Original Article
Factorial study of diesel engine oil
contamination effects on steel and
ceramic sliding contacts
P Ramkumar
1
, TJ Harvey
2
, RJK Wood
2
, AD Rose
3
,
DC Woods
3
and SM Lewis
3
Abstract
The present work investigates the effects of diesel contaminants and their interaction on tribological properties for
bearing steel (En31) and ceramic (Si
3
N
4
) sliding contacts using a factorial study. The contaminants are soot, sulphuric
acid, moisture and oxidation, and each contaminant has three different level of concentration (low, medium and high) in
the test matrix. The factorial test matrix consisted of 20 tests, constructed from a quarter fractional factorial test matrix
with four points at the medium values for the contaminants. Results from this matrix required six further tests to
elucidate aliased pairs of interactions using Bayesian model selection. A pin-on-disc tribometer was used to carry out all
the experiments. All tests were carried out under ambient conditions at 5 m/s sliding speed and contact stress of 1.5–
2.05 GPa to simulate a valve-train in a diesel engine with fully formulated heavy-duty diesel engine oil used as lubricant.
Four different tribological properties were studied. The factorial study showed that charge was influenced by tribocouple
material; the silicon nitride discs produced higher charge than steel discs. However, it was opposite for friction; the silicon
nitride disc gave lower friction and the pins showed higher friction than their steel counterparts. For wear scar and
temperature, soot contaminant was found to be important. The two important interactions were found for the charge
response, with the interaction between sulphuric acid and pin material being more important than sulphuric acid–
oxidation interaction. Similarly to charge, an interaction between sulphuric acid and pin material interaction was
found for friction.
Keywords
Bayesian model selection, ceramic, contamination, factorial design, sliding wear, soot
Date received: 16 December 2017; accepted: 24 July 2018
Introduction
In modern automotive engines, especially in diesel
engines, consumer demand for ever increasing service
intervals has led to longer oil drain periods.
Consequently, without improvements to lubricants,
this will lead to an increase in contamination levels
and in turn, reduced engine efficiency and increased
possibility of system failure (due to increases in viscosity
and the potential of oil starvation leading to scuffing).
There is a wide range of contaminants that are
encountered by diesel engines (see Table 1). From
this list, the top four (soot, water, oxidation and sul-
phuric acid) were chosen for study.
Soot is a major contaminant in diesel engines.
Elemental analysis of particulate matter shows that
oil particulates consist mostly of carbon (88.3%)
with other species
1
as shown in Figure 1.
Concentration levels as high as 9.3% soot by weight
have been reported in oils.
2
Oil can become contaminated by water through
leakage from weak seals and from moisture entering
into the lubricant stream from ambient sources
including combustion and condensation. Typically,
the amount of water contamination varies across the
range 0.2–10% by weight.
3,4
Lubricating oils may contain many corrosive spe-
cies such as naturally occurring sulphur compounds,
acidic combustion products (oxyacids of nitrogen and
1
Department of Mechanical Engineering, IIT Madras, India
2
Faculty of Engineering and the Environment, National Centre for
Advanced Tribology at Southampton (nCATS), University of
Southampton, Highfield, UK
3
Faculty of Mathematical Sciences, Southampton Statistical Sciences
Research Institute, University of Southampton, Highfield, UK
Corresponding author:
P Ramkumar, Indian Institute of Technology Madras, 408, Machine
Design Section, Adyar, Chennai, Tamil Nadu 600036, India.
Email: ramkumar@iitm.ac.in
Proc IMechE Part J:
J Engineering Tribology
0(0) 1–15
! IMechE 2018
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1350650118794730
journals.sagepub.com/home/pij
sulphur), acidic oil oxidation products and anti-wear
and extreme pressure additives. All of these chem-
ically react with metals and this reaction is enhanced
within the tribocontact. Abrasion can promote corro-
sive wear by removing surface tribofilms. Corrosion in
diesels is typically controlled by overbased detergents
that neutralise the (sulphuric) acids produced by com-
bustion. Increased wear in diesel engines using
exhaust gas recirculation method can be associated
with corrosion due to the formation of sulphuric
acid by reaction of sulphur oxides (formed during
combustion) with condensed water. When the acid
reaches the oil sump it reduces the total base
number (TBN) of the lubricating oil and thereby
affects the properties of the lubricating oil.
6
Sulphuric acid is also known to breakdown tribofilms
on cam material surfaces and causes extreme corro-
sive wear.
7
Akiyama et al.
7
showed how the TBNs of
oils decrease with increasing sulphur content in the
oils in the range 0.1–0.4% w/w. They also reported
experiments to evaluate the rate at which oils neutral-
ise the addition of 0.1 ml sulphuric acid. The type of
metal detergent present was found to greatly influence
neutralisation rates.
Oxidation is a natural phenomenon and is con-
sidered to be the leading indicator of oil degradation.
8
In this work, the model diesel lubricant was oxidised
for periods of 2, 4 and 10 h in a bulk lube oil oxidation
bench test as described by Yamaguchi et al.
9
In
this test the rate of oxygen uptake by a given
volume of oil, with added metal catalyst, is monitored
at constant pressure and temperature (171
C and
2 psig O
2
).
An experimental study using a two-level fractional
factorial design
10
was planned to investigate these
four contaminants (soot, oxidation, water and acid),
together with two tribocontact materials (steel and
silicon). A 20-run test matrix was employed, consist-
ing of a six-factor one-fourth replicate fractional fac-
torial design with four additional model checking runs
using mid-levels of the contaminants. Subsequently,
six further tests were performed to enable pairs of
interactions that were aliased in the initial experiment
to be distinguished.
Experimental
Tests were performed using an instrumented pin-
on-disc (PoD) tribometer, as shown in Figure 2.
A button-type inductive electrostatic sensor was
used in all experiments (details are given in Harvey
et al.,
11,12
Wang et al.
13
and Morris et al.
14
).
The sensor had a sensing area of 7.85 10
5
m
2
(10 mm diameter sensing face) and was positioned
approximately 0.5 mm above the disc surface. The
Table 1. Lubricant contaminants types and their origins.
5
Type Primary sources Major problems
Soot Combustion blow-by Interfere with additives, abrasive wear,
heavy deposits, oil thickening/gelation
Water Combustion blow-by,
coolant leakage
Metal corrosion, promotes lubricant breakdown
Oxidation of oil Thermal degradation/
contact with atmospheric air
Oil thickening
Acids Combustion blow-by,
lubricant breakdown
Metal corrosion, catalysis of lubricant breakdown
Metallic particles Component wear Abrasion, surface roughening leading to adhesion,
catalysis of lubricant breakdown
Metal oxides Component wear,
oxidation of metallic particles
Abrasion, surface roughening leading to adhesion
Minerals (i.e. silica sand)
and dirt
Induction air Abrasion, surface roughening leading to adhesion
Exhaust gases Combustion blow-by Acids promoting lubricant breakdown
Glycol Coolant leakage Lubricant breakdown
Fuel Blow-by-rich mixture Lubricant breakdown and dilution of engine oil
Carbon
88.3%
Sulphur
2.5%
Oxygen
4.9%
Nitrogen
0.5%
Hydrogen
2.6%
Metals
1.2%
Figure 1. Elemental composition of the typical diesel
particulate matter by mass.
1
2 Proc IMechE Part J: J Engineering Tribology 0(0)
button-type electrostatic sensor was connected to a
signal-conditioning unit, with switchable high- and
low-pass filters, set at 1 Hz and 10 kHz, respectively.
The incorporation of a high pass was designed such
that it monitored only dynamic charge events and was
insensitive to static/constant charges.
Electrostatic sensing technique is well established
for monitoring wear in lubricated contacts. The mea-
sured charge is directly indicating the deterioration of
the contact surfaces and it is sensitive to steel, ceramic
contacts and quality of lubricants.
15,16
Further, impe-
dance-based sensors are effective to detect changes in
oil quality and make them active lubrication
monitoring.
17
A force transducer was used for measuring friction
and an infrared pyrometer was used to monitor the
temperature of the wear track on the disc as close as
possible to the tribocontact. A linear variable differ-
ential transformer (LVDT) was used to monitor the
linear wear of the pin and disc.
A standard Zeitfuchs cross-arm type (ASTM D
445 and ISO 3104) size 6 viscometer, supplied by
Cannon Instrument Company (USA), was employed
to measure kinematic viscosity of the contaminated
test lubricants. The conductivity measurements were
performed with a Wolfson Electrostatic liquid L30
conductivity meter.
18
Test conditions
The tests used 6 mm diameter balls and 100 mm diam-
eter discs; material properties are provided in
‘Material selection’ section. All tests were carried
out under ambient conditions (temperature ¼ 18–
28
C, relative humidity ¼ 40–70%) at a sliding
speed of 5 m/s and a load of 30 N (2.05–2.55 GPa
initial Hertzian contact pressure, depending on mater-
ial combination). These conditions are mildly acceler-
ated compared with typical valve-train entrainment
velocities and contact pressures. Contact stresses in
the range 1.7–2.07 GPa have been reported for low-
emission diesel engines.
19
Entrainment velocities have
been reported as high as 7 m/s for a 2.0 l four-cylinder
GM Ricardo Hydra Gasoline engine
20
and 4.8 m/s for
a Sequence VE engine tappet system.
21
Material selection
The bearing steel used in testing is EN31 (B.S.
534A99, AISI E 52100 steel), which has an elemental
composition of 0.95–1.10% C, 0.10–0.35% Si, 0.40–
0.70% Mn, 1.20–1.60% Cr, remainder Fe. The silicon
nitride bearing balls were obtained from Spheric
Trafalgar Ltd and the discs were obtained from
H.C. Starck Ceramics, Germany. The material prop-
erties are shown in Table 2.
Test procedure
Prior to loading the disc was rotated (at the desired
speed) for 3 min and then the lubricant (see
‘Uncontaminated lubricating oil’ section) was applied
for 5 min, and background measurements were per-
formed during this period. An initial load of 1.76 N
was increased at a rate of 0.0714 N/s to a final load of
30 N over a period of 7 min; tests were run for a fur-
ther 1 h duration as a steady-state period.
Uncontaminated lubricating oil
The uncontaminated lubricating oil is a typical com-
mercially available heavy-duty diesel engine oil, con-
sisting of succinimide dispersants, sulphonate and
phenate detergents, secondary zinc dithiophosphate
Table 2. Properties and dimensions of bearing steel balls (pins) and discs used in experiments.
Material Bearing steel (EN31) Silicon nitride
Component Ball (pin) Disc Ball (pin) Disc
Density (g/cm
3
) 7.8 3.22–3.25 3.22
Poisson’s ratio 0.30 0.28 0.28
Young’s modulus (GPa) 210 300 290
R
a
(mm)
a
0.050 0.006 0.050 0.002 0.050 0.008 0.050 0.002
Hardness (Hv
50
)
a
980 44 218 5 1800 45 1550 26
a
Measured value.
Electrostatic sensors
LVDT
Pyrometer
Oil Spray
Force Transducer
Figure 2. Instrumentation employed in monitoring PoD
tribometer. LVDT: linear variable differential transformer.
Ramkumar et al. 3
amine antioxidant, phenol antioxidant, foam inhibi-
tor, viscosity index improver and group 1 base oils.
Contamination mixing procedure. The mixing of the con-
taminants into the uncontaminated oil is described
below. The order in which the contaminants were
added to both the oxidised and uncontaminated oil
(both supplied by Chevron Oronite) is the same as the
order in which they are presented below.
Oxidation: The oil was heated to temperature
171
C and constant O
2
pressure was maintained
in the presence of a metal catalyst for the
appropriate time.
Diesel soot mixing: The diesel soot was added to the oil
while mechanical stirring took place and homoge-
nised in an ultrason ic bath for an hour. The soot
was obtained from the overhead soot recovery
system of the Chevron Oronite engine testing facility
in Omaezaki, Japan. Average particle size was 10–
40 nm and chemical composition was 80% carbon,
17% oxygen, 1% nitrogen, 1% zinc and 1% sulphur.
Water mixing: The water was added to the oil during
mechanically stirring and homogenised in an ultra-
sonic bath for an hour.
Sulphuric acid mixing: The test oil was heated to 80
C
and the required amount of acid added drop by
drop, while the oil was continuously stirred using a
magnetic stirrer.
Processing of online data
Online measurements. Online measurements were taken
with a PC data acquisition system, using a Data
Translation DT321 16-bit eight-channel A/D card.
Data from the strain gauge, pyrometer, LVDT and
electrostatic sensor were acquired at a rate of 4 kHz.
These details were processed to produce an average
(root mean square for the electrostatic sensor) at a
rate of one point per second.
Specific wear rates (SWRs). The SWR was calculated
using post-test measurements (for details, see ‘Post-
test analysis’ section).
The volume loss of the disc, VL
disc
, was calculated
using equation (1)
VL
disc
¼ DA ð1Þ
where D is the wear track diameter of the disc and A is
the average track cross-sectional area.
The height of the worn ball (h) can be calculated by
equation (2)
h ¼ R R
2
0:25
d
2
0:5
ð2Þ
where R is the radius of the ball and d is the wear scar
diameter of the ball. From this the volume loss of the
worn pin, VL
pin
, can be obtained from equation (3)
VL
pin
¼
3
h
2
3r hðÞ ð3Þ
The SWR of the ball and disc can be calculated
from equation (4)
SWR ¼
VL
F SD
ð4Þ
where F is the force (load) and SD is the sliding
distance. The units used for SWR in this paper are
mm
3
/N m.
Steady-state measurements
Figure 3 shows a typical output from the online meas-
urements. From such outputs, several parameters are
recorded:
. the average for the steady-state period of coefficient
of friction ();
. the average for the steady-state period of
temperature;
. the difference between the averages for the steady-
state period temperature and room temperature;
. the average for the steady-state period of charge;
Post-test analysis
Two optical microscopes, an Olympus BH and
Olympus BH-2, were employed to measure and
image pin wear scars. In addition, a XYRIS
4000WL TaiCaan Technologies 3D profilometer was
employed to examine wear scars. A JSM 6500F ther-
mal field emission scanning electron microscope
equipped with energy dispersive X-ray microanalysis
spectrometer (Oxford Inca 300 EDS) was employed to
analyse wear mechanisms. However, for the statistical
analysis the wear mechanisms from the experiment
were not focused in this paper
Design of test matrix
Factors and levels. A total of six factors were chosen in
the experimental design and these included the four
contaminants (soot, oxidation, sulphuric acid and
moisture), as well as the disc and pin materials;
these have been designated from A to F, as outlined
in Table 3.
In addition, two uncontrollable covariates (viscos-
ity and conductivity) associated with the contami-
nated oil properties were measured. Neither of these
was found to be useful in the statistical modelling.
Experimental constraints. Due to availability of materials,
only tests with steel discs could be performed at first.
4 Proc IMechE Part J: J Engineering Tribology 0(0)
Therefore, the tests were performed in two batches, the
first with steel discs and the second with silicon nitride
discs. The number of tests was limited to 20.
Prior information on interactions. After discussions with
industrial collaborators, likely interactions between
the factors were identified as follows:
. Sulphuric acid and moisture content
. Soot and oxidation
. Soot and sulphuric acid
. Oxidation and sulphuric acid
Design of the first experiment. The constraint of
restricted randomisation was incorporated through
the use of a factorial split-plot design,
22
with one
‘whole-plot’ factor (disc material, labelled A) and
five ‘sub-plot’ factors (B–F).
As only 20 tests were available, and the full factor-
ial design with each factor taking two values would
require 2
6
¼ 64 tests, a one-fourth replicate fractional
factorial design was used with 16 runs. The fraction
was chosen so that the four interactions thought most
likely to be important could be estimated independ-
ently of the main effects of each individual factor and
independently of each other. However, these inter-
actions cannot be estimated independently of the
other interactions. These dependencies must be
taken into account in the analysis.
In addition, four tests were conducted at the mid-
points of the quantitative variables, one at each com-
bination of the disc and pin settings. The factors and
their levels are shown in Table 3 and the test matrix is
detailed in Table 4.
Results and discussion
The structure of the results section will be details of
results from primary testing, identification of second-
ary test matrix and final results (combination of all
factorial tests).
Primary testing
The results for each experimental run given in Table 5
are used in the statistical analysis.
Table 5 shows the overall response of charge, coef-
ficient of friction, wear scar diameter, SWR and
0.010
0.015
0.020
0.025
0.030
0.035
0.040
30 40 50 60 70 80
Charge / pC
30 40 50 60 70 80
0.05
0.06
0.07
0.08
0.09
0.10
µ
0 1020304050607080
24
28
32
36
40
44
48
Steady State
Steady State
Steady State
DT
Steady State
Time / minutes
Temperature / °C
30 40 50 60 70 80
0
5
10
15
20
25
30
35
wear rate (slope)
LVDT / µm
Time / minutes
Figure 3. Measurement of online steady-state parameters (Run 4 is shown for illustration). LVDT: linear variable differential
transformer.
Table 3. Factors and levels.
Factor 1 Midpoint 1
A Disc material Steel Silicon nitride
B Pin material Steel Silicon nitride
C Soot 0 5 wt% 10% wt
D Oxidation 0 5 h 10 h
E Sulphuric acid 0 1.25 mM 2.5 mM
F Moisture 0 1.25% 2.50%
Ramkumar et al. 5
