Sunteți pe pagina 1din 5

EFECTS OF ELEMENT FORMULATIONS OF 8-NODED

HEXAHEDRON ELEMENT ON HDD DYNAMICS UNDER SHOCK


IMPULSE

Arief Yudhanto, Ong Eng Teo∗, Ong Eng Hong


Data Storage Institute A*STAR, Singapore

Abstract

In the finite element modeling of hard disk drive (HDD), components such as
disk, actuator arm or base casting may be modeled using solid element. The 8-
noded hexahedron element is often the preferred choice due to its low computational
cost compared to its counterpart such as 20-noded hexahedron. When transient
analysis is performed, i.e. shock simulation, HDD component assembled with 8-
noded hex may behave differently depending on the element formulation used.
Although element formulation may not be a big issue to the general FEM users,
however, it may give unexpected response pertaining to HDD dynamics. This paper
presents numerical studies on the effect of using different element formulations for 8-
noded hexahedron element in the HDD shock simulations. The Hitachi 1.8 inch HDD
components were used in this study. Two components, namely disk and actuator
arm, were built using 8-node hexahedral element. Shock impulse of half-sine wave
(500G/2ms) was applied to the individual model, while ‘simply-supported’ boundary
condition was applied. Two element formulations are studied: selectively-reduced
integration (SRI) and full integration (FI). FI is studied using ANSYS solver and SRI
is studied using ANSYS/LS-DYNA. Numerical studies show that FI can give
converged results with coarse mesh. On the other hand, SRI needs finer mesh
before converged results can be attained. Comparison was also made between FI
(coarse mesh) and SRI (coarse mesh) in head-suspension-assembly (HSA) – disk
model. The model was also subjected to 500G/2ms shock impulse. Flying height
modulation of slider during operating condition is plotted. Effects of element
formulations on disk-suspension dynamics are discussed.

Keywords: hard disk drive, shock, finite element method, hexahedron element, full integration,
selectively-reduced integration

1. Introduction

As the use of hard disk drives (HDD) in consumer electronics is growing, the shock performance of
such data storage device is of prime concern. Studies of HDD shock response aim to characterize the
mechanical behaviors of HDD components, where the onset of failure may be obtained from this
process. For more than a decade, finite element method (FEM) has been a preferred analysis tool in


Corresponding author (Email: ONG_Eng_Teo@dsi.a-star.edu.sg)
these studies. Many researchers published their works on simulations of HDD at both component and
system levels, and in operating and non-operating conditions (Edwards [1], Chen-Chi Lin [2] and Zeng
and Bogy [3], Aristegui and Geers [4], Jayson et al. [5], Luo et al. [6], Shu et al. [7]).
Flying height modulation of slider, contact force at head-disk interface, displacement of HDD
component during shock application are some of simulation parameters in time-history analysis
describing shock behavior of HDD. The simulation involves geometrical modeling, component
discretization, selection of integration methods and time-stepping method (by which the choice of
solver is governed) and other preprocessing stages. In component discretization stage, selected
element type as well as the corresponding formulation will govern the dynamic behavior of HDD
component.
This paper studies the effects of element formulations of 8-noded hexahedron on HDD dynamic
behavior under shock impulse. This type of element is used to build actuator arm, disk and slider. Two
element formulations were utilized, namely selectively-reduced integration (SRI) and full integration
(FI). Eight-noded hexahedron with FI (named SOLID45) will be solved using ANSYS, while 8-noded
hexahedron with SRI (named SOLID164) will be solved using ANSYS/LS-DYNA [8]. Finite element
models of disk, actuator arm and HSA-disk are built. The model is subjected to shock impulse of half-
sine-wave (500G/2ms). Dynamic response of structures, i.e. displacement, obtained from time-history
analysis was recorded. Verification is performed to check the convergency of results upon mesh
refinement. For HSA-disk model, simulation is performed to study the effect of using coarse mesh for
actuator arm to the flying height modulation during operating condition.

2. Element Formulations of 8-Noded Hexahedron

2.1 Full integration of 8-noded hexahedron in ANSYS

As mentioned, 8-noded hexahedron element in ANSYS is named SOLID45. By default,


SOLID45 is formulated using full integration. Full integration is defined as a quadrature rule sufficient
to provide the exact integrals of all terms kij in the element stiffness matrix [k] if the element is
undistorted.

[k ] = [B]T k [B]J dξ dη dζ

Full integration is considered to be conservative since it assumes that an element is of arbitrary


geometry. Hence, in order to integrate the volume dV = J dξ dη dζ correctly, 8-noded hexahedron
element uses 2X2X2 order (8 Gauss points). This option comes with extra displacement shape to
alleviate shear locking effect [9]. In addition, time step in ANSYS can be determined by user.

2.2 Selectively-reduced integration in ANSYS/LSDYNA

In ANSYS/LS-DYNA, 8-noded hexahedron element is named SOLID164. By default,


SOLID164 is formulated using reduced integration with ‘LSDYNA hourglass control’. In present
exercise we use selectively-reduced integration which may have faster convergency compared to
reduced integration upon mesh refinement. SRI in ANSYS/LSDYNA incorporates a technique to
control the hourglass effect, namely enhanced assumed strain with physical stabilization. This
formulation was developed by Puso [10]. In ANSYS/LS-DYNA, time step is automatically set by the
software obeying Courant-Friedrich-Lewy stability limit.

3. Numerical Results

Three models were built; they are disk, actuator arm and HSA-disk. Shock impulse of half-
sine wave (500G/2ms) was prescribed in the model, and ‘simply-supported’ boundary condition was
applied in the in-plane directions.

3.1 Shock response of disk

Figure 1-a depicts the disk model used in this study. The disk has an outer diameter (OD) of
48 mm, inner diameter (ID) of 12 mm and thickness of 0.5 mm. The disk material is glass-ceramics
-6 3
with E = 90 GPa, ν = 0.24 and ρ = 2.53 x 10 kg/mm . Shock impulse was applied vertically to the
inner diameter of the disk. In ANSYS, time step is defined to be 10-6 second. This time step is chosen
because the solution is already converged and the simulation is conditionally stable. In
ANSYS/LSDYNA, the time step is automatically determined by the solver. The time step in
-8
ANSYS/LSDYNA is around the order of 10 second.

0.00006

0.00004

Shock impulse 0.00002

0
0.00E+00 5.00E-04 1.00E-03 1.50E-03 2.00E-03 2.50E-03 3.00E-03

-0.00002

Uy (m)
ID OD
-0.00004

-0.00006

FI-coarse (384 elements)


-0.00008
FI-fine (1200 elements)

SRI-coarse (384 elements)


-0.0001
SRI-fine (38400 elements)

-0.00012
time (seconds)

(a) (b)

Figure 1. FE model of disk and time-history response of outer diameter for two formulations (FI of
ANSYS and SRI of ANSYS/LSDYNA)

The dynamics response of a node at the outer diameter (OD) which corresponds to the
maximum displacement of the disk was recorded (Figure 1-b). From Figure 1-b we can observe that
the curves of FI (both coarse mesh and fine mesh) follow closely with each other. It indicates that the
formulation is coarse-mesh-accurate. Using SRI, the displacement at OD of coarse-mesh model
seems to be underestimated. However, upon refinement the displacement of OD using SRI seems to
approach those of FI.

3.2 Shock response of actuator arm

Figure 2-a depicts an actuator arm model, which is built from stainless steel with E = 200
GPa, ν = 0.32 and ρ = 7.80 x 10-6 kg/mm3. Likewise, half-sine sinusoidal acceleration impulse
(500G/2 ms) was applied vertically to the pivot hole of the arm. The dynamics responses of a node at
arm tip were noted for different meshes and element formulations. The time history response curves
were plotted in Figure 2-b. It is again observed that the curves for the FI formulations follow closely
with each other, which indicates the coarse-mesh-accuracy of the element formulation. On the other
hand, the significant difference in the response curves of SRI formulation was noted between the
coarse and fine mesh cases. It is also noted that the SRI-fine curve seems to approach the FI curves,
which suggests that the solutions of SRI formulation is likely to converge to the FI results if further
mesh refinements are carried out.
0.00004

0.00002

0
0.00E+00 5.00E-04 1.00E-03 1.50E-03 2.00E-03 2.50E-03 3.00E-03
Pivot -0.00002

hole
-0.00004
Uy (m)

-0.00006

-0.00008

FI-coarse (327 elements)


-0.0001
FI-fine (662 elements)

SRI-coarse (327 elements)


-0.00012
SRI-fine (2659 elements)

-0.00014
time (seconds)

(a) (b)

Figure 2. FE model of actuator arm and time-history response of the arm tip for two formulations (FI of
ANSYS and SRI of ANSYS/LSDYNA)
3.3 Shock response of HSA-Disk

Based on the shock simulations of disk and actuator arm, it can be concluded that FI
formulation in ANSYS is coarse-mesh accurate compared to SRI in ANSYS/LSDYNA. The latter
needs finer mesh to obtain converge result. The consequence of using coarse mesh for both actuator
arm and disk is tested using head-suspension-assembly-disk (HSA-disk) model (Figure 3).

Figure 3. Head-suspension-assembly (HSA)-disk model

HSA-disk model is subjected to shock impulse 500G/2ms at outer diameter and pivot hole.
Air-bearing stiffness is represented by five spring elements between slider and the disk. Flying height
modulation of a point at the trailing edge of slider is recorded and time-history response is plotted in
Figure 4. Based on the observation of the slider flying height in Figure 4, FI formulation in ANSYS
gives physical understanding that during shock (between 0 to 2 milliseconds) slider is moving away
from the disk and upon the termination of the shock (2 milliseconds and above) slider is stabilizing to
its original steady state position (16 nm). In our simulation, ANSYS/LSDYNA was self-terminating at
1.1 milliseconds due to convergency problem. This is probably due to the SRI formulation of actuator
arm and nonlinear behavior during contact between slider and disk. The contact between slider and
disk should have not happened since at 500G/2ms HDD is normally still can function normally, which
means there will not be any contact between them.
70

60 SRI coarse
(ANSYS/LSDYNA)
50
FI coarse
(ANSYS)
flying height (nm)

40

30

20

10

0
0.00E+00 5.00E-04 1.00E-03 1.50E-03 2.00E-03 2.50E-03 3.00E-03
-10
time (second)

Figure 4. Flying height modulation of slider in HSA-disk model


4. Concluding Remarks

From the numerical experiments, it can be seen that the FI (with extra displacements) can be
coarse-mesh-accurate for shock problem, where the component behavior is dominated with bending
problems. On the other hand, SRI requires finer mesh to give converged results, which leads to
higher computational time. Hence, it is recommended that for bending-dominated problems, the 8-
noded hexahedron with full-integration plus extra displacement shape (to avoid shear locking) should
be used.

References

[1] Edwards J.R. “Finite Element Analysis of the Shock Response and Head Slap Behavior of a
Hard Disk Drive”, IEEE Trans. Magnetics, Vol. 35, No. 2, pp. 863 – 867, March 1999.
[2] Chen-Chi Lin. “Finite Element Analysis of a Computer Hard Disk Drive under Shock”, J.
Mechanical Design ASME, Vol. 124, pp. 121-125, March 2002.
[3] Zeng Q.H. and Bogy D.B. “Numerical simulation of shock response of disk-suspension-slider
air bearing systems in hard disk drives”, Microsystem Technologies, Vol. 8, pp. 289-296,
2002.
[4] Aristegui J.L. and Geers T.L. “Shock Analysis of A Disk-Drive Assembly”, J. Inform. Storage
and Process Systems, Vol. 2, pp. 25 – 31, 2000.
[5] Jayson E.M., Smith P. W. and Talke F.E. “Shock Modeling of the Head-Media Interface in an
Operational Hard Disk Drive”, IEEE Trans. Magnetic, Vol. 39, No. 5, pp. 2429 – 2431,
September 2003.
[6] Luo J., Shu D.W., Shi B.J., Ng Q.Y., Zambri R. and Lau J.H.T. “Study of the Shock Response
of the HDD with ANSYS-LSDYNA”, J. Magnetism and Magnetic Materials, Vol. 303, pp. e57-
e61, 2006.
[7] Shu D.W., Shi B.J., Meng H., Yap F.F., Jiang D.Z., Ng Q., Zambri R., Lau J.H.T and Cheng,
C.S. “Shock Analysis of a Head Actuator Assembly Subjected to Half-Sine Acceleration
Pulses”, Int’l J. Impact Engineering, Vol. 34, pp. 253 – 263, 2007.
[8] “ANSYS LS-DYNA user’s guide – ANSYS Release 10.0”, ANSYS Inc., August 2005.
[9] Taylor R.L., Beresford P.J. and Wilson E.L. “A non-conforming element for stress analysis”,
Int. J. Numer. Meth. Engng, Vol. 10, pp. 1211 – 1219, 1976.
[10] Puso M. A. “A Highly Efficient Enhanced Assumed Strain Physically Stabilized Hexahedral
Element”, Int. J. Numer. Meth. Engng, Vol. 49, pp. 1029 – 1064, 2000.

S-ar putea să vă placă și