Characterization of Lubricating Oil Using Ultrasound

Characterization of Lubricating Oil Using Ultrasound

Ricardo Tokio Higuti

Dep. Eng. Elétrica, Fac. Eng. Ilha Solteira, Unesp

Avenida Brasil, 364

15385-000 Ilha Solteira, SP. Brazil

[e-mail]

Celso Massatoshi Furukawa

Julio Cezar Adamowski

Dep. Eng. Mecatrônica e de Sist. Mecânicos

Escola Politécnica da USP

Avenida Prof. Mello Moraes 2231

05508-900 São Paulo, SP, Brazil [e-mail], [e-mail]

This paper presents an ultrasonic method to measure small concentrations of water in lubricating oil. It uses an ultrasonic measurement cell composed by a piezoceramic emitter (5 and 10 MHz), and a large aperture PVDF receiver that eliminates diffraction effects. The propagation velocity, attenuation coefficient and density of several samples of water-in-oil emulsion were measured. The concentrations of water of the samples were in the range of 0 to 5% in volume, and the results showed that these low concentrations can be discriminated within a resolution of 0.2% in the studied range, using the measurement of the propagation velocity.

Keywords: Ultrasound, liquid characterization, lubricating oil

Introduction

The characterization of liquids has a wide range of applications in industry and science. In many cases it is necessary to monitor some variables during the processing of a product, such as concentration, crystallization and polymerization (Povey, 1997). Some specific applications are referred to the determination of fat content in bovine milk (Higuti et al., 1999) and the determination of water contamination in lubricating oil, which is a problem found in hydroelectric power plants. In the chemical, biochemical and biology fields, studies of chemical and structural relaxation processes are of interest in pure liquids and solutions (Kaatze et al., 1993).

In hydroelectric power plants, the generators require huge quantities of lubricating oil. The oil can be contaminated with water due to ambient moisture and malfunctioning of devices such as heat exchangers. Small quantities of water, in the order of 1% in volume, can alter the lubricating characteristics of the oil. That means that it is necessary to have a continuous monitoring of the oil, so that preventive maintenance procedures (such as changes or cleaning of the oil) can be planned, reducing costs and risks of non-programmed halts.

A liquid can be characterized by its physical properties, such as density, viscosity, conductivity or dielectric permittivity. The classical methods of viscosity measurement require moving parts that are inadequate to be applied on-line. Capacitive methods have not shown good sensitivity for low water content. Other methods such as chemical reaction and centrifugation (Pal, 1994) can be very accurate, but require laboratorial analysis of samples.

Other possibility is to measure acoustic parameters such as the propagation velocity, attenuation coefficient and acoustic impedance, using ultrasound. These parameters can be used to non-destructively evaluate variables of industrial processes, such as the concentration of particles in emulsions and suspensions.

In this work it is used a measurement cell with a double-element transducer (Adamowski et al., 1995) to measure the density, acoustic attenuation and propagation velocity of water-in-oil emulsions. Due to the small water content in the oil-water mixture, the density variations are very small, being of the same order of the measurement errors. Variations in the attenuation coefficient as a function of water content are also very small due to the small number of water droplets. On the other hand, the propagation velocity in the emulsion shows a clear dependence on the water content.

Nomenclature

a_i(t) = echo signal, i=1, 2, 3

A_i(f) = frequency spectrum of echo a_i(t)

c₁ = propagation velocity in the delay line I, m/s

c₂ = propagation velocity in the liquid, m/s

L = sample chamber length, m

L = delay line I length, m

R_ij = reflection coefficient at the interface between media i and j

Greek Symbols

Dt = time delay, s

a = attenuation coefficient, Np/m

b = adiabatic compressibility, m.s²/kg

r = density, kg/m³

f = volume concentration, %

Sound Waves in Non-Homogeneous Media

A sample of oil contaminated with water can be considered an emulsion, which is a two-phase system where droplets of water are dispersed in the oil – the continuous phase. When a sound wave propagates through an emulsion, the amplitude and the phase of the wave change, depending on the concentration of the dispersed phase and the properties of the constituent phases. These changes are caused by viscous and thermal effects in the boundaries of the droplets of water (Epstein and Carhart 1953, McClements and Povey 1989).

The viscous effect is related to the difference between the densities of the dispersed and the continuous phases. Consider a particle immersed in the continuous medium. When a sound wave propagates through this system, the density difference causes a relative movement between the particle and the continuous medium, generating a shear wave that is rapidly attenuated within a distance d_n, called viscous depth. The viscous depth depends on the kinematic viscosity of the continuous phase and the frequency of the incident wave.

In the same way, the discontinuity of the thermodynamic properties of the medium causes the thermal losses. The coupling between pressure and temperature is responsible for the generation of small temperature oscillations that are inhomogeneous around the dispersed particles, generating a heat flux in these regions. Such effect can be seen as a thermal wave, which is rapidly attenuated within a distance d_t, called thermal depth. The thermal depth depends on the thermal diffusivity of the continuous phase and the frequency of the acoustic wave.

The viscous and thermal effects remove part of the energy of the incoming wave and change its phase. This effect can be observed by measuring the variation of the attenuation coefficient or the propagation velocity as a function of the concentration and/or the frequency of the wave.

Since the particle concentration is low and their size is small (around 1 mm) when compared to the sound wavelength, the emulsion can be approximated by a homogeneous medium. This approximation is employed in the model of propagation proposed by Urick (1947). The adiabatic compressibility b and the density r can be written as effective values in terms of the volume concentration of water f and the physical properties of the water and the oil.

The propagation velocity in the non-homogeneous medium results

which is called Urick Equation. For low concentrations, equation 3 is a linear function of the water concentration.

The Measurement Cell

The measurement cell used in this work is shown in figure 1. It is composed of three main parts: a double-element transducer, a liquid chamber and a reflector.

The double-element transducer has a piezoelectric ceramic emitter, a PMMA (Poly Methyl Metacrylate) delay line (I), a P(VDF-TrFE) (PolyVinylidene Fluoride TriFluoroEthylene) receiver and a second delay line (II) made of a glass with low acoustic attenuation, allowing the use of frequencies up to 10 MHz. The P(VDF-TrFE) receiver, 52 mm-thick, is metallised in both faces and has a large diameter (60mm) when compared to the diameter of the emitter, so it can intercept the acoustic field completely, eliminating diffraction losses.

The reflector is made of stainless steel, being part of the cell body. A peristaltic pump keeps the flow of the sample through the chamber. The most important factor for the design and the construction of the cell is the parallelism of all interfaces.

Measurement of Acoustic Parameters

The Multiple Reflection Method (MRM) is used, and a detailed description of the method is given elsewhere (Adamowski et al., 1995). As shown in figure 1, the emitted signal a_T propagates through the delay line I and reaches the receptor. The P(VDF-TrFE) receiver measures the signal a_T, and the multiple reflections in the various interfaces, represented by the echoes a₁, a₂ and a₃. From the magnitudes of the Fourier transforms of these echoes at a fixed frequency (A₁, A₂ and A₃), one can calculate the reflection coefficient in the interface delay line II / liquid (R₁₂), with

The propagation velocity in the liquid, c₂, is calculated from the time interval between the echoes a₁ and a₂, using the cross correlation technique improved with the Hilbert transform (Cabot, 1981). The velocity is given by

where L is the liquid sample length.

From the values of R₁₂ and c₂, the liquid density is determined by

where r₁ is the delay line II density, that is known in advance, and c₁ is the delay line II propagation velocity, measured in the same way as c₂ but using the echoes a_T and a₁.

The attenuation coefficient of the liquid is given by

where R₂₃ is the reflection coefficient of the liquid/reflector interface. It is assumed that the impedance of the reflector is known and does not vary considerably in the temperature range of operation. As it is shown in the experimental section, the temperature remained constant during the measurements.

Experimental Setup

Figure 2 shows the experimental setup. The piezoelectric emitter is driven with a tone burst obtained from a programmable function generator (Tektronix AFG 5102) and a power amplifier (Amplifier Research 150A100A). An ultrasonic analyzer (Panametrics 5072PR) amplifies the echoes, which are digitized by a digital oscilloscope (Hewlett-Packard HP54820A) and transferred via a GPIB interface to a computer, where the data are processed.

Experimental Results

Samples of oil with known water content were prepared in laboratory, employing oil used in velocity regulators (Mobil DTE), obtained from the hydroelectric power plant of Ilha Solteira, Brazil.

Several concentrations in volume of water in oil (f) between 0 and 5% were obtained by mixing distilled water with the oil, without any emulsifier. Before the experiments, the samples were homogenized with a disperser (Ika Ultra-Turrax T25) for 2 minutes at 13500 rpm, and left to rest for a few hours to eliminate air bubbles.

The experiments were conducted at two frequencies: 5 and 10 MHz. For each frequency, a specific emitter transducer was used. After mounting each transducer, the measurement cell was calibrated with distilled water and the samples were measured in sequence. The characteristics of the transducers and the sample temperature for each set of measurements are shown in Table 1.

The transducers were excited with tone bursts, containing two or three cycles in the respective central frequencies. Five acquisitions for each experiment were made to estimate the means and standard deviations of the propagation velocity, density and attenuation coefficient. The oscilloscope was used in average mode to increase the signal-to-noise ratio of the acquisitions.

The measurement cell was calibrated with distilled water to measure the sample length and the alignment of the parallel interfaces. This calibration was made at room temperature In all cases, the density of the distilled water measured during the calibration procedure showed errors smaller than 1% compared to tabulated values.

Measuring the density, the results remained nearly constant with the concentration, due to the low concentration and the proximity of the densities of water (997 kg/m³) and oil (872 kg/m³), as can be observed in figure 3. It is also shown the density of the oil without water (0%), measured with a pycnometer, at a temperature of 24^oC.

Figure 4 shows the attenuation coefficient as a function of the water concentration at 5 MHz of frequency. The attenuation coefficient increases for low volume fractions and decreases for higher concentrations. A similar behavior is observed when it was used the frequency of 10 MHz. As it can also be observed in figure 4, the additional attenuation due to water droplets is small in comparison with the intrinsic attenuation of the oil. As these variations are of the same order of the experimental errors, these attenuation measurements can not be used to characterize the samples. On the other hand, experiments conducted to determine the fat content in milk, which can be considered a kind of oil-in-water emulsion (the inverse problem), showed that small percentages of fat can increase substantially the attenuation in the emulsion (Higuti et al, 1999).

However, useful results were obtained measuring the propagation velocity, as illustrated in figures 5 and 6. These curves show also the propagation velocity in the delay line II. Obviously it should not vary with the water concentration of the sample, and any variation in the velocity in the glass would be caused by temperature fluctuations. As these velocities remained constant (it should be noted the different scales for the sample and delay line), we can conclude that the temperature also remained constant during the measurements for a given frequency. It is an important observation, which allows to conclude that the small variations in the propagation velocity in the samples are in fact due to the varying water content, and not to a variation in the temperature during the experiment.

Using equations 1 to 3, and the compressibility and densities of water and oil, one can conclude that the curve of velocity increases linearly with the concentration of water. Table 2 shows the properties of the constituent phases, measured in laboratory, and figure 7 shows the experimental and theoretical values of the propagation velocity.

The differences between the experimental and theoretical values are very small – less than 0.070%, but it should be noted that the total variation in the propagation velocity is also very small, of the order of 0.1%. Therefore, although the differences seem to be small, they are in fact substantial when they are compared to the total variation of the velocity.

The slope of the curve of velocity shows that it is possible to determinate different water concentrations. This slope is about 0.23 m/s per water concentration (in percent) for concentrations above 1%, and the errors in the measurements of velocity are around ± 0.003% (± 0.04 m/s). Thus, measuring the propagation velocity, it is possible to discriminate water concentrations with a resolution of 0.2%. However, temperature variations can mask these effects. If the temperature gradient is small, this effect can be corrected by an appropriate measurement of the propagation velocity in the glass delay line.

Conclusions

Low concentrations of water in oil introduce small changes in the acoustic parameters to be measured, and a very precise method of measurement is needed in order to characterize the samples. The attenuation could not be used to determine very small water concentrations in the oil because it was only possible to measure it with a precision of 1%, which is of the same order of variation of water concentration. Improvements in the cell are being conducted to obtain more precise measurements of attenuation.

This measurement cell allows a very precise determination of the propagation velocity, which showed to be the best parameter that can be used to determinate water contents in oil. The curve of velocity as a function of water concentration shows a slope of 0.23 m/s per percent of water in the range from 1% to 5% of water content. The sensitivity is even higher when the water concentrations is less than 1%. Using the velocity measurement method described in this paper, it is possible to discriminate water contents in the order of 0.2%.

Care must be taken regarding to temperature variations, which can cause substantial changes in the propagation velocity, due to the small concentrations involved. However, in a hydroelectric power plant, the volume of oil is large, and it should not have sudden temperature variations.

In a future work, using an ultrasonic emitter with broader bandwidth, the frequency dependence of the measured parameters (e.g. dispersion) can give rise to a higher sensitivity. For example, in figures 5 and 6, for low concentrations, the slope of the velocity curve for 10 MHz is higher than for 5 MHz. The effect of the particle size is also important, which should also be studied within a broader frequency range.

Acknowledgements

The authors would like to thank Capes, FAPESP and CNPq for providing financial support, and Eng. Kishi from CESP - Ilha Solteira, for providing the samples of oil.

Paper originally presented at the 15^th Brazilian Congress of Mechanical Engineering (XV COBEM), São Paulo, November 22-26, 1999.

COBEM Editors: R. G. dos Santos, M. H. Robert, A. C. Dannwart, J. R. B. Cruz.

Associate Editor: J. R. F. Arruda.

• Adamowski, J. C., Buiochi, F., Simon, C., Silva, E. C. N. and Sigelmann, R., 1995, "Ultrasonic Measurement of Density of Liquids", The Journal of the Acoustical Society of America, vol. 97, pp. 354-361.
• Cabot, R. C., 1981, "A Note on the Application of the Hilbert Transform to Time Delay Estimation", IEEE Transactions on Acoustics Speech and Signal Processing, vol. ASSP-29, no. 3, pp. 607-609.
• Epstein, P. S. and Carhart, R. R., 1953, "The Absorption of Sound in Suspensions and Emulsions. I. Water Fog in Air", The Journal of the Acoustical Society of America, vol. 25, n. 3, pp. 553-565.
• Higuti, R. T., Bacaneli, F., Furukawa, C. M., Adamowski, J. C., 1999, "Ultrasonic Characterization of Emulsions: Milk and Water-in-Oil", in Proc. IEEE Ultrasonics Symposium, Lake Tahoe, USA.
• Kaatze, U., Kühnel, V., Menzel, K. and Schwerdtfeger, S., 1993, "Ultrasonic Spectroscopy of Liquids. Extending the Frequency Range of the Variable Sample Length Technique", Measurement Science and Technology, vol. 4, pp. 1257-1265.
• McClements, D. J. and Povey, M. J. W., 1989, "Scattering of Ultrasound by Emulsions", Journal of Physics D, vol. 22, n. 1, pp. 38-47.
• Pal, R., 1994, "Techniques for Measuring the Composition (Oil and Water Content) of Emulsions - a State of the Art Review", Colloids and Surfaces A: Physicochemical and Engineering Aspects, vol. 84, pp. 141-193.
• Povey, M. J. W., 1997, "Ultrasonic Techniques for Fluids Characterization", Academic Press, San Diego, USA.
• Urick, R. J., 1947, "A Sound Velocity Method for Determining the Compressibility of Finely Divided Substances", Journal of Applied Physics, vol. 18, pp. 983-987.

Publication Dates

• Publication in this collection
  24 Sept 2002
• Date of issue
  2001