15 Relation between total pressure coefficient ψt and
total pressure efficiency ηt for tongues A, D and I
Fig.16 SPL and specific noise level K for tongues A, D and
I
-11-
5.2.2 Fan noise performance
tongues D and I for φ= 0.3 are shown in Fig.17 (a) and
The spectral analysis of the acoustic signals from dividing
Fig.17 (b), respectively. Tongues D and I result in almost 17
dB noise reduction at the blade passing frequency relative to
that of tongue A. The overall noise level for tongue D was
59.7 dB and that for tongue I was 63.1 dB. There was some
reduction in SPL for the other tongues as well. The skew
tongue I reduced not only the rotation noise level but also
second rotation noise level as shown Fig.17 (b).
The relationship between the overall noise level and the
in Fig.18. Fig.18 (a) shows the overall noise level for φ=φ
rotation noise level for the nine different tongues are shown
max differs more than 17 dB from rotation noise level. At this
flow rate, turbulence noise (sometimes called the broad
Fig.18 (b) shows the noise data for a low flow rate of φ=
band noise [1, 8]) is the main noise source. Meanwhile,
0.3. The rotation noise levels of flat tongues A, B and C are
nearly equal to the overall noise level. However, the
rotation noise levels of the tongues E to I are vastly reduced.
Accordingly, in this case the rotation noise is the main noise
source. The alteration of tongue shapes therefore is very
effective.
(a) Step tongue D (b) Skew tongue I
Fig.17 Spectra SPL for tongue D and I (φ=0.3)
(a) φ=φmax
Fig.18 Overall noise level and rotation noise level for nine
types of tongue
 K. NISHIHARA, Y. NAKAHATA, C. W. KNISELY and M. IGUCHI

http://ecim.co.kr/down/Aeroacoustic%20noise%20sou
rce%20of%20Air-
conditioner%20including%20CFF.PDF as downloaded
(2005).
[2] Yong J. M., Cho, Y. and Hyunsik, N.: Numerical
Prediction of the Cross-Flow Fan Performances and
Noise Characteristics by Unstructured Flow Solver
Algorithm, Proceedings of 4th KSME-JSME Fluids
Eng., (1998), 185-188.
[3] Yong, C. and Young, J. M.: Discrete Noise Prediction
of Variable Pitch Cross-Flow Fans by Unsteady
Navier-Stokes Computations, Trans, ASME J. Fluids

(b) φ=0.3
Eng., 125(2003), 543-550.
[4] Fukano, T. Hara, T. Yamashita, Y. Kinoshita, K. and
Nomiyama, S.: Reduction of Noise Generated by a
Fig.18 Overall noise level and rotation noise level for nine Cross Flow Fan (Part 1: The Effect of Geometries of a
types of tongue
Tongue and a Rotor) (in Japanese), Turbomachinery,
20-8(1992), 464-470.
6. Conclusions [5] Fukano, T. Hara, Y. Yamashita, Y. and Kinoshita, K.:
Experimental and numerical investigations were curried out Reduction of Noise Generated by a Cross Flow Fan
to reveal the relationship between the flow pattern and the
(Part 2: The Effect of the Clearance between a Tongue
fan noise of cross flow fan. The main findings obtained in and a Rotor) (in Japanese), Turbomachinery, 21-
this study can be summarized as follows:
6(1993), 350-357.
(1) The characteristics of internal flow inside impeller
[6] Fukano, T. Hara, Y. Yamashita, Y. and Kinoshita, K.:
obtained from 3-D numerical analysis are in good
Reduction of Noise Generated by a Cross Flow Fan
agreement with those experimentally measured under
(Part 3: The Effect of Scroll Shape) (in Japanese),
every experimental condition.
Turbomachinery, 21-8(1993), 466-472.
(2) The rotation noise is the main noise source in the low
[7] Hayashi, T., Kobayashi, Y., Nagamori, A. and Horino,
flow rate regime, while the turbulence noise is the main
T.: Low-Noise Design for Cross-Flow Fan Based on
noise source in the high flow rate regime. Change in the
Frequency Modulation (in Japanese), Trans. Jpn. Soc.
main noise source is closely associated with the shift of
Mech. Eng., 62C-601(1996), 3446-3451.
the center location of the eccentric vortex.
[8] Lee, D. S., Chen, P. H. and Miao, J. M.: Noise
(3) In the case that the rotation noise is the main noise
Reduction of a Cross-Flow Fan, J. Chin. Soc. Mech.
source, the alteration of tongue shapes is very effective
Eng., 20-3(1997), pp.265-273.
for reducing the noise level.
[9] Hyoung, M. K.: Discrete Frequency Noise Reduction
of the Cross-Flow Fan of the Split Type Room Air-
Nomenclature Conditioners Using the Skewed Stabilizers, JSME
D diameter of impeller, m
International Journal, 43-1(2000), 104-109.
K specific noise level, dB
[10] Nishihara, K., Knisely, C. W., Nakahata, Y., Kashima,
L impeller width, m
K. and Iguchi, M.: Study of Noise Reduction in Cross
pt total pressure, Pa
Flow Fans with Annular Bands, J. JSEM Special Issue,
ps static pressure, Pa
(2009), in press.
Q flow rate, m3/s
[11] Nakahata, Y., Knisely, C. W., Nishihara, K. and Iguchi,
SPL sound pressure level, dB
M.: Development of Noise Reduction Method and
u rotational velocity, m/s
Visualization Studies in Cross-Flow Fan, J. JSEM, 7-
x axial coordinate 4(2007), 349-354.
y vertical coordinate [12] Murata, S. and Nishihara, K.: An Experimental Study
z horizontal coordinate

φ
of Cross-Flow-Fan (1st Report, Effects of Housing
ρ density, kg/m3 Geometry on the Fan Performance) (in Japanese),

ψt
flow coefficient Trans. Jpn. Soc. Mech. Eng., 41-347(1975), 2062-
total pressure coefficient 2075.
[13] Murata, S. and Nishihara, K.: An Experimental Study
References of Cross-Flow-Fan (2nd Report, Fan Performance
[1] Jeon, W. H. and Cho, J.H.: Analysis of Aeroacoustic Curve and inside Flow) (in Japanese), Trans. Jpn. Soc.
Noise Source of an Air-Conditioner Including Cross Mech. Eng., 41-347(1975), 2076-2089.
Flow Fan”,

-12-