Upload
s-s
View
213
Download
0
Embed Size (px)
Citation preview
http://pid.sagepub.com/Engineering
Engineers, Part D: Journal of Automobile Proceedings of the Institution of Mechanical
http://pid.sagepub.com/content/224/7/927The online version of this article can be found at:
DOI: 10.1243/09544070JAUTO1395
2010 224: 927Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile EngineeringC D Bannister, C J Chuck, J G Hawley, P Price and S S Chrysafi
Factors affecting the decomposition of biodiesel under simulated engine sump oil conditions
Published by:
http://www.sagepublications.com
On behalf of:
Institution of Mechanical Engineers
can be found at:Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile EngineeringAdditional services and information for
http://pid.sagepub.com/cgi/alertsEmail Alerts:
http://pid.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://pid.sagepub.com/content/224/7/927.refs.htmlCitations:
What is This?
- Jul 1, 2010Version of Record >>
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
Factors affecting the decomposition of biodieselunder simulated engine sump oil conditionsC D Bannister1*, C J Chuck1, J G Hawley1, P Price2, and S S Chrysafi2
1Department of Mechanical Engineering, University of Bath, Bath, Avon, UK2Ford Motor Company Limited, Dunton, UK
The manuscript was received on 2 October 2009 and was accepted after revision for publication on 8 April 2010.
DOI: 10.1243/09544070JAUTO1395
Abstract: Biodiesel is a renewable fuel derived from plant, waste, or algal oils. It is synthesizedby the transesterification of triglycerides with an alcohol to yield fatty acid alkyl esters. Theseesters are prone to oxidative deterioration, yielding a variety of products which increase theviscosity of the fuel beyond acceptable levels. A proportion of the fuel used will always find itsway into the vehicle’s lubricating oil with dilution becoming increasingly significant on vehiclesequipped with a particulate filter because late injections lead to increased wall wetting.Biodiesel will also accumulate in the lubricating oil but, unlike mineral diesel, the biodiesel willnot evaporate at the normal operating temperature of the oil and will instead accumulate. Asthe biodiesel oxidizes within the oil, degradation products will eventually form, leading to asignificant increase in the oil viscosity with a potential impact on base engine durability. Thisstudy investigates the relative effects of a number of factors likely to be relevant to biodieseloxidation within a simulated engine lubrication oil environment. The findings of this studysuggest that determining the point at which oxidation occurs is inherently difficult and theimpact of various factors can vary depending on the oxidation indicator examined. The mostsignificant factors in total oxidation were found to be the temperature and air flowrate withother factors, such as the presence of iron and pro-oxidant, having a significant impact onlyafter initial oxidation reactions had occurred.
Keywords: biodiesel, oxidation, engine sump oil, fatty acid methyl ester
1 INTRODUCTION
Fossil fuels are a finite resource sourced from
potentially unstable geopolitical areas, the combus-
tion of which creates large volumes of various
greenhouse gases. To combat these issues, various
alternative fuel sources, such as biodiesel, are being
sought. The term biodiesel refers to the fatty acid
alkyl esters derived from plant, animal, or waste oil
feedstocks and is produced by the reaction of the
glyceride starting material with an alcohol over an
acid or base catalyst [1].
This alternative fuel is increasingly becoming a
significant part of the transport sector with over
176106 t sold worldwide in 2008. Biodiesel, derived
from rapeseed, soybean, waste oils, or, more re-
cently, algal oils is composed of a range of saturated
and polyunsaturated esters and, on exposure to air,
these esters are prone to oxidative degradation [2].
The eventual products of this degradation are
polymeric gums and solid material that can, if
present in the fuel prior to combustion, have adverse
affects on fuel system performance or, commonly,
fuel starvation caused by clogged filters. Various
properties of biodiesel, such as a higher surface
tension, lower volatility, and higher density, lead to
the formation of larger drop sizes when injected into
the cylinder. This leads to more impingement on the
cylinder wall and, especially in the case of on-road
diesel vehicles fitted with diesel particulate filters
(DPFs) requiring regeneration via late-injection
strategies, to an increased amount of biodiesel in
the sump oil (often termed oil dilution) [3]. As bio-
diesel is generally not as volatile as diesel, little of
*Corresponding author: Department of Mechanical Engineering,
University of Bath, Bath, Avon, BA2 7AY, UK.
email: [email protected]; [email protected]
927
JAUTO1395 Proc. IMechE Vol. 224 Part D: J. Automobile Engineering
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
the biodiesel evaporates from the crankcase and
instead will start to degrade. This causes the vis-
cosity of the sump oil to increase significantly,
resulting in a loss of performance and an increase
in engine wear, and generally necessitates a pre-
mature oil change [4].
Currently, biodiesel is primarily produced from
rapeseed, palm, and soybean oil, although numerous
other vegetable oils as well as waste cooking oils are
also suitable feedstocks. Biodiesel produced from
tropical oils mainly consists of saturated esters such
as palmitic acid (C16:0), biodiesel produced from
rapeseed and jatropha oils mainly consists of
monosaturated fatty acids, whereas biodiesel pro-
duced from soybean, sunflower, and corn oils is high
in polyunsaturated fatty acids such as linoleic acid
(C18:2) [5, 6].
Numerous studies have been undertaken into the
oxidation of biodiesel. Originally studies detailed the
degradation of normal biodiesel under storage con-
ditions over a period of one year [7, 8] Although
arguably the most realistic test possible, the extended
timeframe of this type of study is clearly not practical
in assessing the oxidative stability of a fuel after
production for certification purposes. For this, many
studies have used the Rancimat test or oxidative
stability index to evaluate the stability of the biodie-
sel samples [9–13]. In these tests the oxidation is
monitored by assessing the acidity of the volatile
oxidation products. It is also possible to link oxidation
to a change in the viscosity, density, acid value, and
peroxide value of the biodiesel sample [14, 15], as well
as measuring the decline in the antioxidant level [16–
18]. More in-depth studies have employed various
analytical techniques such as nuclearmass resonance
(NMR), mass spectrometry, Fourier transform infra-
red spectroscopy, and gas chromatography–mass
spectrometry to elucidate further the mechanism of
degradation and to examine the formation of the
various oxidation products [19–21]. An overview of
themajor oxidation process is presented in Fig. 1. The
fatty acid alkyl chains have variable levels of un-
saturation. The susceptibility of the different fatty
acid esters to oxidation differs according to the
number of double bonds in the sample. Polyunsatu-
rated compounds are more prone to oxidation than
are the monounsaturated esters, which in turn are
more reactive than the saturated esters [19]. Oxi-
dation happens via a radical mechanism, where a
hydrocarbon free radical is formed on the bisallylic
carbon, the double bonds which lie around this
radical are able to reorganize (isomerize) to a more
stable structure, and the resulting radical species then
reacts with oxygen to form a peroxide species [22]. As
the most likely site for this to happen is the bisallylic
carbon, it is the triunsaturated (three-double-bond)
components that are more susceptible to attack than
are the diunsaturated components. The recently
formed peroxide species can then proceed to react
further, forming acids, ketones, and aldehydes. It is
also possible for the peroxide to react with another
fatty acid chain to form oligomers (chain consisting of
a limited number (10–100) of monomer units), which
eventually are deposited in the oil or react with fatty
acid methyl ester (FAME), resulting in peresters and
alcohols [21].
In this study, a technical-grade FAME dissolved in
dodecane was oxidized over 6 h in the presence of
iron and a fuel-borne pro-oxidant usually added to
aid particulate filter regeneration. The levels of these
factors were investigated as well as the flowrate into
the system, the reaction temperature, the amount of
FAME in the dodecane, and the type of FAME. 1H
NMR was used to analyse samples from the reaction
mixture with the pH of the collected volatile com-
ponents also being measured. As biodiesel derived
from natural sources contains antioxidants (such as
vitamin E) the examination of degradation would be
difficult over a feasible NMR timescale. To address
this issue, model solutions, made up of the pure
FAME starting materials, were used and a radical
initiator (2-ethylhexyl nitrate (2-EHN)) was added.
Fig. 1 Simplified flow diagram of the major oxidation process in the degradation of biodiesel
928 C D Bannister, C J Chuck, J G Hawley, P Price, and S S Chrysafi
Proc. IMechE Vol. 224 Part D: J. Automobile Engineering JAUTO1395
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
2 EXPERIMENTAL METHODS
2.1 Materials
Dodecane, methanol (reagent grade; 99.8 per cent
pure minimum), iron filings, methyl oleate (oleic
acid methyl ester (OAME); technical grade; 70 per
cent pure minimum) and linoleic acid (technical
grade 74 per cent pure minimum) were purchased
from Sigma-Aldrich Chemicals. In order to promote
DPF regeneration at lower temperatures, a fuel-
borne organometallic pro-oxidant has historically
been used to dose the fuel [23, 24]. The impact of a
common cerium-based pro-oxidant (referred to as
Pro-ox in the tables and figures) is investigated.
Linoleic acid was converted to methyl linoleate
(linoleic acid methyl ester (LAME)) according to
standard literature procedures [25]. 1H NMR analysis
of the technical-grade methyl oleate and linolenic
acid was undertaken according to the method
reported by Knothe and Kenar [26] with the fatty
acid profile of methyl oleate revealed to be 78 wt %
OAME, 13 wt % LAME, and 9 wt % saturated methyl
esters. The fatty acid composition of LAME (techni-
cal grade) was found to be 26 wt % OAME, 70 wt %
LAME, and 4 wt per cent saturated esters. None of
the FAME samples was treated with additional
antioxidants and all the FAME samples were stored
at 28 uC for the duration of the testing to minimize
oxidation during storage.
2.2 Test procedure
A sealed three-necked flask was filled with 100ml of
the relevant FAME sample in dodecane, which was
used as a simplified substitute for engine lubricating
oil. Engine oil itself is a complex mix of hydro-
carbons with various chain lengths, as well as a
number of additives such as viscosity modifiers and
detergents. Analysis of biodiesel decomposition pro-
ducts would be substantially more complicated if
engine oil was used owing to interference between
the NMR signatures of the oxidation products and
oil constituents. The use of dodecane, while not
corresponding to the main constituents of engine
oil, provided a hydrocarbon environment with a
clearly discernible NMR peak which would not mask
the presence of oxidation products. The flask was
placed in a silicone oil bath heated to the required
temperature. The solution was mechanically stirred
at a constant rate while air was bubbled into the
solution via a needle inserted through a sealed
rubber stopper. Subsequent volatile oxidation pro-
ducts, namely organic acids, were collected, via a
reflux condenser, into a vessel containing deionized
water. 1ml samples were taken from the reaction
mixture at 0min, 15min, 30min, 45min, 60min,
120min, 180min, 240min, 300min, and 360min with
a syringe. The pH of the water was also measured
at these time intervals (Fig. 2).
The samples taken were dissolved in deuterated
chloroform and an NMR spectrum was taken. The
spectrum was referenced to the residual solvent
peak. The 1H NMR spectrum was analysed according
to the method set out by Fang and McCormick [21]
in order to obtain concentrations for various oxida-
tion indicators. 1H NMR is a powerful technique
used in quantifying, as well as qualifying, chemical
compounds. A sample is held in a uniform magnetic
field; an electromagnetic wave, normally of a radio
frequency, is applied which causes the nuclei to
absorb the energy, and this energy is radiated back
out at a specific resonance as the protons relax to
Fig. 2 Schematic diagram of the apparatus set-up
Decomposition of biodiesel under simulated engine sump oil conditions 929
JAUTO1395 Proc. IMechE Vol. 224 Part D: J. Automobile Engineering
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
their original energy state. This resonance is in part
relative to the specific molecular environment that
the proton exists in, whereas the intensity of the
peak is directly proportional to the number of pro-
tons. By this method the conjugation of the double
bonds (6.0–6.6 p/min), the formation of peroxides
(3.9–4.2 p/min) and aldehydes (9.4–10 p/min), as
well as the reduction in the amount of double bonds
(5.5 p/min) and the reduction in bisallylic sites
(2.7 p/min) could be quantified.
2.3 Experimental design
The first run was undertaken under the most severe
oxidative conditions. Subsequent experiments were
undertaken, reducing the impact of one of the
variables (Table 1).
Within a running engine, the oil will experience a
number of different environmental conditions such
as cyclic temperature changes, exposure to metals,
and various levels of aeration. The oil temperature
increases as it passes through the bearings and is
sprayed onto the underside of the pistons to provide
cooling. The oil then experiences a reduction in
temperature as it is passed through a heat exchanger
and is cooled by the engine coolant. The general
operating temperature range of the oil is between
90 uC and 140 uC, although higher local temperatures
can be experienced under certain tribological re-
gimes. The temperatures chosen for this study
represent the worst-case operating condition of
140 uC and a 50 per cent reduction corresponding
to a temperature just below the normal operating
range.
As the oil is pumped around the engine circuit,
there are a number of opportunities for aeration to
occur. The agitation of the oil via chain-driven front-
end ancillaries, as well as the presence of balancer
shafts, lash adjusters, and piston cooling jets can
all contribute to the total volume of air held in
suspension, or dissolved within the oil. In compar-
ison with the Rancimat test, a relatively low air
flowrate of 25 l/h was chosen for this study to
represent an engine speed condition with moderate
oil agitation.
As well as the environmental factors to which the
oil is exposed, it comes into contact with various
materials within the engine. Metals are known to
catalyse oxidation reactions with copper, iron, alu-
minium, and zinc present in abundance within
the oil circuit. Although copper is often considered
the most effective catalyst in the oxidation process,
ferrous material is present in far larger quantities
and, as such, was included as a factor in this study.
Although 25 per cent dilution rates, while very
high, are not impossible to observe under certain
conditions [27], 50 per cent is significantly more
dilution than is likely to occur during normal vehicle
use. Unfortunately the techniques used to determine
the products of degradation were not sensitive
enough to detect, without significant error, the
decomposition products formed from the biodiesel
unless it was present at these elevated levels.
3 RESULTS AND DISCUSSION
3.1 Primary oxidation reaction
3.1.1 Total number of double bonds
Other studies have shown that, in the general
oxidation of FAME, the number of double bonds
decreases throughout the reaction, eventually form-
ing saturated oligomers and volatile oxygenated
species [21]. For this reason the decrease in the
number of double bonds was measured over the
reaction time with the aim of quantifying the general
rate of oxidation (Fig. 3).
Reducing the amount of pro-oxidant had a similar
effect on the oxidation as removing the oxidant
altogether and in both reactions the total amount of
double bonds having reacted is less than in the full
baseline testing run. Removing the iron filings from
the reaction produced a similar level of double
bonds to removing the pro-oxidant; however, a 50
Table 1 List of experiments
Run Dodecane (ml) FAME (ml) Iron (g) Pro-oxidant (ml) 2-EHN (ml) Temperature (uC) Air flowrate (l/h) Sample labelling
1 50 50 (LAME) 0.445 40 2.33 140 25 FULL2 50 50 (LAME) 0.222 40 2.33 140 25 1/2 Iron3 75 25 (LAME) 0.222 20 1.17 140 25 B254 50 50 (LAME) 0.445 20 2.33 140 25 1/2 Pro-ox5 50 50 (LAME) 0.445 40 2.33 140 12.5 1/2 Air6 50 50 (LAME) 0.445 40 2.33 70 25 1/2 Temp7 50 50 (LAME) 0.445 0 2.33 140 25 No Pro-ox8 50 50 (LAME) 0 40 2.33 140 25 No Iron9 50 50 (OAME) 0.445 40 2.33 140 25 OAME
930 C D Bannister, C J Chuck, J G Hawley, P Price, and S S Chrysafi
Proc. IMechE Vol. 224 Part D: J. Automobile Engineering JAUTO1395
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
per cent reduction in the iron filings had little effect
when compared with the full test. The largest effects
on the level of double bonds are the air flowrate and
the temperature, which, coincidently, lead to similar
levels of double bonds throughout the reaction time.
3.1.2 Bisallylic site
The auto-oxidation reaction follows a radical me-
chanism and the most favoured site for radical
formation is the carbon group situated between two
double bonds, namely the bisallylic site. On forma-
tion of the radical, the double bonds immediately
reorder into a more stable conjugated structure. The
rate of radical formation can be measured by observ-
ing the decrease in the NMR signal relating to the
bisallylic protons (Fig. 4).
Removing the iron or the pro-oxidant from the
sample slows the radical formation over the course
of the reaction when compared with the full test. A
similar level of decrease is observed when the pro-
oxidant is reduced by half. There are considerably
fewer bisallylic sites available in technical grade
OAME and the B25 blend; however, beyond this
there is no discernible change in the rate of decay
compared with the full condition. The largest
observable effects on the level of bisallylic sites are
the air flowrate and the temperature of reaction,
with the rate of oxidation being slower and the final
level of bisallylic sites higher than when any other
factor is removed. The rate of bisallylic decay is fairly
smooth under all examined conditions, suggesting
that the highly variable rate of conjugation forma-
tion is caused by the latter oxidation reactions.
3.1.3 Conjugated double bonds
The conjugated radical reacts with oxygen to form
peroxides; these peroxides are still conjugated,
however, and the conjugated bonds decompose in
subsequent chemical reactions to give smaller-chain
oxygenated compounds. The observed conjugated
bonds in the reaction are a balance between the rate
of formation (which as seen above is reasonably
smooth) and the rate of the numerous chemical
reactions involved in their decay. All these rates will
Fig. 3 Decrease in the number of double bonds throughout the reaction time. All trend lines aresecond-order polynomials and have R2. 0.99
Decomposition of biodiesel under simulated engine sump oil conditions 931
JAUTO1395 Proc. IMechE Vol. 224 Part D: J. Automobile Engineering
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
vary greatly over time depending on the reactions of
the various oxidation products formed and the loss
of volatile products.
Figure 5 shows the level of conjugated double
bonds observed throughout the reaction time.
Reducing the amount of iron or pro-oxidant has
little effect on the number of conjugated bonds in
the reaction. However, after 240min, there is a
substantially larger amount of conjugated bonds
when all the pro-oxidant has been removed. There
are approximately half as many conjugated bonds
when the pro-oxidant is reduced by 50 per cent,
suggesting that the pro-oxidant has a large effect on
the breakdown of the oxygenated FAME later in the
oxidation process.
Removing iron from the reaction vessel has a
similar effect to removing the pro-oxidant, and there
are a significantly larger proportion of conjugated
bonds after 6 h. Reducing the metal by 50 per cent
has very little effect compared with the full run; this
follows the consensus that iron behaves catalytically
in the oxidation of biodiesel and was in abundance.
The temperature of the reaction and the air flowrate
in the reaction vessel also have significant impacts
on the conjugation of the methyl linoleate and,
although the initial levels of conjugation are similar,
the numbers of conjugated bonds in the two runs
are much higher at the end of the reaction than in
the full test. The rates of change in the levels of
conjugated bonds in the B25 and full test are highly
similar, suggesting that the blend level has little
effect on the primary reaction mechanism other
than the fact that there is less oxidizable material.
3.2 Subsequent oxidation products
3.2.1 Soluble oxygenated product formation
The newly conjugated radical species will then react
with oxygen to form peroxides and hydroperoxides.
Subsequently these peroxides will break down
through numerous reactions to form aldehydes,
ketones, acids, and water. Because of the complexity
of the 1H NMR spectrum and the relatively small
amounts of soluble organic oxidation products
formed, it was impossible to measure the level of
ketone or acid formation; however, it was possible to
quantify the levels of peroxide and aldehydes pres-
ent (Fig. 6).
Fig. 4 Decrease in the bisallylic protons throughout the reaction. All trend lines are second-order polynomials and have R2. 0.99
932 C D Bannister, C J Chuck, J G Hawley, P Price, and S S Chrysafi
Proc. IMechE Vol. 224 Part D: J. Automobile Engineering JAUTO1395
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
There was very little peroxide observed under the
full testing procedure, suggesting that, under the
conditions imposed, the peroxide is degrading into
the various decomposition products quickly. No
clear trends were observed for the levels of alde-
hydes or peroxides in the reaction mixture under any
of the testing conditions. A large number of reac-
tions are taking place in the formation and decay
of both of these species and, as the rates of all these
reactions vary, the total quantity of peroxides and
aldehydes present is cyclic, allowing no clear inter-
pretation of the results.
3.2.2 Volatile oxygenated products
Peroxides and aldehydes eventually break down into
volatile acids. The volatiles given off throughout the
experiments were passed through distilled water and
the pH of this solution was measured (Fig. 7).
On removal of half the iron, using B25, or
substitution of the LAME with OAME the final pH
values are all similar to those of the full oxidation
run. This suggests that, in the production of volatile
acids, 50 per cent of the iron is having the same
effect as when the full amount is present. There is
little observable change in the pH when the blend
level is reduced or when OAME is used in place of
LAME. The temperature and the air flowrate have
the largest effect in the production of volatiles,
closely followed by removing all the pro-oxidant or
iron. Reducing the pro-oxidant, rather than remov-
ing it entirely, results in a pH value that falls between
the full and the no pro-oxidant reactions.
3.3 General oxidative trends
3.3.1 Oxidation rate
As with Fig. 3, Fig. 8 shows the decrease in the NMR
signal representing double bonds for the full testing
procedure. As discussed previously, the oxidation of
biodiesel is complex with primary, secondary, and
tertiary reactions occurring simultaneously. For this
reason it was decided to split the experimental data
into three smaller time periods with the intention of
gleaning additional information pertaining to the
Fig. 5 The number of conjugated double bonds observed throughout the reaction. All trend linesare second-order polynomials and all R2 values are greater than 0.90 unless statedotherwise
Decomposition of biodiesel under simulated engine sump oil conditions 933
JAUTO1395 Proc. IMechE Vol. 224 Part D: J. Automobile Engineering
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
impacts of the factors under investigation on
primary and secondary oxidation processes. The
data were split into three time periods, namely 15–
60min (initial), 120–240min (midtest), and 300–
360min (final), as shown in Fig. 8, with a linear
least-squares fit applied to each. The gradient of the
fit line represents the rate of change in the number
of double bonds (termed the oxidation rate) during
the initial, midtest, and final phases.
Figure 9 shows the results of a partial least-
squares multi-variate analysis examining the impact
of each factor on the rate of change in the number of
double bonds. The horizontal axis in Fig. 9 repre-
sents the magnitude of the coefficients attributed to
Fig. 7 pH of the deionized water, after the full 6 h reaction time. Shaded areas show significantlyhigher values than does the FULL test run
Fig. 6 The level of aldehydes and peroxides observed for the reactions based on the iron leveland environmental factors
934 C D Bannister, C J Chuck, J G Hawley, P Price, and S S Chrysafi
Proc. IMechE Vol. 224 Part D: J. Automobile Engineering JAUTO1395
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
each factor in the equation for the fit obtained
during multi-variate analysis. The greater the mag-
nitude of the coefficient, the greater the impact that
factor will have on the output, in this case the rate
of decrease in the number of double bonds. In
addition, interaction terms are also presented to
demonstrate whether combinations of factors cause
a greater or smaller impact than would be expected
by the summation of their individual constituents.
It can be seen in Fig. 9 that, of the primary (first-
order) factors, the temperature has the most domi-
nant effect on the initial rate of oxidation. An
increase in temperature causes the gradient of the
fit lines, shown in Fig. 8, to become more negative;
this implies a faster rate of decrease in the number of
double bonds or an increase in the oxidation rate.
Similarly, the air flowrate through the sample and the
saturation of the FAME also had significant impacts
on the oxidation rate, causing an increase in oxida-
tion at higher air flowrates and higher levels of
unsaturation. Interestingly, an increase in the quan-
tity of iron added, normally considered to catalyse
the oxidation process, actually led to a decrease in
the rate of oxidation during this initial period of the
test.
More significant than the primary factors are the
interactions between them, with combinations such
as an unsaturated FAME with a high temperature or
air flowrate vastly increasing the rate of oxidation. As
with the primary factors, the interaction between a
high temperature and air flowrate also resulted in an
increased rate of oxidation.
The second subplot of Fig. 9 shows the impact of
each factor and interaction on the rate of oxidation
during the middle of the test (120–240min). As with
the initial data, the air flowrate remains significant
during this portion of the test but, unlike initially, the
pro-oxidant concentration becomes dominant. This
suggests that while the pro-oxidant may have
relatively little impact on primary oxidation, it does
increase the rate of secondary oxidation processes.
The third subplot of Fig. 9 shows the impact of
each factor and interaction on the rate of oxidation
during the final phase of the test (300–360min). In a
reversal of trends seen during the previous time
periods, many of the factors which had previously
increased the rate of oxidation actually result in a
reduction in rate. For example, an increase in the air
flowrate during the initial and middle sections of the
reaction caused a large increase in the rate of
oxidation; however, during the final phase of the
test a higher air flowrate caused a reduction in the
rate. This can be explained by considering the test as
a whole, with a finite number of double bonds being
oxidized at a given rate. If the initial rate is high, as
the number of bonds remaining decreases, so too
does the rate of oxidation of those remaining bonds.
This effect is highlighted by the increasing impor-
tance of the blend ratio as the test time increases,
with higher blends sustaining higher oxidation rates
near the end of the test. Similarly, if the initial rate is
low, such as in the case of iron content, then the
oxidation rate may increase as there are still suf-
ficient double bonds available for oxidation.
Fig. 8 Reduction in double bonds with test time for the FULL sample
Decomposition of biodiesel under simulated engine sump oil conditions 935
JAUTO1395 Proc. IMechE Vol. 224 Part D: J. Automobile Engineering
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
Fig. 9 Impacts of factors on the initial, midtest, and final rates of change in the number ofdouble bonds. FAME(Mono) represents when OAME was substituted for LAME within theFAME
936 C D Bannister, C J Chuck, J G Hawley, P Price, and S S Chrysafi
Proc. IMechE Vol. 224 Part D: J. Automobile Engineering JAUTO1395
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
When Fig. 9 is examined as a whole, it is possible
to observe how the balance of factor importance
shifts as the reaction progresses. It should be noted
that the axis scales are not the same but instead
decrease with time, suggesting that the impacts of all
factors diminish gradually. Generally speaking, the
rate of a reaction, under constant environmental
conditions, is dependent on the relative concentra-
tions of the reactants and the products. As the
reaction progresses, the concentrations of products
increase and this results in a reduction in the rate of
reaction. This scenario can be clearly observed in
Fig. 9 when examining the effects of factors such as
the temperature and air flowrate. Higher tempera-
tures and air flowrates result in significantly in-
creased rates of reaction in the early stages of the
test but a complete reversal by the end as the
concentration of reactants has decreased and the
products increased. In contrast, it is suggested that
the presence of iron and the cerium-based pro-
oxidant only becomes significant for secondary
oxidation reactions once the concentrations of the
products of primary oxidation reactions have in-
creased sufficiently.
It should be noted, however, that the data
presented in Fig. 9 are intended to highlight the
impact of the different factors on the rate of
oxidation as the reaction progressed and does not
necessarily indicate the degree of oxidation or the
quantities of oxidation products which have been
produced.
3.3.2 Relative impact on oxidation
As has been discussed in sections 3.1 and 3.2 the
oxidation of FAME is a complex process with varying
interpretations of what constitutes ‘oxidized’. In
practice, limits are often set for acceptable levels of
viscosity, acid value, peroxide value, and induction
time; however, owing to the cyclic nature of the
formation of some products (such as peroxide
shown in Fig. 6), these measures may not be suffi-
cient to indicate the degree of oxidation. In section
3.3.1 the impacts of different factors on the rate of
oxidation were examined but, in order to quantify
how far the oxidation reactions have progressed, it is
necessary to define oxidation as either the total
decrease in initial reactants (i.e. FAME) or the
absolute increase in oxidation products.
Table 2 summarizes the relative impact on FAME
oxidation as determined by examining five different
indicators of initial reactant reduction or product
formation.
1. Final pH value, i.e. the pH of the distilled water
containing volatile compounds at the end of the
test period. This is a useful indicator as it will
quantify the amounts of volatile compounds
released during the course of the reaction in a
similar way to the Rancimat test.
2. Percentage decrease in the number of double
bonds, i.e. the decrease in the double-bond
NMR equivalent at the end of the test period as
a proportion of the value at the start of the
experiment. The oxidation process involves radi-
cals interacting with double bonds within the
FAME molecule, causing them to reorganize into
a more stable compound. When this happens, the
number of carbon–carbon double bonds is de-
creased. The percentage change in the number of
double bonds should give an indication of the
degree of oxidation as well as accounting for any
differences in initial levels.
3. Absolute decrease in the number of double bonds,
i.e. the absolute reduction in NMR equivalence
between the beginning and end of the experi-
ment. As with the percentage decrease in the
number of double bonds, the absolute decrease
should also give an indication of the total
progression and rate of the oxidation process;
however, this measure may be more susceptible
Table 2 Factor oxidation rank orders based on different indicators (using the sample labelling in Table 1). Theoverall rank order was obtained by averaging the rank values for all indicators
Final pH value
Percentagedecrease in thenumber of doublebonds
Absolute decreasein the number ofdouble bonds
Final conjugatedvalues
Averagealdehydevalue
Overall rank order
Averageranking
More oxidation 1/2 Iron B25 FULL B25 1/2 Iron FULL 2.2
Q
FULL FULL 1/2 Iron OAME FULL 1/2 Iron 2.2OAME OAME 1/2 Pro-ox 1/2 Iron 1/2 Pro-ox B25 3.8B25 1/2 Iron No Pro-ox FULL No Iron 1/2 Pro-ox 4.21/2 Pro-ox 1/2 Pro-ox No Iron 1/2 Pro-ox 1/2 Air OAME 5No Iron No Pro-ox B25 No Iron No Pro-ox No Iron 5.6No Pro-ox No Iron 1/2 Air 70 uC B25 No Pro-ox 6.41/2 Air 1/2 Air 70 uC 1/2 Air OAME 1/2 Air 7.2
Less oxidation 70 uC 70 uC OAME No Pro-ox 70 uC 70 uC 8.4
Decomposition of biodiesel under simulated engine sump oil conditions 937
JAUTO1395 Proc. IMechE Vol. 224 Part D: J. Automobile Engineering
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
to differences in the initial number of double
bonds present in the sample.
4. Final conjugated values, i.e. a measure of the
number of conjugated double bonds at the end of
the test. During the first stage of the oxidation
process the double bonds within the FAME
molecule will conjugate to form a more stable
compound containing alternating single and
double bonds with delocalized electrons. As the
reaction progresses, conjugation will increase and
then decrease as the conjugated double bonds are
broken down. The final conjugated value should
give an indication of how far this process has
progressed.
5. Average aldehyde value, i.e. the average aldehyde
value during the course of the experiment. The
average value was used because of the parabolic
response. As oxidation progresses, aldehydes are
formed but later react further, leading to a
reduction in the total aldehyde concentration. It
is expected that an average value for the reaction
would give an indication of the progression of the
oxidation process.
The B25 and OAME factors have been highlighted
as these, by definition, contained smaller numbers of
double bonds from the outset of the experiment and,
as such, this difference could have an impact on the
rank order determined using certain indicators. It
can be seen that, depending on which oxidation
indicator is used, varying rank orders are obtained,
thus highlighting the difficulty in determining when
oxidation has occurred and the relative impact of
each factor.
The implications of this are that using only one
indicator as a guide to biodiesel oxidation may not
yield the necessary information, especially if a
history of the changes in that indicator over time
was not available. For example, the aldehyde con-
centration increases over time to a maximum
followed by a gradual decrease; if a single discrete
test was performed, there would not be sufficient
information to establish whether the reading was
early or late in the oxidation process during the rise
or fall in the concentration. If, however, a pH test
was conducted at the same time as determining the
aldehyde concentration, this may yield sufficient
information to infer the advancement of the oxida-
tion process.
Although each indicator does suggest a slightly
different rank order, there are trends which can be
observed. Generally the samples labelled FULL and
1/2 Iron lay at the top of the rankings, demonstrating
that the most oxidation is observed when halving the
air flowrate through the sample and that reducing
the temperature results in reduced oxidation. An
overall rank order was determined by averaging the
rank positions obtained from all the indicators for
each factor. The overall ranking suggests that the
temperature is the most dominant factor, followed
by the air flowrate, fuel-borne pro-oxidant, and
finally iron concentration.
4 CONCLUSIONS
Defining when oxidation has occurred is challenging
as there is no total measurement for the oxidation of
biodiesel. Many indicators fluctuate over time as
they are formed and subsequently break down;
however, the number of double bonds remaining
does seem to be a good measure of the progress of
oxidation. It is difficult to separate the role of each
component in the reaction, owing to the interactions
between factors; however, on the basis of the data
collected, and shown in Table 2, it appears that the
air flowrate into the system and the reaction
temperature have the largest effects on the total
oxidation of FAME. The presence of air and elevated
temperatures leads to an increase in the rate of
oxidation in the early stages of the reaction, but a
decrease in the oxidation rate later in the test as the
concentration of reactants (and absolute number of
double bonds available) decreases.
The presence of both pro-oxidant and iron filings
has an effect on the oxidation, where the removal of
these components from the reaction mixture re-
duces the total level of oxidation. Unlike with the
cerium-based pro-oxidant, merely reducing the
amount of iron in the reaction does not always yield
a significant reduction in oxidation, possibly because
the iron acts catalytically and is present in abun-
dance. The presence of iron and pro-oxidant does,
however, lead to an increased rate of oxidation in the
later stages of the reaction, implying that both
factors promote secondary oxidation processes.
Blend level has little effect on oxidation when
the data have been normalized for the amount of
LAME in the specific samples. Lower blend levels
do, however, mean that less oxidizable material is
available which, in turn, leads to lower absolute
quantities of oxidation products. The results of this
study suggest that a reduction in biodiesel oxidation
within the sump oil could be achieved by reducing
the temperature and aeration of the sump oil;
however, in reality, this is not a feasible solution.
The addition of antioxidants to the sump oil, tailored
938 C D Bannister, C J Chuck, J G Hawley, P Price, and S S Chrysafi
Proc. IMechE Vol. 224 Part D: J. Automobile Engineering JAUTO1395
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
to negate the formation of the initial radical species,
would still seem to be the most effective method.
ACKNOWLEDGEMENT
The authors would like to thank the Ford MotorCompany for their financial support through theAcademic Fellowship of the first author. The thirdauthor holds the Medlock Chair of Engineering. Thiswork was undertaken using facilities within thePowertrain and Vehicle Research Centre at theUniversity of Bath, UK.
F Authors 2010
REFERENCES
1 Schuchardt, U., Serchelia, R., and Vargas, R. M.Transesterification of vegetable oils: a review. J.Braz. Chem. Soc., 1998, 9, 199–210.
2 Falk, O. and Meyer-Pittroff, R. The effect of fattyacid composition on biodiesel oxidative stability.Eur. J. Lipid Sci. Technol., 2004, 106, 837–843.
3 Fang, H. L., Alleman, T. L., and McCormick, R. L.Quantification of biodiesel content in fuels andlubricants by FTIR and NMR spectroscopy. SAEtechnical paper 2006-01-3301, 2006.
4 Fang, H. L., Whitacre, S. D., Yamaguchi, E. S., andBoons, M. Biodiesel impact on wear protection ofengine oils. SAE technical paper 2007-01-4141,2007.
5 Knothe, G. ‘Designer’ biodiesel: optimizing fattyester composition to improve fuel properties.Energy Fuels, 2008, 22, 1358–1364.
6 Gunstone, F. D., Harwood, J. L., and Padley, F. B.The lipid handbook, 2nd edition, 1994 (CambridgeUniversity Press, Cambridge).
7 Bondioli, P., Gasparoli, A., Bella, L. D., Taghlia-bue, S., and Toso, G. Biodiesel stability undercommercial storage conditions over one year. Eur.J. Lipid Sci. Technol., 2003, 105, 735–741.
8 Bouaid, A., Martinez, M., and Aracil, J. Longstorage stability of biodiesel from vegetable andused frying oils. Fuel, 2007, 86, 2596–2602.
9 Dunn, R. O. Oxidative stability of soybean oil fattyacid methyl esters by oil stability index (OSI). J. Am.Oil Chem. Soc., 2005, 82, 381–387.
10 Dunn, R. O. Effect of temperature on the oilstability index (OSI) of biodiesel. Energy Fuels,2008, 22, 657–662.
11 Ferrari, R. A., Oliveira, V. D., and Scabio, A.Oxidative stability of biodiesel from soybean oilfatty acid ethyl esters. Sci. Agric., 2005, 62, 291–295.
12 Mittelbach, M. and Schober, S. The influence ofantioxidants on the oxidation stability of biodiesel.J. Am. Oil Chem. Soc., 2003, 80, 817–823.
13 Schober, S. and Mittellbach, M. The impact ofantioxidants on biodiesel oxidation stability. Eur. J.Lipid Sci. Technol., 2004, 106, 382–389.
14 Dunn, R. O. Effect of oxidation under acceleratedconditions on fuel properties of methyl soyate(biodiesel). J. Am. Oil Chem. Soc., 2002, 79, 915–920.
15 Bouaid, A., Martinez, M., and Aracil, J. Productionof biodiesel from bioethanol and Brassica carinataoil: oxidation stability study. Bioresource Technol.,2009, 100, 2234–2239.
16 Bostyn, S., Duval-Onen, F., Porte, C., Coic, J.-P.,and Fauduet, H. Kinetic modelling of the degrada-tion of the alpha-tocopherol in biodiesel–rapemethyl ester. Bioresource Technol., 2008, 99, 6439–6445.
17 Frohlich, A. and Schober, S. The influence oftocopherols on the oxidation stability of methylesters. J. Am. Oil Chem. Soc., 2007, 84, 579–585.
18 Xin, J., Imahara, H., and Saka, S. Kinetics on theoxidation of biodiesel stabilized with antioxidant.Fuel, 2009, 88, 282–286.
19 Knothe, G. Analysis of oxidized biodiesel by 1H-NMR and effect of contact area with air. Eur. J.Lipid Sci. Technol., 2006, 108, 493–500.
20 Ogawa, T., Kajiya, S., Kosaka, S., Tajima, I.,Yamamoto, M., and Okada, M. Analysis of oxida-tive deterioration of biodiesel fuel. SAE technicalpaper 2008-01-2502, 2008.
21 Fang, H. L. and McCormick, R. L. Spectroscopicstudy of biodiesel degradation pathways. SAE tech-nical paper 2006-01-3300, 2006.
22 Monyem, A., Canakci, M., and Van Gerpen, J.Investigation of biodiesel thermal stability undersimulated in-use conditions. Appl. Engng Agric.,2000, 16, 373–378.
23 Farfaletti, A., Astorga, C., Martini, G., Manfredi,U., Mueller, A., Rey, M., De Santi, G., Krasen-brink, A., and Larsen, B. R. Effect of water/fuelemulsions and a cerium-based combustion im-prover additive on HD and LD diesel exhaustemissions. Environ. Sci. Technol., 2005, 39(17),6792–6799.
24 Stratakis, G. A., Pontikakis, G. N., and Stamatelos,A. M. Experimental validation of a fuel additiveassisted regeneration model in silicon carbidediesel filters. Proc. IMechE, Part D: J. AutomobileEngineering, 2004, 218(7), 729–744. DOI: 10.1243/0954407041580111.
25 Freedman, B., Pryde, E. H., and Mounts, T. L.Variables affecting the yields of fatty esters fromtransesterified vegetable-oils. J. Am. Oil Chem. Soc.,1984, 61, 1638–1643.
26 Knothe, G. and Kenar, J. A. Determination of thefatty acid profile by 1H-NMR spectroscopy. Eur. J.Lipid Sci. Technol., 2004, 106, 88–96.
27 Morcos, M., Parsons, G., Lauterwasser, F., Boons,M., and Hartgers, W. Detection methods foraccurate measurements of the FAME biodieselcontent in used crankcase engine oil. SAE technicalpaper 2009-01-2661, 2009.
Decomposition of biodiesel under simulated engine sump oil conditions 939
JAUTO1395 Proc. IMechE Vol. 224 Part D: J. Automobile Engineering
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from
APPENDIX
Notation
Bx biodiesel blend ratio (where x is the
weight percentage of biodiesel)
DPF diesel particulate filter
2-EHN 2-ethylhexyl nitrate
FAME fatty acid methyl ester
LAME linoleic acid methyl ester
NMR nuclear magnetic resonance
OAME oleic acid methyl ester
940 C D Bannister, C J Chuck, J G Hawley, P Price, and S S Chrysafi
Proc. IMechE Vol. 224 Part D: J. Automobile Engineering JAUTO1395
at Universitats-Landesbibliothek on January 20, 2014pid.sagepub.comDownloaded from