ISSN 1068-798X, Russian Engineering Research, 2017, Vol. 37, No. 4, pp. 287–291. © Allerton Press, Inc., 2017. Original Russian Text © N.V. Sokolov, T.V. Maksimov, M.B. Khadiev, V.A. Futin, 2017, published in Vestnik Mashinostroeniya, 2017, No. 1, pp. 28–32.

Sliding Thrust Bearings in a Centrifugal Compressor with a Semiclosed Impeller N. V. Sokolova, *, T. V. Maksimova, M. B. Khadieva, and V. A. Futinb aKazan

National Research Technological University, Kazan, Russia NIIturbokompressor im. V. B. Shneppa, Kazan, Russia *e-mail: [email protected]

bAO

Abstract—The equipment for dynamic tests of sliding thrust bearings is considered. The test data are analyzed for a thrust bearing with motionless pads in a centrifugal compressor with a semiclosed impeller in pumping mode. Keywords: sliding thrust bearing, dynamic operation, centrifugal compressor, measurement system, pumping mode, vibration frequency, data analysis DOI: 10.3103/S1068798X17040207

Hydrodynamic sliding bearings are widely used in compressor design, on account of their high performance over a wide range of speeds and loads and their relatively simple structure. That ensures satisfactory compressor operation. Despite extensive study and use of sliding bearings, little is known about their operation under dynamic loads; in particular, practical data are unavailable. For example, we lack data regarding the operation of sliding bearings in a real compressor, during pumping phases. Since dynamic loading is associated with high speeds and considerable loads on the compressor components, while long pumping periods are impossible, it is difficult to determine the bearings’ operational parameters. In the present work, we investigate a sliding thrust bearing with immobile pads during the operation of a centrifugal compressor in pumping mode. The results confirm the feasibility and importance of such research. Pumping mode is impermissible in centrifugal compressors. In pumping mode, the operation of a centrifugal compressor with periodic gas supply is characterized by disruption of the flow [1]. Vibration develops in the whole mass of gas filling the compressor and the supply network. Ultimately, inverse flow forms. The vibration frequency is 1–10 Hz. This leads to axial and radial vibration of the rotor and to failure the bearings and seals. Therefore, research in such phenomena is of great interest in order to improve the reliability of centrifugal compressors. Thrust bearings are studied experimentally at AO NIIturbokompressor im. Shneppa, on equipment for gas-dynamic tests of centrifugal compressors [2]. The test bench (Fig. 1a) includes an electric motor 1, a support 2, a centrifugal compressor 3, and an experimental thrust bearing 4 (Fig. 1b), as well as pipelines,

power lines, gates and regulators, lubrication and automation systems, a line supplying air to a labyrinthine seal, and measuring instruments. A two-sided thrust bearing is produced for the experiments. On each side of the bearing, the immobile pads include inclines parallel to the radial channel between the pads (Fig. 2) [3]. We also use consumption rings, a thrust collar, and a bracing ring, determining the total axial gap hs = 0.34 mm in the bearing. The geometric dimensions on each side are as follows: internal and external pad diameters D1 = 65 mm and D2 = 100 mm; number of pads z = 8; pad thickness Hp = 3 mm; angular thickness of pad θp = 38.05°; width of incline ηin = 18 mm; depth of incline δin = h1 – h2 = 0.07 mm. The centrifugal stage contains a semiclosed impeller (external diameter D2 = 266 mm), a bladeless diffuser (width b3 = 6.9 mm), and an annular chamber. The working gas is air. The maximum exit pressure from this stage is 0.25 MPa. The test bench is driven by an MP-700-3000 dc motor (rated power 700 kW), with an adjustable speed (150–3000 rpm). By means of the speed control system for the dc motor’s rotor and a booster with a helical gear transmission (g ear ratio i = 8.5), the speed of the high-speed rotor in the test bench may be smoothly adjusted in the range 1300–25 000 rpm. Tp-22S oil (Technical Specifications TU 38.101821–83) is used in the lubrication system. A standard data-collection system is used to measure the gas-dynamic parameters of the centrifugal stage. To monitor the vibration and parameters of the oil, a standard automatic system is employed. The test bench is remotely controlled [4].

287

288

SOKOLOV et al.

(a)

4

(b)

3 1

ω1

2

Fig. 1. Test bench (a) and cross-sectional view of compressor (b).

′

M4-7N

diam. 8

22° ± 30

diam. 65 diam. 100

p

diam. 82.5

A A

16 max 8 min

diam. 1

A–A (enlarged)

5 ± 0.15 8 slots

0.5 × 45° 4 25

Fig. 2. Thrust bearing and location of TDAS AT-60-1.0 sensor (section A–A): p is the measured pressure.

The standard measurement system is augmented by a channel for measuring the lubricant-layer parameters in the thrust bearing. To measure the pressure of the lubricant layer in two opposing pads of the thrust bearing at a moderate distance from the loaded working end of the bearing, we use TDAS AT-60-1.0 piezoresistive sensors (OOO NPF Intelsens) [2, 4]. The sensors permit measurement of pressures up to 6 MPa. They are small and able to operate over a broad temperature range: from –60 to +150°C. In the thrust bearing, two sensors S1 and S2 are used to measure the pressure. Their configuration permits comparison of the data. The sensors are screwed into two special sockets (with M4 thread) at the mean

bearing radius in the region of the presumed maximum pressure pmax. Loctite 542 sealant prevents leakage over the sensor thread. The dependence of the sensors’ electrical output signal on the measured pressure constitutes a linear calibration curve. According to the sensor specifications, the normalized output signal is 0.335 mV/(V kgf/cm2) for sensor S1 and 0.090 mV/(V kgf/cm2) for sensor S2. The nonlinearity of the output signal is no more than 0.72%. The signal is amplified by means of a tensometric station ensuring a stabilized 5-V supply to the sensors, with 100 : 1 amplification of the output signal. The signal is displayed on a digital oscillograph; the recorded signals are analyzed by means of a personal computer.

RUSSIAN ENGINEERING RESEARCH

Vol. 37

No. 4

2017

SLIDING THRUST BEARINGS IN A CENTRIFUGAL COMPRESSOR

Preliminary static tests of a thrust bearing with steady operation of the centrifugal compressor indicate that there is considerable margin for increase in the load at the bearing [4]. For example, when Mu = 0.907, with minimum lubricant-layer thickness h2 = 25 μm, the theoretical carrying capacity of the bearing is 3266 N, according to Sm2PxT software [5]. By contrast, the experimental value of the axial load at different working points of the characteristic for the centrifugal stage is 1597.69–2379.3 N—that is, it is 48.9– 72.9% of the maximum value. Hence, the actual minimum lubricant-layer thickness is h2 = 50–60 μm. This is associated with considerable damping capacity of the thrust bearing. On account of the instability of the lubricant layer at the given h2 values, it varies as a function of the axial load—that is, as a function of the time. In Fig. 3, we show the test results for the sliding thrust bearing in steady operation, in the form of oscillograms taken directly from pressure sensors S1 and S2 after signal amplification, without any mathematical analysis. The time axis is plotted in scale increments of 2.5 ms, and the voltage (U) axis in increments of 200 mV. For sensor S1, the mean output voltage is about 1150 mV, which corresponds to p1 = 0.69 MPa; for sensor S2, the corresponding figures are about 560 mV and 1.24 MPa. The pressure difference between opposing pads, in contrast to the parallel working surfaces of the bearing and thrust collar (Fig. 4a), is most likely associated with the inclination of the bearing surface relative to the surface of the thrust collar when the bearing is installed in the compressor (Fig. 4b), in the absence of an equalizer. In accordance with the variation in U over time τ (Fig. 3), we also note antiphase pressure pulsations whose frequency corresponds to the shaft speed: fp = 1/τ1 = 1/τ2 = no = 317 Hz, where τ1 and τ2 are the oscillation periods of the pressure at sensors S1 and S2 when Mu = 0.884. The amplitude of the pulsations is 0.14 MPa for sensor S1 and 0.49 MPa for sensor S2. This is thought to be associated with the unavoidable wobble of the thrust disk (Fig. 4c). Together with the underloading of the bearing, this leads to considerable periodic pressure fluctuation. The larger amplitude RUSSIAN ENGINEERING RESEARCH

Vol. 37

No. 4

U, mV ΔU1

S1 1200 1000

τ1 ΔU1me

800

S2

600

ΔU2

400 200 –2.5

ΔU2me

In the experiments, the pressure p of the lubricant layer is plotted as a function of the time τ, in different operating conditions: Mu = 0.707, 0.884, and 0.907 at the impeller output, where the Mach number Mu = u2/a; u2 is the azimuthal velocity at the periphery of the impeller; and a is the velocity of sound. With variation in air temperature at the input to the stage, Mu is maintained constant by adjusting the speed of the high-speed rotor. By gradually closing the electric gate at the compressor, the operating conditions of the stage may be adjusted from maximum productivity to pumping mode. In pumping mode, the continuous supply to the compressor becomes unstable and then periodic.

289

0

τ2

2.5

5.0

7.5

τ, ms

Fig. 3. Oscillograms of the pressure at the pads in a thrust bearing in steady operation.

(a)

(b) hme ± hst

hme

p2

p1

p2

p1

(c) hme ± hdyn

p2

p1

Fig. 4. Position of bearing and thrust collar relative to the rotor axis: (a) no inclination; (b) bearing inclination; (c) inclination of thrust collar; hme, hst, hdyn, mean, static, and dynamic components of the lubricant-layer thickness.

seen for sensor S2 than for S1 also confirms the influence of the inclination of the bearing surface relative to the surface of the thrust collar, in the absence of the skewing predicted from the mean pressure values for the sensors. In Figs. 5 and 6, we show the test results for pumping mode of the centrifugal compressor at 290 Hz, corresponding to operation with Mu = 0.707. The mean pressure is p1 = 0.69 MPa according to sensor S1 and p2 = 1.11 MPa according to sensor S2. The amplitude of the pulsations is 0.14 MPa for sensor S1 and 0.55 MPa for sensor S2. In Fig. 5, we see the pressure variation in pumping mode before and during the first disruption. The time axis is plotted in scale increments of 20 ms, and the voltage (U) axis in increments of 200 mV. According to 2017

290

SOKOLOV et al.

U, mV U1max

2000 1600

S1

1200 800 U2max

U1me

S2 U2me

400 –20

0

20

40

60

80

100 τ, ms

Fig. 5. Oscillograms of the pressure at the pads in a thrust bearing in pumping mode of the compressor at the first disruption.

U, mV 1000 800 U2max

S2 600 400 200

0

U2me

τ0 τ1 50

100

150

200 τ, ms

Fig. 6. Oscillogram of the pressure at the pads in a thrust bearing in pumping mode of the compressor at successive disruptions: τ0, duration of the disruption τ1, duration of the pumping mode.

Fig. 5, the pressure increases relatively rapidly (τ ≈ 10 ms) and falls slowly (τ ≈ 100 ms), with considerable pressure drop after the disruption. The maximum pressure is p1 = 0.96 MPa according to sensor S1 and p2 = 2.4 MPa according to sensor S2. In Fig. 6, we see the oscillogram from sensor S2 for the subsequent disruptions. The time axis is plotted in scale increments of 50 ms, and the voltage (U) axis in increments of 200 mV. An analogous oscillogram is obtained for sensor S2. According to Fig. 6, the disruptions appear at a frequency of 2.5 Hz; each one lasts 190 ms. The maximum pressure during the disruption is p1 = 1.04 MPa according to sensor S1 and

p2 = 2.8 MPa according to sensor S2. The minimum pressure after the disruption is p1 = 0.57 MPa according to sensor S1 and p2 = 0.67 MPa according to sensor S2. In Fig. 6, in contrast to Fig. 5, we may distinguish another periodic process, which occurs at a frequency of about 20 Hz not only during the disruptions but between them. It may be due to vibration of the entire impeller–motor system under the action of the exciting vibrations of the first and subsequent disruptions. Another relevant factor here is the low stability of the lubricant layer in the thrust bearing. When the productivity of the centrifugal compressor falls below the calculated value, an eddy is formed in the impeller channel [1]. That displaces the main flux, and the active cross section of the channel is reduced. The flux velocity in the active cross section is determined by the cross-sectional area and productivity of the compressor. With decrease in the active cross section, the flow velocity in the channel increases, despite the decrease in productivity of the compressor, as confirmed by the pressure rise at the impeller exit according to the Bernoulli equation. According to Figs. 5 and 6, the voltage U increases at disruption by an amount corresponding to increase in the lubricant-layer pressure by a factor of 1.4–2.2. That indicates increase in the axial load at the bearing, evidently on account of the semiclosed impeller design: at breakaway of the rotating eddy, the gas pressure p2 in the impeller channel falls sharply, while the gas pressure beyond the main impeller disk and the housing of the centrifugal stage are not reduced. That leads to an additional pressure difference between the chamber beyond the main disk and the space between the impeller blades and hence an axial force acts until all the gas beyond the impeller has passed to the flow section of the compressor. Gradual decrease in lubricant-layer pressure may be ensured both by the additional pressure difference and by the damping properties of the thrust bearing. To clarify the influence of the semiclosed and closed impeller designs on the dynamic component of the axial force, we recommend comparative tests of the thrust bearing in a centrifugal compressor with a closed impeller. In that case, the pressure beyond the basic disk and casing disk of the impeller should reach equilibrium, and the basic load on the bearing at disruption of the flow should decline. CONCLUSIONS (1) According to our test data for a thrust bearing operating within a centrifugal compressor with a semiclosed impeller, the manufacturing precision and quality of assembly have a considerable influence on bearing operation. Even with slight skewing of the bearing and fluctuation of the thrust disk, the load at the bearing pad is nonuniform. This is accompanied by increase in the dynamic load: greater pressure at the

RUSSIAN ENGINEERING RESEARCH

Vol. 37

No. 4

2017

SLIDING THRUST BEARINGS IN A CENTRIFUGAL COMPRESSOR

pad leads to greater pressure increase at compressor disruption and to pulsations of greater amplitude. (2) In pumping mode, three types of periodic oscillation may be noted in the centrifugal compressor: (1) with the fluctuation frequency of the thrust disk; (2) with the frequency of the gas disruptions; (3) with the frequency of the mechanical vibration of the mobile part of the compressor. (3) In pumping mode, the axial dynamic load at the thrust bearing sharply increases. This may be attributed to the design of the semiclosed impeller and the pressure drop in the flow section of the impeller. (4) For the thrust bearing considered, the lubricant-layer pressure increases by a factor of 1.4–2.2 at flow disruption. (5) The pressure drop after disruption is gradual. That may be attributed to the influence of the additional pressure difference and the damping properties of the thrust bearing.

RUSSIAN ENGINEERING RESEARCH

Vol. 37

No. 4

291

REFERENCES 1. Khadiev, M.B., Zinnatullin, N.Kh., and Nafikov, I.M., The surging mechanism in centrifugal compressors, Vestn. Kazan. Tekhnol. Univ., 2014, no. 8, pp. 262–266. 2. Khadiev, M.B., Sokolov, N.V., and Serazutdinov, M.N., Description of the stand for analysis of dynamic characteristics of thrust sliding bearing with fixed cushions in transient processes, Vestn. Kazan. Tekhnol. Univ., 2012, no. 16, pp. 151–153. 3. Khadiev, M.B., Sokolov, N.V., and Fedotov, E.M., Hydrodynamic, heat, and deformation characteristics of lubrication layers of thrust bearings with parallel bevel to radial inter-cushion channel, Vestn. Mashinostr., 2014, no. 6, pp. 54–59. 4. Sokolov, N.V., Khadiev, M.B., Maksimov, T.V., and Futin, V.A., Testing of thrust sliding bearing with the parallel bevel to inter-cushion channel in centrifugal compressor, Vestn. Kazan. Tekhnol. Univ., 2014, no. 7, pp. 239–244. 5. Fedotov, E.M., Khadiev, M.B., and Sokolov, N.V., RF Inventor’s Certificate no. 2013615 688, 2013.

Translated by Bernard Gilbert

2017