DoctorKnow® Application Paper

Title: CSI Model 5100

Source/Author:Drew Carey
Product: OilView
Technology: Tribology
Classification:

Introduction

The CSI Model 5100 measures oil quality giving an indication of the contamination and oxidation of
industrial oils. The CSI OilView software collects the raw data from the CSI Model 5100 and converts it
into an easy to understand form so that customers can make informed decisions about the health of
their machinery.

This paper attempts to clarify how the CSI Model 5100 works and how the data is interpreted. For
those who wish to read an in-depth technical description of how the unit works, an appendix is
provided for further study.

How the CSI Model 5100 Works

The CSI Model 5100 measures the dielectric property of oils over time. The dielectric property
changes as the oil becomes contaminated or oxidized. The CSI Model 5100 compares a reference
[good, clean oil] with a used oil sample. The dielectric property of a clean oil is a property that can be
measured easily like the freezing point, melting point, and boiling point.

Think of a capacitor. The capacitor is an electric device that "stores" an electric charge and then
discharges. A capacitor is made up of two thin metal plates. The space between these plates is filled
with an oil. The oil has a high dielectric constant. This means that the oil impedes the flow of electrons
from one plate to another. Over time, the charge builds up on one plate until the electrons can
overcome the high dielectric constant and then electrons flow from one plate to another. Hence, the
capacitor charges and discharges. The CSI Model 5100 measures this capacitance/dielectric constant
of the oil.

The CSI Model 5100 measures the dielectric constant using a grid made of concentric circles. Think of
the rings around a bull's eye target. The rings are metal and the space between fills with the oil.

As the instrument goes through a measurement cycle, several things happen. The oil comes into
contact with the grid and a dielectric measurement is made. The contamination particles in the oil are
pulled by gravity and begin to settle down to the grid surface. The contamination may be made up of
water, ferromagnetic particles, dirt, and non-ferrous particles. A ferromagnetic particle is one that is
attracted to a magnet. To speed up the settling of the ferromagnetic particles, the CSI OilView Model
5100 uses an electromagnet to pull the ferromagnetic particles to the grid. When any metal particle
touches the metal rings of the grid, it changes the dielectric reading. The instrument then reads a very
high dielectric. If we are using an electric magnet, the ferrous metal will settle quicker than other non-
ferrous particles such as brass or copper. The presence of ferrous particles (as well as other metal
particles) signify machine wear.

Since water is heavier than most oils, water will also settle to the bottom of the grid. If the water is not
chemically combined with the oil and the water droplets are relatively large, they will settle onto the
grid causing a high dielectric reading. Water that is emulsified [water that is broken into tiny droplets
suspended in the oil like fat in homogenized milk] causes the dielectric reading of the oil to change
over the time of the instruments measuring cycle.

Since both the ferrous metal and water particles seem to cause the same reading as they hit the grid,
how does the CSI Model 5100 tell the difference between the two contaminants? The answer is that
the CSI Model 5100 cycles the polarity of an electromagnet from north to south. This causes the
ferrous metal particles to "stand up" under one polarity and "lay down" during the other polarity. The
CSI Model 5100 looks at the change in the dielectric reading during the two polarities. If there is no
difference the particles are water. If there is a difference, the particles are ferrous.

The CSI Model 5100 measures the change of the dielectric over the sample cycle. The first dielectric
reading gives us a measurement of the oxidation of the oil. The settling of particles onto the grid over
time which changes the dielectric reading gives an indication of the types of contamination in the oil.
Metals in the oil such as copper, lead, etc. that are not attracted by a magnet are estimated from
changes in the capacitance of the grid.

The data collected by the CSI Model 5100 is stored in a personal computer for recall and display by
the OilView software. By looking at the graph of the dielectric versus time, we can tell what is
happening during the instrument's measurement cycle. Large peaks show the metal particles hitting
the grid. A steep slope indicates the presence of water. Once the dielectric graph is understood, then
a diagnosis is made.

This is similar to the process used in vibration analysis: the frequency plot gives information about the
machine's mechanical condition. In addition, the OilView software has a built in diagnostic system that
looks at the dielectric readings and alarm values to make a recommendation.

APPENDIX

Technical Questions and Issues

The CSI Model 5100 measures the time resolved dielectric spectroscopy of oils. The settling of
particles, under the influence of gravity, changes the oil's dielectric constant in the vicinity of the grid.
The settled ferromagnetic particles are magnetically cycled through an angle of 90 degrees altering
the grid capacitance.

This magnetic cycling allows the determination of iron in the oil. Other metals (such as copper, lead,
etc.) are also estimated from changes in capacitance of the grid. Generally, non-metallic materials
(such as metal oxides, silica etc.) cannot be reliably determined using the CSI Model 5100 because
they have a relatively low dielectric constant (4 to 10).

Dielectric constant

"So, what is this dielectric constant and why does anyone care?”

A dielectric constant is a fundamental property of a material such as boiling point, melting point,
viscosity, and refractive index. In fact, all these properties are strongly influenced by their dielectric
constants. Other synonyms for dielectric constant are relative permittivity and impedance.

"So what is this dielectric constant?"

A dielectric constant is a measure of the interaction of an oscillating electric field (changes sign from
positive to negative etc. at a certain frequency) with a molecule. A molecule contains a positive
charge, the nucleus, and the negative electrons which surround the nucleus. The electrons are
involved in forming chemical bonds. The negative electrons move near the speed of light
around/between atoms. The molecule produces its own electric field which interacts with the electric
field generated in the CSI Model 5100.

Infrared spectroscopy (IR) also measures the interaction of an oscillating electric field (infrared light)
with a molecule.

"This sounds like scientific trickery! I don't believe it."

A paper ( T Tjomsland et al. Comparison of Infrared and Impedance Spectra of Petroleum Fractions,
Fuel V(75) 1996 p 322-332) shows 16 of 17 parameters correlate between infrared modes and static
permittivity. This is a scientific paper in a refereed journal published by academics. Copies of this
paper are available from CSI.
The CSI Model 5100 measures the static permittivity of oils. Thus measuring the dielectric constant
with the CSI Model 5100 gives much the same information as a $30,000 FTIR (Fourier Transform
Infrared) spectroscopy instrument. Remember, IR spectroscopy is an oil industry standard for
measuring oil quality.

"I read somewhere that none of the 130 commercial oil analysis labs in North America are known to
use dielectric devices for monitoring the health of lubricating oils."

The oil labs are using scientific instruments developed 30 or more years ago for measuring chemical
parameters of oil (such as viscosity, TAN, TBN, metals IR etc.). The only new technique is the addition
of FTIR coupled with computers for ease of use. Measuring the dielectric constant with a CSI Model
5100 (at time = 0), gives almost the same information as FTIR.

Labs that certify transformer oils such as UPSI in Nashville, Tennessee, and PDMA in Tampa, Florida
measure dielectric constant and dielectric strength using ASTM procedures.

"You still have not explained what a dielectric constant means."

A dielectric constant (permittivity) represents the sum of two different properties of oils: the dipole
moment (small in mineral oils) and its "polarizability." When oils oxidize or the water concentration
increases, then the dipole contribution of the dielectric constant of the oil increases.

"So what is polarizability?"

Polarizability is a fundamental property of every atom or molecule and therefore all material objects.
All atoms are composed of a positive nucleus and negative electrons. Imagine a single hydrogen atom
(one proton and one electron) between two parallel plates. When you put an electric potential between
the two plates, the nucleus is displaced towards the negative plate and the electron is displaced
towards the positive plate. Oscillate this electric field and you can get the hydrogen atom to oscillate
like water in a bucket (or bathtub). The displacement is proportional to the electric field until dielectric
breakdown occurs. The magnitude of this proportionality is called the Polarizability.

:induced = "E

Where :induced represents the induced dipole moment, " is the polarizability, and E the electric field.
3
The polarizability " has dimensions of cm .

"What is a dipole moment ?"

In a compound like water (H2O), the oxygen atom shares electrons with the hydrogen atoms.
However, the oxygen atom holds these negative electrons more strongly than the hydrogen atoms.
Thus, the oxygen atom has a more negative charge and the hydrogen atoms more positive.

A dipole moment measures the center of gravity of the positive and negative charges. If these two
centers do not coincide, then the molecule is electrically nonsymmetrical and has a net polarity. The
dipole moment is defined as the magnitude of a unit charge e (in esu) times the distance r between
charge centers (in cm), and is expressed in Debyes.

When the dipole moment interacts with an oscillating electric field, the whole molecule has to rotate in
order to be realigned with the charge reversal in the capacitor. The molecule is several thousand times
more heavy than electrons and does not respond to high frequency electric fields.

"This sounds like scientific trickery! First you say its polarizability that contributes to permittivity, then
dipole moments. Isn't it convenient that you have two explanations for the same phenomenon?"

The Debye equation relates the permittivity (,) with the polarizability("), and dipole moment (:) of oils:

(, - 1)/(, +2) = (" + :2/3kT)(LD/3MW)

, = permittivity (dielectric constant) of oil
" = polarizability of oil
: = dipole moment of oil
k = Boltzman constant = 1.31*10-23 joules/degree Kelvin
T = temperature degrees Kelvin = 273 + degrees centigrade
L = Avogadro number = 6.02*1023 molecules of oil/mole
D = density of oil = grams/cm3

MW = molecular weight of oil = grams/mole

As you can see, this is not scientific trickery. Every term in measuring permittivity can be exactly
defined using the Debye equation.

"Well, no one ever uses or cares about permittivity in all the oil labs in North America."

Actually, a lot of them do without knowing it. Measuring the refractive index (n) of an oil gives a value
for the high frequency permittivity. The refractive index of an oil is used to measure oil quality:
, = n2

The refractive index measures the interaction of visible light with a material. Visible light contains
oscillating electric fields at a frequency of 1014Hz. At this high electric field frequency, the dipole
moment contribution to permittivity cannot respond because it requires the rotation of the whole large
oil molecule. However, the polarizability part of the equation (caused by less massive electrons
moving at nearly the speed of light causing an induced dipole moment) does respond The result is that
the polarizability (") is related to the refractive index via the Clausius-Mosotti equation:

" = (n2 -1)/(n2 +2)(3MW/LD)

MW = molecular weight of oil = grams/mole
L = 6.023*1023 molecules/mole
D = density = grams/cm3

People using refractive index to measure oil quality measure only the polarizability portion of the
permittivity (at optical frequencies). Note, the refractive index of materials shows a slight variation with
frequency. Usually refractive indices are reported at a particular wavelength (Sodium D line). The
dielectric constant does not vary as much as refractive index because it depends on the square root
term (n = ,1/2).

Stokes Law

"You said that the CSI Model 5100 performed time resolved dielectric spectroscopy. What do these
fancy words mean?"

The CSI Model 5100 measures changes in the dielectric in the vicinity of the grid caused by settling
particles. The changes in dielectric are measured over a specified time interval. The settling of a
material like water causes a change in dielectric because water has a much higher dielectric constant
(78) than oil (2.2). Secondly, metals that fall in the vicinity of the grid change the capacitance of the
grid (by changing the grid spacing) and this is shown as an apparent change in dielectric constant in
the time spectrum output of the CSI Model 5100.

"There are a lot of factors which can affect the CSI Model 5100 output aren't there?"

The settling of particles in a viscous medium is governed by Stokes law. When a small sphere falls
under the action of gravity through a viscous medium, it ultimately acquires a constant velocity:

V = 2ga2(d1 - d2)/90
V = particle settling velocity cm/sec
a = particle diameter in cm
2
g - gravitational acceleration 980 cm/sec
d1 = particle density in g/ml
d2 = oil density in g/ml
0 = viscosity in poise not centipoise (in order for the equation units to work)

The settling velocity determines how much suspended material in the oil falls on the grid. There are
only two factors which determine how much material settles on the grid, they are:

(1) The most important factor is particle size, large particles settle very much faster than small ones.
(2) The difference in density between the particle and the oil.

A hydrocarbon oil has a density of about 0.87 g/ml (0.083 to 0.92 mg/ml). An oil diluted with kerosene
has a density of close to 0.85 mg/l. If water is the particle suspended in this medium, the difference in
density between \water an oil is only 0.15 g/ml, whereas the difference in density between iron and oil
is about 7 g/ml. Copper settles a little faster than iron, and lead about 50% faster than iron.

"Can you be more specific?"

The recommended procedure with the CSI Model 5100 is to dilute the used oil sample with kerosene.
A 50:50 mixture of gear 680 (one of the highest viscosity oils in common use in industry) produces a
viscosity at 250o C of 22 cSt, whereas a C-46 oil produces a viscosity of 9 cSt at the same
temperature.

Let's look at iron first. The particles are assumed to be evenly distributed (shaken) in the oil sample,
and fall a maximum of 4cm in the 250 second test.

All the iron particles greater than 16 microns will reach the grid during the test for the 22 cSt oil,
whereas all the particles greater than 10 microns reach the grid for the 9 cSt oil. It is easy to draw a
family of curves to include every oil.

"Fine so what?"

Let's look at a real lab example. A gear box oil has the following measurements from an oil lab:

40o C viscosity = 407.1 cSt

100o C viscosity = 43.5 cSt
Iron = 35 ppm
Lead = 1 ppm
Copper = 14 ppm
Water (Karl Fischer) = not detected (< 300 ppm)
Particle size distribution: >2 :m = 105,552, >5 = :m = 84,503, >15 :m = 16,994 > 25 :m = 1583, > 50 :m
= 41, and >100 :m = 0.

We can make an approximate calculation of the amount of iron in the sample. We know the number of
particles that will fall on the grid and their size. Mathematically integrating this from the formula:

E Mass = E (number of particles)*(probability of hitting the grid)*(density)*(4/3)Br3

3.
The density of iron is 7.7 g/cm For the radius, we use the particle size (note that this is an
approximation, since we do not expect iron particles to be spherical). Solution of the above equation
shows about 4 mg of iron/copper falls onto the grid. Assuming a sample volume of 32 ml gives 140
ppm of metal. This agrees surprisingly well with the measured values for iron/copper/lead of 60 ppm
listed above.

"I notice that you use these 'indices' instead of real scientific values like ppm (mg/l)."
True, the oil lab spectrometer uses units like ppm, but it only measures particles less than 8 microns.
When you look at a case, you can see that it may miss quite a lot of metal particulates present in the
oil sample. The use of absolute units with a spectrometer requires the understanding that it applies
only to particulates in a certain size range (< 8 microns). Note: the metal additives in oil such as
calcium stearate and ZnDDP are quantitatively measured since they are soluble or highly dispersed.
The CSI Model 5100 measures the size of metal particulates that are greater than 8 microns.

Water

"I heard that the CSI Model 5100 measures water using a convoluted hodgepodge of science."

The science behind the CSI Model 5100 water measurement falls into three categories:

1) dissolved water,
2) emulsified water, and
3) free water.

We will discuss these three categories before analyzing their affects with the CSI Model 5100.

Dissolved water: Although oil and water do not mix, oils can (and do) contain a small amount of
dissolved water. The maximum solubility of water in paraffinic oils is a few ppm. Napthenic oils may
dissolve up to several hundred ppm water. The presence of polar additives in oil (most additives in oil
are polar) increases the solubility of water in the oil. If you dry oil (using a drying agent) and then
expose the oil to air, it will adsorb moisture because of the relative humidity in air. The dissolved water
content in hydrocarbon oils is relatively benign. The dissolved water is estimated as a change in
dielectric constant (time = 0) between a reference oil and a used oil. Note: the presence of
microemulsions will also be shown as a change in dissolved water.

Emulsions: When you mix a pure mineral oil (no additives) and water, the two liquids rapidly separate
into two phases. Water, which is more dense than mineral oil migrates to the bottom of the container
(with settling velocity determined by Stokes law) and the oil to the top of the mixture. The separation
into two phases is the most stable state in all oil water mixtures. However, emulsions perform a "trick"
which manages to keep emulsions stable indefinitely.

All emulsifying agents (which are frequent oil additives) have a similar generic structure. They contain
a long non-polar chain with a polar end. The non-polar chain is oil soluble, the polar end is water
soluble. Since 'like dissolves like,' one part of the molecule is oil soluble and the other part is water
soluble. An example of an emulsifying agent (also known as a surfactant, which means surface active
- +
agent) is an organic ionic salt, sodium stearate (C15H31CO2 Na ). The 15 carbon chain (C15H31) is
-
anchored in the oil phase and the carboxyl chain (CO2 ) in the water phase.

The positive sodium ions (Na+) are dispersed in the water phase to maintain overall electroneutrality.
Consequently, at the oil/water interface, negative charges exist due to the presence of the carboxyl
group. Thus two emulsion particulates (called micelles) repel each other at close distances. Since
there is an electrostatic repulsion to overcome for an emulsion to coalesce, emulsions can be stable
'indefinitely' (that is, for a long time).

Surface charge repulsions by micelles is the 'trick' that keeps emulsions stable indefinitely. Water in oil
emulsions become milky as the water concentration increases. The permittivity of water in emulsion is
likely different than neat water, because emulsions have a more polar ordered structure than neat
water.

There is a certain class of emulsions called microemulsions which have a different behavior in the CSI
Model 5100. Microemulsions are smaller (100 to 500 A0) than the wavelength of light, appearing
transparent and homogenous. Microemulsions do not settle in the CSI Model 5100 because of their
small size. Ambient thermal energy (Brownian motion) is enough to overcome the gravitational settling
affects of microemulsions in oil. Oil in water microemulsions may be formed by the addition of hexanol
to the emulsion.

Free Water: Water in oil which does not contain a micellar structure is free water. Water particles in
this form have little or no electrostatic charge (its zeta potential is small or zero) and consequently
falling water particles will coalesce into larger drops. The collision and coalescence of water particles
can be caused by two methods (orthokinetic and perikinetic encounters). Perikinetic encounters are
caused by Brownian motion of the particles. Perikinetic encounters are caused by hydrodynamic
forces of particles in fluids. Equations are available for determining the coalescence of particles as a
function of temperature, particle size distribution, viscosity, differential settling, and Van der Waals
forces. In the CSI Model 5100, the appearance of large droplets is the result of coalescence of free
water in the oil water mixture.

Water in the CSI Model 5100

"So shake the oil/water mixture in the CSI Model 5100, then what happens?"

The water settles according to Stokes law equation 5. Figure 3 shows the percent water settled as a
function of particle size. The figure gives values for two viscosities (22 cSt and 9 cSt) assuming a 1:1
dilution with kerosene and 680 or C-46 oil. As you can see only very large water particles settle on the
grid during the CSI Model 5100 test. Emulsified and free water will behave very differently in the CSI
Model 5100 and this difference is easy to see. Free water will form larger droplets (caused by
perikinetic and orthokinetic behavior of the water) that coalesce under gravitational settling, than
emulsified water. Thus the CSI Model 5100 test gives an indication of the presence of emulsifying
agents in water- this can be a very useful tool. Emulsified water will form smaller particulates that
settle more slowly on the grid.

"There is no real scientific basis for measuring water in the CSI Model 5100."

The composition of an emulsion can be found from the permittivity of the layer using a dielectric model
of a heterogenous system (Pal, R., 1994, Techniques for measuring the composition (oil and water
content) of emulsions - a state of the art review, Colloids Surfaces, A84 141-93). In this case, the
dielectric constant near the grid at the end of the test can be used to measure the volume fraction (N)
of the oil water mixture:

1 - N = (,disp - ,1)/(,disp - ,cont)*(,cont/ ,1)1/3 6

N = volume fraction of water at the grid,

disp = dielectric constant of dispersed phase = water = 80
,1 = dielectric constant of oil at end of test
,cont = dielectric constant of oil at start of test

Thus this equation can be used to estimate the volume fraction of water in the vicinity of the grid
during the CSI Model 5100 test.

All contents copyright © 1998 - 2006, Computational Systems, Inc.

All Rights Reserved.