You are on page 1of 11

Gil Jun Lee

Department of Mechanical and


Materials Engineering,
College of Engineering and Applied Science,
Application of Computational
University of Cincinnati,
584D Rhodes Hall, 2600 Clifton Avenue, Fluid Dynamics Simulation to
Cincinnati, OH 45221
e-mail: leeg4@mail.uc.edu Squeeze Film Damper Analysis
Jay Kim1 Squeeze film dampers (SFDs) are used in high-speed turbomachinery to provide external
Fellow ASME
Department of Mechanical and
damping to the system. Computational fluid dynamics (CFD) simulation is a highly effec-
Materials Engineering,
tive tool to predict the performance of SFDs and obtain design guidance. It is shown that
College of Engineering and Applied Science,
a moving reference frame (MRF) can be adopted for CFD simulation, which saves com-
University of Cincinnati,
putational time significantly. MRF-based CFD analysis is validated, then utilized to
589 Rhodes Hall, 2600 Clifton Avenue,
design oil plenums of SFDs. Effects of the piston ring clearances, the oil groove, and oil
Cincinnati, OH 45221
supply ports are studied based on CFD and theoretical solutions. It is shown that oil ple-
e-mail: jay.kim@uc.edu
num geometries can significantly affect the performance of the SFD especially when the
SFD has a small clearance. The equivalent clearance is proposed as a new concept that
enables quick estimation of the effect of oil plenum geometries on the SFD performance.
Tod Steen Some design practices that have been adopted in industry are revisited to check their
Mem. ASME
validity. Based on simulation results, a set of general design guidelines is proposed.
GE Aviation,
[DOI: 10.1115/1.4036511]
1 Neumann Way,
Cincinnati, OH 45215
e-mail: tod.steen@ge.com

Introduction CFD has been adopted for SFD analysis, which is not subject to
the limitation of the Reynolds solutions. CFD analysis numeri-
Squeeze film dampers (SFD) have been widely used in high-
cally solves the full Navier–Stokes equation (NSE); therefore, the
speed turbomachinery such as aircraft jet engines to provide exter-
inertia effect of the fluid is considered. Tucker and Keogh [9] and
nal damping to the rotor-bearing system to reduce resonance
Keogh et al. [10] calculated flow and temperature of the journal
responses and improve stability of the system [1,2]. As it is shown
bearing-shaft system by using a flexible CFD modeling approach,
schematically in Fig. 1, an SFD consists of a rolling element bear-
and compared the results with experimental results. Chen and
ing floating in a film of lubrication oil. An anti-rotating pin is used
Hahn [11] studied the suitability of using CFD software to solve
to prevent the outer race of the bearing from rotating with the
steady-state hydrodynamic lubrication problems related to slider
shaft to make the SFD go through a curvilinear translational
bearings, step bearings, journal bearings, and SFDs. Chen and
motion along the trajectory of the shaft whirling. Figure 2 illus-
Hahn [12] used CFD analysis to investigate the effects of the end
trates the motion of the SFD when the shaft is in a circular whirl-
seal clearance of an SFD on the stiffness and damping coeffi-
ing motion. This SFD motion squeezes the lubricant film in the
cients, and found that SFD performance is highly sensitive to the
direction of the instantaneous translational velocity, producing
axial clearances. Guo et al. [13] used CFD to study hydrodynamic,
damping force.
hydrostatic bearings and SFDs, and compared the CFD results to
Della Pietra and Adiletta [1,3] provided a comprehensive
the results obtained from simple computer programs used in
review of analytical and experimental works related to SFD. In
industry. Xing et al. [14,15] developed homogeneous cavitation
earlier works, the Reynolds equation and the continuity equation
models based on the experimental data, and solved the full NSE
have been used to obtain the pressure field and forces of journal
coupled with the cavitation model to obtain the pressure and flow
bearings (JBs) and SFDs, which provides accurate results in the
of an SFD with cavitation. Xing et al. [16] investigated the damp-
low speed range. Gunter et al. [4,5] derived the SFD solution
ing and inertia force coefficients for short and long SFDs by solv-
based on the Reynolds equation and a short bearing approxima-
ing 3D full NSE, and compared the results to those obtained from
tion, and reported that the radial damping coefficient plays an
a modified Reynolds equation that considers the inertia effect.
important role as an equivalent stiffness for SFDs in a circular
They showed that the results from the two methods agree well
orbit. Hamrock et al. [6] derived the SFD solution for a long SFD
only when the Reynolds number (Re) is relatively small, but devi-
based on the Reynolds equation.
ate significantly when the Re becomes moderate or high. Gehan-
Solutions from the Reynolds equation start to become inaccu-
nin et al. [17] reported SFD analysis based on the bulk flow
rate as the speed of the SFD increases; therefore, the Reynolds
equations, and showed that the difference between the solutions
number increases. Tichy [7] examined the pressure distribution of
from the CFD analysis and the Reynolds equation, in terms of the
an SFD and reported that the pressures due to the viscous effect
radial and tangential forces, increases as the eccentricity of the
and due to the inertia effect are out of phase from each other. San
SFD increases. Delgado and San Andres [18] developed an
Andres and Vance [8] drew a similar conclusion and found that
improved bulk-flow model of SFD considering the inertia effect,
the inertia effect of fluid is significant even for moderate Reynolds
and compared force coefficients obtained from the model to
numbers.
experimental results.
Study of effects of oil plenum geometries on the SFD was the
main purpose of this work. The effects of the end seal clearance,
1
Corresponding author. the oil groove, and inlet and outlet port areas were studied system-
Contributed by the Structures and Dynamics Committee of ASME for publication atically for the first time in this paper. A MRF was adopted for
in the JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received
July 18, 2016; final manuscript received March 29, 2017; published online May 16, CFD analysis because it can reduce computation time signifi-
2017. Assoc. Editor: Alexandrina Untaroiu. cantly, thus helping in implementation of numerous simulations

Journal of Engineering for Gas Turbines and Power OCTOBER 2017, Vol. 139 / 102501-1
C 2017 by ASME
Copyright V

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


needed for design studies. In order to validate the CFD analysis, adopting the MRF significantly reduces the computation
the simulation results were compared to those obtained from the time when compared to the analysis adopting an adaptive
Reynolds equation for a low speed case. Using the validated meshing [19].
method, the effects of the geometry of the oil groove, end seal,
and inlet and outlet ports on the SFD performance were examined. Validation of the CFD by Comparing the Results With The-
Some design practices widely used in industry, which were oretical Solutions. The Reynolds equation is obtained by ignor-
largely based on the Reynolds solution, were revisited to check ing the inertia effect in the NSE and further specializing the
their validity. A concept of the equivalent clearance was pro- equation for a fluid bearing that has one of the dimensions much
posed, which allows a quick estimation of the effect of design smaller than the other two. The solution from the Reynolds equa-
changes in the plenum geometry with relatively good accuracy. tion is accurate if the Re is small (Re  1) [7,8]. The Reynolds
Based on the results from this study, a set of guidelines was pro- equation becomes one-dimensional if there is no axial variation in
posed for engineers who design SFDs. pressure as in a very long bearing or a bearing with both ends
sealed. The solution of such a system is called a long bearing solu-
tion. The long bearing solution of the SFD in a circular, synchro-
CFD Simulation of a Squeeze Film Damper Using a nous whirling is obtained as
Moving Reference Frame
12lrb2 ex sin h ð2 þ e cos hÞ
Kinematics of a Squeeze Film Damper in a Circular Whirl- p  ps ¼ (2)
C2 ð2 þ e2 Þ ð1 þ e cos hÞ2
ing Motion. An SFD undergoing a circular whirling motion is
shown in Fig. 2(a). The shaft center O’ whirls around the center
of the housing O with an angular velocity x, and the shaft is rotat-
ing about O’ with an angular velocity xs. X–Y is the inertial coor-
dinate and x–y is a moving coordinate rotating about O with an
angular velocity x. The inner race of the rolling element bearing
mounted on the shaft is fixed to the shaft and the outer race,
shown as an annulus in Fig. 2(a), rotates with respect to the shaft
with an angular velocity xR/s. The velocity at point P on the sur-
face of the outer race can be represented as

vP ¼ vO0 þ vP=O0 ¼ xk  rOO0 þ vP0 =O0 þ vP=P0


¼ exj þ xs k  rb þ xR=s k  rb ¼ exj þ ðxs þ xR=s Þk  rb
(1)

where e is the radius of the whirl orbit and P’ is a hypothetical


point moving with the shaft located instantaneously at the same
position as P.
In an SFD, an anti-rotation pin is used to prevent the outer race
from rotating with respect to the inertial coordinate; therefore,
xR/s ¼ xs and vp ¼ exj. As a result, the outer ring of the SFD
describes a rigid body translational motion. That is, all points on
the outer ring move with the same instantaneous velocity exj
describing a circular path of radius e shown as dashed-line circles
in Fig. 2(b). Notice that the geometry of the oil cavity of the SFD
seen from the moving coordinate remains unchanged. Therefore,
the fluid motion in the SFD can be described by time-invariant,
steady-state equations by adopting the moving coordinates.
Whereas, the direction of the surface velocity of the rotor and the
geometry of the oil cavity change as a function of time if fixed
coordinates are used, the model has to be remeshed at each time
step. In this work, a commercial software FLUENT was used for
CFD analyses. FLUENT allows an MRF option to handle a problem
with a geometry that deviates slightly from a standard MRF, for
example, an SFD with fixed inlet and outlet ports. Analysis

Fig. 2 (a) Geometry of the SFD in a synchronous circular whirl-


ing orbit and the coordinate system and (b) the motion of the
outer surface of SFD. (e: eccentricity, x: whirl speed, xs: angu-
lar velocity of a shaft, xR/s: angular velocity of an outer race of a
bearing, O: center of a housing, O’: center of a shaft, rh: radius
of a housing, rb: radius of a bearing, X–Y: inertia coordinate,
Fig. 1 Schematic of a typical squeeze film damper x–y: moving coordinate rotating about O.)

102501-2 / Vol. 139, OCTOBER 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


Table 2 Pressure and forces from CFD with Re 5 1 3 10211 and
Reynolds equation for long bearings

CFD Reynolds Difference


results equation (%)

Max pressure, Pmax (Pa) 8.50  106 8.54  106 0.52


Min pressure, Pmin (Pa) 8.53  106 8.54  106 0.08
Radial force, Fr (N) 5.81  106 0 —
Tangential force, Ft (N) 1.82  105 1.83  105 0.41

Fig. 3 Schematic of a squeeze film damper without grooves

Table 1 Dimensions of SFD and oil properties

Radius of the rotor (rb) 100 mm


Radial clearance (C) 0.1 mm
Eccentricity (e ¼ e/C) 0.5
Length (L) 40 mm
Whirl speed (x) 1000 rad/s
Viscosity (l) 0.01 kg/m s
Density (q) 1000 kg/m3

where ps is the supply pressure, l is the viscosity, e is the eccen-


tricity defined as
e e
e¼ ¼ (3)
rh  rb C Fig. 5 The ratio of the results from Reynolds long bearing
equation to those from CFD simulation. (Ft, Reynolds: tangential
rb and rh are the radii of the bearing and the bearing housing (see force obtained from Reynolds equation, Ft,CFD: tangential force
Fig. 2), h is the angle measured from the position of the minimum obtained from CFD, DPReynolds: the difference between maxi-
oil film thickness, C is the clearance. The long bearing solution of mum and minimum pressure obtained from Reynolds equation,
the journal bearing in a circular synchronous whirling is DPCFD: the difference between maximum and minimum pres-
sure obtained from CFD.)
6lrb2 ex sin h ð2 þ e cos hÞ
p  ps ¼ (4)
C2 ð2 þ e2 Þ ð1 þ e cos hÞ2 A three-dimensional model of an SFD sealed by piston rings at
both ends shown in Fig. 3 was built as the baseline SFD in this
It is noted that the pressure induced by the journal bearing is a study. The geometry and the material property of the oil are sum-
half of that of the SFD that has the same geometry. marized in Table 1.
The Reynolds number of SFD, also known as the squeeze num- The Reynolds solution given in Eq. (2) becomes very close to
ber, is defined as the exact solution when Re is very small; therefore, the Reynolds
solution can be used to check the validity of the CFD model and
qxC2 solutions. The CFD solution obtained for a very low Re,
Re ¼ (5)
l Re ¼ 1  1011, was compared with the Reynolds solution in Fig.

Fig. 6 Circumferential pressure distributions of an SFD


Fig. 4 Circumferential pressure distribution for long SFD with obtained from Reynolds equation and obtained from CFD with
Re 5 1 3 10211 five different Reynolds numbers (Re)

Journal of Engineering for Gas Turbines and Power OCTOBER 2017, Vol. 139 / 102501-3

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


4. The figure shows that two solutions match almost exactly. The groove varies but only slightly. Also, it is important to notice that
radial (Fr) and tangential (Ft) forces are obtained by integrating the pressure amplitude decreases significantly, from 8.21 bar to
the pressure as follows: 0.32 bar, because the groove effectively increases the clearance.
ð 2p In design of journal bearings or SFDs, it is often assumed that a
center groove acts as a pressure relief volume, which means that
Fr ¼ L pðhÞcos hrb dh (6) the gauge pressure at the center of the groove is zero; therefore,
0
the pressure distribution of the long bearing is approximated as
ð 2p that of the so-called short bearing. Figure 9 illustrates the concept
Ft ¼ L pðhÞsin hrb dh (7) behind the practice. However, the simulation result in Fig. 8(b)
0 shows that the pressure at the groove center does not become
zero, and the pressure amplitude changes significantly by adding
The tangential force acts as the damping force to the rotor. The the groove; therefore, the short bearing approximation is not valid.
damping coefficient due to the SFD (c) is calculated as follows: Figure 10 shows the simulation results of the SFD and the journal
bearing of the same geometries for a very low Reynolds number
Ft Ft (Re ¼ 1  1011) and a very deep central groove (C/h ¼ 50). As it
c¼ ¼ (8)
v ex is expected from Eqs. (2) and (4), the pressure of the SFD is two
times larger than that of the journal bearing. It is seen that the
Table 2 compares two solutions in terms of the maximum and pressure at the center of the groove becomes very small for both
minimum pressures, and the radial and tangential forces. The journal bearing and SFD. Therefore, the short bearing approxima-
comparison validates the CFD analysis. tion of SFD is valid only when the speed of the SFD is very low.
Numerical simulation was repeated for a range of Re ¼ 0.1, 0.5, Even in this case, the pressure amplitude has to be corrected by
1.0, 2.0, 5.0, 10, 20, 50, 100 by changing the angular velocity of using the equivalent clearance to be defined below.
the rotor (x) while keeping all other parameters the same. Figure The circumferential pressure distributions for different groove
5 shows the ratio of the tangential force and the pressure ampli- sizes are shown in Fig. 11. The pressure distribution changes in a
tude (DP ¼ Pmax  Pmin) obtained from the Reynolds equation to similar way to the change observed in Fig. 6 due to the increase of
those obtained from CFD simulation for different Reynolds num- Re. A larger groove height effectively increases the C, thus
bers. It is seen that the tangential force (Ft) calculated from the increasing Re (see Eq. (5)).
Reynolds equation is relatively accurately up to Re ¼ 10, while Changes in the radial force (Fr) and tangential force (Ft) of SFD
the pressure amplitude shows much bigger error, which is almost with respect to the groove depth are shown in Fig. 12. Both forces
20% error at Re ¼ 10. Figure 6 shows the pressure distribution in decrease significantly as the groove depth increases, which is
the circumferential direction for Re ¼ 0.1, 0.5, 1, 2.0, 5.0 com- expected from Fig. 11 that shows the pressure decreases significantly
pared with the Reynolds solution. It shows that the error in the as the groove size increases. Obviously, the decrease of Ft results in
pressure distribution increases as Re increases. the decrease of the damping force and coefficient of the SFD.

Effects of Piston Ring Grooves on SFD Performance. Figure


Application of CFD Analysis to Design Parameter 13 shows a cross-sectional view of the oil plenum of an SFD that
Study employs piston-ring end seals. Ca and Cr are the axial and radial
clearances that are needed to enable assembly, but were ignored in
Effects of the Center Groove on SFD Performance. A center the simulation in Effects of the Center Groove on SFD Performance
groove is adopted in some SFDs to simplify the oil supply system section. In SFDs with a small clearance, the volume formed by the
as shown in Fig. 7. A model was created by adding a center piston ring groove is not negligible when compared to the volume
groove of depth h and width w ¼ 10 mm to the SFD studied in formed by the clearance. To study effects of the piston ring groove,
Validation of the CFD by Comparing the Results with Theoretical simulations were conducted for different Ca and Cr of a piston ring
Solutions section. Simulations were conducted for 12 different groove while keeping all other dimensions the same and Re ¼ 1.
groove depths while keeping all other dimensions the same for the First, Ca was changed while keeping Cr ¼ 0.5 mm. Simulation
angular velocity of 1000 rad/s. was conducted for three cases, CaLh/CL ¼ 0.0125, 0.125, 12.5. The
Pressure distributions on the surface of the rotor without the pressure distributions in the circumferential direction at the center
groove (h/C ¼ 0) and with a groove of h/C ¼ 65.3 are compared in of the SFD and in the axial direction at the maximum circumferen-
Fig. 8. Figure 8(a) shows that there is no axial variation of the tial pressure location are shown in Fig. 14. It is seen that the
pressure when there is no groove as expected. Figure 8(b) shows increase of Ca decreases the pressure amplitude, therefore decreas-
that the pressure in the axial direction of the SFD with a center ing both Fr and Ft. The shape of circumferential pressure distribu-
tion changes in a similar way to the case when Re increases (see
Fig. 6). The pressure in the axial direction is no more constant
when Ca is large. When Ca is very small, the pressure is nearly
constant along the axial direction as the plenum formed by Cr and
Lw (see Fig. 13) is isolated from the main plenum.
The effect of change of Cr was studied while Ca was fixed
(Ca ¼ 0.1 mm) for five cases (CrLw/CL ¼ 1, 2, 5, 10, 15). Figure 15
shows the pressure distribution both in circumferential and axial
directions at different locations. Figure 15(b) shows that the axial vari-
ation of pressure increases as Cr becomes larger. The maximum pres-
sure, consequently Fr and Ft, also decreases significantly as Cr
increases.
The width of the piston ring Lw was changed while Ca, Cr and
Lh were fixed (Ca ¼ C, Cr ¼ 2C, Lh ¼ 10 C). Figure 16 shows the
pressure distribution in circumferential direction of SFD with four
different Lw values (Lw/C ¼ 0, 5, 10, 20). It can be seen that the
pressure decreases as Lw increases. Overall, increase of dimen-
Fig. 7 Schematic of a squeeze film damper with a center sions of the piston ring grooves has similar effects to increase of
groove the clearance.

102501-4 / Vol. 139, OCTOBER 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


Fig. 8 Three-dimensional plots of pressure field on the rotor of SFD. (a) h/C 5 0 (no groove)
and (b) h/C 5 65.3. h and C are the depth of the center groove and the clearance of SFD,
respectively.

Equivalent Clearance. As it is seen in Effects of the Center 0P 11=2


Groove on SFD Performance and Effects of Piston Ring Hi2 Li
B iP C
Grooves on SFD Performance sections, small oil plenums such Ceq1 ¼@ A (9)
Li
as the oil groove or piston ring seal clearance can make a sig- i
nificant effect on the performance of an SFD. A cross-
sectional view of the oil plenum of an SFD with a center P
Hi Li
groove is shown in Fig. 17. Equivalent clearances of two defi-
nitions, Ceq1 and Ceq2, were tested, which are defined as Ceq2 ¼ iP (10)
Li
follows: i

Fig. 9 Short bearing approximation for a SFD with a center groove

Journal of Engineering for Gas Turbines and Power OCTOBER 2017, Vol. 139 / 102501-5

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


Fig. 10 (a) Three-dimensional plot of the pressure field of an SFD and a JB with a center groove (h/C 5 50, h:
the depth of the center groove, C: clearance, Re 5 1 3 10211), (b) axial pressure distribution at 90 deg of circum-
ferential location, and (c) axial pressure distribution at 270 deg of circumferential location

The oil cavity shown in Fig. 17 is approximated as the cavity Figure 18 compares the pressures calculated from the CFD
without any groove as shown in Fig. 3 with C ¼ Ceq1 or C ¼ Ceq2. models with the simplified geometry with Ceq1 and Ceq2 to those
Ceq1 is a squared average of clearance and Ceq2 is a linear average calculated from the CFD model with the original geometry. As it
of clearance. is seen in Fig. 18, the model that employs Ceq2 provides much

Fig. 11 Circumferential pressure distribution on the rotor at


the midplane (L 5 0) with various groove depths. (h: depth of Fig. 12 Radial and tangential forces of SFD with respect to
the groove, w: width of the groove, C: clearance of SFD, L: groove depth (h). C, L, and w are the clearance of SFD, the
length of SFD.) length of SFD, and groove width.

102501-6 / Vol. 139, OCTOBER 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


more accurate approximation than the model that employs Ceq1. Reynolds equation obtained with Ceq2 provides quite an accurate
Figure 19 compares the forces obtained from the model with the estimation of the damping force.
actual geometry, the models with two equivalent clearances from It is somewhat unexpected that the model with Ceq2 gives much
CFD analysis and the Reynolds solution. better approximation than the model with Ceq1. The Reynolds
Because the Reynolds equation ignores the inertia of fluid, Fr is
always zero. Ft, which is the most important because it is the
damping force, is estimated quite accurately from the model that
employed Ceq1. In the given example, the solution from the

Fig. 13 Schematic of the SFD with piston ring seals. L is the


length of the SFD, C is the clearance, Lw and Lh are the width of
a piston ring and the height of a piston ring, Ca and Cr are axial
and radial clearance of the piston ring groove, respectively.

Fig. 15 Pressure distribution of the SFD with different radial clear-


ance (Cr) of piston ring grooves (a) in circumferential direction at
the midplane (L 5 0) and (b) in axial direction at the circumferential
location of maximum pressure. Lw, C, and L are the width of the pis-
ton ring, clearance of the SFD, and length of the SFD, respectively.

Fig. 14 Pressure distribution of the SFD with different axial


clearances (Ca) of the piston ring grooves (a) in circumferential
direction at the midplane (L 5 0) and (b) in axial direction at the
circumferential location of the maximum pressure. C, L, and Lh Fig. 16 Pressure distribution of the SFD with different widths
are the clearance, the length of an SFD, and the height of a pis- of a piston ring (Lw) in circumferential direction at the midplane
ton ring, respectively. (L 5 0)

Journal of Engineering for Gas Turbines and Power OCTOBER 2017, Vol. 139 / 102501-7

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


equation (Eqs. (2) and (4)) shows that the pressure is inversely linearly averaging out the pressures in the sections of different
proportional to the square of C, thus suggesting the squared aver- depths.
age of C will work better. This may be explained by the fact that
the pressure gradient in the axial direction is much bigger than Effects of the Port Size on Mass Flow Rates and SFD
that in the circumferential direction; therefore, the oil in the sec- Performance. To understand the effect of the mass flow rate of
tion of small clearance (L1 and L3 sections in Fig. 17) quickly the SFD on its performance, the CFD model of the SFD with oil
moves into the section of larger clearance (L2 section in Fig. 17), inlets and outlets was established as shown in Fig. 20. Three inlets
were evenly spaced along the midplane of the SFD. Two outlets
are located on two end surfaces of the SFD. Ra, the ratio of the
total inlet area and the outlet area, is defined as below equation

P
Ainlet 3Ai
Ra ¼ P ¼ (11)
Aoutlet 2Ao

where Ai and Ao are the area of the single inlet port and the area of
the single outlet port, respectively. In the CFD models, the bound-
ary condition was defined as the supply pressure at the inlets and
as the ambient pressure at the outlets. In order to find a good ratio
Ra, ten cases were simulated by changing Ai while Ao was fixed
for Re of 1.
The circumferential pressure distributions at the center line for
Fig. 17 Cross section of the general SFD with a groove different values of Ra are shown in Fig. 21(a). Since Ao is fixed, Ai

Fig. 18 (a) Circumferential pressure distribution of the SFD Fig. 19 Forces of SFD with a center groove and corresponding
with a center groove and corresponding equivalent clearances equivalent clearances (Ceq1 and Ceq2, no groove) and Reynolds
(Ceq1 and Ceq2, no groove) at the midplane (L 5 0), and (a) hw/ long bearing solution with respect to nondimensional groove
CL 5 3.50 and (b) hw/CL 5 14.0. h, w, L, and C are groove depth, depth (hw/CL). (a) Normalized radial force and (b) Normalized
groove width, the length of SFD, and the clearance of SFD, damping coefficient. h, w, L, and C are groove depth, groove
respectively. width, the length of SFD, and the clearance of SFD, respectively.

102501-8 / Vol. 139, OCTOBER 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


increases as Ra increases. It is seen that the maximum pressure function of Ps. The mass flow rate increases significantly as the
decreases as Ra increases. From Fig. 21(b), it is also seen that the supply pressure increases; however, the damping coefficient
mass flow rate increases and the tangential force decreases when changes only slightly, decreasing by less than 2%.
Ra (thus Ai) increases. Observations on effects of the mass flow areas can be summar-
The decrease of the SFD pressure shown in Fig. 21 was more ized as follows.
than expected, which prompted the question whether the change
was attributed to the increase of the effective clearance due to the
enlarged port areas or due to the increased mass flow. To answer
this question, the SFD model shown in Fig. 20 was modified by
closing the far end of the inlet pipe shown in Fig. 20(b), which
eliminates the net mass flow, while including portions of the inlet
pipes (20 mm in this case) to the model. The pressure distributions
at the center line of the SFD with two different sizes of the inlet
port (diameter of 2.22 mm and 5.55 mm) are compared as shown
in Fig. 22. Both the maximum pressure and the pressure amplitude
decrease as the size of the inlet port increases. This indicates that
a portion of the inlet pipe effectively acts like an additional clear-
ance volume of the SFD. The circumferential pressure distribu-
tions at the center line are shown as the function of Ao in
Fig. 23(a) when Ra (¼2.0) is fixed. As Ao increases, the mass flow
rate increases and the tangential force decreases (Fig. 23(b)).
To understand the effect of the mass flow alone, the mass flow
rate was changed by changing the supply pressure (Ps) at the inlets
while keeping all other parameters the same. The circumferential
pressure distributions at the midplane of the SFD with respect to
Ps are shown in Fig. 24(a). It is seen that, when compared to the
SFD without mass flow, the pressure of the SFD drops when there
is mass flow due to the effect of increased effective clearance vol-
ume. However, the pressure profiles of SFDs with mass flow with
different supply pressure look nearly the same. Figure 24(b)
shows the damping coefficient and the mass flow rate as a

Fig. 21 (a) Circumferential pressure distribution at the mid-


plane (L 5 0) with different ratios of the area (Ra) between inlets
(Ai) and outlets (Ao) and (b) corresponding normalized mass
flow rate and damping coefficient

Fig. 22 Circumferential pressure distribution at the midplane


Fig. 20 (a) Schematic of the SFD with oil inlets and outlets and (L 5 0) of an SFD with the volume of inlets and without oil flow.
(b) its 3D CFD model di is a diameter of an inlet.

Journal of Engineering for Gas Turbines and Power OCTOBER 2017, Vol. 139 / 102501-9

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


(1) Mass flow rate increases as the cross-sectional areas of the Parameter study based on CFD analysis was conducted to study
inlet port and the exit area increase. the effect of the center groove and piston ring groove on the per-
(2) Increase of the port size decreases the SFD pressure, there- formance of the SFD. The results showed that the center groove and
fore the damping performance, because it increases the piston ring groove can reduce damping force significantly for the
effective clearance of the SFD. Therefore, inlet and outlet SFDs with a small clearance because they effectively increase the
ports larger than necessary cooling requirement will sacri- clearance volume between a rotor and stator. Increased clearance
fice the performance of the SFD. volume decreases the pressure induced by the squeeze action, there-
(3) Increase of the supply pressure with the same outlet pressure fore decreasing the damping force. Simulation results showed that
results in the increase in the mass flow, but no significant the practice in industry that uses a short bearing solution for the
changes in the pressure and damping performance of the SFD. bearing or SFD with a center groove is not always applicable.
That is, the mass flow itself does not affect the performance of It was shown that the axial and radial clearance of the piston
the SFD. Therefore, increasing supply pressure may provide ring groove of the SFD should be minimized to obtain a high
more effective solution when higher flow rate is required. damping force because increase of the axial clearance or the radial
clearance of the piston ring could decrease the damping force gen-
erated by the SFD.
Discussion A concept of the equivalent clearance was discussed. Two defi-
The effects of oil plenum geometry such as grooves, piston ring nitions, Ceq1 and Ceq2, a squared average, and a linear average of
clearances, and mass flow ports on the SFD performance were multiple sections of different clearances, were tested. The results
investigated in this work by using CFD analysis. In order to vali- indicated that Ceq2, the linear average, provides much better
date the analysis, CFD results obtained for very low Re were com- approximation than Ceq1.
pared to the analytical solution obtained from the Reynolds The effect of the size of the inlet and outlet port areas was also
equation. The pressure distribution and forces obtained from the studied. The result showed that the mass flow rate increases and
simulation were very close to those obtained from the Reynolds the damping force decreases as the flow port size increases. Simu-
equation. CFD solutions deviated from the Reynolds solution as lation results showed that the decrease of the damping force is
the speed of the SFD; therefore, Re increased. When Re was mainly caused by the increase of the effective clearance due to the
higher than 10, the error in the performance (damping force) esti- increased port sizes. It was observed that increase of the supply
mated from the Reynolds solution was higher than 20%, showing pressure increases the mass flow rate, while it does not change the
the limitation of the Reynolds solution. SFD performance.

Fig. 23 (a) Circumferential pressure distribution at the mid- Fig. 24 (a) Circumferential pressure distribution at the mid-
plane (L 5 0) force with different area of outlets (Ao) and (b) cor- plane with different supply pressure (Ps) and (b) corresponding
responding normalized mass flow rate and damping coefficient normalized damping coefficient and mass flow rate

102501-10 / Vol. 139, OCTOBER 2017 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a


In this work, cavitation effect of the oil was not considered. ps ¼ supply pressure, Pa
The effects of the cavitation on the SFD performance can be simu- rb ¼ radius of the bearing, mm
lated and investigated by CFD [14,15], which is currently being rh ¼ radius of the housing, mm
studied by the authors. The whirl orbit of the SFD used in this Ra ¼ ratio of area between inlets and outlets; (3Ai/2Ao)
work was assumed to be circular while the orbit is noncircular in Re ¼ Reynolds number
practice. General conclusions obtained from design parameter SFD ¼ squeeze film damper
studies are expected to be valid for noncircular orbit cases, which w¼ width of a center groove of the SFD, mm
may be confirmed by CFD analysis of SFDs with a noncircular e¼ eccentricity
orbit which requires adaptive meshing method [19]. l¼ viscosity, kg/m s
q¼ density, kg/m3
Conclusions x¼ whirl velocity of the shaft center, rad/s
The effects of oil plenum geometry including grooves, piston
References
ring clearance, and oil ports on the SFD performance were studied
[1] Della Pietra, L., and Adiletta, G., 2002, “The Squeeze Film Damper Over Four
in this work by utilizing CFD analysis. Before applying CFD anal- Decades of Investigations—Part I: Characteristics and Operating Features,”
ysis to SFDs operating at realistic speeds, CFD analysis was vali- Shock Vib. Dig., 34(1), pp. 3–26.
dated by comparing CFD solutions to analytical solutions [2] San Andres, L., 2012, “Damping and Inertia Coefficients for Two Open Ends
obtained for SFDs operating at very low speeds, which are the Squeeze Film Dampers With a Central Groove: Measurements and Pre-
dictions,” ASME J. Eng. Gas Turbines Power, 134(10), p. 102506.
cases when both solutions should be identical. [3] Della Pietra, L., and Adiletta, G., 2002, “The Squeeze Film Damper Over Four
Several observations obtained from the simulation results in this Decades of Investigations—Part II: Rotordynamics Analyses With Rigid and
work can be summarized for SFD designers. For SFDs that use pis- Flexible Rotors,” Shock Vib. Dig., 34(2), pp. 97–126.
ton rings as the end seal, the axial and radial clearances of the pis- [4] Gunter, E. J., Barrett, L. E., and Allaire, P. E., 1977, “Design of Nonlinear
Squeeze-Film Dampers for Aircraft Engines,” J. Lubr. Technol., 99(1), pp.
ton ring groove should be minimized because larger clearances 57–64.
reduce the damping performance. In the SFD with a center groove, [5] Gunter, E. J., Barrett, L. E., and Allaire, P. E., 1975, “Design and Application
the cross-sectional area of the groove should be determined by con- of Squeeze Film Dampers for Turbomachinery Stabilization,” Fourth Turboma-
sidering not only the oil flow requirement but also its effect on the chinery Symposium, College Station, TX, Oct. 14–16, pp. 127–141.
[6] Hamrock, B. J., Schmid, S. R., and Jacobson, B. O., 2004, Fundamentals of
damping performance. The inlet port size also has to be determined Fluid Film Lubrication, 2nd ed., Marcel Dekker, New York.
by considering its effect on the damping performance of the SFD. [7] Tichy, J. A., 1983, “The Effect of Fluid Inertia in Squeeze Film Damper Bear-
The equivalent clearance, a new concept proposed in this work, ings: A Heuristic and Physical Description,” ASME Paper No. 83-GT-177.
can be highly useful for designers to make a quick prediction of the [8] San Andres, L. A., and Vance, J. M., 1987, “Effect of Fluid Inertia on Squeeze-
Film Damper Forces for Small-Amplitude Circular-Centered Motions,” ASLE
effect of the oil plenum geometry on the performance of the SFD. Trans., 30(1), pp. 63–68.
Finally, the SFD should be designed to provide not the largest but [9] Tucker, P. G., and Keogh, P. S., 1995, “A Generalized Computational Fluid
an optimal amount of damping because excessive damping can Dynamics Approach for Journal Bearing Performance Prediction,” Proc. Inst.
result in locking up of the rotor system. In such cases, a center Mech. Eng., Part J, 209(J2), pp. 99–108.
[10] Keogh, P. S., Gomiciaga, R., and Khonsari, M. M., 1997, “CFD Based Design
groove can be employed as an optimal compromise. Techniques for Thermal Prediction in a Generic Two-Axial Groove Hydrody-
namic Journal Bearing,” ASME J. Tribol., 119(3), pp. 428–435.
[11] Chen, P. Y. P., and Hahn, E. J., 1998, “Use of Computational Fluid Dynamics
Acknowledgment in Hydrodynamic Lubrication,” Proc. Inst. Mech. Eng., Part J, 212(J6),
Financial support on this work by GE Aviation is highly pp. 427–436.
[12] Chen, P. Y. P., and Hahn, E. J., 2000, “Side Clearance Effects on Squeeze Film
appreciated. Damper Performance,” Tribol. Int., 33(3–4), pp. 161–165.
[13] Guo, Z., Hirano, T., and Gordon Kirk, R., 2005, “Application of CFD Analysis
for Rotating Machinery—Part 1: Hydrodynamic, Hydrostatic Bearings and
Nomenclature Squeeze Film Damper,” ASME J. Eng. Gas Turbines Power, 127(2),
Ai ¼ area of an inlet of the SFD, mm2 pp. 445–451.
[14] Xing, C., Braun, M. J., and Li, H., 2009, “A Three-Dimensional
Ao ¼ area of the outlet of the SFD, mm2
Navier–Stokes-Based Numerical Model for Squeeze Film Dampers—Part 1:
c¼ damping coefficient; (Ft/ex), N s/m Effects of Gaseous Cavitation on Pressure Distribution and Damping Coeffi-
C¼ clearance of a SFD; (rh  rb), mm cient Without Consideration of Inertia,” Tribol. Trans., 52(2), pp. 680–694.
Ca ¼ axial clearance of a piston-ring seal clearance, mm [15] Xing, C., Braun, M. J., and Li, H., 2009, “A Three-Dimensional Navier–
Stokes-Based Numerical Model for Squeeze Film Dampers—Part 2: Effects of
Ceq1 ¼ square-averaged equivalent clearance, mm Gaseous Cavitation on the Behavior of the Squeeze Film Damper,” Tribol.
Ceq2 ¼ linear-averaged equivalent clearance, mm Trans., 52(5), pp. 695–705.
Cr ¼ radial clearance of a piston-ring seal clearance, mm [16] Xing, C., Braun, M. J., and Li, H., 2010, “Damping and Added Mass Coeffi-
CFD ¼ computational fluid dynamics cients for a Squeeze Film Damper Using the Full 3-D Navier–Stokes Equation,”
e¼ eccentric Tribol. Int., 43(3), pp. 654–666.
[17] Gehannin, J., Arghir, M., and Bonneau, O., 2009, “Complete Squeeze-Film
Fr ¼ radial force, N Damper Analysis Based on the ‘Bulk Flow’ Equations,” Tribol. Trans., 53(1),
Ft ¼ tangential force, N pp. 84–96.
h¼ depth of a center groove of the SFD, mm [18] Delgado, A., and San Andres, L., 2010, “A Model for Improved Prediction of
Force Coefficients in Grooved Squeeze Film Dampers and Oil Seal Rings,”
L¼ length of the SFD, mm ASME. J. Tribol., 132(3), p. 032202.
Lw ¼ width of a piston ring, mm [19] ANSYS, 2013, “ANSYS FLUENT Theory Guide,” ANSYS, Canonsburg, PA.

Journal of Engineering for Gas Turbines and Power OCTOBER 2017, Vol. 139 / 102501-11

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jetpez/936205/ on 05/19/2017 Terms of Use: http://www.asme.org/a

You might also like