Tribology International 44 (2011) 963–970

Contents lists available at ScienceDirect

Tribology International

0301-67

doi:10.1

n Corr

E-m

journal homepage: www.elsevier.com/locate/triboint

On-line acoustic viscometry in oil condition monitoring

L.V. Markova a,n, N.K. Myshkin a, H. Kong b, H.G. Han b

a V.A. Belyi Metal-Polymer Research Institute of Belarus National Academy of Science, 32A Kirov Street, Gomel 246050, Belarusb Korea Institute of Science and Technology, P.O. Box 131, Cheongryang, Seoul, Republic of Korea

a r t i c l e i n f o

Article history:

Received 11 August 2010

Received in revised form

17 March 2011

Accepted 22 March 2011Available online 31 March 2011

Keywords:

On-line condition monitoring

Lubricating oil viscosity

Rheology

Magnetoelastic viscometer

9X/$ - see front matter & 2011 Elsevier Ltd. A

016/j.triboint.2011.03.018

esponding author. Tel.: þ375 242 774635; fa

ail address: lvmark@mail.ru (L.V. Markova).

a b s t r a c t

The paper describes the theoretical standpoints of developing magnetoelastic viscometers and a

concept of viscosity measurement. The magnetoelastic viscometer has shown the readings close to the

capillary viscometer. Testing of the oils with PMMA viscosity-index improvers by viscometers has

indicated changes in rheological properties observed in the non-Newtonian behavior of the oils. With

increase in content or molecular weight of the improver, the non-Newtonian behavior of the oil

appeared at lower frequencies of viscosity measurements.

& 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Safety and reliability of tribosystems during operationdepends much on properties of the lubricant as well as oilcondition monitoring and diagnostic means. Viscosity is a keyphysicochemical parameter of oil quality and ability to providethe effective lubricating layer thickness between friction surfaces,so as to prevent severe wear [1–3].

Viscosity of oils and their viscosity–temperature properties aredependent on the boiling temperature of the oil fraction, its apparentmolecular weight and its hydrocarbon structure. Viscous (thickening)additives in the form of polymer compounds are very often used toimprove the viscosity–temperature properties of the oils. To namebut a few: polymethacrylates, polyisobutylenes, products of vinyl–butyl ether polymerization (vinypol) and some other.

Along with the correct choice of the oil, it is important tomaintain the prescribed viscosity of the oil during service life, whichrequires regular monitoring of viscosity. If a change in the oilviscosity is detected, subsequent analysis of the oil can identifythe cause of the disturbance of its properties. Any increase ordecrease in viscosity may result in oil bearing capacity variations.The increase in viscosity can be a sign of oil oxidation, polymeriza-tion or contamination by soot or other substances. Lowered viscositycan be caused by the fuel ingress into the oil, decomposition ofviscosity-index (VI) improvers or thermal decomposition of the oil.Simultaneously running processes are able to compensate viscosityvariations, e.g. diesel oil might be contaminated by fuel, which

ll rights reserved.

x: þ375 242 775211.

reduces viscosity, or by soot leading to its increase. These changes inoil viscosity are often the first sign of a serious problem in thetribosystem. For instance, when the oil looses antioxidative stability,its viscosity increases and eventually disturbs normal operationunless the condition monitoring warns about the problems in thelubricating system. In this connection, knowledge of oil viscositycharacteristics is critical in design and prediction of the behavior oflubricated mechanical systems.

Oil viscosity is usually estimated by the absolute viscosity Z(viscosity defined as a constant of proportionality between thestrain rate and generated thereby shear stress) or a kinematicviscosity n (flow resistance of the fluid under gravitation effect),which is equal for Newtonian fluid to the absolute viscosity ratioto density of the fluid rf:

n¼ Zrf

: ð1Þ

When measuring viscosity one should bear in mind that thelubricant may display either Newtonian or non-Newtonian prop-erties depending on its composition and state. For the Newtonianfluid the absolute viscosity is independent of the shear rate,whereas for non-Newtonian fluid the viscosity is a function ofthe shear rate: Z¼Z(de/dt), where de/dt is the shear strain rate(e is the shear strain, t the is time). Both the base mineral oil andsynthetic ones mostly show the properties of the Newtonian fluid.Oxidation and contamination of oils in the course of operationdeviates viscous properties from the Newtonian ones. Introduc-tion of VI improvers into the base oil or/and formation of thewater–oil emulsion transforms the oil to the non-Newtonianfluid. Besides, if the oil is sheared between two parallel surfaces,one of which is sliding against the other, the shear stress increases

L.V. Markova et al. / Tribology International 44 (2011) 963–970964

linearly with velocity at low sliding velocity. When velocity isincreased to a certain value, the stress in the oil reaches the limitof the Newtonian behavior and the shear stress increases at lowershear rate [4]. To estimate viscosity of the non-Newtonian oils thetrue parameter of interest for the oil analyst should be theabsolute viscosity, which determines the oil film thickness andthe degree of the friction surface protection. However, in view ofeconomy and simplicity, kinematic viscosity of the oil is typicallyused for lubrication monitoring.

Measurement of viscosity (viscometry) is at present based onfinding the resistance experienced by a body moving in amedium, in which viscosity is being studied or when the con-sidered fluid is flowing through a channel of the given geometry.Most spread procedures are the capillary, rotation and vibrationviscometry, and the falling ball method.

A wide range of high-accuracy viscometers has been createdfor laboratory analysis. However, rather large size and high costprevent their usage in the on-line measurement.

In many cases the continuous condition monitoring of oil isnecessary to ensure safe and reliable operation of machines.Therefore, the creation of new efficient methods of viscositymonitoring in a real-time and devices built into the oil line isan urgent task intended to raise reliability of equipment, elim-inate periodic oil sampling for laboratory analysis and cut main-tenance expenses.

The purpose of the paper is to study the peculiarities of newacoustic on-line techniques for viscosity monitoring based onmagnetoelastic effects.

2. On-line oil viscometry

Viscosity monitoring devices built into the oil circulation lineshould obey specific requirements, namely: give reliable informa-tion on the lubricant viscosity, eliminate frequent calibration ormaintenance and show a long-term durability in hostile environ-ment of machines experiencing high temperature, pressure andvibration. At the same time, they must be compact and inexpensive.

2.1. On-line viscometry techniques

On-line viscometers are designed using different approaches:macro-displacement of solid body in fluid, as well as vibrationand acoustic methods.

The methods using macro-displacements are based on mod-ifications of laboratory measurement procedures. The capillary,rotary and the falling ball viscometers belong to the meansmountable in the oil line of various machines. These viscometersare rather complex and bulky, and their moving components areimpairing reliability.

The vibration and acoustic viscometers without macro-displace-ments have been developed as an alternative to the on-linemonitoring. Vibration viscometers are based on analyzing torsionoscillations [5], or those of vibrating cantilever [6], or tuning fork [7].

2.2. Solid-state microacoustic technique

Most promising among the built-in devices, in particular, incor-porated in the lubricating system of vehicles, turned out to bethe microacoustic solid-state viscometers. They operate by excitingthe elastic compression waves (acoustic waves) in fluid and esti-mating their parameters. Piezoelectric materials are used in theseviscometers to generate the acoustic waves by the electric field.

The devices generating the waves with the thickness shear mode(TSM) through the crystal are most simple and widespread in thelubricant viscometry. The viscometers of this type commonly

include the thin AT-cut quartz disk having two circular electrodeson its faces. The voltage applied between the electrodes producesshear strain in the crystal. By varying voltage frequency it is possibleto find the mechanical resonance of the resonator immersed in oil.The resonance frequency shift and the oscillation amplitude varia-tions give information on the oil state [8–10]. The TSM viscometer ischaracterized by design simplicity, stability in severe conditions andtemperatures. However, high operating frequencies (1–200 MHz)require the use of intricate electronic techniques. The devices aresensitive only to the properties of thin oil layer at the interface (oil-sensitive surface of the sensor) because the acoustic wave penetra-tion depth d into the fluid is inversely proportional to a square rootof frequency f [11]:

d¼Z

prf f

!1=2

: ð2Þ

When the elastic high-frequency waves propagate through thehigh-molecular fluids, the fluids behave like gels since oscillationfrequency of the large molecules either coincides or is lower thanthe oscillation frequency of the sensor. Therefore, the viscometerreadings might reflect inadequately viscous properties of the oil.Besides, the output signal is strongly dependent not only onviscosity but also on the oil oxidation degree [10].

Along with the solid-state viscometers based on acousticwaves, we can name magnetoelastic viscometers operating atlower frequencies [12,13].

2.3. Magnetoelastic technique

The phenomenon of magnetoelastic interactions, meaning themutual effect of magnetization and elastic strains of the medium,makes the base of magnetoelastic viscometry. This phenomenonappears in response to variations in dimensions and shape of thespecimen being magnetized (magnetostriction), and to thechanges in magnetization of the specimen subjected to deforma-tions (magnetoelastic or Villary effect) [14].

The magnetoelastic technique provides exciting the longitudi-nal standing elastic wave in a strip by generating the alternatingmagnetic field H

!f in the presence of a dissipative force, i.e.

friction force F!

fr acting on the strip immersed in the viscousfluid being monitored (Fig. 1a).

In addition to longitudinal waves excited by oscillations of thestrip in the Y direction, rapidly decaying transversal or shearwaves exist in the viscous fluid, which propagate along the X-axis.Notice that, the particles in these waves move along the Y-axis[15]. The shear wave amplitude decays with distance from theoscillating surface according to the exponential law, while thepenetration depth of the oscillations into the fluid reduces withincreasing frequency and increments with increasing viscosity(formula (2)). Fig. 1b shows the dependence of the penetrationdepth of the shear wave into the oil (Z¼40 mPa s, rf¼870 kg/m3)on the oscillation frequency.

Natural longitudinal oscillations of the strip immersed inviscous fluid and oscillating along the Y-axis are the harmonicoscillations with exponentially decaying amplitude (Fig. 2)expressed by the following equation [15]:

yðtÞ ¼ Ae�zo0t cosðodtþjÞ, ð3Þ

where Ae�zo0t is the oscillation amplitude, z is the viscous

damping factor, o0 ¼ 2pf0 is the angular frequency of natural

oscillations of the strip in air (without friction), od ¼ 2pfd the

angular frequency of the strip oscillations in fluid, od ¼o0

ffiffiffiffiffiffiffiffiffiffiffiffi1�z2

qand j the phase shift.

0

5

10

15

20

25

30

0.01

f, MHz

Magnetoelastic viscometer

Solid-state viscometer

Y

X

Ffr

2

1

Viscous fluid

�

Hf Z

ds

L

δ, μ

m

0.1 1 10 100 1000

Fig. 1. Magnetoelastic strip under the effect of AC magnetic field H!

f and friction

viscous force F!

fr (a): 1 – strip, 2 – shear wave damped by viscous fluid; dependence

of shear wave penetration depth d in a fluid on operation frequency (b).

Time

Y

t2

t1

y2 = A·exp(-ξ2ωnt)

y1 = A·exp(-ξ1ωnt)

y = y0

Fig. 2. Damped oscillations of a strip along Y-axis in viscous fluid.

L.V. Markova et al. / Tribology International 44 (2011) 963–970 965

The rate of the amplitude decay of longitudinal oscillations ofthe strip immersed in viscous fluid is described by the dampingviscosity factor z of the fluid.

Frequency f0 of the fundamental harmonic of natural oscilla-tions of the magnetoelastic strip placed in non-viscous medium(air) is expressed by the equation derived from the theoretical

model of oscillations of thin elastic strip [11]:

f0 ¼1

4Ls

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiEs

rsð1�ss2Þ

s: ð4Þ

where Ls is the strip length, rs the material density of the stripand Es, ss the Young modulus and Poisson’s ratio of the strip,respectively.

Thus, the natural frequency of the magnetoelastic stripdepends on its dimensions and material properties.

When the strip is immersed in the fluid being monitored, thefrequency fd of its natural oscillations (resonant frequency)reduces by a value Df [12]:

Df ¼ f0�fd ¼

ffiffiffiffiffiffiffipf0

p2ffiffiffi2p

prsds

ðZrf Þ1=2: ð5Þ

Thus, by measuring shift Df, it is possible to find acousticviscosity AV, which is the product of viscosity by density:

AV ¼ Zrf ¼Df 2 8pr2s ds

2

f0: ð6Þ

In contrast to viscosity, oil density varies only slightly duringoperation. Hence, the following relation can be used to estimatedynamic viscosity:

Z¼ AV

rf

: ð7Þ

When developing magnetoelastic facilities for viscosity mon-itoring, it is important to consider the dependence of temperatureversus natural frequency of the strip as far as density rs(T) anddimensions of the strip Ls(T), ds(T) are a function of temperature.In this connection, one should also envisage temperature com-pensation in these techniques.

3. Experimental details

3.1. Magnetoelastic viscometer

Relatively cheap amorphous ferromagnetic alloys Fe40Ni38-

Mo4B18 (Metglas 2826MB) and Fe81B13.5Si3.5C2 (Metglas 2605SC)with high mechanical tensile strength (1000–1700 MPa) are mostfrequently used in miniature sensors. They possess the property ofmagnetostriction (elongation l¼DLs/L under magnetic field effect)about 10�5 and high coefficient of magnetoelastic coupling 0.98(the efficiency of magnetic energy transformation into the elasticone). In contrast, the coefficient of magnetomechanical coupling ofwidely applied nickel magnetostrictive alloys is about 0.4. The use ofthis material allows the designers to reduce the size of sensors [16].It should be noted that, elasticity of these materials depends onstrength of the magnetic field. In order to optimize the operatingpoint of the sensitive element, i.e., to obtain the maximal change inrelative elongation under the effect of alternating field, an additionalstationary bias magnetic field is usually applied.

The Metal-Polymer Research Institute (MPRI NASB, Belarus)jointly with Korean Institute of Science and Technology (KIST,Korea) has developed a magnetoelastic viscometer for conditionmonitoring of lubricating oils.

The viscometer consists of a probe screwed in the oil tank oroil pipe and an electronic module (Fig. 3).

The probe consists of viscosity and temperature sensors. Thefirst one includes the sensitive magnetoelastic (37 mm�6 mm�0.03 mm) strip 1 made of the amorphous metal glass Metglas2826MB, and electromagnetic coil 2 (Fig. 3a). The coil servessimultaneously to generate the signal that excites mechanicaloscillations of the strip and to measure the signal induced by theseoscillations damped by the viscous oil 4. The second sensor 3

Input-output device

Low-pass filter

Amplifier

Excitation pulse voltage

Source of DC voltage of bias field

MixerMicrocontroller

ELECTRONIC MODULE

Viscositysensor

Temperaturesensor

PROBE

Fig. 3. Probe design (a) and block diagram (b) of magnetoelastic viscometer: 1 – magnetoelastic strip; 2 – electromagnetic coil; 3 – temperature sensor; 4 – lubricating oil.

L.V. Markova et al. / Tribology International 44 (2011) 963–970966

(platinum thermometer) is used to measure the oil temperature,which is important in interpreting viscosity readings and performingtemperature compensation. The magnetic strip, electromagnetic coiland temperature sensor are installed in nonmagnetic housing withtwo holes through which the oil flows to temperature sensor 3 andmagnetoelastic strip 1. To protect the magnetoelastic strip frommechanical damage the housing face immersed in the oil isequipped with a protective mesh.

The probe is fit with an electrical connector adapted to providethe electrical interface of the oil viscosity and temperaturesensors to the electronic module. The electronic module isoperated by microcontroller with processor programmed to con-trol the viscometer and to process its output data. Besides, themicroprocessor generates the excitation pulse voltage, sends andreceives data to and from the input–output device, which con-tains a display, a keypad and alarm indicator. The electronicmodule additionally contains low-pass filter and amplifier topreprocess the output signal of the viscosity sensor. Besides, DCvoltage source of the bias field and mixer are used to form theexcitation voltage applied to the magnetoelastic strip. The micro-controller stores such initial parameters as the fresh oil viscosityand lubricant temperature as well as threshold values of themeasured parameters to determine the condition of the oil.

3.2. Measurement techniques

Determination of viscosity consists of finding resonant oscilla-tions of the strip immersed into the oil followed by the analysis ofobtained signal parameters. The magnetoelastic viscometer realizesthe method of viscosity estimation by the resonant frequency shiftand the technique of analyzing the damped amplitude curve.

The control software performs the following measurementalgorithm.

To optimize the performance of the magnetoelastic stripthe excitation pulse voltage is added to the DC voltage inthe mixer. The formed excitation voltage is fed to electromagneticcoil 2 of the viscosity sensor. Forced oscillations of the magne-toelastic strip are excited by the electromagnetic field of the coil.The DC passing through the electromagnetic coil creates thestationary magnetic field that magnetizes the magnetoelasticstrip, providing thereby the conditions for the efficient operation.Simultaneously, the AC passing through the coil induces analternating magnetic field exciting the longitudinal elastic oscilla-tions in magnetoelastic strip 1 (Fig. 3). The AC is then switchedoff while the magnetoelastic strip continues oscillating at reso-nant (natural) frequency with decaying amplitude. The elasticoscillations of the magnetoelastic strip generate in the coil the ACof frequency fd corresponding to the oscillation frequency of thestrip in the oil. After a low-pass filtration and amplification,the response signal is transferred to the microcontroller.

To determine frequency fd the frequency fexc of the excit-ation pulse voltage is varied within the fmin–fmax range. Theextreme frequencies of this range are found from the limitingvalues Zmin and Zmax of the test oil viscosity by formula (5):

fminðZmaxÞ ¼ f0�Df ðZmaxÞ ¼ f0�

ffiffiffiffiffiffiffipf0

p2ffiffiffi2p

prsds

ðZmaxrf Þ1=2,

fmaxðZminÞ ¼ f0�Df ðZminÞ ¼ f0�

ffiffiffiffiffiffiffipf0

p2ffiffiffi2p

prsds

ðZminrf Þ1=2:

5

15

25

35

45

18

Kin

emat

ic v

isco

sity

, cSt

Oil temperature, oC23 28 33 38 43

L.V. Markova et al. / Tribology International 44 (2011) 963–970 967

3.2.1. The technique based on measurement of resonant frequency

shift

This technique of determining viscosity is based on measure-ment of the oscillation frequency fd of the strip and its comparisonto frequency fexc of the excitation pulse voltage, which is scannedwithin the range fmin–fmax. The resonant frequency fd is recordedin the case equality fd¼ fexc is true. The temperature dependenceof the natural frequency is compensated as follows. The naturalfrequency f0(T) of the magnetoelastic strip in air is found from theanalytic dependence stored in the microcontroller, which corre-sponds to the measured temperature T. This value of thefrequency is used to correct frequency shift fd(T) of dampedoscillations in the oil:

DfdðTÞ ¼ f0ðTÞ�fdðTÞ: ð8Þ

Acoustic viscosity AV is calculated by formula (6).

45

55

65

75

85

95

Kin

emat

ic v

isco

sity

, cSt

3.2.2. The technique based on analysis of amplitude decay curve

This technique involves determination of time t lasting tilltermination of the oscillation decay process at the resonantfrequency (Fig. 2). The decay time is defined by the number ofoscillations N of the strip (recorded pulses) whose amplitudeexceeds the given threshold y0: Aexpð�zontÞ4y0. The indicatorof the resonance in this technique is the maximal number of striposcillation pulses determined at excitation frequency in the rangefmin–fmax. Viscosity is found from prerecorded calibration depen-dence AV(N).

The output device of the electronic module shows viscosityand temperature of the oil at which this viscosity has beenmeasured.

25

Theoretical Experimental

Oil temperature, oC30 35 40

Fig. 4. Theoretical and experimental temperature dependencies of kinematic

viscosity obtained by the estimation of resonant frequency shift (a) and by the

analysis of the amplitude decay curve (b).

4. Results and discussion

4.1. Verification of data measured by two developed techniques

We have analyzed the theoretical and rated temperaturedependencies of kinematic viscosity of synthetic oils.

The theoretical temperature dependencies were found inaccordance with ASTM D341 for the fluids with above 2 cStviscosity using the formula:

lgðlgðnþ0:7ÞÞ ¼ a�b lgðTþ273Þ: ð9Þ

where T is the oil temperature (1C), n the kinematic viscosity (cSt)and a, b are the coefficients found from kinematic viscosity valuesat two specified temperatures (40 and 100 1C, respectively).

The experimental data on kinematic viscosity were obtainedby recalculation of the measured acoustic viscosity using therelations derived from Eqs. (1) and (7):

n¼ AV

r2f

: ð10Þ

Fig. 4 illustrates the theoretical and experimental temperaturedependencies of the kinematic viscosity obtained by the techni-que based on the resonant frequency shift (data for synthetic oilPAO 4 are presented in Fig. 4a) and by the analysis of theamplitude decay curve (for synthetic oil PAO 8, Fig. 4b).

The results have proved that the technique of viscositymeasurement based on the oscillation decay rate yields moreaccurate data (75% error) than that based on the resonantoscillation frequency shift (715% error).

4.2. Estimation of correlation between the readings of

magnetoelastic, capillary and solid-state acoustic viscometers in

testing base synthetic and mineral oils

Viscosities of four synthetic (PAO 4, PAO 6, PAO 9 and PAO 40)and two mineral base oils (P-96 and P-480) were evaluated.

Table 1 presents the kinematic viscosity values measuredunder 40 1C by capillary viscometer Cannon–Fenske Viscometer(Glass Capillary Viscometer, ASTM D 446) following the standardmethod (ASTM D445), acoustic viscosity values measured bymagnetoelastic viscometer using the procedure of analyzing thedecay rate of oscillations and by solid-state acoustic viscometer(ViSmart, Vectron Int.). The table also lists the correspondingvalues of kinematic viscosity calculated by formula (10) withaccount of oil density measured under ASTM D 891-09.

The comparison of measurement results shows that the dataobtained by the developed viscometer are the closest to theviscosity measured by the capillary technique within the wholemeasurement range 17–500 cSt. In contrast, the solid-state acous-tic viscometer has shown less validity in measuring viscosityabove 300 cSt. Its readings are much higher than the expectedones, evidently because the layer of viscous oil on the sensingsurface of the viscometer behaves like a gel under high operatingfrequencies (5.3 MHz).

Table 1Measurement results of oil viscosity values by capillary, magnetoelastic and solid-state viscometers at 40 1C.

Viscometer Measured parameter Oils under test

PAO 4 PAO 6 PAO 8 P-96 PAO 40 P-480

Capillary viscometer n (cSt) 17.3 30.3 47.7 96.2 392.6 496.6

Magnetoelastic viscometer AV (cP kg/cm3) 12.7 21.8 30.0 70.0 299.6 417.6

n (cSt) 19.3 32.6 44.2 92.1 423.6 533.2

Solid-state viscometer AV (cP kg/cm3) 12.8 20.2 31.2 79.2 310.0 468.2

n (cSt) 19.5 30.2 45.9 104.2 438.3 597.8

ASTM D 891-09 rf (g/cm3) 0.811 0.818 0.824 0.872 0.841 0.885

Table 2Viscosity monitoring in the course of oil aging.

Oil sample code Sample#1 Sample#2 Sample#3 Sample#4

Duration of oxidation

Hours 0 8 26 35

Days 0 1 3 4

Capillary viscometer ASTM D 446

n (cSt) 236 239 245 247

Solid-state viscometer

AV (cP g/cm3) 201 205 210 212

n (cSt) 248 253 259 262

Magnetoelastic viscometer

AV (cP g/cm3) 197 198 199 204

n (cSt) 243 244 246 251

Table 3Kinematic viscosity of oil samples with VI improvers.

VI improver Mwa (Da) Cb

¼0% C¼3% C¼6% C¼9%

Kinematic viscosity (cSt) at 40 1C

PMMA-1 40,311 30.40 34.18 38.16 42.62

PMMA-2 60,795 30.40 33.40 37.14 41.78

PMMA-3 155,259 30.40 34.40 39.04 44.55

PMMA-4 530,537 30.40 33.31 37.16 42.42

Kinematic viscosity (cSt) at 100 1C

PMMA-1 40,311 5.98 6.68 7.54 8.47

PMMA-2 60,795 5.98 6.68 7.54 8.60

PMMA-3 155,259 5.98 7.04 8.34 9.94

PMMA-4 530,537 5.98 7.15 8.67 10.50

a Molecular weight.b Concentration of VI improver.

L.V. Markova et al. / Tribology International 44 (2011) 963–970968

4.3. Changes in oil viscosity during artificial aging

Oil samples were prepared by artificial aging. The gear oil GSCaltex ‘‘Meropa 220’’ was oxidized at a temperature of 148 1C. Theoxidation process was stopped for the nighttime. In the course ofoil oxidation, the samples were taken after 0, 8, 26 and 35 h.

Viscosity was measured using magnetoelastic, solid-state andglass capillary viscometers at 40 1C (Table 2).

The comparison of test results proves that the solid-stateviscometer shows higher viscosity values than those measuredby the magnetoelastic and capillary viscometers. This can beattributed to variations in chemical structure during artificialaging and increased polarity of molecules, which changes inter-facial interactions between the oil and sensitive surface of thesolid-state viscometer. Since the viscometer is sensitive to varia-tions in properties within a very thin oil layer (the order of amicron), these interactions at the interface can affect the readings.

4.4. Effect of PMMA viscosity-index improver on viscometer readings

To study the effect of PMMA VI improver on viscometerreadings we have prepared the samples of synthetic base oilPAO 6 with polymethyl methacrylate (PMMA) VI improvers inconcentrations 3, 6 and 9 wt%. Four different modifications ofPMMA (PMMA-1, PMMA-2, PMMA-3 and PMMA-4) with differentmolecular weights were under testing to estimate the influenceof molecular weights of the VI improver on viscometer data(Table 3). The molecular weights of the VI improvers weremeasured using GPC (Gel Permeation Chromatography, HLC-8320GPC) at 40 1C temperature.

The improver molecular weight effect on kinematic viscosity at40 and 100 1C was estimated by Capillary Viscometer CannonFenske (ASTM D445) (Table 3). As it was expected, kinematicviscosity increased at 100 1C with increase in molecular weightand content of the VI improver. Meanwhile, at 40 1C viscosityincreased with growing concentration of the VI additive but without

any correlation between the increment in molecular weight andviscosity. Obviously, the VI improvers have different temperaturebehaviors, which is evident at lower temperature (40 1C), while theirapplication is intended for higher temperature (100 1C).

Viscosity of the test samples was also measured by themagnetoelastic and solid-state viscometers at 40 1C. The absoluteviscosity was determined from the measured acoustic viscosityvalues using Eq. (7).

The analysis of measurement results of the absolute viscosity ofthe test oils with addition of PMMA-1 and PMMA-2 in concentra-tions 3, 6 and 9 wt% shows that the data determined using thecapillary viscometer coincide with the readings measured bythe magnetoelastic viscometer (Fig. 5a and b). This fact proves thatthe test oil samples preserve the properties of the Newtonian fluid atthe operating frequency o30 kHz (at shear rateo103 s�1). Viscos-ity values of these oils measured by the solid-state viscometer aremuch lower than that obtained by the capillary and magnetoelasticviscometers. This confirms that the oils behave as non-Newtonianfluids at the operating frequency (5.3 MHz) of the solid-stateviscometer (at shear rate about 105 s�1). It means that the oil PAO6 with polymer additives has a pseudoplastic behavior characterizedby decreasing viscosity with increasing shear rate.

As the molecular weight of the additive increases (introduction ofPMMA-3) the non-Newtonian behavior of the test oils is observedalready at the operating frequency of the magnetoelastic viscometer(Fig. 5c). Further increase in the molecular weight of the additive(PMMA-4 VI improver) results in a situation when the apparentviscosities of the oils coincide at the frequencies 30 kHz and 5.3 MHzor are close to the base oil values without additives (Fig. 5d).

5. Conclusions

Both piezoelectric and magnetoelastic types of the on-lineacoustic viscometers have proved to be promising in the real-timemonitoring of lubricating oils.

PMMA-1

0

2

4

6

8

10

12

0.00C , wt%

ΔηΔη, c

PΔηΔη

, cP

ΔηΔη, c

PΔηΔη

, cP

PMMA-2

0

2

4

6

8

10

12

PMMA-3

0

2

4

6

8

10

12PMMA-4

0

2

4

6

8

10

12

Capillary Magnetoelastic Solid-State

3.00 6.00 9.00 0.00C , wt%

3.00 6.00 9.00

0.00C , wt%

3.00 6.00 9.00 0.00C , wt%

3.00 6.00 9.00

Fig. 5. Changes in absolute viscosity with increase in VI improver concentration at increasing molecular weights of the VI improvers measured by capillary, magnetoelastic

and microacoustic viscometers.

L.V. Markova et al. / Tribology International 44 (2011) 963–970 969

The analysis of correlations between the readings given by themagnetoelastic, capillary and solid-state acoustic viscometers forthe tested base mineral and synthetic oils has indicated that thedeveloped viscometer shows the data close to the capillarymethod within the range 17–500 cSt. The solid-state acousticviscometer gives the least validity in measuring viscosity above300 cSt. Its readings exceed considerably the anticipated data,evidently because the oil layer on its sensitive surface behaveslike a gel at high operating frequency of the viscometer.

Monitoring of artificially aged mineral oil has revealed that thereadings of the solid-state viscometer are much higher than theviscosity measured by the magnetoelastic viscometer. This canresult from the changes in chemical structure of the oil duringartificial aging, and intensified chemical activity (increased polar-ity of its molecules) leading to the change in the interfacialinteractions between the oil and sensitive surface of the solid-state viscometer. Since the detector is sensitive to the propertiesin a very thin layer of the oil (a few microns thick), evidently,these interfacial interactions affect the readings.

Investigations of the effect of PMMA VI improvers on theviscometer readings have indicated that with increase in mole-cular weight of PMMA under the same concentration in thesynthetic base, the oil starts to behave like the non-Newtonianfluid at lower frequencies of viscosity measurements. As themolecular weight of the additive increases till Mw¼530,537 Da,the apparent oil viscosities measured at frequencies 30 kHz and5.3 MHz coincide and approach the viscosity values of the base oilwithout additives. The results obtained agree well with theconclusions presented by Stachowiak and Batchelor [17]. Theyhave proved that mineral oils behaving as Newtonian fluids underlow velocities start to behave like non-Newtonian fluids underthe shear rates above 105–106 s�1 frequently occurring in engi-neering applications.

Evidently, to make a correct choice of the lubricating oil withaccount of predicted rheological behavior in a tribosystem and itsmonitoring during operation, it is expedient to measure viscosity

at the shear rates close to those used in the real tribosystems. Inparticular, it is important to estimate viscosity of non-Newtonianengine oils at medium shear rates (103–104 s�1) at low tempera-ture, and high shear rates (105–107 s�1) at high temperature tomake the best correlations with the engine performance [18].

Acknowledgments

This work was supported partly by Korea Institute of Scienceand Technology (Project no. 2V01500).

References

[1] Serrato R, Maru MM, Padove LR. Effect of lubricant viscosity grade on mechanicalvibration of roller bearings. Tribology International 2007;40(8):1270–5.

[2] Truhan JJ, Jun Qu, Blau PJ. The effect of lubricating oil condition on the frictionand wear of piston ring and cylinder liner materials in a reciprocating benchtest. Wear 2005;259(7–12):1048–55.

[3] Ciantar C, Hadfield M, Smith AM. Swallow A. The influence of lubricantviscosity on the wear of hermetic compressor components in HFC-134aenvironments. Wear 1999;236(1–2):1–8.

[4] Jacobson BO. Rheology and elastohydrodynamic lubrication. In: Tribologyseries, vol. 19, New York: Elsevier; 1991.

[5] Gaglione R, Attane P, Soucemarianadin A. A new high frequency rheometerbased on torsional waveguide technique. Review of Scientific Instruments1993;64:2326–33.

[6] Wright HA. Viscosity measurement by means of damped resonant vibrationnormal to an approximate rigid plate. Acoustical Society of America Journal2004;115(6): p. 2694.

[7] Watson J. Real time in-process viscosity measurement. Hydrocarbon Engi-neering 2000.

[8] Durdag K. Solid state acoustic wave sensors for real-time in-line measure-ment of oil viscosity. Sensor Review 2008;28(1):68–73.

[9] Lee RM, Zhang XJ, Hammond JM. A remote acoustic engine oil quality sensor.In: Proceedings of the IEEE Ultrasonics Symposium. Toronto, Canada; 1997, p.419–22.

[10] Agoston A, Otsch C, Jakoby B. Viscosity sensors for engine oil conditionmonitoring—application and interpretation of results. Sensors and Actuators2005;121:327–32.

[11] Landau LD, Lifshitz EM. Fluid mechanics. 2nd ed.Course of theoreticalphysics, 6. . Butterworth-Heinemann; 1987.

L.V. Markova et al. / Tribology International 44 (2011) 963–970970

[12] Stoyanov P, Grimes C. A remote query magnetostrictive viscosity sensor.Sensors and Actuators 2000;80:8–14.

[13] Jain M, Schmidt S, Grimes C. Magneto-acoustic sensors for measurement ofliquid temperature, viscosity and density. Applied Acoustic 2001;62:1001–11.

[14] Jiles DC. Introduction to magnetism and magnetic materials. 1st ed. London:Chapman and Hall; 1991.

[15] Mason WP. Physical acoustics. Principles and methods, Vol. 1. New York:Academic Press; 1964. part A.

[16] Grimes C, Mungle C, Zeng K, et al. Wireless magnetoelastic resonancesensors: a critical review. Sensors 2002;2:294–313.

[17] Stachowiak GW, Batchelor AW. Engineering tribology. 3rd ed. Butterworth-Heinemann; 2005.

[18] Alexander DL. Change of high-shear rate viscosity of engine oils during use: areview. In: Spearot JA, editor. High-temperature, high-shear (HTHS) oil viscosity:measurement and relationship to engine operation, ASTM STP 1068. Philadel-phia: American Society for Testing and Materials; 1989. p. 60–73.