You are on page 1of 29

2

Ethanol was the largest contributor to alternative


road transport fuel in 2009 with a consumption of 38.7
Mtoe/a (1 Mtoe/a = 1 3 106 t of oil equivalent per
annum), representing 2.3% of the total global fuel consumption1 of 1701 Mtoe/a. In the USA the role of ethanol as an oxygenate source in reformulated gasoline
began to rise in the early 2000s when methyl tert-butyl
ether was phased out. (The use of reformulated gasoline is mandated in some areas of the USA in order to
reduce the emissions of unburned hydrocarbons from
older vehicles and to help to reduce smog (ground-level
ozone) formation.) Ethanol consumption has risen
strongly since 2005 owing to the federal policies which
encouraged its use. The original Renewable Fuels
Standard (RFS)2 was created under the Energy Policy
Act of 2005 and required 7.5 3 109 US gal (1 US gal
= 3.785 l) of renewable fuel to be blended into gasoline
by 2012. As a result of the Energy Independence and
Security Act of 20073 the RFS was expanded to
Renewable Fuels Standard 2 (RFS 2)4 to include diesel
as well as gasoline, mandated an increase of renewable
fuel from 9 3 109 US gal in 2008 to 36 3 109 US gal in
2022 and established new categories of renewable fuels,
setting separate targets for volume and GHG reduction
for each fuel. The target for biomass-based diesel is a
minimum of 1 3 109 US gal from 2012 to 2022 (with
the exact target to be set by future rule making) while
that for cellulosic biofuel was set to rise from 0.5 3
109 US gal in 2012 to 16 3 109 US gal in 2022, most
of which is expected to be in the form of cellulosic
ethanol.4 In total, 21 3 109 US gal of advanced biofuels (including cellulosic biofuels) are required in
2022, leaving 15 3 109 US gal to be supplied by firstgeneration fuels, mostly in the form of corn ethanol.
Many problems have been encountered in meeting the
targets for cellulosic biofuels, and the 2012 target has
recently been revised downwards to 8.65 3 10 US gal;
this will represent less than 0.006% of US fuel usage
that year, compared with 9.23% overall for renewable
fuels, which consist mostly of corn ethanol.5 (More
precisely, the figure of 0.006% represents cellulosic
biofuels as a fraction of non-renewable gasoline and
diesel use.)
Anderson et al.6 has provided an excellent account
of the introduction of ethanol into the gasoline fuel
pool, including details of the changes in the gasoline
blend stock, i.e. the reformulated blend stock for oxygen blending (RBOB), in to which the ethanol is
blended. Between 2000 and 2010, US ethanol consumption grew from 1.6 3 109 US gal/year to 13 3 109 US
gal/year; the latter value represents a hypothetical
nationwide uniform gasolineethanol blend level6 of
almost 10 vol %. In fact, virtually all the ethanol used
in US transport is used in the form of E10 which has
been available in the US since the 1980s but is now
widespread owing to the political and environmental
developments described above. (EX represents a blend
where X denotes the volumetric concentration of ethanol in gasoline, e.g. E10 is a mixture of 10 vol %

Proc IMechE Part D: J Automobile Engineering


ethanol in gasoline.) In order to address the blend
wall which has arisen the US Environmental Agency
(EPA) has granted two partial waivers7,8 which
allow but do not mandate the introduction of E15 for
use in light-duty vehicles from model year 2001. The
approval process is inherently slow since it involves
extensive testing and is open to challenge by vehicle
manufacturers.
Anderson et al.6 examined various scenarios for
ethanol introduction, including the most optimistic,
where the RFS 2 targets are met. This latter scenario
would lead to notional uniform ethanol blend levels of
E24 by 2022 and of E29 by 2035 (assuming a 2%
annual growth post-2022). In the absence of significant
growth in E85 sales this could not be implemented with
new waiver approvals since these levels of ethanol are
well beyond the tolerance capability level of current
conventional vehicles. New vehicle and engine specifications would be required, together with the maintenance of a protection-grade fuel (E10 or E15) for
existing vehicles. In the European Union (EU) a similar
situation may exist in 2020 in order to satisfy the
Renewable Energy Directive9 and the Fuel Quality
Directive10 which together require that 10% of transport energy is supplied in renewable form and that the
overall GHG intensity of fuels should be reduced by
6%. With diesel penetration at approximately 50%
across the EU it has been suggested11 that, because of a
lack of sufficient supplies of sustainable vegetable oils
for biodiesel manufacture and some issues of achieving
emissions compliance of modern vehicles using more
than 7 vol % biodiesel, these targets may need to be
met by supplying base fuel with ethanol levels of the
order of E20.
Flexible-fuel vehicles (FFVs) are capable of using
ethanol in concentration levels of up to 85 vol % (E85).
However, despite the registration of over 9 3 106 such
cars at the end of 2011, incentivised by the rating of
FFVs under corporate average fuel economy (CAFE)
legislation,12 representing 4% of the US light-duty
vehicle fleet, only 1% of the total ethanol use has been
in the form of E85 sales.6 Ethanol in this highconcentration form, E85 (now defined by ASTM
D5798 as 5183 vol % ethanol in gasoline), has suffered from both limited availability and uncompetitive
pricing on an energy basis in the USA. The requirement for FFVs to run on any concentration of ethanol
in gasoline, from 0 vol % to 85 vol %, also means that
the vehicles are not capable of exploiting the high
octane numbers of the higher-level blends; this prevents
such vehicles from being able to offset the reduction in
the volumetric energy content of the fuel by increasing
the thermal efficiency of the engine. Conversely, if
FFVs were to be optimised to have high compression
ratios in order to reduce their volumetric fuel consumption on E85, they would give an unattractive increase
in the fuel consumption when operating on fuel with a
lower ethanol concentration such as E10 if the octane
level of the RBOB is maintained.6 Although this may

Pearson et al.
be viewed as a mechanism to incentivise the FFV customer to use E85, the proposition is only reasonable if
there are a sufficient number of fuel stations offering
the fuel.
In contrast with the CAFE regulations which make
it attractive to manufacture FFVs in the USA, the fiscal
penalties for GHG emissions from cars sold in the EU
is based on only tailpipe carbon dioxide (CO2) emissions. This gives vehicle manufacturers no incentive to
spend even the small extra amount required in order to
produce an FFV (about e100 per vehicle). It has been
suggested11 that attributing some CO2 benefit to manufacturers will provide a compelling reason for original
equipment manufacturers (OEMs) to make FFVs and
thus to produce a greater outlet for ethanol as an automotive fuel. The EU vehicle tailpipe CO2 penalty system does, however, at present allow a 5% reduction in
the tailpipe CO2 emissions to be claimed for any FFV
that an OEM sells, provided that the one third of the
fuel stations in the country in which the vehicle is sold
has at least one E85 pump. 13 Furthermore, for a vehicle emitting the 2011 EU average of 135.7 (g CO2)/km,
and at the highest proposed fine rate in 2015 of e95 per
(g CO2)/km, this represents a saving to the manufacturer of e541 per car (with the benefit limited to 5.7 (g
CO2)/km in this instance since the target would be
achieved), which the present authors contend is significantly greater than the additional costs of producing a
flexible-fuel-capable E85 vehicle.14
The use of an additional lower-cost blend component which is miscible with gasoline may provide an
opportunity to supply higher levels of ethanol in fuels
for FFVs which remain cost competitive for the consumer in terms of the distance travelled per unit fuel
cost. Recent work by Turner et al.14,15 has shown that
it is possible to introduce an additional light-alcohol
blend component, in the form of methanol, into
gasolineethanol blends in order to extend the displacement of gasoline and to allow cost competitiveness for
high-alcohol blends, thereby incentivising their use in
the growing FFV vehicle pool which can use higheralcohol blends and further incentivising the production
of such vehicles via consumer demand. Appropriate
formulation of three-component (ternary) blends of
gasoline, ethanol and methanol endows them with the
potential to serve as drop-in fuels for vehicles which
are able to operate with a given level of ethanol in gasoline. (It is recognised that gasoline is itself a complex
mixture of many hydrocarbons. It is mostly treated
here as a single-blend component with properties which
are representative of the mixture of hydrocarbons that
it contains.) In such blends the ethanol level can be varied, and cost can be made attractive to the customer by
increasing the amount of methanol used. A sensitivity
analysis of the cost of the three blend components was
undertaken by Turner et al.15 showing the conditions
under which gasolineethanolmethanol (GEM) blends
approach and drop below the energy-equivalent costs

3
of gasoline. This has to be balanced against the requirement to supply sufficient renewable ethanol.
Initial vehicle operational and material compatibility
test results, including the tail-pipe CO2 and pollutant
emissions, using such fuels have been reported in earlier
publications,14 18 together with simple economic analysis. While the initial material testing carried out by
Turner et al.18 indicated that elastomeric swelling of
the ternary GEM blends was worse than for E85, none
of the blends gave a worse performance than M85.
(MX represents a blend where X denotes the volumetric
concentration of methanol in gasoline.) This indicates
that the materials specified for the M85 vehicles which
have been used extensively in fleet trials in the USA in
the 1980s and 1990s and which are being used extensively in the Peoples Republic of China today (see the
work by Pearson and Turner19) would be compatible
with GEM blends at no significant vehicle on-cost.
The use of these iso-stoichiometric ternary blends as
transition vectors to a future economy where it may be
possible to provide renewable alcohols to supply a large
proportion of ground transportation has been discussed in several previous publications.14,19 21 This
paper focuses on the formulation and analysis of these
ternary blends and examines some of their physicochemical properties.
The original concept of ternary GEM blends was
based around the objective of formulating threecomponent blends with identical stoichiometric air-tofuel ratios (AFRs) as target two-component (binary)
blends.15 This property is essential if the drop-in fuel is
not to cause engine operation to stray outside the predetermined limits of the gravimetric AFR for which it
and its emissions treatment system were designed to
operate. The adjective iso-stoichiometric was adopted
to describe the resulting ternary blends in order to convey the objective behind their formulation.15 For the
initial blends considered in the previously reported
vehicle test programmes, it was also found that one of
the other key properties which the fuel must have in
order to avoid immediate operational problems in an
existing pre-calibrated engine, namely an approximately equal volumetric energy content (so that fuel
injector pulse durations do not have to be recalibrated
to achieve given load levels), was satisfied to a more
than acceptable degree. It should be noted that, while
much of the discussion relates to ternary blends which
are targeted at E85, it is equally possible to formulate
ternary blends to replace any other gasolineethanol
blend such as E10, E15, E20 or E50, as shown in the
second section.
In this paper the order in which the blend properties
were originally designed was inverted in order to convey the concept in a concise analytical manner. Simple
expressions are derived which quantify the volume fractions of gasoline, ethanol and methanol required to
generate a mixture having the same volumetric energy
density as a target binary blend of gasoline and

4
ethanol. The divergence from the iso-stoichiometry
condition is then quantified.
The formulation of blends where water is introduced
via some of the alcohol components is considered.
Maiorella et al.22 stated that during the ethanol fermentation process the specific ethanol productivity is
reduced with increasing ethanol concentration and is
completely halted at 87.5 g/l; if the rest of the mixture
is assumed to be water, this corresponds to a mass fraction of about 9% ethanol91% water (a volume fraction of 11%89%). This mixture can be subsequently
concentrated to mass fractions of 95.6% ethanol and
4.4% water by fractional distillation, at which point an
azeotrope is formed where the vapour in equilibrium
with the liquid has the same composition and so no further separation of the distillate can be achieved. To go
beyond this level of ethanol concentration, technologies
which use molecular sieves, membranes, solvent extraction or reduced or elevated pressure distillation are
required, adding significantly to the process energy
intensity, the well-to-tank CO2 emissions (depending
on the energy source used) and the capital cost.23,24 (At
pressures less than 9.33 kPa there is no azeotrope.)
Collura and Luyben25 stated that separation of the
water and ethanol can be responsible for up to 50% of
the process energy. As a consequence of the above factors, hydrous ethanol can be obtained at a lower cost
than that of the anhydrous product. Data given by
Makkee et al.26 showed that between 2004 and 2009
the market price of hydrous ethanol varied between
9% and 15% lower than that of anhydrous ethanol.
The effects of introducing water, at or above the
azeotropic concentration level, as a fourth component
to the ternary blend system described above, to form
what are effectively iso-stoichiometric quaternary
blends, are considered. Such mixtures will be subsequently referred to as hydrous ternary blends. It is
intended that any water present in the blend would be
introduced together with the ethanol or the various isomers of propanol and butanol which can also form the
basis of iso-stoichiometric ternary blends. These alcohols also form azeotropes with water, which prevent
them being completely dried by conventional fractional
distillation. Methanol, however, does not form an azeotrope with water. The phase equilibria of both the
anhydrous blends and the hydrous blends are discussed, and a limited set of measurements is presented
Further properties, the constancy of which would be
desirable for drop-in fuels, are also examined.
Measurements and predictions of the vapour pressure
and distillation characteristics of various blends targeted at both E85 and E15 are presented. FFVs compliant with current US and EU V emissions requirements
use an alcohol sensor in the fuel line, in addition to
software algorithms based on the vehicles tank level
sensor, in order to infer the ethanol concentration of
the fuel about to enter the engine. The response characteristics of the physical alcohol sensors to both ethanol and methanol are presented and discussed. Finally,

Proc IMechE Part D: J Automobile Engineering


the measured octane numbers of E85-equivalent and
E15-equivalent iso-stoichiometric blends are presented,
together with predictions based on a simplified
approach using mole fractions rather than liquid volume fractions.27,28

Basic formulation of the isostoichiometric ternary blends


In this section a simple approach to the formulation of
ternary blends which have the potential to be drop-in
fuels for nominal binary mixtures is presented. The fundamental properties which are necessary, but not sufficient, for such blends is that they should have the same
stoichiometric AFR and volumetric energy density as
the target binary blends which they seek to supplement
or replace. The iso-stoichiometry property is required
(matched to the lowest AFR of the limiting target binary blend) so that the exhaust oxygen sensor does not
produce a signal which is outside the bounds of that
which the engine control unit is expecting. If the volumetric energy density of the fuel is the same as that
which it is replacing, the mixture can be said to be also
iso-energetic with the latter, and the opening duration
of the fuel injectors can remain the same so that no
recalibration of the engine is required.
In an gasolineethanol FFV the lowest AFR and
volumetric energy density which the engine would be
required to accept would be that of E85, i.e. that of the
blend with the highest concentration of the component
(ethanol) with the lowest AFR and volumetric energy
density (in fact, 83 vol % ethanol in gasoline according
to ASTM D5798). Since FFVs can also cope with pure
gasoline, they can cope with any mixture of ethanol and
gasoline between this and E85. Such blends are created
as a matter of course during the use of the vehicle when
it is fuelled with various quantities of E85 and the standard available gasoline fuel (which may be currently
anything up to E10 in Europe and the USA, as discussed above).
This section shows how the volumetric energy densities of ternary blends can be matched to those of target
binary blends, to a first degree of approximation. The
deviation of the stoichiometric AFR from that of the
target binary blend is also quantified. Appendix 2
shows that, to create a ternary GEM blend having a
target volumetric energy density of rQLHV , the volume
fraction Vm =V of methanol required in the hydrous
alcohol for a given ethanol concentration Ve =V is
given by
rg QLHVg  rQLHV
Vm
Ve re QLHVe  rg QLHVg
=
+
V
V rg QLHVg  rm QLHVm
rg QLHVg  rm QLHVm

With the aim of formulating ternary GEM blends having a constant volumetric energy density (which could
be that of a binary blend of ethanol and gasoline, such
as E85), equation (1) can be used to calculate the

Pearson et al.

Figure 1. Variations in the components for anhydrous ternary


GEM blends to maintain the E85 volumetric energy density.

volume fraction of methanol required, as the volume


fraction of ethanol is varied. Figure 1 shows the variations in the components of gasoline, ethanol and
methanol in order to maintain the same volumetric
energy density as E85. Any vertical line on this diagram
specifies a ternary blend of the three components with
the same volumetric energy density as E85. Clearly E85
is the limit case where the blend contains no methanol.
Conversely, the limit case where all the ethanol has
been removed from the blend is the vertical line corresponding to the ordinate axis position (where the ethanol volume fraction is zero); this blend has volume
fractions of approximately 44% gasoline and 56%
methanol.
Equation (1) can be rewritten as
Vm
Ve
=A
+B
V
V

where
re QLHVe  rg QLHVg
rg QLHVg  rm QLHVm
rg QLHVg  rQLHV
B=
rg QLHVg  rm QLHVm
A=

Now, since
re QLHVe  rg QLHVg \ 0
rg QLHVg  rm QLHVm . 0

and

 

 r QLHV  r QLHV  .  r QLHV  r QLHV 
g
m
e
g
g
m
e
g

A is negative and less than 1. Thus, starting with a binary gasolineethanol blend which is to be turned into a
ternary GEM blend with identical volumetric energy
density rQLHV , adding methanol enables ethanol to be
removed. The methanol must be added at a slower rate
than that at which the ethanol is removed in order to
allow the volume fraction of gasoline to be increased,

Figure 2. Variation in the stoichiometric AFR in ternary blends


and deviation of the stoichiometric AFR from that of anhydrous
E85 at an equivalent volumetric energy density for various
ternary blend ethanol concentrations.
AFR: air-to-fuel ratio.

thereby compensating for the lower volumetric energy


density of the methanol component.
The B term in equation (2) is constant for a given
target binaryternary blend and gives the volume fraction of methanol when all the ethanol is removed from
the ternary blend, i.e. when Ve =V = 0, as
rg QLHVg  rQLHV
Vm
=
V
rg QLHVg  rm QLHVm

Figure 2 shows that the AFRs of the range of mixtures delineated by all vertical lines in Figure 1 vary by
a maximum of 0.12% relative to that of E85 across the
range of ethanol volume fraction levels in the ternary
blend mixtures. Appendix 2 gives the relationships used
to plot these data.
Figure 3(a) to (d) shows examples of isostoichiometric blends which can substitute E85
(repeated from Figure 1 for comparison), E50, E20 and
E10. Similar blends can be formulated for butanolgasolineethanol mixtures. Again, any vertical line on the
diagrams represents a mixture with the same volumetric
energy content and essentially the same AFR, and the
limit cases of no methanol and the maximum amount
of methanol (no ethanol or butanol) permissible in the
blend are represented by the compositions given by the
vertical lines on the far right and far left of each diagram respectively. In these blends, in order to replace
ethanol with methanol, which has a lower volumetric
energy density and stoichiometric AFR, it is necessary
to replace every unit of the original alcohol which is
removed by a unit of a mixture of methanol and gasoline specifically formulated to preserve the volumetric
energy density and stoichiometric AFR. The latter
component is necessary because it has a higher volumetric energy density and stoichiometric AFR than
those of ethanol and can thus compensate for the
reduction in either property owing to the introduction
of methanol.

Proc IMechE Part D: J Automobile Engineering

Figure 3. Iso stoichiometric ternary GEM blends, formulated as drop in blends for (a) E85, (b) E50, (c) E20 and (d) E10.

While the specification of the gasoline used to make


these blends will of course influence their detailed compositions but not the principle of being able to match
one of the iso-energetic or iso-stoichiometric properties.
Matching one of these properties means that the accuracy of matching the other is subject to the variation in
the gasoline specification but, with the exception of the
vapour pressure, such variation is well within the variation in market fuels caused by the tolerance in the
EN228 fuel specification. The variation in the vapour
pressure of the mixture is discussed in the fourth
section.

Quaternary blends of gasoline, ethanol,


methanol and water
Quaternary
gasolineethanolmethanolwater
(GEMW) blends in which the water component is, for
example, introduced in the form of hydrous ethanol,
can be referred to as hydrous ternary blends. Hydrous
ethanol is used in nominal E100-compatible (totalflex) cars made for the Brazilian market. These vehicles have a small separate fuel tank containing gasoline
which is used to start the vehicle. Keuken23 and
Keuken and de Jager24 discussed blends of gasoline
and ethanol made using hydrous ethanol with a view to
improving the carbon intensity of the fuel by reducing

the energy required to dry the ethanol beyond its azeotropic point. This concept takes wet ethanol and uses it
to blend hydrous gasolineethanol mixtures, from E1
to E95. At volumetric ethanol concentrations of up to
85% these hydrous blends are potential drop-in fuels
for FFVs. Thus, hydrous E10 (hE10) made with wet
ethanol of ethanolwater of mass composition 95.6%
4.4% would contain about 0.5 mass % water; hE85
would contain about 3.5 mass % water. The concept
proposed here would be to use hydrous ethanol to produce hydrous ternary (quaternary) blends as extensions
of the type described in the second section, containing
gasoline, ethanol, methanol and water.
The analysis given in Appendix 2 shows how the
basic concept described in the second section is easily
extended to account for the use of hydrous alcohols
and focuses on obtaining blends of nominally equal
volumetric energy contents while maintaining the stoichiometric AFR of the blend within acceptable limits.
The important issue of phase separation which is often
encountered when water is present with hydrocarbon
fuels is discussed in the seventh section.
Mixtures of water in ethanol (hydrous ethanol; usually hydrous ethanol is defined as ethanol with a water
concentration greater than or equal to that at which it
forms an azeotrope (4.4 mass %)) defined by the mass
concentration as

Pearson et al.

Figure 4. Variations in the blend components to maintain the


E85 volumetric energy density using hydrous ethanol with 4.4
mass % water (hw, e = 0:044).

mw
= hw, e
me + mw

Figure 5. Variations in the blend components to maintain the


E85 volumetric energy density using hydrous ethanol with 10
mass % water (hw, e = 0:10).

can be incorporated to give an expression which is


equivalent to equation (1) for hydrous ternary blends
in the form
Vm
V

Ve re QLHVe rg QLHVg f1 + re =rw hw, e =1


V
rg QLHVg rm QLHVm
rg QLHVg rQLHV
+
rg QLHVg rm QLHVm

hw, e g

Figure 4 and Figure 5 show how Figure 1 is modified


when hydrous ethanol containing 4.4 mass % and 10
mass % water respectively is used to replace the anhydrous ethanol component of the ternary blend. From
these figures and Figure 6 it can be seen that, as the
water component increases with increasing ethanol concentration, proportionally more gasoline must be added
to the blend to maintain the volumetric energy concentration since there is no energy contribution from the
water in the mixture. Figures 7 and 8 show the impacts
of using various concentrations of water in hydrous
ethanol on the stoichiometric AFRs of hydrous ternary
blends having equal volumetric energy densities to that
of E85 made from anhydrous ethanol. It can be seen
that even using hydrous ethanol with 10 mass % water
in ethanol, it is possible to blend hydrous ternary blends
with energy densities equivalent to that of anhydrous
E85 which deviate by less than 1% in their stoichiometric AFRs from the latter. This is within the tolerance of the usable signal from an exhaust gas oxygen
sensor.
The perception of the levels of water concentration
in fuels is influenced by the parameter which is used to
quantify the concentration level. The concentration of
fuels in blends is usually quantified in terms of the volume fraction, as used so far in the present work. The
mole fraction of a mixture component is a measure of
the ratio of the number of its molecules to the total

Figure 6. Reduction in the maximum ethanol concentration


and increase in the minimum gasoline concentration as functions
of the water mass fraction in ethanol (the E85 equivalent
volumetric energy).

number of molecules in the mixture and thus should


provide the most meaningful insight into the variation
in the physicochemical properties caused by varying
the concentration levels of the mixture. This parameter
is used surprisingly infrequently in the engineering literature which deals with fuel mixtures but has recently
been used by Andersen et al.,29 Anderson et al.27,28 and
Chen and Stone.30 In the following, the results
described are replotted in terms of the mixture mole
fractions, while Appendix 3 describes more generally
how the concentration of a range of alcohols in gasoline varies when the volume fractions and the mole
fractions are used as the metrics. The influence of using
mole fractions when predicting the octane numbers of
alcohol blends is discussed in the ninth section.
The mole fraction of any component in a mixture
can be obtained from the mass fraction using the
expression
ni
mi M
=
n
m Mi

Proc IMechE Part D: J Automobile Engineering

Figure 7. Variation in the stoichiometric AFR with ethanol concentration for various water concentrations in hydrous ternary
blends at an E85 equivalent volumetric energy density.
Stoich. AFR: stoichiometric air-to-fuel ratio; vol. frac.: volume fraction.

ni
ni
= Pk
n
i 1 ni
= Pk
i

= Pk
i

mi =Mi
mi =Mi
mi =m1=Mi

11

mi =m1=Mi 

Equation (11) applies to mixtures of both gases and


liquids as it is not derived using an equation of state,
with the caveat that, for liquids, the mass fractions are
calculated from the volume fractions which are those
of the constituents before mixing via the relationship
Figure 8. Deviation of the stoichiometric AFR from the
anhydrous E85 value with the ethanol concentration for various
water concentrations in hydrous ternary blends at an equivalent
volumetric energy density to that of E85.
AFR: air-to-fuel ratio.

mi
r Vi
= i
m
r V

Alternatively, equation (11) can be combined with


equation (12) to give
xi =

where ni and Mi are the number of moles and the


molar mass of component i respectively, while n and M
are the total number of moles of the mixture and the
molar mass of the mixture respectively. This relationship, however, requires a priori knowledge of the molar
mass of the mixture, which is calculated via knowledge
of the mole fractions of the mixture constituents via the
relationship
M=

k
X
ni
i

Mi

10

To avoid this circularity, equation (9) can be reexpressed in the form

12

ni
n

= Pk
i

= Pk
i

ri Vi 1=Mi
ri Vi 1=Mi 
ri Vi =V1=Mi
1

13

ri Vi =V1=Mi 

From this equation it is clear that mixtures of components where there is a large difference in the ratio of the
component density to its molar mass will exhibit a large
difference in the values of the concentration expressed
in terms of the mole fraction and the volume fraction.
Figure 14 shows that, owing mainly to the large difference between the molar masses of the two substances,
10 vol % water in gasoline represents a mole fraction
of almost 50%, i.e. almost 50% of the molecules

Pearson et al.

Figure 9. Variation in the water concentration expressed in


terms of the mole fraction. with the concentration expressed in
terms of the volume fraction for water in gasoline and for water
in ethanol mixtures.

Figure 10. Variations in the blend components to maintain the


E85 volumetric energy density using hydrous ethanol with 4.4
mass % water (hw, e = 0:044). The concentrations are expressed
as mole fractions.

forming the mixture are water molecules. In ethanol,


10 vol % water represents a mole fraction of 26%.
Figures 9 to 11 show the component concentrations
expressed in terms of the mole fractions for the hydrous
ethanol blends discussed above.

Test fuels
The anhydrous mixtures studied in the present work are
those used in the vehicle test work described by Turner
et al.14,15,18 Measurements of specific hydrous blends are
not described below; only the water tolerance (to phase
separation) of the anhydrous blends is studied. More
detailed measurements of the hydrous ternary blends
described above will be presented in a further publication. Measured data for hydrous (binary) gasoline
ethanol blends have been presented by Keuken.23
Four E85- and four E15-equivalent GEM blends
were specially prepared for physicochemical analysis
and are detailed in Table 1. Ideally a RBOB should be
utilised for the gasoline fraction. An RBOB is a gasoline blending component suitable for addition of

Figure 11. Variations in the blend components to maintain the


E85 volumetric energy density using hydrous ethanol with 10
mass % water (hw, e = 0:10). The concentrations are expressed
as mole fractions.

oxygenates to formulate a finished fuel, and the RBOB


is usually formulated specifically to have volatility
properties such that the final fuel meets the regional
volatility specifications. RBOBs frequently also have
reduced octane as the oxygenate components typically
have high octane values. However, first, the results generated with RBOB blending components would have
been specific to that RBOB and, second, different
RBOBs may well have been required for the E85- and
E15-equivalent blends and possibly also as the methanol content changed. It was thus decided to use a
known entity as the gasoline blending fraction and a
batch of CEC RF-02-03 was acquired for the purpose.
CEC RF-02-03 is the low-oxygen-content European
emissions certification fuel. (For gasoline CEC RF-0203, the lower heating value is 42.76 MJ/kg, the
hydrogen-to-carbon molar ratio is 1.8 and the oxygen
content is less than or equal to 0.1 mass %.) High-purity ethanol and methanol were acquired from within
Sasol, both having a purity better than 99.98%.
The density data presented in Table 1 were obtained at
the industry standard 15 C in a routine fuels analysis
laboratory. A more detailed experiment was performed to
investigate the blending effects. The densities of the individual components and the blends were all measured at
20 C using a pycnometer, which works on a fundamental
measuring principle. The pyncnometer was calibrated
with de-ionised water. The measured densities were
observed to be nearly linear combinations by volume of
the individual component densities, suggesting that the
density of any final blend can be directly calculated from
the component densities and the blend proportions.

Vapour pressure and distillation


characteristics: simulations and
measurements
Theory and model formulation
For a mixture of substances A and B in the vapour and
liquid phases in a closed vessel, the Raoult law states

G90:E0:M10
757.1
33.5
196.8
98.3
1.0
46.5
56.6
61.9
84.8
87
88
G88.5:E4:M7.5
758.5
33.6
194.8
97.8
1.0
49.3
57.5
61.9
84.5
83
84

pA
= xA
pA

G85:E15:M0
759.8
34.8
190.6
99.1
0.9
49.8
64.1
66.3
86.4
71
72
GEM: gaso ineethano methano ; G: gaso ine; E: ethano : M: methano ; RVP: Reid vapour pressure; NA: not avai ab e.

G44:E0:M56
776.7
35.7
194.2
99.1
0.1
81.2
81.2
81.2
90.2
NA
81
G37:E21:M42
778.0
37.4
184.3
98.6
0.9
70.2
93.0
93.0
93.9
72
72
G29.5:E42.5:M28
781.5
40.0
172.5
98.1
1.8
32.0
95.6
95.6
95.7
61
61
G15:E85:M0
787.6
50.6
102.7
99.0
0.9
4.4
99.0
99.0
99.0
40
37
kg/m3
C
C
vo %
vo %
vo %
vo %
vo %
vo %
kPa
kPa
GEM component rat o
Dens ty at 15 C
In t a bo ng po nt
F na bo ng po nt
Recovered
Res due
(Vrec/V)70
(Vrec/V)100
(Vrec/V)110
(Vrec/V)150
Measured RVP
Pred cted RVP

G100:E0:M0
756.9
34.5
194.4
99.0
0.8
32.3
52.9
58.9
84.7
65
65

B end D
B end B

B end C
RF-02-03 gaso ne

Un ts

Va ue for the fo ow ng

that the ratio pA =pA of the partial vapour pressure of


each component to its vapour pressure as a pure liquid
is approximately equal to the mole fraction of A in the
mixture so that

G87.5:E7.5:M5
758.1
33.6
196.8
98.1
1.0
52.0
59.3
62.7
84.9
78
80

B end H
B end E
B end A

B end F

E15 so-sto ch ometr c b ends


E85 so-sto ch ometr c b ends

B end G

Proc IMechE Part D: J Automobile Engineering

Property

Table 1. Dens t es and vo at ty parameters of the base RF-02-03 gaso ne and the GEM b ends, where (Vrec/V)X represents the vo ume fract on recovered at XC.

10

14

Mixtures of chemically similar substances obey the


Raoult law well throughout the composition range and
are called ideal solutions. The Raoult law implies that
the presence of another component reduces the rate at
which molecules of substance A escape from the surface of the mixture, contributing to the partial vapour
pressure of substance A, but does not impede the rate
at which they return. The rate at which molecules of A
leave the surface is proportional to their population at
the surface; this is proportional to the mole fraction of
component A. The rate at which molecules condense is
proportional to their concentration in the gas phase,
the latter being proportional to the partial pressure. In
a perfect gas there are no interactions between the
molecules. In an ideal solution of liquids there can be
interactions but they must be equal for all combinations of molecule types.
Non-ideal solutions do not obey the Raoult law and
can depart significantly from it. Solutions of methanol
or ethanol in gasoline behave as non-ideal mixtures in
which the alcoholalcohol interactions differ from the
gasolinealcohol or gasolinegasoline interactions. In
the case of methanol and ethanol the polarity of the
molecules gives rise to extensive hydrogen bonding
where, for example, methanol forms a quasi-supermolecule known as a cyclic tetramer in which four
methanol molecules form a super-structure via hydrogen bonds between the individual molecules. These cyclic tetramers have an effective molecular mass of 128
(four times that of an individual molecule), rendering
the vapour pressure of the pure methanol relatively
low.31 (Indeed, the manifestation of the hydrogen bonding in this way is the reason that methanol and ethanol
have high boiling points relative to their molar masses.
Water is the extreme example of this interaction.) The
hydrogen bonds are progressively weakened and
become less extensive when the alcohol is mixed with
increasing quantities of a non-polar solvent such as gasoline, making them behave as low-molecular-mass components (32 in the case of methanol) which increase the
vapour pressure of the mixture. The effect of the lowmolecular-mass component eventually falls away as its
concentration drops to zero. Conversely, its effect
becomes dominant at higher concentrations. In addition to affecting the cold-start performance of an
engine, this behaviour illustrates how evaporative emissions using methanol and ethanol at high concentration
levels can be lower than those of gasoline while they
can be higher using low-concentration alcohol blends.31
Kretschmer et al.32 and Chen and Stone30 analysed the
vaporisation of iso-octaneethanol blends, showing the

Pearson et al.
rate program to be more suitable for a blend with a
large quantity of a single low-molecular-mass component returned the results presented in Figure 12, which
are considered to be reliable. Some of the other results
presented by Turner et al.18 are further discussed below
for completeness.
The inclusion of large quantities of a single component in a blend is known to distort the distillation profile
near to the boiling point of that single component. The
formation of azeotropes can cause further distortions,
as is known to be the case with ethanol.37 Figure 12
indicates the distillation profiles for the E85-equivalent
blends. The severe distortion of the predominantly ethanol blend of blend A is readily apparent in the figure.
The very low front-end volatility (low temperatures) is
the reason that during winter the ethanol content in E85
blends is reduced to typically 70 vol %. An adequate
cold-starting performance is thus ensured as there is a
sufficiently high gasoline component which will vaporise
at lower temperatures. With the addition of methanol
and gasoline components in the GEM blends, the distillation profile approaches that of the base gasoline. The
front-end volatilities of the methanol-only blend (blend
D) and blend C are almost identical with that of the gasoline for the first 20% recovery.
The RVP, which is an indication of the presence of
very light fractions that vaporise at low temperatures,
are also influenced by the azeotrope formed between
ethanol and the base gasoline, resulting in an increased
RVP even though the boiling point of ethanol is higher
than that of the test temperature for the RVP procedure.37 However, blend A, the high-ethanol-content
blend, has a significantly lower RVP than that of the
base gasoline. Increasing the gasoline and methanol
contents results in increases in the RVP. Blends C and
D have RVP values above that of the base gasoline. An
increased RVP is known to increase the evaporative
emissions; however, the vapour recovery systems that
were applied to the logistics chain and the use of vehicle
evaporative emissions controls, in the form of carbon
canister vapour traps, are very effective at controlling
these emissions. This small increase in the RVP for the
E85 GEM blends is not likely to impact the overall evaporative emissions severely, although the extent of this
should be considered. GEM fuels will be used in FFVs
which should not suffer from hot-weather handling
issues and this increased RVP should not negatively
impact driveability in hot conditions. Specially formulated low-volatility RBOB blending components could
of course be used to counteract these issues if required.
The E15-equivalent blends produce somewhat different trends. Figure 13 indicates that the deviations for
the iso-stoichiometric GEM blends from the base gasoline are less pronounced. All the iso-stoichiometric
GEM blends indicated higher volatilities than that of
the base gasoline with the high-methanol-content blend
H having the largest front-end deviation. Cold-weather
operation should therefore not be negatively impacted
by any of the blends.

13
The most significant impact on these isostoichiometric GEM blends is seen for the RVP. From
Table 1 it is apparent that all the alcohol-containing
blends have increased RVPs. Blend E, which contains
only ethanol, has an RVP which is 6 kPa above that of
RF-02-03 gasoline. As discussed above, an impact on
the evaporative emissions may result, but the extent of
this is likely to be limited. However, the methanolcontaining blends have significantly elevated RVP
results, which may lead to increased emissions and
increased loading of carbon canister vapour traps.
The application of gasoline fuels containing ethanol
at levels of 10% (E10) and more recently 15% (E15)
are targeted for standard vehicle applications. Modern
fuel-injected vehicles are typically immune to hotweather fuel-handling concerns. However, older vehicles with carburetted engines and some applications
with early fuel injection may be negatively affected.
The fuel volatility index (FVI), which is a composite
index of the front-end volatility parameters, is used to
control this phenomenon. The FVI takes the form

Vrec 
FVI = pRV + 0:7
16
V 70
where pRV is the RVP and Vrec =Vj70 represents the
volume fraction recovered at 70 C. (Note that the usual
notation for this parameter is E70 but this has been
avoided in order to prevent confusion with the ethanol
fraction.) An alternative form of this same index is
sometimes used, namely the vapour lock index (VLI),
which is numerically 10 times greater than the FVI and
is given by

Vrec 
VLI = 10pRV + 7
17
V 70
Table 2 presents the FVIs of the RF-02-03 and E15equivalent iso-stoichiometric GEM blends. Significant
increases in the FVI are apparent with the addition of
alcohol and, as the methanol content increases, the FVI
increases further. While the critical value at which hotweather handling issues would become apparent is climate dependent, the methanol-containing blends would
potentially cause some problems in regions with hot climates owing to these elevated levels. The methanolcontaining blends may be trouble free in colder climates
and, as indicated above, these effects could possibly be
counteracted to some extent with the use of specially
formulated low-volatility RBOB blending components.

Storage and oxidative stability


The results previously presented by Turner et al.18 indicated that all the blends performed very well from a
storage and oxidative stability point of view. The induction period (ASTM D525) returned values greater than
360 min, and both the existent gum (ASTM D381) and
the potential gum (2.5 h at 100 C) (ASTM D873)
returned values of less than 1 mg/100 ml. (The potential

14

Proc IMechE Part D: J Automobile Engineering

Table 2. FVI values of the RF 02 03 gasoline and the E15 equivalent iso stoichiometric GEM blends.
Value for the following

GEM component ratio


FVI

RF 02 03

Blend E

Blend F

Blend G

Blend H

G100:E0:M0
88

G85:E15:M0
106

G87.5:E7.5:M5
114

G88.5:E4:M7.5
118

G90:E0:M10
120

GEM: gasoline ethanol methanol; G: gasoline; E: ethanol: M: methanol; FVI: fuel volatility index.

Figure 16. General volume fractions of ethanol and methanol in an admixture with gasoline, showing the boundary between the
blend stability and the unstable region in which phase separation can occur at 15 C (data corrected and replotted from the work
by Qi et al.38).

gum (ASTM D873) is an accelerated ageing method


commonly used as a simple method to rate a fuels
long-term storage stability.) Analyses for the inherent
storage stability have since been completed and have
further indicated that these blends have an excellent
long-term storage capability. Tests similar to ASTM
D4265, used for the testing of middle distillate fuels,
were executed whereby the fuel is exposed to a temperature of 43 C in amber closed Schott bottles. The acid
number, the existent gum content and the potential
gum content were measured at weeks 0, 4, 8 and 12.
The 12 week exposure period is considered to be
roughly equivalent to 12 months. All blends returned
acid numbers at or below 0.001 (mg KOH)/g and existent and potential gum values of or below 1 mg/100 ml.
The results indicate that all fuels performed excellently
in terms of the inherent storage stability, confirming
the results of the induction period tests previously
reported.

Phase separation: anhydrous and


hydrous blends
The co-blending of gasoline, ethanol and methanol to
avoid the phase separation of methanol and gasoline

was investigated by Qi et al.38 (note that importantly,


and in contrast with the present work, Qi et al. were
not blending for constant stoichiometry, but only to
investigate phase separation). Some of the data
obtained by Qi et al.,38 corrected and replotted in
Figure 16, show that there exists a boundary between
stable and unstable mixtures. (The correction applied
to the data obtained by Qi et al.38 is related to the fact
that there is an ambiguity in the values in their Table 2,
which they then use as the abscissa of their Figure 1.
Investigation of the data in that paper shows that,
whereas the labelling of the horizontal axis of their
Figure 1 reads Volume fraction of methanol, the data
used to plot the figure were in fact the sum of the ethanol fraction and the methanol fraction. In Figure 16 of
the present work, only the methanol fraction is used as
the abscissa.) (Note that the gasoline which Qi et al.38
used to obtain their data was a blend with a research
octane number (RON) of 70. The low octane number
should not materially affect the phase separation characteristics of GEM blends.) This boundary moves with
the ambient temperature; in the work caried out by Qi
et al.38 the data are said to pertain to 15 C.
Nevertheless, from the data in Figure 16 it can immediately be seen that the straight E85 blend A (G15 E85

Pearson et al.

15

Figure 17. Ternary GEM blend test fuel data superimposed on the data obtained by Qi et al.,38 showing the possibility of phase
separation at 15 C with very low ethanol concentrations.

M0) would be expected to be stable, but not blend D


(G44 E0 M56). Since the blend relationships are linear,
the data for the iso-stoichiometric E85-equivalent family of blends can be plotted, as shown in Figure 17,
which suggests that the region where phase separation
for such blends might occur starts with an ethanol concentration of about 8.5% or less at 15 C. From
Figures 16 and 17 it could also be suggested that a binary blend of M70 could be used as a high-alcoholcontent fuel since this would avoid issues of phase
separation. However, E85 FFV manufacturers would
not generally be expected to make their current production vehicles capable of operating at the stoichiometric
AFR for M70 of 8.7:1.
With regard to the water tolerance, Skinner and
Viljoen39 indicated the sensitivity of gasolinealcohol
mixtures towards water-induced phase separation and
described the tendency of water-induced phase separation to be a function of the type of alcohol, the amount
of water, the alcohol-to-gasoline ratio and the temperature. Other work, referred to below, has established that
the composition of the gasoline is a significant factor.
Qi et al.38 produced critical phase separation temperatures for a range of gasolinemethanol blends as a function of the water content. A comment was made that
the addition of ethanol will reduce the phase separation
temperature, but no data were presented.
Recent studies40 by the oil companies European
association for environment, health and safety in refining and distribution, namely Conservation of Clean Air
and Water in Europe (CONCAWE), have shown that
ethanol can increase the concentration level of dissolved water in the blend which can be tolerated before
phase separation (turbidity) occurs, and Keuken23 and
Keuken and de Jager24 proposed to use wet ethanol
(ethanolwater of composition 95.6 mass %4.4 mass
%) at higher levels of ethanol concentration (E15 and

above) on the basis that the higher level of ethanol will


increase the water tolerance level sufficiently to avoid
any phase separation issues even at low ambient
temperatures.
Donnelly et al.41 showed the effect of temperature
on the phase separation of gasolinemethanol mixtures
in the absence of water, indicating that such mixtures
become more stable as the temperature rises. Their
data are in reasonable agreement with those obtained
at 15 C by Qi et al.38 They also presented data, showing how the phase stabilities of gasoline and methanol
are made significantly more unstable by the presence of
water but that the sensitivity to water concentration
reduced with increasing methanol concentration over
the range from M5 to M25. Neagu et al.42 showed that
there is increasing water tolerance with increasing ethanol concentration and increasing methanol concentration (with the former increasing at a faster rate) but
reduced water tolerance for low-concentration blends
of methanol in gasoline compared with that of ethanol
in gasoline. The increased water tolerance produced by
using isopropanol (propan-2-ol) as a co-solvent, compared with tert-butanol (2-methyl-2-propanol) was also
demonstrated.
Data presented by Owen and Coley43 illustrated the
significantly increased water tolerance to phase separation in M10 and E10 when gasoline with a higher aromatic content was used. The phase separation
temperature was reduced by approximately 20 C by
increasing the aromatic content of the gasoline blend
stock by about 10%. This sensitivity was also reported
by Green and Yan44 who rank the water tolerances of
aromatics, olefins and saturates in the order aromatics
 olefins  saturates.
Green and Yan44 predicted the water tolerance of
model M5 fuel where the gasoline is composed
entirely of one of a generic aromatic, olefin and alkane

16

Proc IMechE Part D: J Automobile Engineering

Figure 18. Water tolerances for the RF 02 03 gasoline and the


E85 equivalent iso stoichiometric GEM blends from the work by
Turner et al.18

Figure 19. Water tolerances for the RF 02 03 gasoline and the


E15 equivalent iso stoichiometric GEM blends from the work by
Turner et al.18

(saturate) compound. They showed that the use of an


aromatic-rich gasoline not only improves the water tolerance for a given gasolinemethanol blend but also
minimises the co-solvent volume required to attain a
given level of water tolerance. The relative ordering of
the sensitivity to gasoline composition was claimed to
persist in the presence of co-solvents in methanol
blends.44 It was found that, while ethanol is not as
effective in M5 at stabilising the water content, it is significantly better, at a given volumetric concentration,
than methyl tert-butyl ether is. Iso-butyl alcohol and
iso-propanol are shown to have similar degrees of effectiveness, both showing slightly higher water tolerances
than that of ethanol. It was shown that about
30% more ethanol than iso-butyl would be required to
stabilise 0.15 vol % water in M5 at this temperature
with the particular gasoline used (which had an intermediate effectiveness between those of gasoline based
on aromatics and gasoline based on saturated
hydrocarbons).
The blend stabilities were evaluated18 for all the
anhydrous GEM blends by storing 100 ml of fuel in a
fridge at 18 C for 24 h and visually rating any separation. All the ethanol-containing blends were seen to
remain stable. The gasolinemethanol only blends
(blends D and H) indicated separation at this temperature. Further water tolerance assessments were made in
the temperature range from 10 C to 30 C as presented
in Figures 18 and 19. The water tolerance of the base
RF-02-03 hydrocarbon-only gasoline is very low, as
expected, with only 0.02% water at 30 C, decreasing
to 0.01% at 10 C. The gasolineethanol-only blends
have significantly the highest water tolerance, which
decreases monotonically as the methanol content and
the gasoline content increase in the blends.
Unfortunately, lower temperatures could not be
investigated at this time. It is intended that the water
tolerances at lower temperatures will be addressed in a
future publication.

Response of the alcohol fuel


concentration sensor
Gasolineethanol FFVs are required to operate on
blends having between 0 vol % and 85 vol % ethanol
concentrations (in the USA the maximum ethanol concentrations can be between 83 vol % and 51 vol % in
the summer and the winter respectively). The two main
ways of inferring the concentration of alcohol in
flexible-fuel production vehicles involve, in addition to
the exhaust oxygen sensors, either a dedicated alcohol
sensor in the fuel line or the use of an algorithm which
utilises information from the fuel tank level sensor; the
latter is called a virtual sensor. Although virtual sensors
vary in detail, generally they consist of an algorithm in
the energy management system (EMS) which, after a
refuelling event, calculates the approximate values of
the two possible new stoichiometric AFRs, when the
fuel added was either gasoline or E85. The EMS can do
this because the fuel tank level sensor is used to obtain
an approximate value for the volume of fuel added,
and the EMS knows the value of stoichiometry that
the vehicle was operating at when the vehicle was
switched off. On restart, the EMS monitors the oxygen
sensors and looks for a perturbation in the direction of
each possible new calculated AFR. Once it has ascertained the direction of the AFR swing (rich or lean of
where it was operating when switched off), it then
adjusts the fuel injector pulse widths gradually in that
direction until the system locks on to the actual AFR,
and the vehicle has been conditioned. Such virtual
sensors are clearly cheap to employ and function
reliably.
Recent requirements for system performance diagnosis have required the employment of a so-called
physical sensor which directly measures the properties
of the fuel flowing through it. Originally, flexible-fuel
cars employed gasolinealcohol sensors because virtual
sensors had not then been developed. Physical sensors
generally measure the electric permittivity or the

Pearson et al.

17

Figure 20. Physical sensor responses measured during the work conducted by Pearson et al.45 Note the different responses for
binary blends of ethanol and methanol in bulk gasoline due to the different levels of polarity of the two molecules; also note that
blends A (G15 E85 M0) and D (G44 E0 M56) have substantially the same responses.

resistance of the fuel flowing through them; these values vary widely between those for non-polar pure
hydrocarbons (of the type generally consisting of gasoline) and those for compounds with a high polarity
such as the alcohols. The performance of the virtual
sensor is used to monitor and diagnose the physical
sensor and vice versa.
Since their stoichiometric AFRs and volumetric
energy contents are essentially constant (the latter
meaning that the injector pulse widths would be identical for any calculated AFR in the EMS), ternary blends
of the type described in the present work could be
expected to be undetectable in a vehicle employing such
sensors. However, from tests on a prototype tri-FFV
employing a physical sensor, it was known that methanol and ethanol cause different responses in a physical
sensor. This is illustrated in the data presented in
Figure 20, which was gathered on the vehicle used by
Pearson et al.45 when it was fuelled with binary
gasolineethanol or gasolinemethanol fuels of various
compositions. The measured ethanol data correlate
well with information published by the supplier of this
sensor.46 From this, and other data in the literature,47
it can be discerned that, while a virtual sensor was
expected to work with the fuels in an invisible manner,
it would be dangerous to make such an assumption
automatically with a physical sensor, because the
response of the sensor for each alcohol is different
(unsurprisingly, the methanol molecule with a greater
polarity produces a greater response in the sensor than
that of the ethanol molecule).
However, inspection of Figure 20 from the standpoint of the ternary blends discussed above suggests
that the responses may not be very different for blends
with the same stoichiometry. Consider the two limiting
cases at the same stoichiometry as E85: blend A (G15

E85 M0) and blend D (G44 E0 M56). From the data


used to plot Figure 20, the sensor responses for these
two binary blends would be 134 Hz and 129 Hz respectively, i.e. a difference of the order of 4%, and thus
extremely close (see the horizontal dashed lines on
Figure 20). Indeed from this observation it would not
be unreasonable to suppose that all the potential isostoichiometric blends at 9.7:1 stoichiometry would have
sensor outputs between these two limit values. The
proximity of the limit cases suggest that vehicle tests
with a control system utilising a physical sensor would
be expected to yield no issues; this was indeed found to
be the case in the work reported by Turner et al.14

Octane number: anhydrous blends


Some relevant octane number effects were presented
and discussed by Turner et al.18 and, because of the
importance of these issues, they are presented and discussed again below. It had previously been reported
that the octane numbers for iso-stoichiometric GEM
blends are similar across the entire range.14,15 This finding was based on analyses on GEM blends performed
in routine fuel analysis laboratories. The accurate determination of the octane number according to ASTM
D2699 and D2700 (for the RON and the motor octane
number (MON) respectively) for high-alcohol-content
blends is complicated by two factors.
1.

The standard reference fuels required at octane


numbers above 100 require the inclusion of tetraethyl lead (TEL). TEL is extremely toxic and therefore most routine laboratories will not keep TEL
in stock and therefore will either use the compression ratio guide table only, which is considered to

18

2.

Proc IMechE Part D: J Automobile Engineering


be not sufficiently accurate, or will utilise secondary references such as the toluene standardisation
fuels.
The standard test procedure is unsuitable for fuels
with high levels of oxygenates. The standard carburettor settings provide insufficient adjustment to
correct the AFR in line with the procedure.
Hunwartzen48 proposed modifications to the
engine hardware required to perform these measurements correctly.

The details of the tests reported by Turner et al.14,15


were not available and thus it is not clear how reliable
or accurate they might be. These results are plotted as a
function of the methanol content in Figure 21 for comparison. The results do not have good absolute value
agreement even though the same base gasoline and
alcohol blending components were used for both sets of
blends.. This inconsistency may be indicative of some
of the points raised above regarding the difficulties of
measuring these types of blend. It is also apparent that
there is a trend of a slight decrease in the octane number with increasing methanol content. As noted previously, the range of the differences is much less than
would initially be expected given that the relatively lowoctane-number gasoline fraction is significantly higher
for the high-methanol-content blends.
In the work by Turner et al.,18 E15-equivalent isostoichiometric blends were prepared using a base

Figure 21. Octane number as a function of the methanol


content for E85 equivalent iso stoichiometric GEM blends as
reported by Turner et al.14,15

gasoline with an RON of 93. This was done in order to


ensure that the final blend RON values would lie below
100, and therefore they would not present the pitfalls
discussed above and allow robust measurement in a
routine laboratory. The RONs and MONs of these
blends were analysed and the results are given in
Table 3 and plotted against the methanol content in
Figure 22. The range of octane numbers is relatively
small, although there is a generally decreasing trend as
the methanol content increases.
Anderson et al.27,28 presented a compelling case for
the use of molar blending rules for ethanol and methanol in gasoline. It is therefore possible to use the molar
blending rule for the RON in the form
NRON =

ng
ne
nm
NRONg +
NRONe +
NRONm
n
n
n

18

and the analogous version of this equation for the


MON, to study further the octane number of isostoichiometric GEM blends.
The most reliable values for true octane numbers for
ethanol and methanol are considered to be those presented in Table 4,49 and it is clear that the two alcohols
have very similar octane number ratings. However,
their molar masses are significantly different with a
value of 32.0 for methanol versus a value of 46.1 for
ethanol, while typical values for gasoline are in the
range 105120. These differences tend therefore to
counteract each other when using the molar blending
rules with the resulting RON and MON values as all
the blends are quite similar.
Figure 23 presents the octane numbers of the isostoichiometric E85-eq21 blends, utilising the molar
blending rules to calculate the octane numbers. It is
again clear that the octane number decreases slightly
with increasing methanol content; however, the
decreases in the RON and the MON over the entire
blending range are only 1.1 and 1.2 respectively. RBOB
fuels used to manufacture E85-equivalent blends may
have a lower octane number than that used in this
study. Molar blending rules were then used to calculate
the RONs and the MONs for E85-equivalent isostoichiometric GEM blends with the gasoline fraction
which have RON values of 85 and 75 and MON values
of 75 and 65; these results are presented in Figure 24. It
is apparent that, the lower the octane number of the
gasoline fraction, the larger the effect of the decrease in
the octane number with increasing methanol content.

Table 3. Octane number measurements for the E15 equivalent iso stoichiometric GEM blends with a low octane number gasoline
blending component.
Value for the following GEM component ratios

RON
MON

G100:E0:M0

G85:E15:M0

G87.5:E7.5:M5

G88.5:E4:M7.5

G90:E0:M10

93.3
84.0

98.0
85.8

97.8
85.5

97.6
84.8

97.0
84.5

GEM: gasoline ethanol methanol; G: gasoline; E: ethanol: M: methanol; RON: research octane number; MON: motor octane number.

Pearson et al.

19

Figure 22. Octane number as a function of the methanol


content for E15 eq19 iso stoichiometric GEM blends from the
work by Turner et al.18

Figure 24. Octane number as a function of the methanol


content for E85 equivalent iso stoichiometric GEM blends
calculated with molar blending rules with a range of hypothetical
lower octane number gasoline components from the work by
Turner et al.18

Table 4. Octane number ratings for ethanol and methanol.


Value for the following

RON
MON

Ethanol

Methanol

108.6
89.7

108.7
88.6

RON: research octane number; MON: motor octane number.

Figure 23. Octane number as a function of the methanol


content for the E85 equivalent iso stoichiometric GEM blends
calculated with molar blending rules from the work by Turner
et al.18

Importantly, even with the highest-methanol-content


blend, the octane number remains high and this effect
is not expected to have a significant impact on the vehicle performance or the fuel consumption.

Conclusions
Constraints on the supply of sustainable feedstocks
limit the degree to which biofuels are able to displace
fossil fuels. The miscibility of methanol with both ethanol and gasoline enables it to be blended into mixtures
which might facilitate the increased penetration of

alcohols in the transport fuel pool. Specific GEM


blends can be designed to be iso-stoichiometric and isoenergetic replacements for mixtures of gasoline and
ethanol. A simple analytical approach based on the
volumetric energy densities of the pre-blended components enables the formulation of these ternary blends
to be determined and allows the concept to be extended
to specifying quaternary GEMW blends (the so-called
hydrous ternary blends) which are also isostoichiometric and iso-energetic.
For the anhydrous blends, the distillation and vapour
pressure characteristics of E85- and E15-equivalent
ternary blends are considered. The limit-case binary
E85 blend considered has a significantly lower RVP
than that of the base gasoline. The RVP level increases
as the gasoline content and methanol content increase,
with some E85-equivalent blends having RVP values
above that of the base gasoline. The implied increase
in the vehicles evaporative emissions is likely to be
containable using the vapour recovery systems in
place in the logistics chain and carbon canister vapour
traps fitted to modern vehicles. FFV technology
implies that these vehicles should not suffer from hotweather handling issues, and this increased RVP
should have no effect on driveability in hot conditions. Specially formulated low-volatility RBOB
blending components could be used to counteract
these issues to some extent.
For the E15-equivalent blends the deviations of the
iso-stoichiometric GEM blends from the base gasoline
are less pronounced, as would be expected from the
lower total amount of alcohol included. All the isostoichiometric GEM blends were more volatile than
the base gasoline is and had a higher RVP, with the
highest-methanol-content blend having the largest
front-end deviation. Cold-weather startability should
not be expected to be impacted negatively on any of
these blends. The methanol-containing blends have a
significantly elevated RVP and thus may lead to

20
increased emissions; also the carbon canister vapour
traps may be overloaded, leading to fugitive hydrocarbon emissions.
All the fuels tested performed excellently in terms of
their inherent storage stabilities, confirming the results
of the induction period tests previously reported. The
blend stability against phase separation was evaluated
for all the anhydrous GEM blends. The blends with
only gasolineethanol have the highest water tolerance,
which decreases monotonically as the ethanol is displaced by increasing amounts of methanol and
gasoline.
The octane numbers of the ternary blends vary much
less than expected on the basis of the volume fractions
of the components. As the volume fraction of the
methanol component is increased, so the gasoline volume fraction, which has by far the lowest octane number, increases. The octane number response, however,
is much better represented using the mole fractions of
the components. According to the molar blending rules,
as the methanol and gasoline volume fractions increase
for the iso-stoichiometric GEM blends, the significant
difference in their molar masses results in a lower variation in the total alcohol mole fraction. In the current
work, E15-equivalent iso-stoichiometric blends were
prepared in order to ensure that the final blend RONs
would lie below 100 and therefore would be more accurately measured in a routine laboratory. The resulting
range of octane numbers is small, although there is a
generally decreasing trend as the methanol content
increases.

Proc IMechE Part D: J Automobile Engineering

3.

4.

5.

6.

7.

8.

9.

Acknowledgement
The authors would also like to acknowledge the useful
discussions held with Richard Stone (University of
Oxford) regarding several aspects of this work.
10.

Declaration of conflicting interests


The authors declare that there is no conflict of interest.
Funding
The authors under the affiliation of the University of
Bath would like to acknowledge the funding from the
Home Grown Cereals Authority (grant number) and
the UK Department for Environment, Food and Rural
Affairs during the HOOCH project in which some of
the concepts discussed above were developed during
their employment by Lotus Engineering.

11.

12.

References
1. IEA. Advanced motor fuels annual report 2010, Interna
tional Energy Agency, Paris, France, http://www.iea.org/
impagr/cip/pdf/AMF_2010_Annual_Report.pdf (2010,
accessed 2 January 2012).
2. US Environmental Protection Agency. 40 CFR Part 80.
Regulation of fuels and fuel additives: Renewable Fuel

13.

Standard program: final rule. Fedl Register 1 May 2007;


72(83): 23 900 24 014.
US Government. Energy Independence and Security Act
of 2007, Public Law 110 140, 110th Congress. Document
f:publ140.110, 2007.
US Environmental Protection Agency. 40 CFR Part 80.
Regulation of fuels and fuel additives: changes to Renew
able Fuel Standard program: final rule. Fedl Register 26
March 2010; 75(58): 14 670 14 904.
US Environmental Protection Agency. EPA finalises
2012 Renewable Fuel Standards. EPA Regulatory
Announcement EPA 420 F 11 044, Office of Transporta
tion and Air Quality, US Environmental Protection Agency,
Washington, DC, USA, December 2011, http://www.
epa.gov/otaq/fuels/renewablefuels/documents/420f11044.pdf
(accessed 19 February 2012).
Anderson JE, DiCicco DM, Ginder JM, et al. High
octane number gasoline ethanol blends: quantifying the
potential benefits in the United States. Fuel 2012; 97:
585 594.
US Environmental Protection Agency. Partial grant and
partial denial of Clean Air Act waiver application sub
mitted by Growth Energy to increase the allowable etha
nol content of gasoline to 15 percent; Decision of the
administrator. Fedl Register 4 November 2010; 75(213):
68 094 68 150, http://www.gpo.gov/fdsys/pkg/FR 2010
11 04/pdf/2010 27432.pdf (accessed 19/02/2012).
US Environmental Protection Agency. Partial grant of
Clean Air Act waiver application submitted by Growth
Energy to increase the allowable ethanol content of gaso
line to 15 percent; Decision of the administrator. Fedl
Register 26 January 2011; 76(17): 4662 4683, http://
www.gpo.gov/fdsys/pkg/FR 2011 01 26/pdf/2011 1646.pdf
(accessed 19/02/2012).
Directive 2009/28/EC of the European Parliament and of
the Council of 23 April 2009 on the promotion of the use
of energy from renewable sources and amending and
subsequently repealing Directives 2001/77/EC and 2003/
30/EC. Document 32009L0028. Off J Eur Un 2009;
L140: 16.
Amending Directive 98/70/EC as regards the specification
of petrol, diesel, and gas oil and introducing a mechanism
to monitor and reduce greenhouse gas emissions and
amending Council Directive 1999/32/EC as regards the
specification of fuel used by inland waterway vessels and
repealing Directive 93/12/EEC. Directive 2009/30/EC of
the European Parliament and of the Council of 23 April
2009.
Cooper J. Future fuels for transport: regulatory and sup
ply drivers. In: Institution of Mechanical Engineers: Inter
nal combustion engines: improving performance, fuel
economy and emissions, London, UK, 29 30 November
2011.
US Environmental Protection Agency/US Department of
Transportation/National Highway Traffic Safety Admin
istration. 40 CFR Parts 85, 86 and 600; 49 CFR Parts
531, 533 and 536. Light duty vehicle greenhouse gas emis
sions standards and corporate average fuel economy stan
dards: final rule. Fedl Register 7 May 2010; 75(88): 25
324 25 728.
Regulation (EC) 443/2009 of the European Parliament
and of the Council of 23 April 2009 setting emission per
formance standards for new passenger cars as part of the
Communitys integrated approach to reduce CO2

Pearson et al.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

emissions from light duty vehicles. Off J Eur Un 2009;


L140: 1 15.
Turner JWG, Pearson RJ, Dekker E, et al. GEM ternary
blends: testing iso stoichiometric mixtures of gasoline,
ethanol and methanol in a production flex fuel vehicle
fitted with a physical alcohol sensor. SAE paper 2012 01
1279, 2012.
Turner JWG, Pearson RJ, Dekker E, et al. GEM ternary
blends: removing the biomass limit using iso
stoichiometric mixtures of gasoline, ethanol, and metha
nol. SAE paper 2011 24 0113, 2011.
Turner JWG, Pearson RJ, Dekker E, et al. Evolution of
alcohol fuel blends towards a sustainable transport
energy economy. In: Prospects for bi fuel and flex fuel
light duty vehicles, Cambridge, Massachusetts, USA,
MITEI Symposium Series, Cambridge, Massachusetts,
USA, 19 April 2012, https://mitei.mit.edu/system/files/
2012 mitei symposium turner.pdf. Cambridge, Massa
chusetts: MIT Energy Initiative, Massacusetts Institute
of Technology, 2013.
Turner JWG, Pearson RJ, Dekker E, et al. Extending the
role of alcohols as transport fuels using iso stoichiometric
ternary blends of gasoline, ethanol and methanol. J Appl
Energy 2013; 102: 72 86.
Turner JWG, Pearson RJ, Bell A, et al. GEM ternary
blends: investigations into exhaust emissions, blend prop
erties and octane numbers. SAE paper 2012 01 1586,
2012.
Pearson RJ and Turner JWG. The role of alternative and
renewable liquid fuels in environmentally sustainable
transport. In: Folkson R (ed) Alternative fuels and
advanced vehicle technologies for improved environmental
performance: towards zero carbon transportation, Wood
head Publishing Series in Energy. Cambridge: Woodhead
Publishing, 2014: 19 51.
Pearson RJ, Turner JWG, Eisaman MD and Littau
KA.Extending the supply of alcohol fuels for energy
security and carbon reduction. SAE paper 2009 01 2764,
2009.
Pearson RJ, Eisaman MD, Turner JWG, et al. Energy
storage via carbon neutral fuels made from CO2, water,
and renewable energy. Proc IEEE 2012; 100(2): 440 460.
Maiorella BL, Blanch HW and Wilke CR.Economic eva
luation of alternative ethanol fermentation processes. Bio
technol Bioengng 1984; 27: 1003 1024.
Keuken H. Hydrous ethanol for gasoline blending cost
and energy savings. In: 5th annual meeting biofuels 2010,
Amsterdam, The Netherlands, 9 12 November 2010.
London: WRA.
Keuken H and de Jager HC.Environmentally improved
motor fuels. US Patent Application 2010/0325945 A1,
2010.
Collura MA and Luyben W. Energy saving distillation
designs in ethanol production. Ind Engng Chem Res 1988;
27: 1686 1696.
Makkee M, Czerwinski J and Comte P. The effect of
(hydrous) ethanol on the emission and performance of 2
and 4 stroke scooters. In: SAE 2009 Particle emissions of
2 stroke scooters, science, problems, solutions & perspec
tives, Monza, Italy, 11 12 June 2009. Warrendale, Penn
sylvania: SAE International. paper no. 2010 01 0794.
Anderson JE, Kramer U, Mueller SA and Wallington TJ.
Octane numbers of ethanol and gasoline ethanol blends

21

28.

29.

30.

31.
32.

33.

34.

35.

36.

37.

38.

39.
40.

41.

42.

43.

44.

estimated from molar concentrations. Energy Fuels 2010;


24: 6576 6585.
Anderson JE, Leone TG, Shelby MH, et al. Octane num
bers of gasoline ethanol blends: measurements and novel
estimation method from molar composition. SAE paper
2012 01 1274, 2012.
Andersen VF, Anderson JE, Wallington TJ, et al. Vapor
pressures of alcohol gasoline blends. Energy Fuels 2010;
24: 3647 3654.
Chen L and Stone R. Measurements of enthalpies of
vaporisation of isooctane and ethanol blends and their
effects on PM emissions from a GDI engine. Energy Fuels
2011; 25(3): 1254 1259.
Furey RL. Volatility characteristics of gasoline alcohol
and gasoline ether blends. SAE paper 852116, 1985.
Kretschmer CB, Nowakowska J and Wiebe R. Densities
and liquid vapor equilibria of the system ethanol
isooctane (2,2,4 trimethylpentane) between 0 and 50.
J Am Chem Soc 1948; 70(5): 1785 1790.
Lazzaroni MJ, Bush D, Eckert CA, et al. Revision of
MOSCED parameters and extension to solid solubility
calculations. Ind Engng Chem Res 2005; 44: 4075 4083.
Smith J, Nesss HV and Abbott MM. Introduction to
chemical engineering thermodynamics. 6th edition.New
York: McGraw Hill, 2001.
Chueh P and Prausnitz J. Vapor liquid equilibria at high
pressures: calculation of partial molar volumes in non
polar liquid mixtures. AIChE J 1967; 13(6): 1099 1107.
Tsonopoulos C, Dymond J and Szafranski A. Second vir
ial coefficients of normal alkanes, linear 1 alkanols and
their binaries. Pure Appl Chem 1989; 61(8): 1387 1394.
Muzikova Z, Pospisil M and Sebor G. Volatility and
phase stability of petrol blends with ethanol. Fuel 2009;
88: 1351 1356.
Qi DH, Liu ShQ, Liu JC and Bian YZh. Properties, per
formance, and emissions of gasoline ethanol blends in a
spark ignition engine. Proc IMechE Part D: J Automotive
Engineering 2005; 219(3): 405 412.
Skinner W and Viljoen H. Corrosion of carburettor mate
rials. Corrosion Coatings, South Africa 1981; 34: 5 9.
Guidelines for blending and handling motor gasoline con
taining up to 10% v/v ethanol. Report 3/08, prepared for
the CONCAWE Fuels Quality and Emissions Manage
ment Group by its Special Task Force FE/STF 24, CON
CAWE, Brussels, April 2008.
Donnelly RG, Heywood JB, LoRusso J, et al. Methanol
as an automotive fuel: a summary of research in the
M.I.T. Energy Laboratory. Report MIT EL 76 013,
Energy Laboratory, Massachusetts Institute of Technol
ogy, Cambridge, Massachusetts, USA, April 1976, http://
dspace.mit.edu/bitstream/handle/1721.1/27837/MIT EL
76 013 03592022.pdf?s (accessed 03/01/2012).
Neagu M, Rosca P, Dragomir R and Mihai O. The effect
of bioalcohol on the water solubility in reformulated gas
oline. Chem Engng Trans 2010; 21: 1291 1296.
Owen K and Coley T. Automotive fuels reference book.
2nd edition. Warrendale, Pennsylvania: SAE Interna
tional, 1995.
Green GJ and Yan TY. Water tolerance of gasoline
methanol blends. In: ACS Energy and Fuels Division
spring 1990 symposium on oxygenated motor fuels, Bos
ton, Massacusetts, USA, Vol 35(1), pp. 276 284, http://
www.anl.gov/PCS/acsfuel/preprint%20archive/Files/

22

45.

46.

47.

48.

49.

50.
51.

52.

53.

54.

55.

56.
57.

Proc IMechE Part D: J Automobile Engineering


35_1_BOSTON_04 90_0276.pdf (accessed 3 January
2012). Washington, DC: ACS.
Pearson RJ, Turner JWG and Peck AJ. Gasoline etha
nol methanol tri fuel vehicle development and its role in
expediting sustainable organic fuels for transport. In:
Institution of Mechanical Engineers: Low carbon vehicles
conference, London, UK, 20 21 May 2009. Cambridge:
Woodhead Publishing, 2009: 89 110.
Continental Automotive Systems USA. Flex fuel sensor
generation II specification. Working Document, Conti
nental Automotive Systems USA, Auburn Hills, Michi
gan, USA, private communication, 3 February 2010.
Lunati A and Galtier O. Determination of mixture of
methanol and ethanol blends in gasoline fuels using a
miniaturised NIR flex fuel sensor. JSAE paper 20119264,
SAE paper 2011 01 1988, 2011.
Hunwartzen I. Modification of CFR test engine unit to
determine octane numbers of pure alcohols and gasoline
alcohol blends. SAE paper 820002, 1982.
Yates A, Bell A and Swarts A. Insights relating to the
autoignition characteristics of alcohol fuels. Fuel 2009;
89(1): 83 93.
Atkins PW. Physical chemistry. 5th edition.Oxford:
Oxford University Press, 1994.
Vazquez Esparragoza JJ, Inglesias Silva GA, Hlavinka
MW and Bullin JA. Encyclopedia of chemical processing
and design. New York: Marcel Dekker, 1994, pp.
415 424.
Reddy S. Evaporative emissions from gasolines and
alcohol containing gasolines with closely matched volati
lities. SAE paper 861556, 1986.
Reddy SR. A model for estimating vapor pressures of
commingled ethanol fuels. SAE paper 2007 01 4006,
2007.
Greenfield ML, Lavoie GA, Smith CS and Curtis EW.
Macroscopic model of the D86 fuel volatility procedure.
SAE paper 982724, 1998.
Chen K. Fuel volatility modeling. Masters Thesis, Massa
chusetts Institute of Tecnology, Cambridge, Massachu
setts, USA,1994.
Prausnitz J and Shair F. A thermodynamic correlation of
gas solubilities. AIChE J 1961; 7(4): 682 687.
Martini G, Manfredi U, Mellios G, et al. Joint EUCAR/
JRC/CONCAWE Programme on: effects of gasoline
vapour pressure and ethanol content on evaporative emis
sions from modern cars. Technical Report EUR 22713
EN, European Commission, Directorate General Joint
Research Centre, Institute for Environment and Sustain
ability, Ispra, Italy, 2007.

Appendix 1
Notation
G
h
m
mi
M
Mi
n

Gibbs free energy


mass concentration of water in hydrous
ethanol
mass; mass of mixture
mass of component i
molar mass of the mixture
molar mass of component i
number of moles; number of moles of the
mixture

ni
NRON
p
pi
pRV
QLHV
Rmol
S
T
V
Vi
xi
yi

a
b
g
gi
m
r
ri
fi
Cstoich

number of moles of component i


research octane number
pressure exerted by the mixture
partial vapour pressure of component i
Reid vapour pressure
lower heating value
universal gas constant
entropy
temperature
volume; volume of the mixture
volume of component i
mole fraction in the liquid phase of
component i
mole fraction in the vapour phase of
component i
number of carbon atoms
number of hydrogen atoms
number of oxygen atoms
liquid-phase activity coefficient of
component i
chemical potential
density; density of the mixture
density of component i
fugacity of species i
stoichiometric air-to-fuel ratio (by mass)

Subscripts
e
g
m
stoich
w

ethanol
gasoline
methanol
stoichiometric
water

Abbreviations
AFR
CAFE
ECU
EMS
EX
EU
FFV
FVI
GEM
GEMW
GHG
MON
MOSCED
MX
RBOB
RFS
RON
RVP

air-to-fuel ratio
corporate average fuel economy
electronic control unit
energy management system
blend of X vol % ethanol in gasoline
European Union
flexible-fuel vehicle
fuel volatility index
gasolineethanolmethanol
gasolineethanolmethanolwater
greenhouse gas
motor octane number
modified separation of cohesive energy
density
blend of X vol % methanol in gasoline
reformulated blend-stock for oxygenate
blending
renewable fuel standard
research octane number
Reid vapour pressure

Pearson et al.

23

Appendix 2
Theoretical foundations
The density of a mixture consisting of k components
can be expressed as
r=

m
V
Pk

= Pik
i

mi

19

Vg
Ve
Vm
+
+
=1
V
V
V

where the mass of blend component i is given by


mi = ri Vi . The pre-blended volume fractions of the
mixture components clearly must satisfy the equation
k
X
i

Vi = V

20

k
X

Vi
=1
V
1

where k = 3 and k = 4 for ternary blends and quaternary blends respectively. For equation (20) to be valid
for the blend after mixing, the post-mixing volume
must be equal to the linear sum of the volumes of the
pre-blended components. In fact, there can be small
departures from this assumption because of the differences in the way that liquids having different molecular
sizes pack together. This effect can be quantified using
the concept of partial molar volumes discussed in
Appendix 4 but, to a first approximation, equation (20)
is sufficient to estimate the volumetric energy contents
of the mixtures of liquids considered here.
Measurements to support this assumption are discussed
briefly in the fourth section.
Using the above expressions, the density of the mixture can be written as

k 
X
Vi
r=
ri
V
i 1

21

The lower heating value QLHV of the mixture can be


calculated from the mass fractions of the mixture components in the form
QLHV =

k 
X
mi
i

QLHVi

which can be rewritten as


QLHV =


k 
X
ri Vi
QLHVi
r V
i 1

23

This leads to the expression for the volumetric energy


density of the mixture in the form

k 
X
Vi
rQLHV =
ri QLHVi
V
i 1

24

Vg
Ve
Vm
QLHVg + re
QLHVe + rm
QLHVm
V
V
V
26

Substituting for the volume fraction of gasoline in


equation (26) using equation (25) and rearranging give
the volume fraction Vm =V of methanol required in a
ternary blend for a given ethanol concentration Ve =V
and volumetric energy density rQLHV of a target mixture as
rg QLHVg  rQLHV
Vm
Ve re QLHVe  rg QLHVg
=
+
V
V rg QLHVg  rm QLHVm
rg QLHVg  rm QLHVm

27

The stoichiometric combustion of an oxygenated


hydrocarbon fuel in air can be represented in the form



b g
79
C a Hb Og + a + 
O2 + N2 !
4 2
21


28
b
b g 79
a CO2 + H2 O + a + 
N2
2
4 2 21
or, expressed per mole of carbon atoms, as



b
g
79

O2 +
N2
CHb=a Og=a + 1 +
4a 2a
21


b
b
g 79
H2 O + 1 +

N2
! CO2 +
2a
4a 2a 21

29

Thus the stoichiometric AFR is given by


Cstoich =

22

25

Using equation (6), the volumetric energy density of


such a mixture can be expressed as
rQLHV = rg

or

If the density r is expressed in units of kilograms per


litre and the energy content (lower heating value) QLHV
is expressed in units of megajoules per kilogram, the
units of the volumetric energy density are megajoules
per litre.
If the subscripts g, e and m denote the mixture components gasoline, ethanol and methanol respectively,
equation (20) can be written in the form

1 + b=4a g=2aMO2 + 79=21MN2 


MC + b=aMH + g=aMO

30

where Mj is the molar mass of species j . Clearly, from


equation (30), fuels with identical molar hydrogen-tocarbon ratios b=a and identical oxygen-to-carbon
ratios g=a will have identical stoichiometric AFRs. In
order to calculate the species molar ratios for a mixture, the species mass fractions must be calculated from
the volumetric concentrations of the mixture. The mass
fraction of chemical species j in the mixture is given by

k 
X
mj
mji mi
=
m
mi m
i 1

31

24

Proc IMechE Part D: J Automobile Engineering

where mji =mi is the mass fraction of species j in the mixture component i. The mass fraction of component i in
the mixture is given by
mi
r Vi
= i
m
r V

32

where the mixture density is obtained from equation


(21). The molar hydrogen-to-carbon and oxygen-tocarbon ratios can now be evaluated from the expressions

34

and
rQLHV = rg

35

respectively. Mixtures of water in ethanol (hydrous


ethanol; usually hydrous ethanol is defined as ethanol
with a water concentration greater than or equal to that
at which it forms an azeotrope (4.4 mass %)) defined
by the mass concentrations as
mw
= hw, e
me + mw

36

can also be characterised in terms of their volume fractions by introducing the densities of the components so
that equation (36) can be rewritten as
Vw
r Ve hw, e
= e
V
rw V 1  hw, e

hw, e g

38

A generalised version of equation (38) can be written


for hydrous alcohol a by defining the water content as
mw
= hw, a
ma + mw

39

40

and
33

where mH =m and mO =m are given by equation (31).


The analysis below shows how the basic concept is
easily extended to account for the use of hydrous alcohols, giving quaternary blends, and focuses on obtaining blends of nominally equal volumetric energy
contents while maintaining the blend stoichiometric
AFRs within acceptable limits.
Equations (25) and (26) are easily developed to
account for the water content in the mixture so that
they take the forms

Vg
Ve
QLHVg + re QLHVe
V
V
Vm
Vw
QLHVm + rw
QLHVw
+ rm
V
V

rg QLHVg rQLHV
rg QLHVg rm QLHVm

Vw
r Va hw, a
= a
V
rw V 1  hw, a

and

Vg
Ve
Vm
Vw
+
+
+
=1
V
V
V
V

rg QLHVg f1 + re =rw hw, e =1


rg QLHVg rm QLHVm

so that the generalised forms of equations (37) and (38)


are

b
mH =m MC
=
mC =m MH
a

g
mO =m MC
=
mC =m MO
a

Ve re QLHVe
V

Vm
V

37

Combining equations (34), (35) and (37), substituting


for Vg =V and setting QLHVw = 0 give an expression
which is equivalent to equation (27) for hydrous ternary (quaternary) blends in the form

Vm
V


Va
C ra QLHVa
V





r
hw, a
+B
rg QLHVg 1 + a
rw 1 hw, a

41

respectively, where B is given by


B=

rg QLHVg  rQLHV
rg QLHVg  rm QLHVm

42

and C is given by
C=

rg QLHVg

1
 rm QLHVm

43

Note that the assumption of setting QLHVw = 0 in equation (35) above is open to discussion. Because the presence of water in a liquid fuel being measured in a bomb
calorimeter will change the enthalpy of vaporisation of
the mixture but will not contribute to the chemical
availability of the fuel in the test, it can be argued that
the lower heating value of the water fraction should be
negative and equal in magnitude to its enthalpy of
vaporisation. Such a modification would not make a
significant difference to the results shown in the present
work. Knowledge of the variation in the enthalpy of
vaporisation as a function of the mixture composition
would also be required. Because of the significant difference between the densities of water and the components in the blends and the effects of strong hydrogen
bonding in water (much stronger than even methanol
and ethanol), this variation is likely to be non-linear.

Appendix 3
Comparison of the volume fraction and the
mole fraction
Figure 25 shows the relationship between the volume
fraction and the mole fraction for several different alcohol fuels mixed with a standard-specification gasoline
pump fuel. It can be seen that for methanol and ethanol

Pearson et al.

Figure 25. Mixtures of alcohols with gasoline: the mole


fraction of alcohol as a function of the volume fraction.

there is considerable difference between the concentrations defined in terms of the volume fraction and the
mole fraction. For ethanol a 10% volume fraction in
gasoline equates to a 23% ethanol concentration by
mole fraction. In terms of the terminology defined
above, E10 on a volumetric basis could be regarded as
E23 on a molar basis, more than double the concentration implied by the volume fraction. In the same way,
E20, E50 and E85 defined on a volumetric basis become
E40, E73 and E94 respectively on a molar basis. In the
case of methanol, M10, M20, M50 and M85 defined on
a volumetric basis become M30, M49, M79 and M96
respectively on a molar basis. Therefore the mole fraction of what is referred to as M10 by volume fraction is
three times higher than might be inferred from the volume fraction. In a 50:50 volumetric mixture of methanol and gasoline, almost 80% of the molecules are
those of the alcohol.
It is also apparent from Figure 25 that for the higher
alcohols there is significantly less difference between
the mole fractions and the volume fractions. For mixtures of the particular gasoline used here and octanol
(C8H17OH) there is virtually no difference between the
volume fractions and the mole fractions. This is
because the higher density of octanol (the density of
octan-1-ol, or 1-octanol, was used) balances the effect
of the fact that the molecular mass of octanol is higher
than that of gasoline. For decanol the ratio of its density to its molar mass (decan-1-ol was used) is such that
the mole fractions of the alcohol in the gasoline are
lower than the corresponding volume fractions.

Appendix 4
Effect of the partial molar volume: anhydrous
and hydrous blends
Limitations of the simple approach to calculating the
volumetric energy of the blends in the second and third
sections are caused by the non-linear molecular packing
characteristics of the different molecules. This can be
quantified by measuring the partial molar volumes of
the various combinations. The approach defined in the

25
second section, for anhydrous blends, is sufficiently
accurate to prevent any activity of the indicator lights
of a vehicle in a current FFV, as reported by Turner
et al.14,18 The section below gives a brief description of
the concept of partial molar volumes.
The literature quantifying the effects of partial molar
volumes in fuel blends is sparse; many publications do
not acknowledge that it needs to be considered. This
may be the case where blends of fuels use components
with similar molecular sizes, or where the concentrations of the additional components are very low. The
phenomenon is raised as a possible concern in the present work because the ratio of the molecular sizes of
some of the blend components with respect to others is
very large, e.g. gasoline versus water. The effects of this
phenomenon will not be quantified here; a subsequent
publication will aim to analyse the fuel blends considered in the present work. Below, only a brief description of the concept is outlined in order to clarify its
potential importance.
The total volume of the liquid mixture components
after mixing may often depart significantly from the
sum of the volumes of the components prior to mixing.
This is because the volume occupied by a given number
of molecules in a liquid mixture depends on the identity
of the molecules that surround them. For example, when
1 mol of water is added to a huge volume of water at
25 C, the volume increases by 18 cm3. When 1 mol of
water is added to a huge volume of ethanol at 25 C,the
volume increases by only 14 cm3 because of the higher
density packing of the water molecules in this case.50 To
deal with this variation, chemists and chemical engineers
employ the concept of partial molar volumes. If volume
fractions Vi /V, rather than partial molar volumes, are
employed for mixtures of liquids then, owing to the variation in the packing effects discussed above, the partial
volumes Vi considered should refer to the volume occupied by the component in isolation (before the liquids
are mixed) and the total volume V should refer to sum
of the partial volumes in isolation rather than to the volume of the blend after mixing.
The partial molar volume is the change in the volume of a mixture on the addition of 1 mol of a substance to a large excess of the mixture50 defined as
 
V
Vmoli =
44
ni p, T, n9
In this equation, ni is the number of moles of component i and n9signifies that the amounts of all other substances present are constant. The value of the partial
molar volume depends on the composition of the mixture affecting the way in which the molecules being
added pack together with those in the recipient mixture. When the composition of a two-component mixture of substances A and B is changed by the addition
of dnA mol of A and dnB mol of B, the total volume of
the mixture changes by

26

Proc IMechE Part D: J Automobile Engineering



dV =

V
nA


dnA +

p, T, nB

V
nB


dnB
p, T, nA

45

= Vmol A dnA + Vmol B dnB

When the partial molar volumes of the two components


at the composition (and temperature) of interest are
known, the total volume of the mixture can be found
from the equation
V = nA Vmol A + nB Vmol B

46

This equation, defined in terms of partial molar


volumes after mixing, is the equivalent of equation (2)
defined in terms of the partial volumes of the substances before mixing.

For ideal binary solutions the Gibbs energy of mixing is minimum when the mole fractions of the constituents are 0.5 as this corresponds to the state in which
the entropy change of mixing is a maximum. In nonideal solutions such as the gasolinealcohol mixtures
described above, in which the molecular interactions
between the components differ, there may be an
enthalpy change on mixing.30 There is also an additional contribution to the entropy change arising from
the tendency of molecules of one type to cluster instead
of mixing freely with the other molecules.50 In some
such situations the entropy of mixing may become positive, resulting in phase separation, as discussed above.

Appendix 6
Appendix 5

Numerical modelling: methodology


and validation

Thermodynamics of mixing
The chemical potential of mixture component A in the
liquid phase (indicated by (l)) is given by50
 
pA
mA l = mA l + Rmol T ln 
47
pA
where mA l is the chemical potential of A as a pure
liquid. For an ideal solution the Raoult law can be combined with equation (47) to give
mA l = mA l + Rmol T ln xA

48

When A and B are separate, the Gibbs free energy of


the system is given by
Ginitial = nA mA l + nB mB l

49

and, when they are mixed, the Gibbs free energy is


obtained by applying equation (48) to both components to yield


Gfinal = nA mA l + Rmol T ln xA


50
+ nB mB l + Rmol T ln xB
The Gibbs energy of mixing can then be expressed by
Dmix G = Gfinal  Ginitial
= nRmol TxA ln xA + xB ln xB

where n = nA + nB . Because
 
G
= S
T p, n

51

52

it follows from equation (51) that the entropy of mixing


is given by


Dmix G
Dmix S = 
T
53
p, nA , nB
=  nRmol xA ln xA + xB ln xB
As expected, Dmix S . 0 for all compositions because
ln x \ 0.

Numerical prediction of the RVP


Algorithms that model the full ASTM RVP test procedure have been proposed in the literature.51 However,
Reddy,52,53 while acknowledging that the gasoline
vapour pressure at 310.95 K (the ASTM test temperature for RVP) will be slightly higher than the RVP measured using the prescribed ASTM test procedures,
convincingly argued that the RVP is satisfactorily represented by a blends bubble-point pressure at 310.95
K. In the present work, the RVP is calculated by a thermodynamic model in a similar manner to that proposed
by Reddy.52,53 The RVP model then follows the gf
methodology described in the main text with minor
changes to the MOSCED parameters for ethanol and
methanol so as to improve the prediction accuracy for
gasolineethanol and gasolinemethanol blends across
the full range of alcohol contents. The RVP model was
validated against a range of literature data. The absolute average deviation Dabs (%) between the predicted
and the measured values for a sample of 10 fuel compositions was 0.95%, with
Dabs =

100 X pRVpred  pRVexp


n
pRVexp

54

where pRVpred and pRVexp are the predicted value and the
measured value respectively of the RVP. Figure 26
shows the predicted variation in the RVP of indolene
with increasing levels of ethanol and methanol splash
blending. The experimental data plotted in Figure 26
are those from the work by Furey.31

Numerical prediction of the distillation curves


The numerical simulation of the gasoline distillation
curves and the validation of numerical results against
the experimental distillation data present an interesting
technical challenge. First, it should be noted that the
experimental data obtained by either the ASTM D86
test method or the ISO equivalent method are a

Pearson et al.

27

Figure 26. Predicted variation in the RVP of indolene with (a) ethanol and (b) methanol blending, validated against the
experimental data from the work by Furey.31
psi: lbf/in2.

function of the test apparatus, the procedure and the


fuel handling. A number of researchers, including
Greenfield et al.54 and Chen55 have developed models
of the ASTM D86 distillation process. The work of
Greenfield et al.54 is particularly interesting in that a
detailed multi-zone thermodynamic model was developed which considered the dynamics of the ASTM D86
test apparatus and procedure. Nevertheless, despite the
relative sophistication of this approach, it is notable
that the final predictions of the distillation temperatures are still significantly dependent on the accuracy
of the underlying thermodynamic calculations. Second,
it must be recognised that real-world gasolines will contain a large number (greater than 120) of individual
hydrocarbon species, some of which may be listed as
unknown during gas speciation. The exact simulation
of these fuels is therefore impractical. Accordingly, it is
normal practice within the community to adopt a
major-component modelling approach, whereby a
full-boiling-range gasoline is represented by a reduced
number of key hydrocarbon species. The accuracy of
the simulation is then to a significant degree dependent
on the ability of the model fuel to mimic the distillation behaviour of the target gasoline.
In the present work, the CRC RF-02-03 base gasoline was simulated by a 16-component hydrocarbon
mixture as indicated in Table 5, and an equilibriumbased calculation method is used to simulate the distillation process as follows. Given a liquid fuel of known
composition, the model performs vapourliquid equilibrium calculations using the gf methodology to find
the bubble-point temperature and the equilibrium

Table 5. Model RF 02 03 fuel composition.


Species

Amount (mass %)

n butane
Isopentane
Isohexane
Isooctane
Cyclopentane
2,2 dimethybutane
1 hexene
Toluene
o xylene
Ethylbenzene
m propyltoluene
n nonane
n decane
Naphthalene
Methyl tert butyl ether
n dodecane

2.2
13.18
1.66
12.71
23.27
1.97
3.36
10.90
9.90
1.99
7.50
2.93
5.92
0.83
0.07
1.61

vapour composition of the mixture. A small quantity


of the fuel at the equilibrium vapour concentration is
then removed from the liquid mixture. The bubblepoint calculation and the liquid removal process are
then repeated until distillation is complete. At each calculation step, the liquid volume is calculated and the
fuel vapour is removed from the system (this is then
equivalent to a perfect, or infinitely fast, diffusion
process).
The distillation model seeks to mimic the ASTM
D86 process to some extent by controlling the dissolved
gases within the liquid mixture. As described by numerous researchers including Chen55 and Greenfield,54 the
initial boiling point reported by the distillation

28

Proc IMechE Part D: J Automobile Engineering

Figure 27. (a) Predicted indolene, (b) predicted indolene + 10 vol % ethanol and (c) predicted indolene + 10 vol % methanol
distillation curves, validated against the experimental data from the work by Furey.31

calculation is highly sensitive to the mole fraction of


dissolved gases, in this calculation oxygen and nitrogen,
which are assumed to be present in the liquid fuel. In
the present work the mass fractions of the dissolved
gases at the start of the distillation are calculated
according to the thermodynamic correlation reported
by Prauznitz and Shair56 with the partial pressures of
the two gases estimated as 0.79 (101 RVP (kPa)) and
0.21(101 RVP (kPa)) for nitrogen and oxygen respectively. The dissolved air is reduced linearly through the
calculation at a rate proportional to the exponential of
the initial RVP.
The performance of the present authors distillation
model was assessed against a wide range of literature

data for ethanol and methanol blended fuels of various


complexities. Figure 27(a), (b) and (c) show the numerical predictions for the distillation curves of indolene,
indolene + 10 vol % ethanol and indolene + 10 vol
% methanol against the experimental data obtained by
Furey.31
Figure 28 shows the model predictions for the more
complex case of a full-boiling-range gasoline (represented by a 17-component model) with 5 vol % ethanol
and 10 vol % ethanol, validated against the experimental data obtained by Martini et al.57 The results shown
in Figures 27 and 28 indicate that the present authors
zero-dimensional thermodynamic distillation model is
able to represent the ASTM D86 distillation

Pearson et al.

29
characteristics of both the base fuel and the non-ideal
interactions of the base fuel with ethanol and methanol
with good accuracy.

Figure 28. Predicted distillation curves for RF 02 03 summer


gasoline base fuel, base fuel + 5 vol % ethanol and base fuel +
10 vol % ethanol splash blended fuels, validated against the
experimental data from the work by Martini et al.57

You might also like