Applied Thermal Engineering 102 (2016) 302310

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Modelling of evaporative losses in n-alcohol/diesel fuel blends

Juan P. Hernndez a, Magn Lapuerta b,, Reyes Garca-Contreras b, John R. Agudelo a
a
Mechanical Engineering Dept, Universidad de Antioquia (UdeA), Calle 70 #52-21, Medelln, Colombia
b
Escuela Tcnica Superior de Ingenieros Industriales, Universidad de Castilla La-Mancha, Avda. Camilo Jos Cela s/n, 13071 Ciudad Real, Spain

h i g h l i g h t s

Modelled evaporative losses from blends agreed with experimental results.

Diffusion and activity coefficients were proved to have significant effect.

Evaporation rate was proportional to the decrease in the alcohol carbon number.

Paradoxically, evaporative losses are faster in blends with lower alcohol content.

a r t i c l e i n f o a b s t r a c t

Article history: Alcohol/diesel fuel blends allow reducing the formation of soot in diesel engines. However, its use in a
Received 28 October 2015 large scale framework still faces the challenge of high evaporative losses under storage conditions which
Accepted 2 March 2016 is hindered by the high volatility of the alcohols. Since the design of the most adequate fuelling system for
Available online 19 March 2016
each blend requires detailed knowledge about the evaporation losses from the liquid phase, in this work a
method for calculating the evaporation of pure n-alcohols (from C1 to C5 carbon atoms) and blends of
Keywords: these alcohols with ultra-low sulphur diesel (ULSD) was proposed. Alcohols vapour pressure was
Diesel engine
determined with the Antoine equation, while the diffusion coefficient of alcohols in air was calculated
Alcohols
Biofuels
as a function of carbon number and temperature. The coefficients of activity were obtained by the
Thermodynamics combination of continuous thermodynamic (gamma distribution function) and a modified UNIFAC-
Dortmund-Continuous method. The evaporation losses model is based on Ficks law and exhibited good
agreement with experimental data for all alcohols and blends at 20% content of all alcohols and at 2060%
contents of n-butanol with diesel fuel. The evaporation was found to decrease approximately with an
inversely proportional rate with respect to the increase in the carbon chain length of the alcohol. The
paradox of faster evaporative losses from blends with lower alcohol content is explained.
2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction engine loads [4,5]. However, different challenges will have to be

Blending diesel fuels with bioalcohols is one of the most efficient
means to introduce a renewable fraction in the fuel, to improve its
sustainability, and to provide some oxygen content, this being
essential to reduce soot engine emissions. Additionally, compared
with biodiesel/diesel blends, alcohol/diesel (or even alcohol/biodie-
sel/diesel blends) can provide similar renewable fraction, similar
oxygen content and superior soot-reduction potential with much
lower cost and energy consumption. The hydroxyl group of the
alcohol molecules contributes to reduce soot formation, and conse-
quently particulate emissions, even more than other functional
groups with the same oxygen content [13], especially at high
Corresponding author. Tel.: +34 926 295431.
E-mail address: Magin.Lapuerta@uclm.es (M. Lapuerta).
faced for the future extension of the use of these alcohols as diesel
fuel components to a large scale.
First, the energy and environmental life-cycle assessments of
the bioalcohols must be advantageous with respect to other biofu-
els. The recent commitments to further reduce greenhouse emis-
sions from transportation sources make this a crucial factor.
Ethanol has proved to have a significant potential for reducing
life-cycle greenhouse gas emissions, as acknowledged by the Euro-
pean directive 28/2009/CE [6], where the typical greenhouse sav-
ings range from 21% (in the case of using wheat as feedstock) to
75% (in case of sugar cane ethanol) or even to 87% (in case of wheat
straw). Methanol is also gaining interest as a renewable fuel, since
its production derived from gasification of waste biomass followed
by catalytic synthesis at high pressure provides both high yield
(4555% in weight) and energy efficiency (7075%) [7]. Biomass

http://dx.doi.org/10.1016/j.applthermaleng.2016.03.027
1359-4311/ 2016 The Authors. Published by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 J.P. Hernndez et al. / Applied Thermal Engineering 102 (2016) 302310
Nomenclature
A area of the cylindrical vessel
A, B and C coefficients of the Antoine equation
D diffusion coefficient
d diameter of the cylindrical vessel
f fugacity
Hr mean relative humidity
L free height of the vessel with respect to the liquid inter-
phase
m_ mass flux
m00 mass flux per area
n number of carbon atoms
p pressure
R gas constant
T temperature
t time
V volume
v mole volume
W molecular weight
w mass fraction
x mole fraction
sources are even preferable for biomethanol production than for
bioethanol because the last is a high-cost and low-yield product
[8,9]. Butanol has also proved to have an interesting potential for
saving life-cycle greenhouse emissions when it is produced from
acetobutylicum fermentation, as far as the bio-acetone obtained
as co-product is valued as such [10]. Other higher alcohols could
also become a sustainable alternative as far as they can be pro-
duced in biorefineries together with a wide range of commodities.
Second, the fuel storage and handling difficulties derived from
the high volatility of alcohols must be addressed. Previous experi-
ences with ethanoldiesel blends have proved that different tech-
nical modifications in the fuelling system are necessary, such as
the installation of seals for fuel tanks, security valves on both vehi-
cle fuel tanks and fuel station tanks, and special systems for recov-
ering of fuel vapours on vehicles and fuel stations and to avoid
vapour locks. These restrictions constitute a barrier for the massive
use of these blends in private vehicles, and have limited their prac-
tical use to captive fleets, such as those of urban buses or public
construction and civil machinery.
Third, some key properties of the alcohol blends affecting the
injection system may limit their use beyond certain limits. A pre-
vious work was published studying blending stability, lubricity,
viscosity and cold filter plugging point as the key properties of
alcoholdiesel blends [11].
Finally, among the properties affecting the combustion process,
cetane number is certainly the most limiting one. Although diesel
engines need to be fuelled with high cetane number-fuels (above
51 as stated by norm EN 590) to facilitate autoignition and provide
short ignition delay, alcohols exhibit low cetane numbers [12,13].
For this reason, the direct use of alcohols as fuels for unmodified
diesel engines is not recommended, unless they are blended with
diesel fuel. However, long-chain alcohols have higher cetane num-
ber than short-chain alcohols [14,15], this enhancing their interest
for being used in diesel engines.
The work presented here is related to the second challenge
mentioned. The design of the most adequate fuelling system for
each blend requires detailed knowledge about the evaporation
losses from the liquid phase. This work proposes a model to predict
evaporation losses from alcohols (from methanol to n-pentanol)
blended with a diesel fuel under stagnant air conditions, and
303
z volume fraction, compressibility factor, height
/ coefficient of fugacity
c activity coefficient
q density
Subscripts
A alcohol
Air air
b boiling
c critical
D diesel fuel
i instant
L free height of the vessel
0 interphase (z = 0)
v vapour
Superscripts
l liquid
v vapour
describes the guidelines to be followed for modelling evaporative
losses from other base fuels and under more realistic evaporation
conditions.
Although some literature has reported modelling studies on
vapour pressure and evaporation losses in alcoholgasoline blends
[16] or in biodieseldiesel blends [17], very few have been found
addressing this issue in alcoholdiesel blends, and in any case, these
studies have been applied to droplet evaporation in ethanoldiesel
blends from combusting [18] or non-combusting [19] liquid sprays,
rather than to evaporation losses from fuel storage systems.
2. Tested fuels, blends and experimental conditions
2.1. Pure fuels
The main measured specifications of pure alcohols and ULSD
are shown in Table 1. The diesel reference fuel was supplied by
the Colombian petroleum company (Ecopetrol). The alcohols used
in this work were methanol (Honeywell, 99.9% purity), ethanol
(Panreac, 99.5% purity), n-propanol (Merck, 99.5% purity),
n-butanol (Panreac, 99.5% purity), and n-pentanol (Panreac, 98%
purity).
2.2. Blends
Blends of 20% by volume of the five n-alcohols in ULSD fuel
were selected to experimentally validate the evaporation model.
Additionally, blends of n-butanol with 30%, 40%, 50% and 60%
content in diesel fuel were also selected to study the effect of the
alcohol content in the evaporation losses. The volume, mass and
mole fractions of each alcohol (identified with subscript A) in each
of the tested blends are shown in Table 2.
The 20% blends were selected because this concentration is high
enough to modify the blend properties with respect to those of
pure diesel fuel, and additionally, because they might be eventu-
ally of interest to be implemented in the market. For the study
about the effect of the alcohol content, n-butanol was chosen
because it presents good miscibility with diesel fuel at all
304 J.P. Hernndez et al. / Applied Thermal Engineering 102 (2016) 302310

Table 1
Properties of the pure fuels tested.

Properties ULSD Methanol Ethanol n-Propanol n-Butanol n-Pentanol

a
% C (wt.) 86.13 37.48 52.14 59.96 64.82 68.13
% H (wt.) 13.87a 12.58 13.13 13.42 13.60 13.72
% O (wt.) 0a 49.93 34.73 26.62 21.59 18.15
Molecular weight, W (kg/kmol) 211.7b 32.04 46.07 60.09 74.12 88.15
Critical temperature, Tc (K) 512.64 513.92 536.78 563.05 588.15
Critical pressure, pc (bar) 80.97 61.48 51.75 44.23 39.09
Normal boiling temperature (at 1 atm), Tb (C) 64.7 78.3 97.1 117.5 137.9
Mole volume of liquid at boiling temp, vb (cm3/mol) c
41.75 52.10 64.31 79.53 93.24
Paraffins (%) 32.55 0 0 0 0 0
Naphthenes (%) 35.95 0 0 0 0 0
Aromatics (%) 31.50 0 0 0 0 0
Sulphur (ppm w/w) 12 0 0 0 0 0
Purity (% v/v) 99.9 99.5 99.5 99.5 98
Density at 15 C (kg/m3) 835 791.3 789.4 803.7 809.7 814.8
Kinematic viscosity at 40 C (cSt) 2.72 0.58 1.13 1.74 2.22 2.89
Water content (ppm wt.) 57 1382 2024 309 1146 2970
Cetane number 52d 5e 8f 12g 17h 18.2i
a
Calculated by ASTM D5291.
b
Calculated by AspenTech HYSYS software.
c
Calculated from the equation proposed by Yamada and Gunn [20].
d
Provided by the fuel manufacturer.
e
Obtained from Refs. [13] and [14].
f
Obtained from Refs. [13] and [21].
g
Obtained from Ref. [22].
h
Obtained from Refs. [5] and [22].
i
Obtained from Ref. [15].

3. Determination of alcohol parameters

Table 2
Volume, mass and mole fractions of each alcohol in the tested alcoholdiesel blends.
3.1. Alcohol vapour pressures
Alcohol blend Volume fraction, zA Mass fraction, wA Mole fraction, xA
Methanol 20% 0.200 0.178 0.535 Evaporation losses directly depend on the concentration of
Ethanol 20% 0.200 0.178 0.439
vapour on the liquidgas interphase, which are related to the
n-Propanol 20% 0.200 0.180 0.378
n-Butanol 20% 0.200 0.181 0.333 vapour pressure of the volatile compounds (i.e., alcohols). The
n-Pentanol 20% 0.200 0.182 0.296 determination of the vapour pressure (pv,A) was made by means
n-Butanol 30% 0.300 0.284 0.500 of the Antoine equation, with pv,A in Pa and temperature in K:
n-Butanol 40% 0.400 0.382 0.609
n-Butanol 50% 0.500 0.481 0.700 B
log pv ;A A  1
n-Butanol 60% 0.600 0.582 0.778 T C
Coefficients A, B and C are specific for each alcohol and are col-
lected in different data bases, where often different values are dis-
concentrations [11], contrary to light alcohols, and because of its tinguished as a function of the temperature range. The data used in
high future potential as a fuel component, as mentioned above. this work are shown in Table 3. Fig. 1 presents pv,A as a function of
temperature for all alcohols.
2.3. Experimental tests
3.2. Diffusion coefficients
All samples tested were introduced in beakers of 50 ml under a
controlled room temperature of 20 1.0 C. An atmospheric
The correlation proposed by Lapuerta et al. [25] for the determi-
pressure of 0.855 0.5 atm was determined with a conventional
nation of the diffusion coefficients of alcohols in their binary alco-
piezoelectric barometer. A relative humidity of 70 5% (% Hr)
holair gaseous mixtures at reduced pressure, was used (Eq. (2)),
was measured with a 10.2 cm synthetic-hair thermo-hygrometer.
with n being the carbon number of the alcohols. Results showed
Both parameters were constant throughout the test. In all tests,
that diffusion coefficients are higher as carbon chain length
an initial volume of 40 mL was used and, periodically, it was
decreases, as shown in Fig. 2.
weighted in a balance model Shimadzu Auw 120 D with an accu-
racy of 0.001 g and time was registered with a chronometer with ln DAAir 5:386  0:686n ln T  32:138  3:604n 2
an accuracy of 0.01 s. From both accuracies, the experimental
uncertainty for the evaporation rate was calculated as
0.00034 g/h. The beakers and the balance were introduced into 3.3. Activity coefficients
a climatic chamber in order to avoid ambient air flow fluctuations
during measurements. During first 24 h the measurements were Continuous thermodynamic theory and modified UNIFAC
more frequent to evaluate in detail the evolution of evaporation method were combined to calculate the coefficients of activity of
losses, especially for the most volatile alcohols. After that, samples the n-alcohols/diesel blends following the methodology described
were weighted each 24 h. The total duration of the experiments in Ref. [26]. The variation of activity coefficients with of the alcohol
varied from 84 to 96 h. Once experimental results were obtained, content in blends is shown in Fig. 3. From this figure, two conclu-
these were compared with the modelling results. sions can be obtained:
 J.P. Hernndez et al. / Applied Thermal Engineering 102 (2016) 302310 305

Table 3
Antoine coefficients for different alcohols, obtained from Refs. [23,24].

Antoine coefficient Range Methanol Ethanol n-Propanol n-Butanol n-Pentanol

A T < Tb 10.20587 10.32907 10.31384 9.54607 9.68277
T > Tb 10.09503 9.80607 9.59871 9.39031 8.97383
B T < Tb 1582.27 1642.89 1690.864 1351.555 1492.549
T > Tb 1521.23 1332.04 1300.491 1254.502 1106.11
C T < Tb 33.45 42.85 51.804 93.34 91.62
T > Tb 39.15 73.95 86.364 105.246 134.587

Fig. 1. Vapour pressure of alcohols as a function of temperature. Fig. 3. Activity coefficients of alcohols in function of mole fraction of alcohol at
mole fraction in liquid phase of blends.

4. Evaporative losses model

4.1. Vapourliquid equilibrium condition

In multi-component and multi-phase systems, the equilibrium

condition is expressed by the equivalence between the fugacity
of each component in each phase and the fugacity of that compo-
nent in every other phase. In the case of liquidvapour equilibrium
of dieselalcohol blends, since the vapour pressure of diesel fuel is
very low in comparison with the alcohols, only the equivalence
between the fugacity of the alcohol in the liquid and vapour phases
is considered:
v l
fA fA 3
The fugacity of the alcohol component in the vapour phase is
related to the coefficient of fugacity (/vA ) and to its mole fraction
in the vapour phase (xvA) through Eq. (4):
v
f A /vA  xvA  p 4
Fig. 2. Diffusion coefficients of alcohols in alcoholair mixtures as a function of In the liquid phase, the fugacity of alcohol component is related
temperature.
to its activity coefficient (clA ) and to its fugacity when the alcohol is
l;0
not blended but pure (f A ):

f A clA  xlA  f A
l l;0

The values of the activity coefficients decrease, almost exponen- 5
tially, from diesel fuel (nil alcohol content) to blends with
The equilibrium condition (Eq. (3)) leads then to:
twenty percent of alcohol. From this point, the variation of
activity coefficients is smaller as the alcohol content increases,
/vA  xvA  p clA  xlA  f A
l;0
6
and they tend to unity for pure alcohols.

Activity coefficients are higher for longer chain-length alcohol, The fugacity of the liquid pure alcohol depends on its vapour
although the differences between alcohols remain small for pressure. This relationship is shown when the coefficient of fugac-
blends with high alcohol content. ity, expressed here as a function of the compressibility factor (zA),
306 J.P. Hernndez et al. / Applied Thermal Engineering 102 (2016) 302310

is integrated from zero to the ambient pressure surpassing the d

vapour pressure:
Z p
l;0 dp
ln f A ln p zA  1 x=L
0 p
Z pv ;A Z p . .
ln p
dp
zA  1 zA  1
dp
7
mA mA
0 p pv ;A p L gas
Below the vapour pressure (first integral in Eq. (7)) alcohol is in
vapour phase and, at such low pressure, this vapour can be
assumed to behave as an ideal gas (zA = 1), whereas above the x=0
vapour pressure (last integral), alcohol is in liquid phase and it
can be assumed to be incompressible (zA = 0). Therefore, Eq. (7)
becomes:
l;0
ln f A ln p  ln p  ln pv ;A ! f A pv ;A
l;0
8 liquid
Consistently with the last assumptions, the coefficient of fugac-
ity of the alcohol at the vapour phase (with partial pressure
remaining below the vapour pressure) must be considered equal
to unity. The equilibrium condition (formerly Eq. (6)) leads finally
to: Fig. 4. Diffusion of alcohol in air along the vessels height.

c 
l
xlA  pv ;A
xvA  A
9
p
Modelling of the evaporation losses (Section 4.2) requires Initially, the mass flow rate does not remain uniform along the
knowing the mass fraction of alcohol at the interface, wvA. This frac- vessels height because of the mole of alcohol being accumulated.
tion can be obtained by assuming that the vapour phase is just The evaporation losses can be identified just with the transported
alcohol mass flow rate at the interface, m _ A;0 . This looped couple of
composed of alcohol and air:
equations is then solved at each height step for successive time
xvA  W A
wvA 10 steps until z = L, where alcohol is assumed to reach infinite dilu-
W A  xvA W Air  1  xvA tion in air: wvA,L = wvA (z = L) = 0. This boundary condition forces
the concentration profiles to stabilize after a short period of time.
From this instant onwards, the mass flow rate of alcohol trans-
4.2. Modelling of evaporation losses for alcohols
ported along the cylindrical vessel (with diameter d), m _ A,
Mass diffusion of a gaseous component in air is governed by becomes uniform and then it can be identified as the evaporated
Ficks law. Assuming that evaporation from a liquid occurs under alcohol mass flow rate. Form this instant onwards, it can be
uniform temperature and pressure with no border effects, the obtained (in kg/s) as:
approach can be assumed to be one-dimensional. In this case the
q  pd2  DAAir
Ficks law for a single component can be written as [27]: _A
m ln1  wvA;0 14
4L
v
dwA
_ 00A wvA  m
m _ 00A  qv  DAAir  11 However, even after the process has reached the quasi-stable
dz
regime, this mass of evaporated alcohol changes slightly with time
where m _ 00A is the mass flow rate per unit area of alcohol (with units because the depth of the interface in the vessel decreases, and thus
2
kg/s m ) in the vertical direction (z in m), DA-Air is the binary the height L increases, as the liquid is being consumed.
diffusion coefficient (in m2/s) and wvA is the mass fraction The model starts running right after the simple preparation, the
(dimensionless) in the gas phase of the alcohol. qv is the density initial conditions being mA(t = 0) = 0, L(t0) = L, wvA (z = 0, t = 0) = wvA,0
of the gaseous mixture obtained with the ideal gas law (in kg/m3, and wvA (z = z1, z = z2, . . . , z = L; t = 0) (t = 0) = 0. Then the evaporated
Eq. (12)) with R being updated as concentration of alcohol in air mass, the liquid density, the free height of the vessel and the
increases. mass fractions of diesel on the alcohols in the liquid phase are
p successively updated at each instant through:
qv 12
RT mvA t i mvA t i1 m
_ 00A;0 ti1  ti  ti1 15
The model proposed here divides both time and height of the
gas volume into finite steps. Since ambient conditions are sup- 1
posed to remain unchanged along a height step and during a time
ql ti W l t W lA t i
16
D i

step, the product q  DA-Air can be considered constant for integra- qlD qlA
tion. The model integrates along every height step at each time
step, and the result for wvA,z from the previous time step is used mA t i1  mA t i
Lti Lti  1 17
at the current one, starting with wvA,0 at the interface level (z = 0, ql  A
Fig. 4).
This integration provides the mass flow rate transported along a mlD
wlD ti 18
height step of the cylindrical vessel (with diameter d), m _ A;z , at a mlD mlA t 0  mvA ti
given instant:
" # mlA t 0  mvA t i
q  pd2  DAAir 1  wvA;z wlA ti 19
_ A;z
m ln 13 mlD mlA t 0  mvA ti
4  Dz 1  wvA;0
 J.P. Hernndez et al. / Applied Thermal Engineering 102 (2016) 302310
In the following time steps, the calculation has to sweep
increasing number of height steps, until reaching the top of the
vessel, furthermore in the alcohol/diesel blends, the remaining
concentration of alcohol in the blend must also be updated.
As a consequence, the mole fraction (Eq. (9)) and mass flow
(Eq. (14)) of alcohol in the interface decrease in the gas phase.
All these variables must be updated at each time step leading
to changes in the evaporation mass flow rate (Eq. (15)). The steps
followed for the calculation of the evaporative losses are shown
in Fig. 5.
5. Results and experimental validation
Fig. 6 shows a comparison between the experimental results of
evaporative losses from the n-alcohol/diesel fuel blends (20% of
alcohol) and those proposed by UNIFAC-Dortmund-Continuous
method. Additionally, results obtained from the ideal method
(activity coefficient = 1) are included for each alcohol. The mass
of methanol in the blend was completely evaporated at around
10 h, the mass of ethanol at around 20 h and the mass of
n-propanol at around 60 h (see Fig. 6). On the contrary, in the case
of n-butanol and n-pentanol, after 90 h the alcohol mass was not
completely evaporated (scale of y axis is different for the heaviest
alcohols), consistently with their lower volatility.
Comparing the methods used, the UNIFAC-Dortmund-
Continuous method proposed in this work predicts with qualita-
tively good accuracy the evaporative losses for all alcohols, with
no fitting constants. An approximately inversely proportional
Fig. 5. Steps of the modelling of evaporation losses for the n-alcohol/diesel blends.
307
relationship can be observed between evaporation rate and carbon
chain length of the alcohol. This relationship is illustrated in Fig. 7,
where the average evaporation rate along the first eight hours
(before the complete evaporation of the lightest alcohol), obtained
from both the model and the experiments, is shown.
The deviation of the proposed model with respect to the
experimental results observed for the cases of methanol/diesel
(maximum over-prediction of 35%) and n-pentanol/diesel blends
(maximum under-prediction of 38%) could be explained, in the
former case, by the error induced during the mass measurement
of a very volatile material, and in the latter case, by the not-
negligible evaporation of the most volatile fraction of diesel fuel
over a very long period of time. For the rest of alcoholdiesel
blends, errors remain always below 15%. On the contrary, results
provided by ideal model were far from those obtained experi-
mentally, which implies that blends are far from ideal behaviour.
These results are consistent with the high values of activity
coefficient corresponding to blends with 20% content (or less) of
alcohol (Fig. 3).
In order to complete the validation of the model proposed,
different blends with a wide range of alcohol contents (with the
consequent variation in the activity coefficients) were tested.
n-Butanol was chosen for the experiments and it is expected that
rest of alcohols would achieve the same agreement since the diffu-
sion and the activity coefficients for the different alcohols are not
qualitatively different (Fig. 3). Results for 30%, 40%, 50% and 60%
content of n-butanol in blends (additionally to the results for 20%
already shown in Fig. 6) are shown in Fig. 8.
308 J.P. Hernndez et al. / Applied Thermal Engineering 102 (2016) 302310
Fig. 6. Evaporated mass for the 20% n-alcohol/diesel blends.
Fig. 7. Modelled and experimental results for evaporation rate (averaged along the
first eight hours) as a function of the number of carbon atoms in the alcohol.
Evaporation rate, defined as the mass evaporated per unit time
(slope of lines in Fig. 8), appears to be slightly lower as the butanol
content increases. Although this trend could be opposite to the
expected one, it is consistent with the increase of activity values
as the alcohol content decreases. In Eq. (9), lower values of xlA are
compensated by the higher values of clA . Results from the ideal
model in Fig. 8 show that the evaporation rate increases propor-
tionally to the highest butanol content, but this trend is reversed
in the UNIFAC-Dortmund-Continuous model and in the experi-
ments because at high butanol concentrations the activity coeffi-
cients are close to unity, whereas at low concentrations they are
very high (Fig. 3). These results were completed with additional
modelling results for all alcohols (not only n-butanol but for all
alcohols) and the resulting evaporation rates (again averaged along
the first eight hours) are shown in Fig. 9. It can be observed that, for
all alcohols, the mentioned compromise between activity coeffi-
cient and alcohol content leads to a wide region (starting from
10% or 20% contents) in which evaporation losses decrease as the
alcohol content increases. Dotted lines in Fig. 9 indicate unstable
blends [11], for which experimental validation is impossible.
As an intermediate result from the model, Fig. 10 shows, as an
example, that n-butanol evaporates much faster than ideally and
thus, its mole fraction in the liquid phase decreases sharply. The
reason for this is that the mole fraction in the vapour interphase
increases with time (as the alcohol content in the liquid decreases)
rather than decreasing (as predicted by the ideal model), as a con-
sequence of the increase in the activity coefficient.
 J.P. Hernndez et al. / Applied Thermal Engineering 102 (2016) 302310 309

Fig. 8. Evaporated mass for the n-butanol/diesel blends with increasing initial concentrations of n-butanol.

Fig. 9. Effect of the initial alcohol content on the evaporation rate for all alcohols
(modelled results).

Fig. 10. Mole fraction at the liquid and vapour phase for n-butanol in the 60%
6. Conclusions n-butanol/-diesel blends along time.

Evaporative losses of different alcoholdiesel blends were stud-
ied in this work. Results obtained experimentally were compared
with those estimated using a model based on the UNIFAC-
Dortmund-Continuous method, proposed in this work. From the
results, the following conclusions can be drawn:

The higher is the volatility of the alcohol, the faster is the evap-
oration process. The evaporation decreases approximately with
an inversely proportional rate with respect to the increase in
the carbon chain length of the alcohol. While all the content
of methanol, ethanol and n-propanol in blends (20%) was
evaporated in 10, 20 and 60 h, respectively, part of n-butanol
and n-pentanol remains in liquid blends at the end of the test
(almost 100 h).

The evaporation rate results obtained from the model proposed
in this study, which includes no fitting constants, are in a good
agreement with the results obtained experimentally. Only a
slight underestimation of the evaporation rate for n-pentanol
blends can be remarked.

Results were far from the ideal behaviour for all alcohols, espe-
cially for low alcohol contents. As the alcohol content in the
blend increases, the activity coefficients get closer to ideal ones
310 J.P. Hernndez et al. / Applied Thermal Engineering 102 (2016) 302310
(unity) and both results, experimental and modelled become
closer to those corresponding to the ideal model.

The model was helpful to explain the experimentally observed
paradox of low alcohol content blends evaporating faster
than high content blends. This effect can be explained as a
consequence of a compromise between activity coefficients
and alcohol content.
Acknowledgements
Authors wish to thank Universidad de Antioquia for the
financial support through the project CODI 2015-8143 and the
Sostenibilidad program 20142015. Co-author JPH acknowledges
COLCIENCIAS for the doctoral scholarship received. The Colombian
Petroleum Company (Ecopetrol) is acknowledged for supplying the
ULSD fuel. The research leading to these results was co-funded by
the European Union Horizon 2020 Research and Innovation
Programme under grant agreement n 640462.
References
[1] C.D. Rakopoulos, K.A. Antonopoulos, D.C. Rakopoulos, Experimental heat
release analysis and emissions of a HSDI diesel engine fueled with ethanol
diesel fuel blends, Energy 32 (2007) 17911808.
[2] P. Pepiot-Desjardins, H. Pitsch, R. Malhotra, S.R. Kirby, A.L. Boehman, Structural
group analysis for soot reduction tendency of oxygenated fuels, Combust.
Flame 154 (12) (2008) 191205.
[3] M. Lapuerta, O. Armas, R. Garca-Contreras, Lubricity of ethanolbiodiesel
diesel fuel blends, Energy Fuels 23 (2009) 43434354.
[4] Y. Di, C.S. Cheung, Z. Huang, Experimental study on particulate emission of a
diesel engine fueled with blended ethanoldodecanoldiesel, J. Aerosol Sci. 40
(2) (2009) 101112.
[5] J. Huang, Y. Wang, S. Li, A.P. Roskilly, H. Yu, H. Li, Experimental investigation on
the performance and emissions of a diesel engine fuelled with ethanoldiesel
blends, Appl. Therm. Eng. 29 (1112) (2009) 24842490.
[6] Directive 28/2009/CE of the European Parliament and of the Council on the
Promotion of the Use of Energy from Renewable Sources, 2009.
[7] F. Hasegawa, S. Yokoyama, K. Imou, Methanol or ethanol produced from woody
biomass: which is more advantageous?, Bioresour Technol. 101 (2010) S109
S111.
[8] A. Demirbas, Biomethanol production from organic waste materials, Energy
Source. Part A: Recov. Util. Environ. Effects 30 (2008) 565572.
[9] K.J. Ptasinski, Thermodynamic efficiency of biomass gasification and biofuels,
Biofuels, Bioprod. Biorefin. 2 (3) (2008) 239253.
[10] M. Wu, M. Wang, J. Liu, H. Huo, Assessment of potential life-cycle energy and
greenhouse emission effects from using corn-based butanol as a
transportation fuel, Biotechnol. Prog. 24 (6) (2008) 12041214.
[11] M. Lapuerta, R. Garca-Contreras, J. Campos-Fernndez, M.P. Dorado, Stability,
lubricity, viscosity and cold flow properties of alcoholdiesel blends, Energy
Fuels 24 (2010) 44974502.
[12] R.A. Baranescu, Fumigation of Alcohols in Multi-Cylinder Diesel Engine-
Evaluation of Potential, SAE Paper 860308, 1986.
[13] P. Zelenka, P. Kapus, L.A. Milkulic, Development and Optimization of Methanol
Fueled Compression Ignition Engines for Passenger Cars and Light Duty Trucks,
SAE Paper 910851, 1991.
[14] Z. Huang, H. Lu, D. Jiang, K. Zeng, B. Liu, J. Zhang, X. Wang, Combustion
behaviors of a compression-ignition engine fuelled with diesel/methanol
blends under various fuel delivery advance angles, Bioresour. Technol. 95
(2004) 331341.
[15] B. Freedman, M.O. Bagby, Predicting cetane numbers of n-alcohols and
methyl esters from their physical properties, J. Am. Oils Chem. Soc. 67
(1990) 565571.
[16] J.A. Pumphrey, J.I. Brand, W.A. Scheller, Vapour pressure measurements and
predictions for alcoholgasoline blends, Fuel 79 (2000) 14051411.
[17] Z. Yang, B.P. Hollebone, Z. Wang, C. Yang, M. Landriault, Effect of storage period
on the dominant weathering processes of biodiesel and its blends with diesel
in ambient conditions, Fuel 104 (2013) 342350.
[18] S. Huang, P. Deng, R. Huang, Z. Wang, Y. Ma, H. Dai, Visualization research on
spray atomization, evaporation and combustion processes of ethanoldiesel
blend under LTC conditions, Energy Convers. Manage. 106 (2015) 911920.
[19] W.L.H. Hallet, S. Beauchamp-Kiss, Evaporation of single droplets of ethanol-
fuel oil mixtures, Fuel 89 (2010) 24962504.
[20] B.E. Polling, J.M. Prausnitz, J.P. OConnell, The Properties of Gases and Liquids,
fifth ed., Mc-Graw Hill Book Company, New York, 2000.
[21] X. Li, X. Qiao, L. Zhang, J. Fang, Z. Huang, H. Xia, Combustion and emission
characteristics of a two-stroke diesel engine operating on alcohol, Renew.
Energy 30 (13) (2005) 20752084.
[22] H. Abou-Rachid, K. El Marrouni, S. Kaliaguine, DFT studies of the hydrogen
abstraction from primary alcohols by O2 in relation with cetane number data,
Theochem J. Mol. Struct. 631 (13) (2003) 241250.
[23] www.ddbst.com/online.html (last accessed: March 2016).
[24] www.webbook.nist.gov (last accessed: October 2015).
[25] M. Lapuerta, J.P. Hernandez, J.R. Agudelo, An equation for the estimation of
alcohol-air diffusion coefficients for modeling evaporation losses in fuel
systems, Appl. Therm. Eng. 73 (2014) 537546.
[26] J.P. Hernndez, F. Bustamante, J.R. Agudelo, Thermodynamics of
semicontinuous n-alcohol/ultra-low sulfur diesel blends, Fluid Phase Equilib.
401 (2015) 915.
[27] R. Byron Bird, W.E. Stewart, E.N. Lightfoot, Transport Phenomena, second ed.,
John Wiley& Sons, Inc., New York, 2002.