MESO-SCALE OIL CONDITION SENSOR
Bradley Dale McClelland
BSME, University of Colorado Denver, 2009
A thesis submitted to the
University of Colorado Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
2011 by Bradley Dale McClelland
All rights reserved.
This thesis for the Master of Science
Bradley Dale McClelland
has been approved
McClelland, Bradley D. (M S., Mechanical Engineering)
Meso-Scale Oil Condition Sensor
Thesis directed by Associate Professor Ronald A.L. Rorrer
With advances in vehicle electronics and instrumentation, there has become an
increasing demand for sensors to monitor the various systems of vehicles. One
fruitful area for exploration is oil condition monitoring. There are a variety of devices
that have been researched and some commercial sensors based upon indirect
measures of oil condition based upon temperature history and mileage. In this work, a
meso-scale oscillatory device has been built to monitor the condition of oil while in
use. Specifically, the device measures the viscosity, dielectric constant, and
temperature of the oil. These measurements were chosen based on their importance to
the function of oil. This is especially true of viscosity, which is known to be a major
indicator of the overall condition of the lubricant.
Tests showing the operation of the device, including measurements of oil samples
from actual engines, are shown. Other tests include the detection of fuel
contamination and metal wear particle presence. The device presented here could be
easily incorporated into the onboard electronics of a modern vehicle in order to
monitor the condition of the oil and determine when oil replacement is necessary.
This abstract accurately represents the content of the candidates thesis. I recommend
Ronald A.L. Rorrer
I dedicate this thesis to all of my family and friends who supported me throughout the
I wish to express my deepest thanks to my advisor, Ronald A. L. Rorrer, for his
guidance and support to my research. I also wish to thank Luis R. Sanchez and
Randall P. Tagg for their participation on my committee and their contributions to my
research. I would also like to thank Mike Dean from Epilog Laser for allowing us
access to one of Epilog Lasers cutting devices.
TABLE OF CONTENTS
1. Introduction........................................................... 1
1.1 Purpose of the Study................................................... 1
1.2 Introduction to Lubrication and Oils...................................2
1.2.1 Hydrodynamic Lubrication............................................4
1.2.2 Boundary Lubrication................................................6
2. Review of the Literature...............................................7
2.1 Presence of Outside Contaminants.......................................7
2.1.1 Contamination Detection Through Oil Property Measurement.............8
2.1.2 Contamination Detection Through Electrical Measurement...............9
2.1.3 Contamination Detection Through Optical Techniques...................10
2.2 Presence of Wear Particles............................................ 12
2.2.1 Capacitance Based Measurement........................................ 13
2.2.2 Inductance Based Measurement
2.2.3 Permittivity Based Measurement........................................14
2.3 Presence of Additives in the Oil.....................................14
2.4 Physiochemical Properties of Oil.......................................16
2.4.1 Detection of Oil Viscosity and Density................................16
2.5 Relevant Patents.......................................................22
2.6 Existing Commercial Devices............................................23
3.1.1 Viscosity Sensor Design...............................................29
3.1.2 Contamination Detection Sensor........................................37
3.1.3 Integrated Sensor Design..............................................40
3.2 Modeling.............................................................. 42
3.2.1 SolidWorks Model.....................................................42
3.2.2 Two Degree of Freedom Model.......................................... 46
126.96.36.199 Mass............................................................... 48
188.8.131.52 Spring Constant.................................................... 49
184.108.40.206 Damping............................................................ 50
220.127.116.11 Force Applied...................................................... 51
18.104.22.168 Drive Frequency and Time........................................... 51
22.214.171.124 Simulating with the 2 DOF model.................................... 52
3.3 Proof of Concept........................................................56
3.3.1 Contamination Detection Proof of Concept Testing....................56
3.3.2 Viscosity Sensor Proof of Concept Testing...........................61
3 .4 Device Construction and Validation...................................76
3.5 Device Testing........................................................86
3.5.1 Fuel Contamination..................................................86
3.5.2 Coolant Contamination...............................................90
3.5.3 Used Oil Testing....................................................93
4.2 Conclusions......................................................... 95
LIST OF FIGURES
1.1 Stribeck curve........................................................3
1.2 Simple plain bearing..................................................5
1.3 Pressure wedge formed by oil in hydrodynamic lubrication..............5
2.1 TSM resonator.......................................................... 18
3.1 Micro-Scale Oil Condition Sensor Design.................................30
3.2 MEMS Resonator Model................................................... 31
3.3 Device Layout, from top left to bottom right: copper electrode, bond film,
bottom copper resonator layer, kapton layer, top copper resonator layer
3.4 Acoustic Diaphragm Driven by Piezoelectric Film.........................33
3.5 Viscosity sensor design incorporating a piezoelectric diaphragm
oscillator with a capacitance based position sensor.................. 34
3.6 Viscosity sensor design incorporating a piezoelectric diaphragm
oscillator bonded to a piezoelectric diaphragm sensor to
detect the displacement of the device................................ 35
3.7 Final design for viscosity sensor constructed from two
3.8 Contamination Sensing Using LEDs........................................38
3.9 Parallel Plate Capacitance Model........................................39
3.10 Initial oil condition sensor design.....................................40
3.11 Revised oil condition sensor design......................................41
3.12 Piezoelectric diaphragm model in SolidWorks............................. 42
3.13 Piezoelectric diaphragm mounting........................................ 43
3.14 (a) Simple support (b) Fixed support.................................... 43
3.15 Piezoelectric diaphragm model with straight wire leads.................. 45
3.16 Piezoelectric diaphragm model with curved wire leads.................... 45
3.17 2 DOF system model...................................................... 47
3.18 Free body diagram for 2 degree of freedom model......................... 48
3.19 Capacitance proof of concept test device.................................56
3.20 Variable gap capacitance device..........................................59
3.21 Results of variable capacitance test in air..............................60
3.22 Results of variable capacitance test in water............................60
3.23 Results of variable capacitance test in oil..............................61
3.24 RadioShack piezoelectric buzzer..........................................62
3.25 Original viscosity sensor proof of concept test device...................63
3.26 Measured amplitudes of sensor diaphragm for a range of frequencies......64
3.27 Comparison of the driven diaphragm frequency to the sensor
3.28 Results of multi-grade oil sensor test...................................66
3.29 Results of single-grade oil sensor test..................................67
3.30 Viscosity sensor proof of concept device version 2.......................68
3.31 Results of single-grade oil proof of concept test using
2nd version test device................................................69
3.32 Results of multi-grade oil proof of concept test using
2nd version test device................................................70
3.33 3rd version of the viscosity proof of concept device.....................72
3.34 Results of initial testing with the 3rd version
of the proof of concept device.........................................73
3.35 Results of test using modified 3rd version
of the proof of concept device.........................................75
3.36 Oil condition sensor test device design................................. 76
3.37 Epilog laser cutting machine.............................................77
3.38 Test device, version 1 ..................................................77
3.39 Results of Test device version 1 silicone oil test...................... 78
3.40 Piezoelectric diaphragm fixture with (b) and without (a) fluid passages .... 79
3 .41 Test device, version 2...................................................79
3.42 Results of test device version 2 silicone oil test.......................80
3.43 Comparison of viscosity measurement for 3 silicone oils..................81
3.44 Results of viscosity test on three single-grade SAE oils.................82
3.45 Comparison of viscosity measurement for three lubricating oils...........83
3.46 Comparison of viscosity measurement for SAE lubricating oils
and silicone reference oils.............................................84
3.47 Results of heated oil viscosity test.....................................85
3 .48 Results of fuel contamination testing on SAE 30 lubricating oil..........87
3.49 Effect of fuel contamination on viscosity of SAE 30 lubricating oil.....88
3.50 Results of small concentration fuel dilution test on
SAE 30 lubricating oil.................................................89
3.51 Effect of fuel contamination on viscosity of SAE 30 lubricating oil.....90
3.52 Oil-coolant emulsion....................................................91
3.53 Results of coolant contamination test...................................92
3.54 Results of used oil viscosity test......................................94
LIST OF TABLES
2.1 Measured oil condition properties in terms of importance...............7
3.1 CoventorWare simulation results for 5 V input.........................32
3.2 Comparison of simple and fixed support methods....................... 44
3.3 Comparison of piezoelectric diaphragms with straight
and curved wire leads................................................ 46
3.4 Results of 2 DOF model for both damping models....................... 52
3.5 Comparison of the undamped natural frequencies
from the models and experiment..................................... 53
3.6 Comparison of the damped natural frequencies
from the models and experiment....................................... 54
3.7 Comparison of the relative displacement from models
and experimental data................................................ 55
3.2 Capacitance between plates for varying gap distances..................58
3.3 Comparison of measured amplitude of sensor diaphragm
for two oils of different viscosities..................................65
3.4 Results of capacitance measurement under influence of oscillation test.71
3.5 Capacitance of samples during coolant contamination test..............93
The level of technology in modern automotive engines has progressed at a rapid rate
over the past 40 plus years. Innovations in power generation and fuel consumption
have left the power plants of todays motor vehicles more efficient than ever before.
However, there are areas in the power plants of modern vehicles in which little
development has occurred in the past half century. One such area is the monitoring of
oil condition. Other than oil pressure sensors, oil level sensors, and the dipstick, there
are relatively few ways to gauge the condition of the oil used in modern internal
combustion engines. Rather than assess the condition of the oil in an engine,
manufacturers of both vehicles and lubricating oils recommend changing oils in
engines on regular, time based intervals.
While this strategy of oil replacement avoids many wear issues in most
circumstances, it also requires that the oil be changed before its lubricating properties
have been reduced to unsatisfactory levels. In a time when the worlds supply of oil
becomes of increasing importance, it is important not to waste large quantities of this
1.1 Purpose of the Study
The purpose of this study was to design and create a sensor capable of determining
the condition of lubricating oils in internal combustion engines through continuous,
real-time property measurement. To achieve this purpose, it was necessary to first
identify what properties of lubricating oils were most directly related to the condition
of the oil. Fluids such as lubricating oils are defined by numerous properties, so it
was cmcial to determine what properties are most indicative of the degradation of the
oil. After determining the most crucial properties relative to oil condition,
measurement techniques for each property were evaluated. During this process,
common measurement techniques were evaluated to determine their effectiveness and
complexity. After evaluating these measurement techniques, it was decided a new
sensor design would best address the issues at hand.
After determining the properties to be measured, the measurement device was
1.2 Introduction to Lubrication and Oils
In any mechanical device in which moving components make contact friction is
generated. This friction is the cause of wear in any mechanical system. If the friction
between the parts in motion can be reduced, the lifetime of the device can be
increased. Friction can be reduced in a number of ways, but the most common
method is the use of some lubricant between the parts.
Lubricants can be grouped into two major categories, solid lubricants and liquid
lubricants. The most common form of solid lubricants are greases and these are often
used in low speed rotation or sliding applications. Liquid lubrication is the more
prevalent form of lubrication and is most commonly found in the form of oils. These
oils are used more prevalently due to their higher operating temperatures, cooling
properties and their superior lubricating properties.
The behavior of liquid lubricants can be broken into three different parts, boundary
lubrication, hydrodynamic lubrication, and mixed lubrication. Each type of
lubrication has a different effect on the friction generated by the interaction between
the rotating parts. The type of lubrication in a mechanical device at any given time is
dependent on a number of variables, including the speed of the rotation, the load on
the rotating shaft, and the viscosity of the lubricating oil. The relationship between
these variables and the resulting friction is illustrated by the Stribeck curve , shown
in Figure 1.1.
The variable represented on the x-axis of the Stribeck curve shown in Figure 1.1 is
pN/P = oil viscosity x shaft speed / bearing pressure (1.1)
The three zones in the curve represent the three types of lubrication. Zone 1 is
boundary lubrication, zone 2 is mixed lubrication, and zone 3 is hydrodynamic
lubrication. As Figure 1.1 shows, the friction decreases as the conditions increase the
pN/P coefficient. The friction reaches its lowest at the point where mixed lubrication
and boundary lubrication meet, shown as point B in Figure 1.1.
As pN/P increases from point A to point B, the film thickness of the oil between the
surfaces increases. That is to say, the thickness of the oil between the surfaces
decreases from point B to point A. At point A, there is no oil between the surfaces to
reduce the friction produced. As the lubrication gets closer to point B, the oil
thickness increases until the film thickness reaches the point in which the film of oil
is the exact thickness to prevent contact between the surfaces but no thicker. This is
the point at which the friction is lowest (point B). After this point, the film thickness
continues to increase and the friction increases.
When designing mechanical devices, the goal of the engineers designing the device is
to reduce friction while ensuring the device does not wear excessively. In order to
accomplish these goals, typically devices are designed to operate just to the right of
point B on the Stribeck curve. This allows the designers to take advantage of the low
friction in this region as well as ensure there is always a film of oil between the
surfaces to prevent contact, and increased wear, between the parts. This region in
which the designers work is known as the hydrodynamic lubrication region.
1.2.1 Hydrodynamic Lubrication
Hydrodynamic lubrication is lubrication between two surfaces due to movement of a
liquid between the surfaces. In the case of a simple plain bearing, such as the one
shown in Figure 1.2, the oil is moved into the area between the surfaces due to the
movement of the shaft.
Figure 1.2. Simple plain bearing
Due to the load applied to the simple plain bearing, the film formed between the
surfaces is shaped as a wedge. As the oil moves through the wedge between the
surfaces, the pressure increases. This pressure exerted by the oil on the surfaces
supports the load between the surfaces and separates the two surfaces. This wedge
formed between the surfaces is illustrated in Figure 1.3.
Figure 1.3. Pressure wedge formed by oil in hydrodynamic lubrication
The hydrodynamic lubrication in any system is dependent upon only one property of
the oil, the viscosity. This is why viscosity is such an important property of oil.
1.2.2 Boundary Lubrication
Boundary lubrication in a mechanical system is defined as the point in which surface
imperfections from each surface begin to contact each other. This occurs due to the
fact that the oil film thickness between the surfaces becomes too small to prevent
contact between the surfaces. In this type of lubrication, many properties other than
the viscosity of the oil come into play. This type of lubrication typically results in
higher wear of the surfaces due to the contact between the surfaces.
2. Review of the Literature
The scope of existing research in this field is extensive. Many different oil properties
and methods of measurement are suggested and demonstrated in the current literature.
In order to organize the presentation of these different devices, each of the different
oil properties measured will be presented, followed by the corresponding device(s).
Many of the devices developed measure properties in multiple categories. These
devices will be discussed in the category the author of this work deems to be the
largest contributor to the sensor. Table 1 shows the properties measured to determine
oil condition in order of their importance to the determination of oil condition.
Table 2.1. Measured oil condition properties in terms of importance
1 Physiochemical Properties
2 Presence of Wear Particles
3 Presence of Outside Contaminants
4 Presence of Additives
2.1 Presence of Outside Contaminants
The presence of outside contaminants is the first measurement category to be
considered. This consists of the detection of any substance in the oil that did not
originate in the oil. This could include contaminants such as glycol, water, fuel, and
solid contaminants. These contaminants are detected due to the fact that they have a
negative effect on the lubrication properties of the oil. More specifically, such
contaminants lower the viscosity of the oil as well as corrode engine internals. Such
contaminants are measured in a variety of ways.
2.1.1 Contaminant Detection Through Oil Property
One of the ways outside contaminants are detected is through viscosity and/or density
measurement. Contaminants such as water or fuel will alter the viscosity and density
of the lubricating oil. This change in viscosity or density can be detected and
correlated to the concentration of contaminants. One example of such a device is
found in the work of Sparks and his colleagues , In this work, a MEMS-based
resonant microtube was created and used in conjunction with a temperature sensor to
determine the ratio of fuel to water in a fuel cell based on the density of the fluid.
This device was not tested in lubricating oils, but it is believed that such a device
could be adapted to perform in these conditions.
Another example of this method of outside contaminant detection is found in the oil
sensor developed by Bernhard Jakoby and his research team for the Robert Bosch
Corporation and Mercedes-Benz , This device detects the viscosity of oil by
creating surface acoustic waves and detecting the changes in the damping of the fluid
and the frequency of the device. A surface acoustic wave (SAW) device was used to
create the waves. This device is a small piezoelectric device that oscillates in the
MHz range. This sensor effectively detected the presence of outside contaminants.
However, it was found that these types of contaminants, mainly oil and water, are not
a major contributor to the breakdown of the viscosity of lubricating oils. This device
also includes other sensing abilities and will be discussed in greater detail in the
Physiochemical Properties of Oil section.
2.1.2 Contaminant Detection Through Electrical
Contaminants such as water or fuel can also be detected by way of electrical
measurement. This method involves applying an electrical charge to the fluid
between two small plates and measuring the resulting electrical properties of the
fluid. One example of this method is found in the work by Surapol Raadnui and
Srawut Kleesuwan , In this paper, the researchers measure the capacitance
between two plates submerged in oil. This capacitance could then be used to
calculate the dielectric constant of the fluid. Since outside contaminants such as
water, fuel, and silicon change the dielectric constant of the oil, a correlation between
contaminant concentration and dielectric constant was made. This method proved to
be successful for the researchers; however, this sensor is still at a very early stage of
Another method of electrical measurement of oil contamination is electrorheology, as
demonstrated by Xiuyu Wang and Vladimir Alvarado , This method applies a
high voltage, up to 4000 V, between two plates spaced about 1 mm apart. The
current is then measured while dynamic shear is applied to the fluid. From this
measurement, the viscosity is obtained, which can then be correlated to the amount of
water in the oil. This method has drawbacks, however. First, the authors point out
that measuring the viscosity of the oil alone is not enough to accurately gauge the
condition of the oil. Other parameters must be measured. Another limiting factor to
this design is the large voltage input that is required. The generation of these voltage
levels is difficult, if not impossible, in most settings outside the laboratory or
industry. Certainly such voltage levels are impractical for use in an automobile.
The next form of electrical detection of contaminants is the use of a diamond-like
carbon electrode as used in , In this work, authors Gun-Ho Noh, Adela Bordeanu,
Ju-kyung Lee and Jae-Chul Pyun use a diamond-like carbon electrode to measure the
oil-to-water concentration in various brake fluids. This device applied a step
potential to the fluid and measured the time until the potential decayed to a certain
level. This current decay time was then correlated to the viscosity which was in turn
used to determine the amount of water in the oil. The researchers report very accurate
and repeatable results from this work. However, it is unknown how well this
technology would transfer to the measurement of lubricating oils.
Another sensing capability of the sensor designed by the Bosch Corporation , 
is the permittivity of the oil. The permittivity of a fluid is related to the effects of the
fluid on the electric field applied to it. For this sensor, the permittivity is measured by
applying AC potential between two concentric tubes. The permittivity of the oil
increases with the amount of contamination. This technology, in conjunction with the
other sensing abilities of this device, has been tested in automobiles by Mercedes-
Benz. The release of this sensor for mass production was anticipated in the mid to
late 2000s, but no evidence of its use on production vehicles exists.
2.1.3 Contaminant Detection Through Optical
Outside contaminants can also be detected through a variety of optical techniques.
One such technique is outlined by A.J. Scott and his research team in , In this
work, the presence and concentration of outside contaminants is determined by
measuring the turbidity of the oil. This process involves directing a light source at a
sample of the oil and measuring the amount of light that is transmitted through the oil.
Higher amounts of transmitted light correspond to lower contamination levels of the
oil. Currently it is unknown how well this method correlates to the actual decrease in
the condition of the oil. It is thought that there is some correlation, but this method
seems limited when compared to other measurement techniques. However, this
device has been designed and tested on automobiles. This is significant, as very few
devices have reached this level of development.
Another example of contaminant detection using optical techniques is found in ,
In this work, Yonghui Yin, Weihua Wang, Xinpin Yan, Hanliang Xiao, and Chengtao
Wang use an optical sensor in series with an inductive sensor, which will be
discussed later. The optical sensor measures particles in the oil by detecting the
amount of light that is transmitted through the oil, much like the sensor created by
A.J. Scott , However, for this device, fiber optics was employed to deliver and
receive the light from two light sources and detectors to the sensing area. This sensor
was designed to detect the small particles the inductive sensor could not. In
combination, the two devices were combined with the knowledge that no single
sensor could accurately determine the condition of lubricating oils. The device was
also designed to be used in-line, meaning the sensor gathers data as the machine is in
use. Obviously this design would lend itself to use in an automobile. However, this
sensor does have drawbacks. One of the most significant is the inability of the device
to measure the physiochemical properties of the oil. This is a major drawback, since
it is widely held that these properties play a significant role in the performance of
Much like the two previous sensors, the sensor designed by Saurabh Kumar and P S.
Mukherjee in  uses the same optical measurement techniques. A device was
created that continuously measured the condition of the oil. This device was designed
to be retrofitted to an automobile. Testing of the device was conducted on a generator
engine. A light source transmits light through a thin film of oil and the resulting light
transmission is measured. In this work, this light transmission is related to viscosity,
which was measured using standard laboratory tests during the usage of the oil. The
authors show that a drop in light transmitted is proportional to a decrease in
viscosity. In fact, the authors show that the decrease in transmitted light as a function
of time exhibits a linear behavior. This contrasts with the non-linear behavior of the
viscosity as a function of time. A correlation between the two could be made, but it
appears to be a weak one. There are too many other variables that influence the
viscosity to base such a correlation simply off the turbidity of the oil.
Another type of contaminant detection using optical techniques is spectrometry. This
is the focus of the work of V. Macia'n, B. Tormos, P. Olmeda, and L. Montoro ,
This method also uses a light source to transmit light through the oil sample. A
detector senses the transmitted light and identifies the wavelengths the light is
composed of. These wavelengths correspond to different compounds in the oil.
Analysis of these wavelengths can determine what particles are present in the oil.
This has the obvious benefit of knowing exactly what is contaminating the lubricating
oils in question. However, this method has some drawbacks. First, it is pointed out
in  that particles larger than 5 pm are not detected by spectrometry. Since the
paper also points out that harmful particles are > 1 pm, this method is only useful
over a very small range of particle sizes. Even though the author points out that
particles > 10 pm are typically captured by the oil filter, half the theoretical range
proposed (1 pm to 10 pm) is not detected by the device.
2.2 Presence of Wear Particles
The presence of wear particles in lubricating oils is the next measurement category to
be discussed. This category consists of debris in the oil that is a result of component
wear inside the engine. This category deals mainly with ferrous and non-ferrous
metallic particles in the lubricating oils. Many detection methods have been
suggested for this type of particle detection. The following is a brief summary of
these detection methods
2.2.1 Capacitance Based Measurements
The first wear particle detection method is capacitance based sensing. Capacitance is
the measure of the electric potential between two parallel plates. This electric
potential is influenced by the medium in the gap between the two plates. When
metallic particles pass between the plates, the permittivity of the medium separating
the two plates is altered, causing an increase in the capacitance. One device that
operates using these principles was created by Srinidhi Murali, Xingao Xia, Ashish V
Jagtiani, Joan Carletta, and Jiang Zhe , In this device, capacitance is used to
detect ferrous metallic particles in lubricating oils. The device was integrated into a
microfluidic device that located the metal plates 20 pm from one another. The final
device was able to detect particles as small as 10 pm. However, the device was
limited to detecting particles up to 25 pm due to the size of the channels in which the
device was located. Another drawback to this device is the inability to detect non-
ferrous particles. Since modern engines contain non-ferrous metals as well as ferrous
metals, a device that detects both is needed. Further study into this topic is currently
being explored by this group.
A similar device was created by Surapol Raadnui, and Srawut Kleesuwan , As
was discussed in the previous section, the device measured the capacitance between
two parallel plates. This capacitance is then used to calculate the dielectric constant
of the oil flowing between the two plates. Changes in the dielectric constant indicate
the presence of wear particles. As was noted before, the research is still in an early
stage and more work is to be conducted in order to further develop the device.
2.2.2 Inductance Based Measurements
Inductive based sensing of metallic particles in oil is based on the properties of
inductance in a circuit. The principle is that as a particle passes through a coil of
wire, a potential will be created. Yonghui Yin  applied this principle to the
detection of wear particles. As particles pass through the coil, a charge is created.
This method of particle detection has one main advantage over capacitance based
sensing, it can detect both ferrous and non-ferrous particles. However, the method
has its drawbacks. As designed by Yin , this device could only detect particles >
500 pm. This is not nearly small enough, as other sources have claimed engines can
be damaged by particles > 1 pm ,
2.2.3 Permittivity Based Measurements
The permittivity of a fluid is related to the effects of the fluid on the electric field
applied to it. The Bosch Corporation has applied this principle to the detection of
wear particles in oil , , According to Bosch, this measurement was designed to
find the contamination particles discussed earlier. However, since this device senses
changes in the electrical properties of oil, it is assumed that the device would also
sense metallic particles.
2.3 Presence of Additives in the Oil
The breakdown of additives in lubricating oils is one contributing factor to the
breakdown of the oil as a whole. Additives in lubricating oils are meant to do things
such as reduce thermal breakdown, increase the viscosity of the oil at high
temperatures and shear forces, and prevent carbon deposits from forming. Measuring
the actual amount of additives in the oil is not feasible; however, the breakdown of
such additives can be detected. One device that observes these breakdowns is the
corrosion monitor created by Attila Agoston, Edda Svasek, and Bernhard Jakoby
, In this research, a thin metal film was used as a resistive sensor to detect the
corrosiveness of lubricating oils. A rise in the pH and sulfur levels of the lubricating
oil indicates the breakdown of additives in the oil. The thin metal film was made of
copper, which corrodes in an acidic environment. The loss of particles from the thin
film due to corrosion was tracked and correlated to the level of corrosiveness of the
oil. This method accurately tracked the increasing corrosiveness of the lubricating
oils, but, its use in automobile oil sensing might be limited. One of the major limiting
factors of this device is its disposable nature. The thin metal films used to sense the
corrosiveness are consumed throughout the measurement process. This would
necessitate a new sensor every time the oil in the engine was changed. This fact is
acknowledged by the authors, as they state the device is designed for large stationary
engines, not automobiles.
Another device that measures the deterioration of additives in lubricating oils was
created by Simon S. Wang , This device measures oil condition with a Pt sensing
electrode. Basically, two metal plates are placed closely to one another and the
change in voltage over time is measured. This is correlated to the depletion of oil
additives as well as an increase in viscosity. This correlation is based on laboratory
measurements of the oil properties. The device itself does not measure these
properties; the authors simply correlate changes in voltage in their device to these
property changes. Due to variances in oil propertie: between
and the effects of different driving habits on the oil, this does not seem to be a very
czr zi1 ^e^redetion. However, the idea of detecting the
breakdown of additives in the oil is a sound one, and one that deserves further
2.4 Physiochemical Properties of Oil
The final category of oil condition measurement is the measurement of the
physiochemical properties of the oil. This is a very broad category, encompassing the
measure of many different properties of the lubricating oil. Some of the properties
measured include viscosity, flash point, solid point, contents of resins, asphaltenes,
carbenes, carboids, and insoluble sludge. It is not possible or necessary to measure
all of these properties of oil in an onboard measurement device, but some properties
lend themselves well to such tests and provide information about the quality of
lubricating oils that cannot be obtained in any other fashion. A summary of
physiochemical measurement techniques currently employed by researchers follows.
2.4.1 Detection of Oil Viscosity and Density
The measurement of oil viscosity is widely held as a critical measurement of the
condition of lubricating oils. While not an exclusive indicator of oil condition, the
viscosity is certainly a major factor in the condition of lubricating oils. As a result,
many researchers have proposed and developed viscosity measuring devices for use
in on-board diagnostics on automobiles. A brief description of many of these devices
can be found in the remainder of this section.
The first viscosity measurement device to be discussed is found in the work by J.M.
Hammond, R.M. Lee, D.G. Libby, X.J. Zhang, and L A. Prager , This device
uses a shear wave piezoelectric resonator to measure the viscosity of the oil. The
device works by oscillating the piezoelectric to its resonant frequency and measuring
the change in resonant frequency, amplitude, and phase or impedance/admittance.
The changes in these properties can be correlated to the changes in the viscosity of
the oil. The authors show that changes in viscosity occur, but mainly due so between
differing oil types. Only a very small portion of the article was devoted to showing
the changes in the electrical properties of the resonator due to oil wear. Another issue
that arises from this paper is the lack of correlation to macro-scale viscosity testing.
At no point in the article is an attempt made to correlate the electrical properties
measured with an actual viscosity. This makes it impossible for the reader to
determine the accuracy of the device.
The next viscosity sensor lo be discussed Tias been described and tested in a few
different articles, all written in part by Ti. Jakoby , , , , All of these
papers discuss creating a micro-acoustic viscosity sensor. The micro-acoustic device
realized used a thickness shear mode (TSM) resonator. This type of resonator consists
of a thin quartz disk with electrodes on either side The TSM resonator is shown in
Figure 2.1. Changes in viscosity result in changes in both the inductance of the
resonator as well as the 6 MHz resonant frequency. During the testing of the device,
it was discovered that the output of the sensor did not change when changes in
viscosity due to water emulsion occurred Further investigation lead to the discovery
that due tp the small size of the resonator and high resonant frequency, only the
viscosity of the base oil was measured. Because the dimensions of the sensor are not
significantly larger than those of the water emulsions, the water emulsions are not
part of the resulting thin-layer .that is analyzed .Therefore, the sensor does not
correlate with macroscopic viscosity measurements. After coming to these
conclusions, the authors suggest other methods of viscosity measurement that avoid
very small dimensions,-on the order of micrometers, and very high frequencies, in the
Figure 2.1. TSM resonator
In light of these issues discovered with micro-acoustic viscosity sensors, some of the
authors from the previous three works created a device aimed at overcoming these
issues , It was established from the start that the device had to measure at a lower
frequency with larger oscillations. In order to accomplish this, the researchers created
a cantilever sensor. This sensor was actuated by placing the cantilever in a static
magnetic field and passing current through the base of the device. The larger device,
on the order of 1000s of micrometers, was resonated at a lower frequency of 8 kHz
and larger amplitude of 20 pm than the micro-acoustic viscosity sensor. The
oscillations of the device were detected optically. After testing, it was discovered that
the device operated in a domain similar to that of a conventional macroscopic
viscometer. These results confirm the need for a larger resonator to obtain viscosity
measurements comparable to conventional macroscopic viscometer. The device is
not without flaws, however. The testing of this device was conducted in a laboratory
setting. Further development would be necessary before the device could be tested in
an automobile. Another area that needs possible refinement is the oscillation sensor.
Since this sensor is an optical device, its performance is dependent on the clarity of
Other groups of researchers have used TSM resonators for liquid property sensing as
well. S.J. Martin, G.C. Frye and K.O. Wessendorf  used TSM resonators to
measure both the viscosity and density of liquids. In order to measure these two
properties, two TSM sensors were employed. For viscosity sensing, a sensor with a
smooth surface finish was used to examine the response of the fluid when oscillated.
For density sensing, a sensor with a rough surface finish was used. It was believed by
the authors that the rough surface finish would trap the fluid and cause the response to
be affected by the increased mass loading. The average measurement error for
density measurement was reported as 5.3%. The average measurement error for
viscosity measurement was reported as 19.5%. While the density measurement error
seems reasonable, the viscosity error for this device seems high. Another possible
issue with this device is the lack of testing of oils such as those used in internal
combustion engines. The behavior of the device might be altered when exposed to
these types of fluids.
Christian Bergaud and Liviu Nicu also used cantilever beams to measure the
properties of fluids , In this work, arrays of cantilever beams (40 pm x 150-300
pm) are used to sense the viscosity of the fluid using the eigenfrequencies of the
beams. The beams were oscillated using a piezoelectric actuator. Changes in the
measured frequencies correspond to changes in the viscosity of the fluid. The
frequencies of the beams are measured by optical methods. Testing showed that the
device only works in light fluids such as water or air. The device was unsuccessfully
tested in silicon oil. In addition, the accuracy of the device in light fluids was not
very good with an error of 20%. Clearly this device is still in an early stage of
development and would take much more development to refine the device. It is also
difficult to determine how successfully the device would work in lubricating oils,
considering their viscosity when compared to the viscosities of the devices tested.
A similar cantilever beam design was created by Isabelle Etchart and her fellow
researchers , In this work, a MEMS based resonating cantilever beam is used to
detect the viscosity and density of fluids in microfluidic channels. Beams of varying
lengths and widths were tested in microfluidic flows up to 1 pL/min. Two
measurement techniques are presented, optical displacement measurement and strain
gauge measurement. By measuring the frequency change and quality factor of the
beam in a fluid, the density and viscosity of the fluid were determined. However,
these values were determined with large errors present. This was attributed to the
simplicity of the model of the system. With a more sophisticated model, a closer
correlation between measured properties and macroscopic properties could be
achieved. Despite the stated objective of measuring the viscosity of crude oils, it
appears this device would not be very well suited to this type of measurement. Due
to the small fluid capacity of the device (1 pL), any solid particulates that infiltrated
the sensor would quickly damage the sensor.
In this work , C. Harrison used a MEMS based cantilever beam to detect the
density and viscosity of oils. As with the Etchart device detailed in , this
cantilever beam measures the change in resonant frequency and quality factor when
the device is introduced to oil in order to determine the viscosity and density of the
oil. This device differs from the previous device in that the oscillation is driven by a
magnet. The presented results stated that the device could measure densities between
0.6 and 1.5 g/cc and viscosities between 0.4 and 100 cP. This range of viscosity is
not large enough to characterize all lubricating oils used in internal combustion
engines today. In addition, the authors report measurement errors of 1.5% for density
measurement and 10% for viscosity measurement.
A device similar to a cantilever beam is employed by V. Chang, A. Zambrano, M.
Mena, and A. Millan , In this work, the authors use an oscillated device actuated
by a magnetic field to create shear stresses in the oils being measured. The energy
dissipation of the device is then measured, and a corresponding viscosity is
determined. The actual condition of the oil is assessed using a neural network. This
network takes into account multiple engine parameters as well as the viscosity
measurement to determine when the condition of the oil warrants the replacement of
The final device covered here using viscosity sensing to determine the condition of
lubricating oils is found in the work by D.M.G. Preethichandra and K. Shida , In
this work, the authors create an oil condition sensor that monitors three oil properties,
viscosity, capacitance, and cleanness. The device is made up of a motor that spins a
disk in close proximity to a fixed plate. This arrangement shears the fluid between
the disk and plate which in turn causes a resistive torque to be applied to the disk.
This resistance is measured using optical techniques. This optical sensor can also
monitor the cleanness of the oil by measuring the clarity of the oil. The final sensor
included is the capacitive sensor. It is not clearly stated what the purpose of this
sensor is, or what the sensor is intended to measure. Other issues with the sensor
include the size and actuation of the device. Although no dimensions of the device
are provided, it can be assumed the device is not micro-scale due to the presence of a
DC motor. In addition, the presence of a motor draws into question the reliability and
shock resistance of the device. Clearly more work is needed before this device can be
considered a feasible design for oil condition sensing.
2.5 Relevant Patents
In addition to the existing research, many patents relevant to oil condition sensors
exist. These patents cover a variety of different devices and concepts, from methods
of determining oil condition to devices for measuring oil condition. These relevant
patents will be discussed in this section.
US Patent number 6937332 B2  covers a device that measures the quality of
lubricating oils using light sources and light detectors to determine the amount of
particles present in the lubricating oil. This is measured by determining how much of
the light passes through the oil. This measurement method is based on the principle
that particles in the oil will absorb or reflect some of the light passing through the oil.
Much like the other optical techniques discussed, the turbidity of the oil is not
necessarily an indicator of the condition of the oil. As a result this method is not an
accurate means of determining the condition of oil.
US Patent number 5750887  covers a process for measuring the properties of
engine oil and remaining oil life. The process covers the measure of soot, oxidation,
and viscosity. Then the measurements are used to determine the remaining oil life.
The patent does not cover the methods of measurement of any of the properties listed,
nor does it describe any devices.
US Patent number 6718819  covers a device that determines the quality of the oil
by measuring the capacitance between multiple electrodes. The signal is then
conditioned to determine the oil quality. The device is circular in construction and
attaches to the engine block between the engine block and oil filter. The technology
is similar to that found in many of the research devices discussed earlier.
US Patent number 6247354 B1  describes a resonator which senses the properties
of fluids by resonating oscillators at a minimum of two frequencies in air and in the
fluid to be measured. The change in frequency between the two states is then
measured at each frequency. These frequency changes are then used to determine the
properties of the measured fluid. The resonators described in this patent are
US Patent number 7210332 B2  describes a mechanical resonator for use in fluid
measurement. The proposed device is able to measure a range of fluid properties
including viscosity, density, and dielectric constant. In order to avoid the issues
caused by high frequency resonators that have already been described, this patent
proposes an oscillating frequency between 25 and 30 kHz. The patent drawings show
the resonator as a tuning fork design, but the author claims the patent can be extended
to other device designs.
US patent number 2008/0314128 A1 describes a device for measuring the viscosity of
liquid using piezoelectric diaphragms. The device measures the viscosity of a fluid
by exposing a piezoelectric diaphragm to the fluid inside of an amplification
chamber. The piezoelectric diaphragm is then excited to cause the fluid to move and
causes the diaphragm to output a signal proportional to the viscosity of the fluid.
2.6 Existing Commercial Devices
In the marketplace today, there are a range of products designed to measure various
properties of fluids. Some of these products are designed for use in the automotive
world, while others are not. This section aims to briefly introduce some of the
commercial fluid property measurement devices currently available on the market.
The Visyx Tuning Fork Resonator  is a device produced by Visyx that measures
the viscosity, density, dielectric constant, and temperature of fluids. This device is
designed for automotive applications. The sensor is capable of monitoring the quality
of engine oil, fuel, lubricating oils, hydraulic fluid, and cooling systems. The device
monitors the condition of the fluid by continuously monitoring the viscosity, density,
dielectric constant, and temperature of the fluid. All of the electronics are packaged
contained in one module so no external interfacing is necessary.
The Flowtronics FS-3 fluid sensor  is a device designed to determine the
contamination and health of fluids, specifically, automotive fluids. The device uses
electrochemical impedance spectroscopy to measure engine oil properties such as
soot, fuel, water, and coolant. From these measurements, the condition of the oil is
determined using a Windows based program. The device is designed for inline
retrofit into any diesel fueled vehicle.
The Kittiwake Oil Condition Sensor  is designed to continuously monitor the
condition of lubricating oils based on property measurement. The device monitors oil
properties such as moisture content, ferrous metal content, oxidation levels, and
temperature in order to assess the condition of the oil. This condition is then reported
to the user on a scale of 0-100 Oil Quality Units. The device is designed for a range
of applications including automobile engines and transmissions.
The Gill Sensors Oil Quality Sensor  is a solid-state device designed to measure
the quality of oils and other automotive fluids in the racecar environment. The device
senses the dielectric value of the oil to determine the degradation and contamination
of the oil. The device is created in both in-line and in-tank variants. Before use, the
device must be calibrated in the oil that will be monitored using proprietary computer
software. This device can be used to measure a variety of automotive fluids including
engine oil, transmission oil, hydraulic fluids, and diesel fuel.
The GM Oil-Life System [30, 31] is an onboard algorithm that determines oil and
filter change intervals based on some operating conditions of the vehicle.
Specifically, the system attempts to monitor the oil degradation process due to the
combustion event in the engine by observing engine revolutions and operating
The DiamlerChrysler Flexible Service System  is a computer algorithm that
tracks various car operating conditions in order to determine the condition of the oil.
The device measures factors affected by the habits of the drivers. These include
vehicle speed, coolant temperature, load signal, engine rpm, trip distance, engine oil
temperature and level. In addition, the system measures the capacitance of the oil in
the oil pan to determine the level of contamination in the oil.
The Delphi INTELLEK Oil Condition Sensor  measures the oil temperature,
conductivity, water and glycol contamination as well as the oil level to determine the
condition of the oil. In addition, the system uses an algorithm that takes into account
operating conditions similar to the GM and DiamlerChrysler devices. The sensor is
located in the oil pan.
The QLT Oil Condition Sensor  was developed in the mid 1990s to determine
the condition of engine oils through the monitoring of oxidation products, and water
and fuel contamination. These properties are measured by determining the
permittivity of the oil using a capacitor.
The Oil Insyte [30,32] device by Voelker Sensors Inc. uses a unique method in order
to determine the condition of the oil in a system. This device measures electrical
properties of the oil using an oil-insoluble polymeric bead matrix. This detection
technique essentially measures the polarity of the oil in order to determine the level of
oil degradation and contamination. This change in polarity is measured by observing
the change in capacitance of the device.
The Lubrigard Oil Condition Monitoring Sensor [30,33] is designed to detect coolant
contamination, metallic wear debris and oil degradation such as oxidation and soot
accumulation. The system accomplishes this by measuring a dielectric loss factor, or
Tan Delta. The system was designed to be connected to the onboard computer
system of modern cars, or be retrofitted to a standalone display.
The Solid-State Oil Condition Sensor from Symyx Technologies [30,34,35] uses a
solid-state micromechanical resonator to measure the viscosity, density, and dielectric
constant of the oil. This solid-state sensor is based on a crystal tuning fork design.
The company claims the device is more sensitive than TSM resonators on the market
due to its lower resonance frequency and ability to measure multiple fluid properties
simultaneously. Symyx has sold the licensing rights for the device to Hella KG
Hueck and Co. for automotive applications.
The Eaton Fluid Condition Monitor  uses impedance spectroscopy to determine
several electrical properties of the oil. The company claims the device measures both
surface and bulk properties of the oil. These properties include conductivity, and
dielectric constant. These measurements are then correlated to the physiochemical
properties of the oil. The sensor is located in the oil pan.
The Ford Intelligent Oil-Life Monitor (IOLM)  determines when the oil in the
vehicle needs to be changed based on a number of factors. Based on the driving
habits, hours in operation, oil temperature, engine speed and engine torque of the
vehicle, an algorithm calculates the appropriate oil change interval. The system does
not use oil quality sensors it is entirely software-based The system is currently
offered on over 90 percent of 2011 model year Ford, Lincoln and Mercury vehicles.
The IntelliStick Oil Condition Monitoring System  determines the condition of
engine oils using multiple sensors. Specifically, the system uses 3 sensors to
determine the condition of the oil. A temperature compensated conductance
measurement detects the amount of additives and oxidation byproducts in the oil. An
emulsion ratio sensor detects the presence of water or glycol contamination in the oil
as well as sludge formation. The final sensor measures the temperature of the oil.
This combination of sensors is believed to accurately gauge the condition of engine
oils and prevent damage and failures associated with oil breakdown.
The FluiSens system  was developed by Lubrizol and Delta Electronics as an
alternative to other oil condition monitoring techniques. The system uses a
transducer and thermocouple in conjunction with an algorithm to determine the
condition of lubricating oils. The transducer monitors the electrochemical impedance
of the oil to assess the functionality of the oil in 3 categories, friction reduction,
contamination and oxidation, and surface protection. From these measurements, the
device determines the remaining useful life of the oil. This product was scheduled to
be introduced to the market in 2008, but no evidence was found that this occurred.
The Sentelligence Technology Fluid Condition Monitoring Sensor  employs a
universal optical platform to determine the properties of fluids such as lubricating
oils. The major optical phenomenon measured is light absorption. These
measurements can be correlated to changes in the fluid such as soot formation,
degradation, contamination, oxidation, and viscosity. From these measurements the
onboard data handling functions determine the condition of the fluid.
SenGenuity, a division of Vectron International, has developed Fluid Condition
Monitoring system [40, 41] based on surface acoustic wave (SAW) resonator
technology. The SAW resonator detects changes in viscosity due to fluid
contamination by measuring the power loss of the device when exposed to the fluid.
SenGenuity currently offers the ViSmart fluid condition sensor as its only fluid
condition monitoring sensor.
However, the company is currently working on a new sensor that will measure a
broader range of factors that influence the condition of oil and lubricants. In addition
to monitoring the viscosity of oils, the new device will measure the dielectric constant
and conductivity of the oil. This will allow the device to more accurately determine
the source of contamination or degradation of the oil. The dielectric constant sensor
will allow for the precise identification of water, fuel, and glycol contamination. The
conductivity sensor will allow for the detection of soot, metal particles, and other
debris in the oil. Once these measurements have been made, an algorithm will
analyze the data and determine the condition of the oil.
The Bosch GmbH Multifunction Oil Condition Sensor [3, 14] is a joint venture
between Bosch GmbH and researcher Bernhard Jakoby, one of the major contributors
to the oil condition monitoring field. This device uses the solid-state viscosity
sensing technology in addition to permittivity sensing to determine the condition of
lubricating oils. Much of the technology used in this device was derived from the
work of Jakoby and his laboratory.
For this research work, a device was designed, built, and tested. In order to detail the
entire research process, this chapter will be split into three major sections, design,
proof of concept and device construction, and testing.
Once the sources found in the literature review were studied, the design of the device
was initiated. Multiple sensor designs were investigated. In the end, it was decided
that no one property could accurately predict the condition of lubricating oils in
internal combustion engines. Rather, a series of measurements would be necessary to
determine the condition of lubricating oils.
3.1.1 Viscosity Sensor Design
A wide variety of device design concepts were initially considered. One of the first
ideas was to incorporate an oil condition sensor into a microfluidics device. This was
proposed due to the microfluidics background of the research group. A viscosity
sensor based on a combination of microfluidics and MEMS was proposed. This
device is shown in Figure 3.1.
Figure 3.1 Micro-Scale Oil Condition Sensor Design
This design consisted of a patterned glass substrate containing a microfluidic channel
and two pads. An electro-statically actuated MEMS resonator was located in each
pad to measure the viscosity of the oil by means of the increased damping of the
oscillation of the resonator.
While this idea initially seemed attractive, several issues with the design arose. First,
the size of the device was an issue for the proposed environment in which the device
was to operate. Internal combustion engines are designed to circulate oil at pressures
from 20 psi to 100 psi. Based on the flow rates associated with these pressures, the
small cross sectional area of the device channels would create a restriction in the
oiling system. This restriction could interfere with the operation of the oiling system
as well as damage the MEMS resonators. In addition, the small channel size could
also become easily clogged by debris from used oil. Due to these and the difficulty of
manufacturing such a device, another device design was considered.
The next design concept considered for the project was to increase the scale of the
micro device to eliminate the shortcomings of the previous design. By increasing the
size, flow restrictions and manufacturing difficulties could be eliminated. The
proposed MEMS resonator is shown in Figure 3.2. The resonator model was
constructed using printed circuit board (PCB) techniques. The resonator was actuated
using the electrostatic force between two copper plates. The device dimensions were
5 mm x 5 mm.
Once the resonator concept and materials were chosen, the performance of the design
was determined using CoventorWare analysis software. In order to analyze the
design, a model was constructed using PCB materials and process. Figure 3.3 shows
the construction layup of the device.
Figure 3.3. Device Layout, from top left to bottom right: copper electrode, bond film,
bottom copper resonator layer, kapton layer, top copper resonator layer (complete
Using CoventorWare, the resonant frequency and displacement of the device at 5
volts were determined. The results of the simulations are shown in Table 3.1.
Table 3.1 CoventorWare simulation results for 5 V input.
Resonant Frequency 1140 Hz
Amplitude 7 .8 pm
These results showed good agreement with the analytical calculations. This low
resonant frequency and large amplitude when compared to other oscillating viscosity
sensors made this sensor design an attractive solution. However, some issues with
this sensor arose. Once again the ease of manufacture was an issue since the
construction relied on PCB construction techniques. The larger issue with the design
is the durability. Many of the materials used in the construction of the device would
not withstand the environment found in the oiling system of an internal combustion
engine. In addition, it was determined the flexures, made of a thin layer of kapton
polymer, would not have a long enough fatigue life to last a substantial amount of
time. This would require the sensor to be replaced on a regular basis.
It was initially suggested that a device containing multiple sensors be used to
overcome this issue. In this device, the viscosity detection would simply shift from
one sensor to another as the sensors wore out. This would solve the fatigue life issue,
but would not solve other corrosion issues found in the highly acidic environment of
an internal combustion engine oiling system. Due to these issues, other detection
methods were investigated.
In order to solve the corrosion and manufacturing issues found in the previous design,
a new oscillation technique was needed. Due to the novel nature of the measurement
technique (normal oscillation), a device that could oscillate a diaphragm normal to a
planar surface was desired. Two possible solutions to this problem were analyzed,
piezoelectric actuation and electromagnetic actuation. These methods were chosen
due to their common usage in the electronics field in acoustics applications. This
meant both technologies were robust and relatively small. Upon further analysis it
was determined that electrostatic actuation required more space and power than a
similarly sized piezoelectric design.
A piezoelectric material is a material that changes shape when a voltage is applied, or
conversely, creates a voltage when displaced. The material is often used in sensor
applications to detect small loads and displacements. Initially, a piezoelectric film
was to be applied to a diaphragm much like the one shown in Figure 3.2, however,
this design was found to require too much power to actuate. Further research into the
use of piezoelectric materials in acoustics revealed that acoustic diaphragms actuated
using piezoelectric materials were quite common. These diaphragms are often found
in devices such as alarm clocks and cell phones. Such a diaphragm is shown in
Figure 3.4 Acoustic Diaphragm Driven by Piezoelectric Film
Since these piezoelectric diaphragms were commercially produced and relatively
inexpensive, it was decided to design the device around these diaphragms.
Once the decision was made to use piezoelectric diaphragms, multiple sensor
configurations were considered. The first design was a combination of the
piezoelectric diaphragm and the previous electrostatically driven device. Figure 3.5
shows a cross section of the device.
Figure 3.5 Viscosity sensor design incorporating a piezoelectric diaphragm oscillator
with a capacitance based position sensor.
The operation of this device was based upon two main components, the piezoelectric
oscillator and a capacitance electrode. To detect the viscosity of a fluid between the
two components, the piezoelectric diaphragm would be oscillated at a known
frequency and amplitude. The capacitance electrode would then be used to determine
the position of the oscillator based on changes in the capacitance between the
electrode and oscillator. From this position measurement, the maximum
displacement of the diaphragm could be determined. The maximum displacement
could then be compared to the maximum displacement of the diaphragm when no
fluid was present and a relative viscosity could be determined.
This design, although simple in construction, was deemed too complicated for a few
reasons. First, the method of determining the maximum amplitude was unreasonable
considering the resources available to measure the changes in capacitance. Because
the device was to be oscillated in the range of 100 to 10000 Hz, the response time of
the measurement device would have to be very small. This response time was not
feasible using the equipment available. In addition, it was not known how accurately
the viscosity measurements from this device would correlate to more traditional
viscosity measurements. Since the viscosity would be based on the oscillation
behavior in air, it was decided the accuracy of the viscosity measurement would not
correlate well to traditional viscosity measurement techniques.
In order to alleviate the measurement issues detailed above, the design in Figure 3.6
was developed. In this design, two piezoelectric diaphragms would be bonded
together. The upper diaphragm would be excited using a sinusoidal voltage input and
the lower diaphragm would be used to measure the displacement of the sensor. This
design eliminates the complicated capacitance based displacement sensing technique
and instead relies on the voltage created by displacing the piezoelectric film on the
Figure 3.6 Viscosity sensor design incorporating a piezoelectric diaphragm oscillator
bonded to a piezoelectric diaphragm sensor to detect the displacement of the device.
This device obviously addresses the issues with the capacitance based displacement
measurement technique discussed earlier; however, the design still contained some
flaws. Despite simplifying the measurement technique, the device would still
measure the viscosity in much the same way as the previous sensor, therefore, the
device would still have the same issues with correlation that the previous sensor had.
Another possible issue is the added stiffness of the oscillation diaphragm when the
sensor diaphragm is bonded to it. This added stiffness will reduce the displacement
of the device at a given voltage. This reduced displacement will result in smaller
changes in displacement between fluids of different viscosities and as a result will
lower the sensitivity of the device. Because of these issues, it was decided to explore
other sensor configuration options.
The final design iteration for the device aimed to solve the issues of the previous two
device designs while still incorporating piezoelectric diaphragms into the design. The
solution devised for this work is shown in Figure 3.7.
Figure 3.7 Final design for viscosity sensor constructed from two piezoelectric
In this design the oscillator is excited with a sinusoidal voltage input. This
propagates a wave through the fluid in the gap between the oscillator and sensor
diaphragms. When this wave reaches the sensor diaphragm, a force is exerted on the
diaphragm and the diaphragm oscillates at an amplitude and displacement
proportional to the wave generated in the fluid. Depending on the viscosity of the
fluid, the amplitude of the wave propagated through the fluid will change. When this
fluid interacts with the sensor diaphragm, the amplitude of the oscillation induced by
the wave will change based on the viscosity of the fluid.
3.1.2 Contamination Detection Sensor
In addition to designing a viscosity sensor, a sensor capable of detecting
contamination of lubricating oils was also designed. The design of the contamination
detection sensor began shortly after the viscosity sensor design. Both sensors were
designed in parallel since the sensors would be used in conjunction with each other to
determine the overall condition of the lubricating oil.
The first step in the design process for the contamination sensor was to determine
what contamination would be detected. It was initially decided that the
contamination sensor would focus on detecting the presence of solid contamination
particles, such as metal wear particles, rather than liquid contaminants. It was
concluded that liquid contaminants would have an effect on the viscosity of the oil,
and would therefore be detected by the viscosity sensor.
After researching some of the methods used by other researchers to detect solid
contaminants, an optical method was initially explored. A design composed of two
light emitting diodes (LEDs) was proposed. A schematic of the design is shown in
Figure 3.8 Contamination Sensing Using LEDs
In this design changes in the clarity of the oil are determined by measuring the
intensity of the light passing through the sample. As the oil sample becomes more
clouded from contaminants, less and less light passes through the sample to the light
sensing LED. Some preliminary testing of this design was conducted, but it was
quickly discovered that this method had some inherent drawbacks. The largest
drawback was the lack of correlation between oil turbidity and condition. No data
was found that shows that as oil turbidity increases, the lubricating properties of oil
decrease. Based on this finding, a different method of contamination detection was
The next approach taken to detect the presence of contamination in lubricating oils
was electrical detection. Many different approaches to the detection of contamination
using electrical measurement have been explored in the literature. After examining
some of these methods, it was decided that a capacitance based measurement
technique would be used to detect the presence of contamination in the oil. This
method was chosen for a few reasons. First, it is a relatively simple measurement
technique. According to the parallel plate capacitance theory, capacitance between
two plates such as those shown in Figure 3.9 is a function of the area of the parallel
plates (A), the gap separating the plates (D), and the permittivity of the medium
separating the plates (ej) as shown in Equation 3.1.
From Equation 3.1 it can be seen that if the area of the plates and the gap separating
them are held constant, a change in capacitance will indicate a change in the
permittivity, or resistance to the flow of electricity, of the volume between the plates.
When some conductive object passes between the plates, the permittivity will
increase due to a drop in the resistance to the flow of electricity between the plates.
These objects will result in a spike in the measured capacitance and allow the amount
of metallic wear particles in the oil to be monitored.
Figure 3.9 Parallel Plate Capacitance Model
Another reason parallel plate capacitance was chosen as the method of contamination
detection was the ability to detect more than just metal wear particles. By using a
capacitance based sensor, liquid contamination of the oil from glycol, water, fuel and
other sources can also be detected. This allows for the capacitance sensor to aid in
the identification of liquid contaminants that are initially detected by changes in the
viscosity of the oil. Due to large differences in the permittivity of contaminants when
compared to oil, small concentrations of contaminants can be detected.
3.1.3 Integrated Sensor Design
Once the design for the viscosity and capacitance sensors was completed, it was
necessary to combine the sensors into one device design. Initially it was thought that
the two detection methods would be separate from each other in the design. Then, it
was realized that the two detection methods could be measured using only one sensor.
In other words, both the viscosity and the capacitance of a fluid could be measured
using the same sensor structure. This would allow the sensor to be packaged in a
much smaller space than if two dedicated structures were necessary. Once it was
decided that only one sensor was necessary for both measurements, a device design
was created. The initial design of the oil condition sensor is shown in Figure 3.10.
In this design two sensor pairs would be aligned in a small rectangular cross-section
tube through which oil would be allowed to flow. One of the sensors would detect
the viscosity of the fluid while the other would detect the capacitance of the fluid.
When the piezoelectric film on the viscosity sensor pair wore out or was damaged,
the sensors could switch sensing duties in order to extend the lifetime of the device.
Figure 3.10 Initial oil condition sensor design
This design was thought to be feasible but had one flaw, the piezoelectric diaphragms
would have to be custom manufactured for the project in order to be mounted in the
configuration shown in Figure 3.10. Fabricating these diaphragms would simply be
too costly to make the design a feasible path forward.
As a result of the design complications of the initial design, a new design was created.
This design is shown in Figure 3.11. This design consists of the same basic layout as
the previous design but incorporates a more widely available piezoelectric diaphragm
design. By mounting the piezoelectric diaphragm along the outer diameter, common
piezoelectric diaphragms used in sound applications could be sourced for the device.
Figure 3.11 Revised oil condition sensor design
This design was used as the basis for the research and testing of the device.
In order to predict the behavior of the device when exposed to the oil, a model of the
system was created. Specifically, two different models of the system were created, a
SolidWorks model and a 2 degree of freedom (DOF) model based on simple vibration
concepts. The SolidWorks model was used to model the behavior of the piezoelectric
diaphragm as it resonated in air. Information obtained from this model was then used
in the 2 DOF model to approximate the behavior of the complete sensor in a fluid.
3.2.1 SolidWorks Model
The goal of the first model was to approximate the behavior of the piezoelectric
diaphragm and obtain information about the diaphragm needed to model the behavior
of the complete system in a fluid. The modeling began by drawing the piezoelectric
diaphragm in the program using the dimensions provided by the manufacturer.
Figure 3.12 shows the completed drawing of the piezoelectric diaphragm.
Figure 3.12. Piezoelectric diaphragm model in SolidWorks
Once the model was completed, a finite element analysis (FEA) was conducted to
determine the natural frequency of the diaphragm. This natural frequency could then
be compared to the natural frequency reported by the manufacturer as well as the
measured natural frequency to determine the accuracy of the model.
One of the issues encountered early in the modeling process was the method of
restraining the diaphragm. In the actual device, the diaphragm is attached to the
frame with epoxy along the outer edge of the diaphragm, as shown in Figure 3.13.
This mounting configuration could be approximated using FEA in two different ways,
a fixed support or a simple support. Both methods of fixing the diaphragm are
illustrated in Figure 3.14.
Figure 3.14. (a) Simple support (b) Fixed support
In order to determine which method was more accurate, both methods were modeled
and the resulting natural frequencies were compared to the value reported by the
manufacturer (Murata) as well as actual test results. These results are shown in Table
Table 3.2. Comparison of simple and fixed support methods
Model Natural Frequency (Hz)
From Table 3.2 it can be seen that both the fixed and simple support methods of
restraint produce very similar natural frequency values. As a result, fixed supports
were chosen for the modeling due to quicker computation times and less complication
After determining the method of restraint for the diaphragm, further refinement of the
model was conducted. In order to create a more realistic model, the wire leads
attached to the electrodes were added to the model. Since the leads are also vibrated
by the oscillation of the diaphragm, the presence of the wires would have an effect on
the natural frequency of the system. More specifically, the presence of the leads was
expected to increase the stiffness of the system. As a result, the natural frequency
was also expected to increase when compared to the piezoelectric diaphragm alone.
Two different models were considered, these models are shown in Figures 3.15 and
Figure 3.15. Piezoelectric diaphragm model with straight wire leads
Figure 3.16. Piezoelectric diaphragm model with curved lead
Each of the models was analyzed to determine the natural frequency using the
SolidWorks FEA package. The ends of the wires were restrained using a fixed
support about the diameter of the wires. The results of the analysis for each model
are shown in Table 3.3.
Table 3.3. Comparison of piezoelectric diaphragms with straight and curved wire
Model Natural Frequency (Hz)
Straight Leads 11152
Curved Leads 10202
From Table 3.3 it can be seen that the addition of lead wires to the model, whether in
the straight or curved configuration, increases the natural frequency of the model
substantially. In fact, the addition of the lead wires more than doubles the natural
frequency. When compared to the experimental results shown in Table 3.2, it is clear
that the addition of the lead wires causes the model to no longer accurately model the
system. As a result the lead wires were removed from further model simulations.
Once the SolidWorks model was created and validated using experimental data, the
model was used to gather properties needed to create the 2 DOF model for the sensor.
These values and the specific methods by which they were obtained will be discussed
in the 2 DOF model section.
3.2.2 Two Degree of Freedom Model
The two degree of freedom model was created in order to model the entire sensor
system and determine how the system would react to changes in the fluid interacting
with the sensor. The model was based on simple system dynamics principles found
in any engineering vibration textbook. The first step in modeling the system was to
create a simple model to approximate the system. The model chosen for the sensor is
shown in Figure 3.17.
The model consists of 2 masses and 2 springs, which represent the piezoelectric
diaphragms. The 2 piezoelectric diaphragms are coupled by a damper, which
represents the damping in the system due to the fluid between the diaphragms. The
force acting on the lower piezoelectric diaphragm is due to the forcing frequency
created by the voltage input to the oscillator. Once the model was created, the system
was broken into 2 free body diagrams and the forces on each diaphragm were
analyzed. The free body diagrams for each of the diaphragms are shown in Figure
Figure 3.18. Free body diagrams for 2 degree of freedom model
From these free body diagrams, a mathematical model of the system was created
The math model is shown in matrix form in equation 3.2.
ml 0 x\ +
0 m2 *2.
From equation3.2, it can be seen that multiple variables are needed in order to solve
the system. Specifically, the mass (m), damping (c), spring constant (k), applied
force and frequency (F, w), and time. The following sections describe the methods
and values used for each of these variables.
The mass needed for the equation is the mass of the piezoelectric diaphragm. This
value was determined in two different ways. First, the mass of an actual piezoelectric
diaphragm was measured. It was discovered that the mass of the diaphragm,
including the leads and solder, was 0.2 g. The mass of the diaphragm was also
determined from the SolidWorks model. The total mass of the diaphragm was found
to be 0.26 g by SolidWorks. It was decided that the mass determined by SolidWorks
would be used rather than the measured mass due to the fact that the masses are
relatively close and other values determined using SolidWorks are necessary to use
The total mass of the diaphragm was not included in the model due to the fact that not
all of the mass of the diaphragm is oscillating. In order to include only the mass that
is oscillated, an approximation was made. Specifically, only the mass of the
diaphragm included in a 9 mm area from the center was included in the
approximation. This diameter was chosen due to the fact that 9 mm is the area
covered by the piezoelectric element on the diaphragm. Taking only this area into
account, the effective mass of the diaphragm was found to be 0.21 g.
126.96.36.199 Spring Constant
The spring constants for the model represent the stiffness of the piezoelectric
diaphragms. Both ki and k2 represent the same value. The spring constant for the
piezoelectric diaphragms was calculated using the SolidWorks model. A force of 1 N
was placed in the center of the diaphragm while the outer diameter of the diaphragm
was fixed. The resulting deflection was then determined by the program and using
the value obtained, as well as the input force, the spring constant of the diaphragm
could be determined using equation 3.3.
F = k8 (3.3)
F = force
k = spring constant
5 = displacement
By rearranging the equation to solve for k, the spring constant of the diaphragm was
determined. It was discovered that the spring constant was 100000 N/m.
The damping of the system is determined by the properties of the fluid between the
two piezoelectric diaphragms. The damping of the fluid is predominantly related to
the viscosity of the fluid. As a result, changes in viscosity will result in changes in
damping. These changes in damping affect the response of the system to a given
input, allowing the sensor response to be correlated to viscosity. The damping of a
fluid between two closely spaced, moving plates has been modeled by many different
researchers. Many of these mathematical models are very complicated and rely on
many properties of the fluid to determine the damping of the fluid. In order to keep
the system model from growing too complex, two simple damping models will be
employed. The first model for the damping of fluids between two closely spaced,
oscillating plates was created by Axel Berny , In this work, a simple model based
on squeeze film damping is derived from Reynolds equations describing fluid flows.
The model, while based on MEMs devices, can be extended for use in meso-scale
devices as well. Equation 3.4 shows Bernys model for the damping of a fluid
between two closely spaced, oscillating plates.
u 4 i
c =--------nr3 (3.4)
ft3 3 v '
c = damping coefficient
p = viscosity of the fluid
h = initial gap between the plates
r = radius of the plates
The other damping model employed also relies on squeeze film damping to describe
the effect of a fluid between two closely spaced, oscillating plates. This model,
created by Behraad Bahreyni , is based on the more general Navier-Stokes
equations describing fluid flows. This model was also developed for use in MEMs
devices, but is applicable to the meso-scale device described in this work. Equation
3.5 shows the Bahreyni model for the damping of a fluid between two closely spaced,
Both of these models were used in the 2 DOF model and compared to each other as
well as to experimental results.
188.8.131.52 Force Applied
The force applied by the voltage input on the piezoelectric was calculated and used as
the force applied to the system. When a voltage is applied to a piezoelectric material,
a displacement of the material occurs. The resulting force that causes the
displacement can be calculated using the properties of the material. Using equation
3.3, the force can be determined per volt of input. For this system, the force was
calculated to be 7.8 x 10'5 N.
184.108.40.206 Drive Frequency and Time
The drive frequency for the system is the frequency at which the system is oscillated.
This value was chosen based on the experimental natural frequency, 3500 Hz. The
time for the system is the time at which the equations are to be solved. For all of the
simulations the time chosen was 1 second.
220.127.116.11 Simulating with the 2 DOF Model
Once all of the variables were determined, the system could be simulated. In order to
solve the model, a simple Matlab code was created to solve the system. The code for
the system can be found in the appendix. Using the code and the variables, the
system was solved. For each of the damping models, three oils of different viscosities
were simulated. Table 3 .4 compares the results for the two different damping models
Table 3.4. Results of 2 DOF model for both damping models
Bahreyni Dam| [ling Model
Oil Viscosity (m2/s) C W (rad/s) W (Hz) Wd (rad/s) Wd(Hz) Xss
SAE30 8.60E-05 0.0767 21822 3473.08 21821 3472.92 1.673E-06
SAE40 1.40E-04 0.1248 21822 3473.08 21820 3472.76 9.126E-07
SAE50 2.13E-04 0.1889 21822 3473.08 21817 3472.28 4.764E-07
Be my Damping Mode
Oil 2 Viscosity (m /s) C Wn (rad/s) Wn(Hz) Wd (rad/s) Wd (Hz) Xss
SAE30 8.60E-05 0.0778 21822 3473.08 21821 3472.92 1.481E-06
SAE40 1.40E-04 0.1267 21822 3473.08 21820 3472.76 7.507E-07
SAE 50 2.13E-04 0.1927 21822 3473.08 21817 3472.28 3.564E-07
From the results in Table 3.4, it can be seen that the damping models both produce
similar results. The damped natural frequencies (Wd) of both models were identical
for the three different oils simulated. The displacement of the sensor diaphragm (Xss)
is different for the two models, but not significantly so.
One interesting result of the simulation was the damped natural frequency results.
Despite large changes in viscosity, the damped natural frequencies change by only
tenths of a hertz. This is an interesting finding and one that was not expected. One
behavior of the sensor that did change was the displacement of the sensor diaphragm.
Large changes in viscosity resulted in large changes in the displacement of the sensor.
This result was anticipated and the model verified this theory.
After the model was completed, it was compared to experimental results to determine
if the model was valid. First, the undamped natural frequency of the model was
compared to the undamped natural frequency measured experimentally. Both values
are shown in Table 3.5.
Table 3.5. Comparison of the undamped natural frequencies from the models and
Undamped Natural Frequency (Hz)
Bahreyni Model 3473
Bemy Model 3473
Table 3.5 shows that the undamped natural frequencies for both of the models match
the experimental undamped natural frequency very closely. In addition to comparing
the undamped natural frequencies, the damped natural frequencies were compared.
For this comparison, both models were simulated with SAE 30 oil and compared to
experimental tests of the sensor using SAE 30 oil. This comparison is shown in Table
Table 3.6. Comparison of the damped natural frequencies from the models and
Damped Natural Frequency (Hz)
Bahreyni Model 3472.92
Bemy Model 3472.92
Table 3.6 shows a very close approximation of the experimental damped natural
frequency by both of the damping models. It also validates the previous finding that
the natural frequency of the system remains largely the same despite large changes in
the damping of the system.
The last validation of the model was to compare the displacements of the model to the
experimentally measured displacements of the sensor. This last validation is perhaps
one of the most important, since the measurement technique of the sensor is based
upon changes in the displacement of the sensor. Rather than look at the displacement
values produced by the model and measured experimentally, the values will be
compared to each other to determine the relative change between each viscosity. This
comparison for each of the two models and the experimental measurements is shown
in Table 3.7.
Table 3.7. Comparison of relative displacement from models and experimental data
Oil 2 Viscosity (m /s) Viscosity Ratio Xss Displacement Ratio
SAE 30 8.60E-05 1.00 1.67E-06 1.00
SAE 40 1.40E-04 1.63 9.13E-07 1.83
SAE 50 2.13E-04 2.48 4.76E-07 3.51
Oil 2 Viscosity (m /s) Viscosity Ratio Xss Displacement Ratio
SAE 30 8.60E-05 1.00 1.48E-06 1.00
SAE 40 1.40E-04 1.63 7.51E-07 1.97
SAE 50 2.13E-04 2.48 3.56E-07 4.16
Oil , 2 Viscosity (m /s) Viscosity Ratio X* Displacement Ratio
SAE 30 8.60E-05 1.00 6.61E-02 1.00
SAE 40 1.40E-04 1.63 4.87E-02 1.36
SAE 50 2.13E-04 2.48 3.03E-02 2.18
Table 3.7 shows the ratio between the displacement of the sensor in SAE 30 oil and
the displacement of the SAE 40 and SAE 50 oils. Both of the damping models
overestimate the effect of the oils of higher viscosity on the sensor. However, the
Bahreyni model does more closely approximate the experimental data. Both models
do indicate large changes in the displacement due to large changes in the viscosity
interacting with the sensor. The ability of the models to predict these changes,
specifically the Bahreyni model, demonstrates the validity of the model and validates
the measurement technique chosen to monitor the viscosity of the fluid interacting
with the sensor.
3.3 Proof of Concept
Once the design of the sensor was established, some proof of concept tests were
conducted to determine the feasibility of the designs. Because the design relied on a
mix of new and existing technologies, it was crucial to test the concepts before
investing the time and resources to produce the device as designed. The proof of
concept phase can be divided into two parts, contaminant detection proof of concept
testing and viscosity sensing proof of concept testing.
3.3.1 Contaminant Detection Proof of Concept
The first goal of the proof of concept testing was to prove the feasibility of the
contamination detection method. Specifically, a validation of parallel plate
capacitance was necessary for plates submerged in oil. It would also be necessary to
see if the presence of a metallic particle in the gap separating the plates would cause a
noticeable change in the measured capacitance. In order to conduct these tests, the
device in Figure 3.19 was constructed.
The device in Figure 3.19 contains a small oil reservoir. Inside the reservoir, two
copper electrodes are separated by a 10 mm gap. The reservoir is inside a
containment box that prevents spills and eliminates the chance of electrical shock
from the electrodes. Also inside the containment box is a tilting stage for the oil
reservoir to be placed on. This tilting stage allows for metallic particles, such as ball
bearings, to pass between the electrodes while the containment box is sealed.
The first experiment conducted with this device was a measurement of the
capacitance between the two plates in air. When a charge was placed across the two
plates, a small change in capacitance was noted using an Agilent U1732A RLC
Meter. Next, a small ball bearing was placed between the two copper plates to see if
any change in capacitance could be detected. No significant change in capacitance
Once it was confirmed that the capacitance between the two plates could be
measured, 5W-30 weight motor oil was placed in the space between the two plates.
Once again the capacitance was measured when a charge was placed across the
plates. A larger change in capacitance was noted. This was most likely due to the
fact that the dielectric constant of oil (2.1-2.4) is higher than that of air (1.0). After
this confirmation of the proper functioning of the device, a 0.25 in. (6.35 mm)
diameter ball bearing was placed in between the plates of the device.
The capacitance across the plates rose approximately 0.1 pf when the 0.25 in.
diameter ball bearing was present. This appears to be a very small change, but it is
consistent with the results of a similar test described in , Other size ball bearings
were also tested in the device, but proved to be too small to create any significant
changes in capacitance.
After these preliminary tests, it was decided a study of the effect of the gap distance
was necessary. It is well known that the capacitance between two parallel plates will
increase as the distance between the plates decreases. The goal of this experiment
was to determine how closely the theoretical predictions of capacitance matched the
Initially, the device shown in Figure 3.19 was modified to study the effects of gap
distance on capacitance. For these tests, the capacitance was measured while the 0.25
in. diameter ball bearing was placed between the plates. The results of the test are
shown in Table 3.8.
Table 3.8. Capacitance between plates for varying gap distances
Gap Distance (mm) Capacitance (pf)
Table 3.8 clearly shows that the closer the gap distance is to the size of the particle,
the larger change in capacitance the particle will make. This was an expected result,
but a result that has a large impact on the design of the capacitance measurement
portion of the device. Since the device must be able to measure the presence of small
particles (~10pm), a small gap size was necessary for accurate detection.
After completing the initial investigation on capacitance based measurement, the
effect of gap distance on the capacitance between two parallel plates was explored.
In this experiment, the device shown in Figure 3.20 was created to determine the
effects of gap distance on capacitance. The device was constructed to allow the
distance between two parallel plates to be changed between 10 mm and 1 mm.
Figure 3.20 Variable gap capacitance device
For this experiment, the capacitance across the plates was measured in 1 mm
increments from 1 mm to 10 mm in 3 different dielectric mediums, air, deionized
water, and 5W-30 motor oil. The results of the experiments are shown in Figures
3.21 3.23. These results show that as the gap between the plates becomes smaller,
the capacitance across the plates rises exponentially. However, this behavior was not
clearly demonstrated in the oil. The behavior of the oil seemed to be somewhat
erratic. It is not known exactly why this occurred, but it seems that there was not a
large enough capacitance between the plates at the larger distances (>5 mm). This
could be due to the change in plate geometry for this experiment. Instead of using
square plates as the first device did, this device used rectangular plates. This could
account for the change in capacitance at larger gaps between the two devices.
Ciipiicitance (pf) apncitance (pf)
Figure 3.21 Results of variable capacitance test in air
Figure 3.22 Results of variable capacitance test in water
Figure 3.23 Results of variable capacitance test in oil
Based on this series of tests, it was decided that capacitance measurement was an
acceptable method for detecting contamination in lubricating oils. Relative to the
design of the device, it is important to use a gap size as small as possible in order to
have the greatest resolution. However, a gap size that is too small will not allow the
passage of larger particles between the plates and will therefore not be detected. As a
result, it is important to choose a gap size that will maximize the sensitivity of the device
while still allowing larger metal particles to pass through the gap.
3.3.2 Viscosity Sensor Proof of Concept
Once the design was completed, preliminary proof of concept experiments were
conducted to determine the feasibility of the design. An inexpensive piezoelectric
diaphragm was needed for the proof of concept device. An inexpensive source of
such piezoelectric diaphragms is piezoelectric buzzers. Such buzzers are often found
in electronics devices such as computers and calculators. The piezoelectric
diaphragm chosen for this project (RadioShack PN 273-0073) is shown in Figure
Figure 3.24 RadioShack piezoelectric buzzer
These experiments began with the construction of the proof of concept device shown
in Figure 3.25. This device consists of two piezoelectric diaphragms mounted to an
acrylic frame. The two piezoelectric diaphragms were spaced approximately 2 mm
apart. As with the proposed device, the sensor operates by driving the top diaphragm
at some known frequency and measuring the frequency and amplitude of the
oscillation using the lower diaphragm.
Figure 3.25 Original viscosity sensor proof of concept test device
The first experiment using this device was to determine if the lower diaphragm could
measure the frequency and amplitude of the upper diaphragm. This experiment was
conducted in air. The upper diaphragm was driven using an HP 3311A Function
Generator. The voltage output of the lower diapragm was sent to a National
Instruments PXI-4472 board and the data was logged using the National Instruments
Lab VIEW graphical programming program.
The diaphragm was driven at various frequencies to determine the sensitivity of the
lower diaphragm to the frequency changes of the upper diaphragm. It was discovered
that the lower diaphragm was capable of measuring the frequency and amplitude of
the upper diaphragm for all frequencies tested (1 Hz 2500 Hz).
After the functionality of the device was tested, the device was submerged in 5W-30
oil and tested at a range of frequencies. The results of this test are shown in Figure
3.26. Note that the data labels at each point refer to the measured frequency of the
device when driven at the corresponding frequency.
Figure 3.26 Measured amplitudes of sensor diaphragm for a range of frequencies
Figure 3.26 shows a very close match between the driven frequency and measured
frequency of the device. Figure 3.27 shows the relationship between the frequency of
the driven diaphragm and the frequency of the sensing diaphragm. From Figure 3.27
it is clear that the relationship is nearly linear, with a maximum percent difference
between the frequencies of the two diaphragms at 6.25%. This shows that the
frequency of the device does not seem to be affected by the viscosity of the fluid.
This indicates that the viscosity of the fluid does not change the frequency transmitted
to the sensor diaphragm. Therefore, it was proposed that the viscosity of the fluid be
correlated to the measured amplitude.
Figure 3.27 Comparison of the driven diaphragm frequency to the sensor diaphragm
In order to test this proposed measurement, the device was tested in 75W-90 gear
lubricant oil. This oil has a higher viscosity than the 5W-30 oil, so a change in
amplitude given the same frequency is expected. A comparison of the device
performance in the two oils is shown in Table 3.9.
Table 3.9 Comparison of measured amplitude of sensor diaphragm for two oils of
Oil Frequency (Hz) RMS Amplitude (V) Viscosity @ 40 C (cSt)
5W-30 2500 0.011 63
75W-90 2500 0.0093 128.3
% Difference 15 104
Table 3.9 indicates that the viscosity of the fluid has an influence on the amplitude of
the transducer at a given frequency.
After completing the initial proof of concept test, the measurements for SAE 5W-30
and SAE 75W-90 were compared to a measurement of SAE 10W-30 over a
frequency range of 1000-2500 Hz. This test was conducted to determine if the device
could detect the change in viscosity between SAE 5W-30 and SAE 10W-30 oils. The
results of the test are shown in Figure 3.28.
Figure 3.28 Results of multi-grade oil sensor test
In Figure 3.28 it can be seen that the viscosity measurements of the SAE 10W-30 oil
do not match those of the other two oils. There are a few different explanations for
this behavior. First, this behavior could be due to some change in the sensor or the
conditions under which the device was tested. Another possible explanation for this
behavior is the lubricating oils themselves. Since these tests were conducted with
multi-grade oils, their behavior is different than single-grade oils. Since they are
designed to have a more constant viscosity over a wide range of temperatures than
single-grade oils, multi-grade oils do not make an ideal test fluid for the initial stages
of research for the device.
As a result, it was decided to switch to single-grade oils for the rest of the proof of
concept testing of the device. This eliminated many of the possible measurement
issues associated with multi-grade oils.
The next proof of concept test was conducted on 3 single-grade engine lubrication
oils: SAE 30, SAE 40, SAE 50. The tests were conducted using the same procedures
as the previous tests, however, for this test a frequency range from 1000 Hz to 3000
Hz was used. The results of this test are shown in Figure 3.29.
Figure 3.29 Results of single-grade oil sensor test
Figure 3.29 shows that for most frequencies tested, there is a clear delineation
between the measurements for the different oils. However, this relationship is not
consistent. Around 2000 Hz, the amplitude of the SAE 30 sample increased
significantly while the other samples experienced only a small increase. It is also
worth noting that for frequencies below 2000 Hz, the amplitudes of the samples are
proportional to their viscosities. In other words, the sample with the highest viscosity
(SAE 50) also had the highest amplitude. It would seem that this relationship should
be the reverse. A fluid with a high viscosity should have a greater damping effect on
the diaphragm and cause a small amplitude oscillation to occur. However, this was
clearly not the case in this test.
After the single-grade oil test it was decided to build a new device that more closely
resembled the proposed device design. This device is shown in Figure 3.30. The
biggest change in this device from the previous device is the addition of another
sensor pair. This was done in order to allow one of the sensors to be used for
capacitance measurement in addition to viscosity measurement.
Figure 3.30 Viscosity sensor proof of concept device version 2
The first test conducted with the 2nd version device was a test of the three single-
grade oils to compare to the previous test. For this test, both sensors in the 2nd
version device would be tested. The results of the testing are shown in Figure 3.31.
Figure 3.31 Results of single-grade oil proof of concept test using 2nd version test
From the results it can be seen that Sensor 1 exhibited an inversely proportional
relationship between viscosity and amplitude of the sensor diaphragm. This was the
expected behavior from the device. Sensor 1 also shows a virtually linear relationship
between viscosity and amplitude of the sensor diaphragm. Sensor 2 showed a
different behavior, however. Sensor 2 did not have an inversely proportional
relationship between the viscosity and the amplitude of the sensor diaphragm. This is
likely due to the diaphragm being damaged during the manufacture of the device.
In order to confirm the behavior measured by the previous device for multi-grade oils
was valid, a test was conducted with multi-grade oils in the 2nd version proof of
concept testing device. The results of the testing are shown in Figure 3.32.
Figure 3.32 Results of multi-grade oil proof of concept test using 2nd version test
The results from this test show the same inconsistency as the previous tests using the
original proof of concept device. Due to the results of this test, the practice of using
only single-grade oils was continued.
One of the potential drawbacks of the design is the proximity of the two measurement
sensors. The effects of the oil being oscillated by the viscosity sensor on the
capacitance sensor were unknown during the design of the device but were assumed
to be minimal if at all existent. In order to determine the validity of that assumption,
a test was conducted using the 2nd version of the proof of concept viscosity sensor.
For this test, one of the sensors was oscillated at 1500 Hz while the other sensor was
used to monitor the capacitance of the oil. Since the properties of the fluid between
the electrodes were not changing, there should not be any changes in the capacitance
measured between the plates. The test was run for 5 minutes. The results are shown
in Table 3.10.
Table 3.10 Results of capacitance measurement under influence of oscillation test
Time (sec.) Capacitance (pF)
From the data it is clear that the capacitance measurement was not affected by the
oscillations of the viscosity sensor. This validates the design assumption that the
viscosity sensor oscillation has little to no effect on the capacitance measurement
One issue that was discovered during the viscosity sensor proof of concept testing
was the rigidity of the sensor mounting. Neither of the first two viscosity sensor
proof of concept devices had rigid mounts for the frames that held the sensors. This
became a concern since the device is based on vibration and it was unknown if the
mounting configuration used was either positively or negatively affecting the
behavior of the device. As a result, a new proof of concept device was constructed.
This device is shown in Figure 3.33.
Figure 3.33 3rd version of the viscosity proof of concept device
The device shown in Figure 3.33 used a steel frame rather than a polymer frame to
increase the rigidity of the sensor mounting. In addition, the device was placed in a
steel container for testing to avoid any contamination testing in polymer containers
might create. The gap between the resonator and sensor was also changed to 1.27
mm in order to better simulate the actual dimensions of the proposed device. Once
the new device was constructed, a series of tests were conducted to determine the
response of the device when exposed to fluids of different viscosities. The results of
the tests are shown in Figure 3.34.
Figure 3.34 Results of initial testing with the 3rd version of the proof of concept
From the figure it is clear that the behavior of the sensor, while it is affected by the
viscosity of the fluid to which it is exposed, does not accurately determine the
viscosity of the fluid. This is evident by the large changes in the amplitudes of the
sensor at different frequencies as well as the measured amplitudes with respect to the
fluids of different viscosities. As has been stated previously, the viscosity and
amplitude of the sensor diaphragm should be inversely related to one another.
However, this was not the case in this test.
After completing the initial tests using the 3rd version of the device, a few issues were
discovered with the device. It was proposed that there were two possible sources of
the discrepancies of the initial tests. The first was that the piezoelectric diaphragms
used in the sensor were damaged. These sensors were carried over from the 2nd
version of the device and damage could have occurred while the sensors were being
removed from or installed on the device. The sensors could have also been damaged
due to repeated exposure to high temperatures from soldering leads to the
diaphragms. The diaphragms are very thin and very sensitive to heat. When exposed
to high temperatures for even a short amount of time the sensors can become
compromised. Repeated heat cycles could have much the same effect on the sensors.
The second possible source of the discrepancies shown in Figure 3.34 was the
mounting of the diaphragms to the frame. Since the frame was now made of a
conductive material (steel), the sensors could easily be shorted out if allowed to come
into contact with the frame. A non-conductive layer between the frame and
diaphragms would be necessary to ensure proper operation of the device.
In order to resolve the suspected issues with the device, a few changes to the device
were made. First, new pair of piezoelectric diaphragms were mounted in the device.
This would eliminate the possible issues with the old diaphragms. Second, the frame
was modified in order to mitigate the change of shorting out the sensors. Specifically,
the opening in the frame to which the diaphragms were mounted was made larger.
Also, a non-conductive spacer was placed in between the frame and diaphragms to
eliminate any metal-to-metal contact. Once these changes were made, the device was
tested in a range of lubricating oils to determine if the device could sense changes in
viscosity. The results of the test are shown in Figure 3.35.
Figure 3.35 Results of test using modified 3rd version of the proof of concept device
The results of this test demonstrate the expected behavior for the sensor. As the
viscosity of the fluid in which the sensor is located increases, the amplitude of the
sensor diaphragm decreases. This is clearly shown in Figure 3.35. The test also
continues the trend of other tests of the amplitude increasing between 1000-2000 Hz,
then dropping rapidly around 2500 Hz. This indicates a resonant frequency
somewhere around 1000-2000 Hz and as a result, the highest amplitude of the sensor
diaphragm. Due to the high amplitudes, this is also the location of the greatest
resolution for the sensor. As a result, any viscosity measurement using this sensor
should be made around the resonant frequency of the device.
At this point in the research it was decided that the proposed sensor design using
piezoelectric diaphragms was both a feasible and accurate way of measuring the
viscosity of a fluid. As a result, the proof of concept tests were ended and a test
device based on the original design as well as the lessons learned from the proof of
concept testing was created.
3.4 Device Construction and Validation
After completing the proof of concept tests for the device, a test device was
constructed. This test device design was used throughout the testing with only small
modifications. A model of the design is shown in Figure 3.36.
Figure 3.36 Oil condition sensor test device design
The device consists of two 12mm diameter piezoelectric diaphragms (Murata 7BB-
12-9) spaced 1 mm apart by a poly methlyl methacrylate (PMMA) housing. This
diaphragm was chosen due to its small size and resonant frequency (9.0 kHz). The
resonant frequency of the device is important due to the fact that it is the point at
which the viscosity measurement is made. A high resonant frequency allows the
sensor to operate at frequency ranges greater than those typically created by an
automobile under normal operating conditions. However, a resonant frequency that is
too high will not allow large oscillations to occur and could reduce the effectiveness
of the viscosity measurement device. Therefore it is very important to pick a
piezoelectric diaphragm that balances these two performance parameters. The
devices were constructed using the Epilog laser cutting machine. Figure 3.37 shows
the Epilog laser cutting machine.
Figure 3.37 Epilog laser cutting machine
Using the Epilog laser cutting machine, the device shown in Figure 3.38 was created.
Figure 3.38 Test device, version 1
After constructing the final device, a series of tests were conducted to assess
the performance of the device. Before testing the device on lubricating oils, a series
of tests were conducted using a standard set of silicone oils. This was done to test the
response of the device over a wide range of known viscosities. Three silicone oils of
different viscosities were tested in order to analyze the performance of the device.
The results of the testing are shown in Figure 3.39.
Â£ 0 0025
Â£ 0 002
< 0 0005
1 I t i i i i i j i t 1 1 L_ --245cSt
\ i 1 i i m 1 i 1 1 1 1 l : r i i rn 380 cSt A 550 cSt
t 1 1
[ l 1 1 \ ] \ \ \
1 i * 1 4
! L _j 1
0 1000 2000 3000 4000 5000 6000 000 8000 9000 10000
Figure 3.39 Results of Test device version 1 silicone oil test
From the results it can be seen that the sensor did not respond to the fluids of different
viscosities as expected. Due to this inconsistency, the test was repeated to determine
if some error in the measurement was present. The repeat tests yielded similar results
as the ones shown in Figure 3.39. This lead to a question, was there something wrong
with the new device that was constructed or was something wrong with the theory of
In order to answer this question, a new device was constructed. In this device, the
diaphragm was exposed to the fluid on both sides rather than one. This is illustrated
in Figure 3.40. These fluid passages were added to equalize the pressure between the
two sides of the diaphragm. It was proposed that on the diaphragm that had one side
sealed off from the fluid, the pressure of the air trapped behind the diaphragm was
dominating the behavior of the sensor rather than the viscosity of the fluid. By
exposing both sides of the diaphragm to the fluid, the same pressure exists on both
sides of the diaphragm. By exposing both sides of the diaphragm to the fluid, only
the properties of the fluid, specifically the viscosity, will affect the amplitude of the
Fluid Passage | Diaphragm | Fluid Passage
Figure 3.40 Piezoelectric diaphragm fixture with (b) and without (a) fluid passages
A new device was constructed using this design change. This device is shown in
Figure 3 .41.
Figure 3.41 Test device, version 2
Using the device shown in Figure 3.41, a new test using the silicone oils was
conducted. In addition to measuring the response of the sensor diaphragm to the
input, the temperature of the tests was also measured. The results of the test are
shown in Figure 3.42.
Figure 3.42 Results of test device version 2 silicone oil test
From the results shown in Figure 3.42 it is clear that the changes made to the
device, namely the change to the mounting of the diaphragms, were effective in
fixing the measurement issues. The test shows the sensor is capable of distinguishing
the viscosities of fluids through normal oscillation. Figure 3.42 also shows the
effective frequency range of the sensor. At frequencies larger than 5000 Hz, the
sensor can no longer differentiate between the fluids. This limits the measurement
range to frequencies lower than 5000 Hz.
Figure 3.43 shows the viscosity of the fluids compared to the amplitude of the sensor
diaphragm at 4000 Hz. This frequency was chosen due to its location on Figure 3.42
as a compromise in peak amplitude between the three fluids. Figure 3.43 shows the
relationship between the fluids is nearly linear. A linear relationship between the two
would indicate that the amplitude of the sensor diaphragm is proportional to the
viscosity of the fluid.
Figure 3.43 Comparison of viscosity measurement for 3 silicone oils
After verifying the performance of the device with silicone oils, lubricating oils were
tested. Three single grade lubricating oils (SAE 30, SAE 40, and SAE 50) were
tested to determine the viscosity of each using the device. The device shown in
Figure 3.41 was once again used for this test. Figure 3.44 shows the results of the
Figure 3.44 Results of viscosity test on three single-grade SAE oils
Figure 3 .45 shows the response of the sensor for each single-grade lubricating oil
across a frequency range from 0-10000 Hz. The response of the sensor for these oils
is much like the response for the silicone oils. There is very little crossover in
amplitude between the different oils. Once again the amplitude of the sensor in
different oils at the resonant frequency (4000 Hz) differentiates between the
viscosities of the fluids.
Figure 3.45 Comparison of viscosity measurement for three lubricating oils
Figure 3.45 shows the output of the sensor at 4000 Hz compared to the actual
viscosities of the fluids. This frequency was chosen due to the fact that the largest
displacements of the sensor occur at this frequency. This frequency would shift
depending on the dimensions of the actuator and sensor as well as the dimensions of
the cavities surrounding the diaphragms; however, the measurement technique
remains the same. As with the silicone oils in Fig. 3.43, the single-grade lubricating
oils in Fig. 3.45 exhibit a nearly linear behavior, indicating a linear relationship
between the viscosity of the fluids and the amplitude of the sensor diaphragm.
Figure 3.46 compares the measured viscosities of the single-grade SAE lubricating
oils to the measured viscosities of the silicone oils. A few interesting conclusions can
be made from the figure. The first observation that can be made is that the viscosities
behave as predicted across a very wide range of fluid viscosities. Low viscosities
yield high amplitudes while high viscosities yield low amplitudes. It can also be seen
that the two data sets are not linear with respect to one another. As the viscosity of
the fluid increases, the measured amplitude of the device become smaller and less
Â£ 0 06
9 0 04
2 0 03
< 0 02
0 200 400 600 800 1000 1200 1400 1600
Figure 3.46 Comparison of viscosity measurement for SAE lubricating oils and
silicone reference oils
86 c St
\ -SAE Oils Silicone ( Dlls
8 o o i/*.
The final design validation test conducted before the sensor performance was tested
in conditions found in the field was a test to determine the response of the sensor to
changes in temperature. It is well known that when a fluid is heated, the viscosity of
the fluid decreases. As a result, by heating the oil tested in the sensor, the viscosity
measurement of the fluid should change. More specifically, the viscosity of the fluid
should decrease. In order to determine if the sensor was capable of detecting changes
in viscosity due to temperature change, a series of tests were conducted. In these
tests, lubricating oil (SAE 30) was heated to 5 different temperatures (23 C, 30 C,
40 C, 50 C, 60 C) and the viscosity of the oil was measured. The results of the test
are shown in Figure 3.47.
Figure 3.47 Results of heated oil viscosity test
The results shown in Figure 3.47 show that as the temperature of the oil increases, the
viscosity of the oil decreases. This was the behavior expected in this experiment. In
fact, the measured data shows a nearly linear relationship between the measured
amplitude of the sensor diaphragm and the increase in temperature of the oil. This
test shows that the sensor can accurately monitor the viscosity of a lubricating oil in