2. 2396 IEEE TRANSACTIONS ON MAGNETICS, VOL. 48, NO. 8, AUGUST 2012
TABLE I
SIMULATION CONDITIONS
TABLE II
EXPERIMENTAL CONDITIONS
A. Numerical Simulation
The large eddy simulation (LES) with the dynamic
Smagorinsky model was employed as the turbulence model in
order to reproduce precise and accurate air flow behavior inside
the HDD. Fig. 1(a) shows the numerical simulation model. We
modeled most of the key components such as the disk, base,
arm, yoke, bracket, and filter in the CFD model. We imposed
periodic boundary conditions on both the upper and lower
boundaries of the fluid domain. The rotational velocity was
applied to the surfaces of the disks and the spacer between two
disks. Table I summarizes the simulation conditions.
B. Experimental Visualization
To clarify the simulation results and confirm the actual air
flow behavior inside the HDD, we conducted flow visualization
based on a particle scattering technique using a laser sheet. For
the flow visualization between the co-rotating disks, we build a
special drive, as illustrated in Fig. 1(b). This drive has a clear
side window, a clear cover, and clear glass disks. Oil tracer par-
ticles were seeded into the drive using an atomizer. A thin laser
sheet was injected into the drive through its clear side window,
and it illuminated the tracer particles in the mid-plane between
the co-rotating disks. The illuminated tracer particles made it
possible to visualize the air flow, and the flow images were cap-
tured by a charge-coupled device (CCD) through the clear disk
and clear cover. The air was used as the working fluid to mea-
sure the actual air flow behavior inside the HDD. Table II sum-
marizes the experimental visualization conditions.
Fig. 2. Simulated instantaneous velocity magnitude distributions: (a) trian-
gular, (b) tetragonal, (c) pentagonal, and (d) hexagonal flow patterns.
III. RESULTS AND DISCUSSIONS
A. Numerical Simulation
Fig. 2 illustrates the instantaneous in-plane directional ve-
locity magnitudes at the mid-plane of two co-rotating disks after
the flow was fully developed. Turbulent flow behavior was ob-
served in the outer diameter (OD) region. In the inner diameter
(ID) region, the air behaved as a central core body in the manner
of a laminar flow. Between the turbulent region at the OD and
the laminar region at the ID, a polygonal flow pattern rotating
with the disk rotation was observed. Furthermore, we found that
over time the polygonal number varied repeatedly between three
and six in the actual HDD. We observed triangular, tetragonal,
pentagonal, and hexagonal flow patterns in the actual HDD, as
illustrated in Fig. 2. Even though the triangular and tetragonal
flow patterns are relatively not clear compared to other patterns,
in Fig. 2 the polygonal numbers were determined based on the
spectrogram analysis. The flow pattern changed from one pat-
tern to another in a quasi random manner. Duration time of one
pattern typically corresponded to the rotation time for one to
four disk revolutions.
Fig. 3 plots the power spectrum density of the circumferen-
tial directional air velocity at three radius positions. In the ID the
flow was laminar, and the velocity fluctuation was the smallest.
In the OD the flow was turbulent, and the velocity fluctuation
was quite large compared to the ID. Furthermore, sharp peaks
were observed in the mid-diameter (MD) and OD positions. The
first four peak frequencies in the MD were 200, 310, 370, and
3. KUBOTERA et al.: COMPUTATIONAL FLUID DYNAMICS AND EXPERIMENTAL VISUALIZATION OF TIME-VARIABLE AIR FLOW PATTERN 2397
Fig. 3. Simulated power spectrum density.
450 Hz. These frequencies correspond to the triangular, tetrag-
onal, pentagonal, and hexagonal flow patterns observed in the
velocity magnitude contours in Fig. 2.
Here, for the flow between two fully shrouded co-rotating
disks, the relation between the polygonal number and the peak
frequency has been experimentally well studied [7] and is ex-
pressed by
(1)
where is the vortex frequency, is the polygonal number,
and is the disk rotation frequency (Reynolds numbers
, and the ratios of the gap between two disks to
the disk radius ). If we apply the param-
eters used in the simulation into (1), we can expect vortex fre-
quencies of 180, 270, 360, and 450 Hz for the triangular, tetrag-
onal, pentagonal, and hexagonal flow patterns, respectively, for
the fully shrouded cases. These frequencies basically coincide
with those appearing in the power spectrum density of Fig. 3.
The earlier studies found the time-invariant polygonal pattern
with a fixed polygonal number. However, we found that the
polygonal number of the polygonal flow pattern varied period-
ically inside the actual drive structure.
B. Experimental Visualization
To clarify the existence of the time-varying polygonal pattern,
the actual air flow inside the HDD was experimentally measured
under the same conditions as in the numerical simulation. Fig. 4
illustrates the results of instantaneous flow visualization at the
mid-plane between two co-rotating disks. The laser sheet was
injected from the right side of the figures. Because the spindle
motor and spacers are opaque to the laser sheet, the left portions
of the figures are in shadow. Similar to the simulation results,
the flow in the OD was turbulent, and the air was well stirred,
resulting in a uniform and highly dense tracer mist, as shown
in Fig. 4. In the ID the air behaved as a central core body in
the manner of a laminar flow, resulting in nonuniform density
of the mist. Between the turbulence region in the OD and the
laminar region in the ID, the polygonal patterns were observed.
Fig. 4. Experimental instantaneous images of air flow visualization: (a) trian-
gular, (b) tetragonal, (c) pentagonal, and (d) hexagonal flow patterns.
Furthermore, as in the simulation results, the triangular, tetrag-
onal, hexagonal, and pentagonal flow patterns were confirmed
inside the actual HDD. These experimental results revealed the
existence of polygonal flow patterns and the time-variation of
the polygonal pattern with disk rotation inside an actual hard
disk drive.
C. Discussion
The time-variable polygonal flow structure acts as a distur-
bance torque against the arm. It excites the rigid mode vibra-
tion of the arm and degrades the head positioning accuracy in
the wider frequency range less than 1 kHz. A possible solution
for suppressing the polygon structure and its degradation of the
head positioning error might be to put the spoiler between the
disks, which might distort the polygonal structure and reduce
the flow induced arm vibration.
IV. CONCLUSION
The air flow inside an HDD was analyzed using CFD and
experiment. The existence of rotating polygonal flow patterns
inside the actual HDD was directly confirmed both by sim-
ulation and experimental flow visualization. Furthermore, we
found that over time the polygonal number varied repeatedly in
the actual HDD and we observed triangular, tetragonal, pentag-
onal, and hexagonal flow pattern both in simulation and exper-
imental visualization. The outcomes of this study contribute to
build basis of understanding of the actual air flow behavior in-
side the HDD to accomplish the higher areal density HDDs.
ACKNOWLEDGMENT
The authors would like to thank Prof. S. Obi and Mr. T.
Washizu from Keio University, Japan, for their useful sugges-
tions and discussions.
4. 2398 IEEE TRANSACTIONS ON MAGNETICS, VOL. 48, NO. 8, AUGUST 2012
REFERENCES
[1] K. Aruga, “3.5-inch high-perfomance disk drives for enterprise appli-
cations: AL-7 series,” Fujitsu Sci. Tech. J., vol. 37, no. 2, pp. 126–139,
Dec. 2001.
[2] E. Lennemann, “Aerodynamic aspects of disk files,” IBM J. Res. De-
velop., pp. 480–488, 1974.
[3] R. Kaneko, S. Oguchi, and K. Hoshiya, “Hydrodynamic characteristics
in disk packs for magnetic storage,” Rev. Elect. Commun. Labs, Nippon
Telegraph and Telephone Public Corp., Japan, vol. 25, pp. 1325–1336,
1977.
[4] S. D. Abrahamson, J. Eaton, and D. J. Koga, “The flow between
shrouded co-rotaing disks,” Phys. Fluids, vol. 1, no. 2, pp. 241–251,
1989.
[5] J. A. C. Humphrey, C. A. Schuler, and D. R. Webster, “Unsteady lam-
inar flow between a pair of disks corotating in a fixed cylindrical en-
closure,” Phys. Fluids, vol. 7, no. 6, pp. 1225–1240, 1995.
[6] M. Tatewaki, N. Tsuda, and T. Maruyama, “An analysis of disk flutter
in hard disk drives in aerodynamic simulations,” IEEE Trans. Magn.,
vol. 37, no. 2, pp. 842–846, Mar. 2001.
[7] D.-W. Kong, “Behavior of Periodic Flow Disturbances in Shrouded
Co-Rotating Disks,” Ph.D. dissertation, Yonsei University, Korea,
2008.
[8] J. Mizushima, G. Sugihara, and T. Miura, “Two modes of oscillatory
instability in the flow between a pair of corotating disks,” Phys. Fluids,
vol. 21, p. 014101, 2009.
[9] M. Ikegawa, H. Mukai, T. Sugii, and M. Watanabe, “Fluid simulation
and measurement of polygonal airflow pattern in HDD,” in APMRC
2010 Dig.
[10] K. Sundaravadivelu, Q. D. Zhang, N. Y. Liu, E. H. Ong, T. H. Yip, G.
L. Chin, and J. Q. Mou, “Flow-induced slider vibration in a functional
hard disk drive: Influence of air shroud,” IEEE Trans. Magn., vol. 45,
no. 11, pp. 4923–4928, Nov. 2009.
[11] M. Kazemi, “Investigation of fluid structure interaction of a head stack
assembly in a hard disk drive,” IEEE Trans. Magn., vol. 45, no. 12, pp.
5344–5351, Dec. 2009.
[12] M. Ikegawa, H. Mukai, and M. Watanabe, “Airflow-simulation by
voxel mesh method for complete hard disk drive structure,” IEEE
Trans. Magn., vol. 45, no. 11, pp. 4918–4922, Nov. 2009.
[13] H. Min, X. Huang, and Q. Zhang, “Aerodynamic pressure fluctua-
tions associated with flow-induced vibration of the head gimbals as-
sembly inside a hard disk drive,” IEEE Trans. Magn., vol. 48, no. 1,
pp. 101–106, Jan. 2012.