Waste Management 34 (2014) 1501–1509

Contents lists available at ScienceDirect

Waste Management
journal homepage: www.elsevier.com/locate/wasman

Pervaporation of ethanol produced from banana waste

Roger Hoel Bello a, Poliana Linzmeyer b, Cláudia Maria Bueno Franco b, Ozair Souza a,b,c, Noeli Sellin a,b,c,
Sandra Helena Westrupp Medeiros a,b,c, Cintia Marangoni d,⇑
a
Chemical Engineering Department, University of Joinville Region (UNIVILLE), Joinville, SC, Brazil
b
Sanitary and Ambient Engineering Department, University of Joinville Region (UNIVILLE), Joinville, SC, Brazil
c
Masters Program in Process Engineering, University of Joinville Region (UNIVILLE), Joinville, SC, Brazil
d
Federal University of Santa Catarina (UFSC), Campus Blumenau, Blumenau, SC, Brazil

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

Article history: Banana waste has the potential to produce ethanol with a low-cost and sustainable production method.
Received 19 December 2013 The present work seeks to evaluate the separation of ethanol produced from banana waste (rejected fruit)
Accepted 10 April 2014 using pervaporation with different operating conditions. Tests were carried out with model solutions and
Available online 16 May 2014
broth with commercial hollow hydrophobic polydimethylsiloxane membranes. It was observed that
pervaporation performance for ethanol/water binary mixtures was strongly dependent on the feed
Keywords: concentration and operating temperature with ethanol concentrations of 1–10%; that an increase of feed
Banana waste
flow rate can enhance the permeation rate of ethanol with the water remaining at almost the same value;
Bioethanol
Biofuel
that water and ethanol fluxes was increased with the temperature increase; and that the higher effect in
Lignocellulosic residue flux increase was observed when the vapor pressure in the permeate stream was close to the ethanol
Pervaporation vapor pressure. Better results were obtained with fermentation broth than with model solutions,
Polydimethylsiloxane membrane indicated by the permeance and membrane selectivity. This could be attributed to by-products present
in the multicomponent mixtures, facilitating the ethanol permeability. By-products analyses show that
the presence of lactic acid increased the hydrophilicity of the membrane. Based on this, we believe that
pervaporation with hollow membrane of ethanol produced from banana waste is indeed a technology
with the potential to be applied.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction
One of the bases for the economic development of Brazil is
agriculture, and in certain states, such as Santa Catarina, banana
is widely cultivated as a commercial crop. Banana is one of the
most consumed fruits in the world and it is commercially grown
in about 120 countries. Currently, Brazil is the second largest
producer (preceded by India) and is responsible for 7.5% of world
production (about 7.2 million tons per annum, according to the
Center of Socioeconomics and Agricultural Planning for the State
of Santa Catarina). The State of Santa Catarina has approximately
six thousand producers, being the fourth largest banana growing
region in Brazil, with 663,892 tonnes of bunches of bananas
produced per annum (ABIB, 2011).
Commercial banana production generates a large proportion of
waste; there are reports of 30% waste in Australia (Clarke et al.,
2008) and Malaysia (Tock et al., 2010), and 25–50% in Central
⇑ Corresponding author. Address: Rua Pomerode, 710 Salto do Norte, Blumenau,
SC 89065-300, Brazil. Tel.: +55 48 3721-6308.
E-mail address: ci.marangoni@gmail.com (C. Marangoni).
and South America (Hammond et al., 1996). In countries like India,
all kind of banana waste is considered an important urban waste
because the fruit is used in all religious functions, festivals and in
temples (Chanakya and Sreesha, 2011). According to Graefe et al.
(2011), around 20–40% of the bananas produced do not meet
export standards or even the quality demands of spot markets. In
Brazil, particularly in the southern regions, it is estimated that
for every 100 kg of harvested fruit, 46 kg are not used (EMBRAPA,
2006). Further, Souza et al. (2010) indicate that for every ton of
bananas produced approximately 3 tons of pseudostem, 160 kg
of stems, 480 kg of leaves and 440 kg of skins are generated.
Fernandes et al. (2013) found that less than 10% of available bio-
mass as waste (440 million tons) is designated to some application.
Thus, an established commercial use for such residues, as well as
generating extra remuneration for regional farmers, would help
to reduce environmental pollution. Alternative uses for these dis-
cards have to be explored, and in this regard processing to produce
ethanol is seen to have potential from both an environmental as
well as an economic point of view.
Many countries are investing in the development and use of
biofuels as a way of reducing environmental impacts and ethanol

http://dx.doi.org/10.1016/j.wasman.2014.04.013
0956-053X/Ó 2014 Elsevier Ltd. All rights reserved.
1502 R.H. Bello et al. / Waste Management 34 (2014) 1501–1509
is one of the fuels that can be produced from various raw materials
(España-Gamboa et al., 2011). The use of agricultural or agro-
industrial waste is an interesting option in this context. Biofuel
has been produced on a large scale in Brazil for three decades using
sugarcane as feedstock (Soccol et al., 2010), however there are
many criticisms of the practice and an ongoing debate about the
ethical issue of using food (or land available for the cultivation of
food) as an energy source (Sarkar et al., 2012; Swana et al.,
2011). Lignocellulosic material does not play an intrinsic role in
the food chain and this is a fundamental aspect that makes it an
attractive alternative for ethanol production. Besides, the cost
and availability of the feedstock are crucial and can contribute
65–70% to the total ethanol production costs (Kazi et al., 2010).
In this sense, the substitution of biomass wastes for raw materials
such as cane sugar, starch and corn for ethanol production is an
alternative that has shown promising results (Dermibas, 2011;
Mabee et al., 2011). Examples of this residual biomass include
bagasse sugar cane, corn straw and fiber, wheat and rice straw,
eucalyptus wood and crop wastes from commercial cultivation of
fruits such as bananas, grapes and apples (Rivas-Cantu et al., 2013).
Technologies for the conversion of biomass to ethanol are
also under various stages of development. The use of these
lignocellulosic residues requires some separation of cellulose and
hemicellulose from lignin, followed by hydrolysis of sugars, and
this bioconversion has been extensively studied using the different
types of wastes. The potential yield of ethanol from lignocellulosics
varies significantly between feedstocks, so many applications in
alcoholic fermentation are reported in the literature with different
wastes. Specifically in the case of ethanol from bananas, the few
studies that have been published involve the use of the fruit, leaves
and other waste such as the pseudostem. Tewari et al. (1986)
reported the suitability of banana peel for alcohol fermentation.
Hammond et al. (1996) presented ethanol yield (on a dry weight
basis) from ripe bananas as higher than from most other agricul-
tural commodities. Velásquez-Arredondo et al. (2010) investigated
the acid hydrolysis of banana pulp and fruit and the enzymatic
hydrolysis of flower stalk and banana skin, and the results obtained
demonstrated a positive energy balance for the four production
routes evaluated. The study by Graefe et al. (2011) presents results
of a case study in Costa Rica and Ecuador which found that consid-
erable amounts of ethanol could be produced from banana
bunches that do not meet quality standards, as well as from which
are partly left to rot in the fields. Oberoi et al. (2011) also demon-
strated that banana peel could serve as an ideal substrate for the
production of ethanol through simultaneous saccharification and
fermentation. Hossain et al. (2011) evaluated bioethanol from
rotten banana and concluded that this can be used in motor
vehicle engines, producing low emissions, and thus it can be used
as an environmental recycling process for waste management.
Arumugam and Manikandan (2011) explore the potential applica-
tion of pulp and banana peel wastes in bioethanol production using
dilute acid pretreatment followed by enzymatic hydrolysis.
Gonçalves Filho et al. (2013) evaluate the same techniques with
banana tree pseudostem.
Although the lignocellulosic material shows positive results, it
still requires more research to be exploited on an industrial scale.
Great efforts are being undertaken to improve ethanol productivity
and reduce the overall production costs. According to Gaykawad
et al. (2013), one of the ways to achieve these goals is to modify
the configuration of the process and perform process integration.
Traditionally, the recovery of ethanol by distillation is a challenge
because of the high costs and energy expenditure required (Vane,
2008). Toward this end, membrane separation processes such as
pervaporation have been used. The great interest in these pro-
cesses is mainly because of features such as cost-effectiveness,
high energy efficiency and environmental friendliness. Membrane
based separation technologies normally fulfill the criteria for sus-
tainability and energy efficiency (Korelskiy et al., 2013). In addition
to reducing the inhibition of ethanol in the production stage due to
the possibility of its simultaneous use with fermentation
(Lewandowska and Kujawski, 2007), this procedure could replace
a concentration step that is required for recovery because of the
presence of alcohol in small quantities in the broth (Nomura
et al., 2002). As shown by Chovau et al. (2011), the composition
of the fermentation broth influences the separation, and the use
of different substrates leads to the need to reevaluate the process,
even if it is already well established. Also, in multicomponent
systems, the diffusivity of one component is influenced by the
presence of others. The use of lignocellulosic biomass will not only
affect feedstock pretreatment and fermentation process of the
ethanol production but also the downstream processing
(Gaykawad et al., 2013).
In pervaporation, a liquid mixture is fed through a membrane.
The mixture components permeate selectively through the mem-
brane and vaporize on the other side of the membrane where
low pressure is maintained. By this means, there is a selective
removal of organic compounds from dilute aqueous solutions.
There are several studies regarding ethanol pervaporation and they
relate mainly to the use of different membranes. Specifically the
pervaporation of ethanol from lignocellulosic residues is reported
by Gaykawad et al. (2013) with barley straw and willow wood
using commercial polydimethylsiloxane (PDMS) membranes.
Zhang et al. (2012) studied the membrane fouling in pervaporation
of ethanol from food waste after a flocculation–filtration pretreat-
ment. Aroujalian and Raisi (2009) study the effects of various oper-
ating parameters such as feed temperature, permeate side
pressure, and Reynolds number (volumetric flow rate) on the total
flux, and ethanol selectivity of a porous membrane-based pervap-
oration process with 2% aqueous ethanol solutions, simulating an
ethanol content from lignocellulosic residues. O’Brien et al.
(2004) related an efficient system of coupled fermentation and
pervaporation for ethanol from corn fiber hydrolisates.
Studies of pervaporation in ethanol production hitherto have
not used banana waste as a substrate for ethanol production. Thus
the aim of this research is to evaluate if pervaporation can be used
in the production of ethanol from banana and to investigate the
effects of operating variables and of lignocellulosic biomass fer-
mentation by-products on membrane performance for the recov-
ery of ethanol by using pervaporation.
2. Materials and methods
To investigate the membrane behavior, first model solutions of
ethanol/water were separated in pervaporation experiments to
characterize ethanol transfer across the hollow polydimethylsilox-
ane (PDMS) membrane and these results provided the reference for
the broth experiments.
In the first case, feed conditions (flow rate, temperature, ethanol
composition) and permeate pressure were modified. Also, the time
necessary to reach steady state was determined. Then, tests were
performed with the fermentation broth produced using banana
fruit waste as a substrate varying the ethanol feed mass fraction
and feed flow rate. The presence of some byproducts was also
studied.
2.1. Membrane
The pervaporation unit used consisted of a removable perme-
ation module made of polyvinyl chloride (PVC) of 0.2 m internal
diameter containing 50 dense hollow polydimethylsiloxane
(PDMS) membranes (di = 0.6 mm and de = 1 mm), 0.25 m in length,
 R.H. Bello et al. / Waste Management 34 (2014) 1501–1509 1503

covering an area of 0.04 m2. This is a hydrophobic commercial Table 2

membrane, with the membrane and module built by a Brazilian Operating conditions used in the experiments with fermentation broth.

company (PAM-Membranas Selectivas). Experiments with fermentation Operating conditions

broth
Effect of feed flow rate Feed flow rate: 5.5  106 and
2.2. Fermentation
22.2  106 m3 s1
Feed mass fraction: 3 wt%
To obtain the feed from the process of pervaporation, 0.002 m3 Feed temperature: 295 K
of fermentation broth was produced, as described by Schulz Variation of feed weight Feed flow rate: 5.5  106 m3 s1
(2010). Rejected banana fruit (Musa cavendishii) was used as ligno- fraction Feed mass fraction: 2, 2.5 and 3 wt%
cellulosic material (500 g L1 wet mass) and 20% (v/v) of inoculum Feed temperature: 295 K
(Saccharomyces cerevisiae). After a period of 10 h of fermentation,
different ethanol mass fractions were obtained in broths for per-
vaporation tests. The total volume of the broth was centrifuged Eqs. (1)–(5), respectively (Pereira et al., 2006; Gaykawad et al.,
at 3800 min1 (3018g) for 1200 s in a refrigerated centrifuge and 2013; Sun et al., 2013):
then stored at 277 K.
yi =yj
ai ¼ ð1Þ
xi =xj
2.3. Pervaporation

The fermentation broth/standard solution was supplied with

W
J tot ¼ ð2Þ
the aid of a gear pump from a reservoir positioned upstream from At
the membrane. A vacuum pump coupled to the permeate side
J i ¼ J tot yi ð3Þ
of the assembly provided the pressure drop for vaporization of
ethanol. The permeate vapor was directed to a condensation bath
yi
refrigerated with liquid nitrogen. The liquid retentate was bi ¼ ð4Þ
xi
recirculated to the feed tank.
A summary of the operating conditions tested is presented in
PSI ¼ J tot ðai  1Þ ð5Þ
Table 1 for the standard mixture and in Table 2 for the fermenta-
tion broth. All experiments were carried out with condensation where yi and yj are the weight fractions of ethanol and water in the
temperature of 77 K and permeate pressure lower than 667 Pa. permeate, respectively, and xi and xj are the weight fractions of eth-
The tests were carried out in duplicate, and the results presented anol and water in the feed, W is the mass (g) of the permeate, A is
here were consistent with the average of the same. Feed and the effective area (m2) of the membrane, and t is the time interval
permeate samples (upon reaching the steady-state condition), (h) for pervaporation.
were collected for 1–2 during each test to quantify the concentra- The membrane flux was determined gravimetrically using a
tions of ethanol with a gas chromatograph (GC, Agilent model balance with an accuracy of 104 g by weighing the mass of perme-
6890, coupled with autosampler: Agilent model 7683) using a ate obtained during the collecting time. The enrichment factor rep-
Hewlett–Packard HP-1 column with a length of 50 m, an external resents the membrane capacity of concentrating the component i
diameter of 0.32 mm, a stationary phase composed of 100% (Pereira et al., 2006) and is an interesting parameter to evaluate
polydimethylsiloxane and a film thickness of 1.05 lm. the process for multicomponent mixtures.
Temperature dependency of the flux was analyzed using an
2.4. Parameters from the process evaluation Arrhenius-type equation:
 
Ea
The performance of the process of pervaporation has been lnðJ i Þ ¼ J io exp  ð6Þ
RT
expressed in terms of separation factor of the membrane (ai), the
total permeate flux (Jtot), the ethanol permeate flux (Ji), enrichment where Ji is the partial flux of the compound, Jio is the pre-exponen-
factor (bi) and pervaporation separation index (PSI) according to tial factor of the flux, R is the gas constant (J mol1 K1), T is the
temperature (K) and Ea is the apparent activation energy (KJ mol1).
Permeance was estimated according to Eq. (9), based on the
Table 1
molar flux of each component – ji (obtained from Eqs. (7) and
Operating conditions used in the experiments with the model solutions.
(8)), and the selectivity of the membrane was calculated from
Experiments with model Operating conditions the ratio between permeances (Eq. (10)).
solutions
mi
Effect of ethanol feed weight Feed flow rate: 5.5  106 m3 s1 J im ¼ J tot yi ð7Þ
fraction mt
Feed weight fraction: 1–12 wt%
Feed temperature: 295 K J im v Gi
Effect of feed flow rate Feed flow rate: 5.5  106 and ji ¼ ð8Þ
22.2  106 m3 s1 mi
Feed weight fraction: 10 wt%
Feed temperature: 295 K Pi ji
Effect of feed temperature Feed flow rate: 5.5  106 m3 s1 ¼ ð9Þ
l ðxi ci Poi  yi Pp Þ
Feed weight fraction: 3 wt%
Feed temperature: 295, 300 and 303 K
Effect of permeate pressure Feed flow rate: 5.5  106 m3 s1 Pi=l
Feed weight fraction: 3 wt% ai=j ¼ ð10Þ
Pj=l
Feed temperature: 295 K
Permeate pressure: <600, 2600 and
with mi and mt being the molecular weight of the component I and
6000 Pa
the mixture, respectively, tGi is the molar volume of gas, cI is the
1504 R.H. Bello et al. / Waste Management 34 (2014) 1501–1509

activity coefficient for each component, Poi is the vapor pressure and
Pp is the total pressure in the permeate side.
Reynolds number was calculated according to Schnabel et al.
(1998). It was considered an aqueous solution, and because of this,
viscosity and density used was the values described by the authors
with temperature of 298 K (25 °C).

3. Results and discussion

3.1. Experiments with model fermentation broths

3.1.1. Effect of ethanol feed weight fraction

Chovau et al. (2013) reported that the fermentation of lignocel-
lulosic wastes produces, on average, 3–6 wt% of ethanol. Hence we
analyzed the results for total and partial fluxes and separation and
enrichment factors, as illustrated in Figs. 1 and 2, respectively, with
an ethanol weight fraction range from 1% to 5% (values expected
for a second-generation biomass) and then compared them with
9%, 10% and 12% (values related or even superior to first-generation
biomass).
Increasing alcohol concentration increases permeation fluxes of
the mixture of ethanol and water resulting in higher total flux
(Fig. 1). In fact increasing the amount of the component that has
major affinity with the membrane in the feed stream leads to an
increase in the flux of these components in the permeate as
reported in literature (Ortiz et al., 2002; Sommer and Melin,
2005; Wu et al., 2005; Zereshki et al., 2011). A higher ethanol con-
centration in the feed produces a higher ethanol flux. Dobrak et al.
(2010) reported similar results with polydimethylsiloxane com-
posite membranes, where the permeate flux through the mem-
brane increased with increasing concentration of ethanol in the
feed, due to the swelling effects.
The ethanol flux demonstrated an increase even with ethanol
concentrations less than 5 wt% in the feed, but this increase is even
greater at higher concentrations. As reported by Pereira et al.
(2006), the permeate flux of an organic dilute feed solution is gen-
erally a linear function of concentration, and the water flux
remains constant (dependent on the concentration of the organic
component). This behavior is observed in this study when ethanol
content in the feed is less than 5 wt%.
However, water flux increases more than alcohol, resulting in a
separation factor reduction as observed in Fig. 2. According to
Mohammadi et al. (2005), this is a result of an enhanced diffusion
of water into the membrane by the fact that increasing alcohol con-
centration increases membrane-free volume and simultaneously
Fig. 1. Effect of ethanol content in feed on total (j), ethanol (d) and (N) water
fluxes.
Fig. 2. Effect of ethanol content in feed on separation (j) and enrichment (d)
factors.
side chain mobility increases. Consequently, small-sized water
cluster can permeate easily through the membrane-free volume.
However, Van Baelen et al. (2004) reported that this behavior
can be attributed to a dragging effect. In both cases, the final result
is a decrease in separation factor. This also was observed by Zhou
et al. (2011), Lai et al. (2012), Lee et al. (2012), who also analyzed
pervaporation of alcohols in dilute mixtures with hollow poly-
dimethylsiloxane membrane.
3.1.2. Effect of feed flow rate
Table 3 compares the values for the average total and ethanol
permeate mass fluxes, separation and enrichment factor of the
polydimethylsiloxane membrane obtained in pervaporation of
10 wt% of alcoholic mixtures when the feed flow rate was varied.
As expected, ethanol and total flux increased with feed flow rate
as shown in Table 3 due to a decrease of concentration polariza-
tion. These results are in agreement with other research, even
when different membranes were used (Jiraratananon et al., 2002;
Zereshki et al., 2011). As the ethanol is the component having
the greatest affinity with the polydimethylsiloxane membrane, a
reduction of concentration polarization means that ethanol con-
centration near the membrane surface was close to the ethanol
concentration in the bulk. As observed in relation to the partial flux
of ethanol, an increase of feed flow rate can enhance the perme-
ation rate of ethanol with the water remaining at almost the same
value. This behavior is in agreement with studies considering
the effect of Reynolds number in pervaporation of organic com-
pounds (Liang et al., 2004) In fact, the lower flux was observed at
the lowest value of the Reynolds number (feed flow = 5.5  106
m3 s1 – Reynolds number: 18.47; feed flow = 22.2  106 m3 s1 –
Reynolds number = 73.90).
The increase of feed flow rate enables better penetration of eth-
anol into the membrane as the driving force of ethanol increases.
As consequence, the hydrophobic properties of the membrane also
increase, and the process will be more selective for ethanol, as indi-
cated by the separation factor that showed an increase around 30%.
Li et al. (2004) also reported that with increasing flow rate both
total flux and separation factor increase, using a similar membrane
(polydimethylsiloxane supported by cellulose acetate) for a model
solution with 5 wt% ethanol. The authors show values for the
separation factor very close to that indicated in Table 4, with
the same feed flow rate. Finally, an increase of PSI with feed flow
rate indicated that better results will be obtained for the pervapo-
ration system with higher flow rates, as observed in other works
(Jiraratananon et al., 2002).
 R.H. Bello et al. / Waste Management 34 (2014) 1501–1509
Table 3
Results obtained for pervaporation experiments with model solutions (10 wt% ethanol) and flow rates of 5.5 and 22.2  106 m3 s1.
Flow rate (m3 s1) Total flux (g m2 h1) Ethanol flux (g m2 h1)
6
5.5  10 8.07 3.34
22.2  106 8.22 3.92
The improvement in ethanol content in permeate, which occurs
because of the effect of reduced concentration polarization at
higher flow rates that reduces the transport resistance in the liquid
boundary layer, results in a mass fraction of ethanol in permeate of
41.27 wt% with 5.55  106 m3 s1 and 47.24 wt% with 22.2 
106 m3 s1. This is an interesting result when compared with
the conventional ethanol recovery method, which is distillation.
Standard purification involves a first distillation column, the ‘beer
column’, where the top product consists of 37–40 wt% of ethanol
(Chovau et al., 2013). The ethanol content observed was similar,
indicating the potential of pervaporation as a concentration
method for further distillation.
3.1.3. Effect of feed temperature
The effect of the temperature on the total and partial fluxes for
3 wt% ethanol/water mixture is presented in Fig. 3.
As related in other works (Li et al., 2004; Mohammadi et al.,
2005; Dobrak et al., 2010; Luis et al., 2013) increasing temperature
causes an increase in the total permeate flux due to the increase of
diffusion rate of individual permeating molecules by the free
volumes produced because of the thermal motions of polymer
chains. As related by Pereira et al. (2006), a higher temperature
stimulates the driving force due to an increase in vapor pressure
and the activity coefficient of the permeating species (and their
chemical potentials) as temperature grows. Fig. 3 illustrates that
when temperature feed was tested from 295 to 300 K, total flux
was increased by a factor of 1.18, water flux by 1.14 and ethanol
flux by 1.5. This can also be observed from Fig. 4 which illustrates
the effect of feed temperature in separation factor. Different effects
were related in the literature, with some studies demonstrating
that the separation factor increases when temperature is increas-
ing and others showing the contrary. According to Zereshki et al.
(2011), the main reason for the separation factor to be affected
by temperature is because the diffusion and solubility of penetrat-
ing components changes significantly with temperature, but also
depends on many factors such as different organics, different
membrane-preparation method, and different supports of compos-
ite membranes and so on.
The total and partial permeation flux is plotted in logarithmic
scale as a function of the reciprocal temperature in Fig. 5 and
results show that an Arrhenius type relationship exists between
the fluxes and feed temperature, i.e. fluxes decrease with decreas-
ing temperature. These results are in agreement with the literature
(Zhou et al., 2011; Lai et al., 2012; Lee et al., 2012).
The apparent activation energy could be calculated from the
slope of the corresponding curve and Eq. (6) and the value is
summarized in Table 4.
The higher apparent activation energy value for ethanol flux
indicates that it was more affected than water flux as the
Table 4
Apparent activation energy for ethanol and water per-
meation estimated from experiments with ethanol
(3 wt%) and water mixtures.
Apparent activation energy (KJ mol1) Value
Ethanol 83.23
Water 62.64
1505
Separation factor Enrichment factor PSI
6.32 4.13 43.07
8.26 4.72 61.05
temperature increased. These results are in agreement with the
study by Shepherd et al. (2002) that used the same membrane to
separate aroma compounds. Thus the separation factor increases
with the increase of temperature. Similar results are presented
by Dobrak et al. (2010).
3.1.4. Effect of permeate pressure
Three variations of permeate pressure were investigated in the
pervaporation system: 6000, 2600 and less than 600 Pa. The results
obtained for the mass flow of permeate with respect to the differ-
ent applied pressures in the process of pervaporation are shown in
Table 5. Pereira et al. (2006) describe how when the permeate
pressure grows, the chemical potential gradient across the mem-
brane is reduced and, as a consequence, a reduction in transmem-
brane flux is observed. The same result was obtained in this work,
with the higher effect observed when the vapor pressure in the
permeate stream was close to the ethanol vapor pressure.
The results presented in Table 5 show a decrease of 64% at a
pressure of 2660 Pa and 95% at 6000 Pa in relation to a maximum
of 5.85 g m2 h1 were obtained for the permeate flux at the sys-
tem operating with a permeate pressure less than 600 Pa.
Jiraratananon et al. (2002) found that in ethanol–water mixtures,
as permeate pressure increases, the driving force for permeation
of the ethanol molecules decreases, which results in a decrease
in the mass flow of the permeate.
3.2. Experiments with fermentation broth
3.2.1. Effect of feed flow rate
Tests were performed with the fermentation broth that con-
tained about 3 wt% ethanol by weight in the feed mixture, and
the results were compared with those obtained using the standard
solution at the same ethanol weight fraction. The results are sum-
marized in Table 6.
It was observed that all of the parameters increased for the per-
vaporation of the broth at low flow rates compared with the model
solutions. For the specific case of the ethanol produced from the
Fig. 3. Temperature dependency on the total (j), ethanol (d) and water (N) fluxes
for 3 wt% ethanol/water mixture.
1506 R.H. Bello et al. / Waste Management 34 (2014) 1501–1509
Fig. 4. Effect of feed temperature in separation factor for 3 wt% ethanol/water
mixture.
Fig. 5. Arrhenius plot of total (j), water (N) and ethanol (d) fluxes vs. temperature
for ethanol (3 wt%) and water mixtures.
lignocellulosic banana waste, with 5.5x106 m3 s1 of flow rate, an
increase of about 28% in the enrichment factor was obtained, as
well as a 3.5 times increase in ethanol flux, which resulted in a per-
vaporation index (PSI) double that achieved with the standard
mixture (62.18 vs. 29.11). One important analysis consists of the
enrichment factor. This parameter remained almost the same, indi-
cating a possibility that there is no flux coupling mechanism.
Shepherd et al. (2002) reported the same results using the similar
membrane and studying multicomponent model mixtures.
For the flow rate of 22.2  106 m3 s1, contrary observations
were made. All performance parameters decrease for fermentation
broth compared with model solutions. This can be a result of the
presence of by-product that at this flow rate has an influence on
the ethanol diffusion mechanism. As the flux was reduced by a
factor of 0.25 and the separation factor was reduced by a factor
of 0.43, it was concluded that the ethanol pervaporation was
minimized. This behavior could be supported by the results of
Table 5
Effect of permeate pressure in total permeate flux for 3 wt% ethanol/water mixtures.
Pressure (Pa) <600 2660 6000
Total flux (g m2 h1) 5.85 2.1 0.3
enrichment factor, since this parameter relates to the separation
ability by the ratio of permeate and feed weight fractions. A reduc-
tion twice was observed. Also, when comparing the pervaporation
of fermentation broth between different flow rates, all parameters
decrease with the increase of feed flow. In these cases, the effect
observed in the enrichment factor was more pronounced.
3.2.2. Variation of feed weight fraction
Tests were performed with the fermentation broth of varying
ethanol by weight in the feed mixture, and the results were com-
pared with those obtained using the standard solution at the same
ethanol weight fraction. Figs. 6–8 show the comparison of results
obtained with the model mixture and fermentation broth for total
flux, separation and enrichment factors and pervaporation index,
respectively. The results of the parameters presented demon-
strated an increase that was observed with the pervaporation of
fermentation broth. The comparison between model solutions
and broth shows similar results, i.e., when the feed ethanol weight
fraction is increased, we observed an increase of flux, decrease of
enrichment and separation factor (consequently an increasing of
permeate ethanol weight fraction). More interesting, however, is
the increase in pervaporation index (Fig. 8), which indicates that
the pervaporation of the fermentation broth obtained from banana
residue with PDMS membrane has a more pronounced increase
with the increase of feed ethanol weight fraction than does the
model solution.
Figs. 9 and 10 show the comparison of results obtained with
model mixture and fermentation broth for permeance and selectiv-
ity, respectively. According to Luis et al. (2013), these parameters
enable better observation of the process efficiency once the effect
of the driving force and operating conditions is eliminated when
the parameters are calculated. Fig. 9 shows better results for broth
pervaporation, and an increase of both ethanol and water flux as
the feed ethanol weight fraction is increased, as observed in
Fig. 6. However, the permeance analysis shows clearly that the eth-
anol flux increase has a major effect on the total flux increase,
especially for fermentation broth. This indicates that by-products
present in the multicomponent mixtures could be facilitating the
ethanol permeability, and no coupling flux is observed. In fact,
by-products of fermentation, such as carboxylic acids, aldehydes,
alcohols and other salts, may influence the separation process,
especially at lower ethanol feed mass fractions and flow rates. No
glucose was observed in the broth used for the separation, when
analyzed via high-performance liquid chromatography. Glucose
has been shown to affect the performance of pervaporation, as
reported by Chovau et al. (2011).
In testing for selectivity, as observed before and with model
solutions, the membrane shows more affinity for water, mainly
at higher feed ethanol weight fractions. However, is important to
notice that a value of selectivity equal to 1 represents the perfor-
mance of a membrane with no intrinsic membrane selectivity,
achieving the same separation as simple evaporation of the liquid
into the vapor phase (Luis et al., 2013). In this work, selectivity
shows values greater than 1, and higher for fermentation broth
than for model solutions, which indicates that the effect of the
membrane is to enrich the permeate with ethanol.
3.2.3. Influence of by-products
Table 7 presents the average values of the by-products of the
broth in the feed analyzed by HPLC. The results for the centrifuged
broth were compared with those for the standard mixture to deter-
mine the influence of by-products on the permeate flux, the
enrichment factor and the concentration of the permeate.
According to Table 7, the glycerol concentration in the fermen-
tation broth in the present study was 1.67% by weight. Chovau
et al. (2011) have reported that the presence of glycerol in low con-
 R.H. Bello et al. / Waste Management 34 (2014) 1501–1509 1507

Table 6
Comparison of the pervaporation carried out at different feed flow rates with ethanol/water model solutions and fermentation broth with 3 wt% of ethanol.

Experiment Total flux (g m2 h1) Ethanol flux (g m2 h1) Enrichment factor Separation factor PSI
6 3 1
Flow rate 5.5  10 m s
Ethanol/water 3.58 0.37 7.56 8.31 29.71
Broth 5.85 1.06 8.88 10.63 62.18

Flow rate 22.2  106 m3 s1

Ethanol/water 4.80 0.41 6.51 7.06 33.91
Broth 3.60 0.30 3.73 3.97 14.29

Fig. 6. Comparison of the effect of ethanol content in feed on total flux for model
solutions (j) and fermentation broth (h). Fig. 8. Comparison of the effect of ethanol content in feed in pervaporation index
(PSI) for model solutions (j) and fermentation broth (h).

Fig. 7. Comparison of the effect of ethanol content in feed on separation (j) and Fig. 9. Comparison of the effect of ethanol content in feed on permeance of ethanol
enrichment (d) factors for model solutions (closed symbols) and fermentation (d) and water (N) for model solutions (closed symbols) and fermentation broth
broth (open symbols). (open symbols).

centrations does not significantly affect the permeate flux param-
eter or the coefficient of enrichment. In addition, the effect of car-
boxylic acids on the process of pervaporation was studied, and the
presence of lactic acid at a concentration of 10.9 mmol L1 was
observed to increase both permeate and water flux; however, the
coefficient of enrichment was reduced to 12.2%.
As previously discussed, in tests where a flow rate of
5.5  106 m3 s1 was used, a comparison of the pervaporation of
the fermentation broth and that of the standard mixture of ethanol
and water indicates that the best results are achieved with the
fermentation broth. Furthermore, in the specific case of ethanol
produced from lignocellulosic banana waste, greater values were
observed for the pervaporation index than for the standard mix-
ture. We believe that the observed increase in the pervaporation
index was due to the presence of lactic acid in the broth. As a result
of the low concentration of this acid observed in the fermentation
broth at 0.75 mmol L1, a decrease was not observed in the enrich-
ment factor of ethanol, as reported in the work of Chovau et al.
(2011). Thus, because the polydimethylsiloxane membrane is
hydrophobic, the presence of lactic acid increased the hydrophilic-
ity of the membrane, which increased the flow of water, whereas
the increase in the flow of ethanol can be attributed to the effect
of membrane fouling. Bowen et al. (2007) have confirmed these
statements in their studies.
1508 R.H. Bello et al. / Waste Management 34 (2014) 1501–1509
Fig. 10. Comparison of the effect of ethanol content in feed on selectivity for model
solutions (j) and fermentation broth (h).
Table 7
By-products of the fermentation broth that may affect the process of pervaporation.
Byproduct Average concentration (g L1)
Lactic acid 0.068
Acetic acid NDª
Glucose ND
Glycerol 1.67
a 2011.
Not detected.
Finally, in all experiments the flux values obtained (5 g m2 h1)
were inferior to those described in the literature, such as those
mentioned in work by Molina et al. (2002), who obtained
500 g m2 h1 at a concentration of 15% (wt) under similar operat-
ing conditions. This discrepancy indicates the need for optimization
of the operating conditions. However, the values obtained here are
consistent with the data provided by the membrane supplier. In this
regard, the results are considered to be encouraging because the
parameters show substantial increases in the recovery of ethanol
produced by fermentation using pervaporation compared with
using the standard mixture.
4. Conclusions
The use of the pervaporation process for the recovery of ethanol
from fermentation broth produced from lignocellulosic banana
waste demonstrated a very attractive alternative to classical meth-
ods. When tests were carried out varying operational conditions
with model solutions, it was observed behaviors related to
reported in literature: a higher ethanol concentration in the feed
produces a higher ethanol flux and decrease the separation factor;
an increase of feed flow rate can enhance the permeation rate of
ethanol with the water remaining at almost the same value; water
and ethanol fluxes was increased with the temperature increase;
and the higher effect in flux increase was observed when the vapor
pressure in the permeate stream was close to the ethanol vapor
pressure. One interesting point was that the ethanol flux demon-
strated an increase even with ethanol concentrations less than
5 wt% in the feed (concentration expected in second generation
ethanol), but this increase is even greater at higher concentrations,
as expected. However, water flux increases more than alcohol,
resulting in a separation factor reduction.
It was observed that all of the parameters increased for the per-
vaporation of the broth at low flow rates compared with the model
solutions. The process efficiency was evaluated by permeance and
selectivity, showing that ethanol flux increase has a major effect on
the total flux increase, especially for fermentation broth. In fact, in
the specific case of ethanol produced from lignocellulosic banana
waste, greater values were observed for the pervaporation index
than for the standard mixture. This could due to by-products pres-
ent in the multicomponent mixtures facilitating the ethanol per-
meability. By-products analysis show that the presence of lactic
acid increased the hydrophilicity of the membrane, which
increased the flow of water, whereas the increase in the flow of
ethanol can be attributed to the effect of membrane fouling.
Thus, it can be concluded that the pervaporation of ethanol pro-
duced from banana waste is indeed a technology with the potential
to be applied. The results showed a very interesting performance,
highlighting the better separation efficiency for fermentation broth
in relation to model solutions.
Acknowledgements
The authors acknowledge the support received from the Uni-
versidade da Região de Joinville – Univille as well as all of the
resources that allowed for the preparation of this work, and
acknowledge the financial support that was received in the form
of research grants.
References
ABIB – Associação Brasileira Indústrias Biomassa Brasil Biomassa E Energia
Renovável. Inventário Residual Brasil – Wood Pellets - Briquete – Energia,
E-book. <http://pt.calameo.com/read/000200968cc3a949579a0>.
(accessed 27.02.12).
Aroujalian, A., Raisi, A., 2009. Pervaporation as a means of recovering ethanol from
lignocellulosic bioconversions. Desalination 247, 509–517.
Arumugam, R., Manikandan, M., 2011. Fermentation of pretreated hydrolyzates of
banana and mango fruit wastes for ethanol production. Asian J. Exp. Biol. Sci. 2,
246–256.
Bowen, T.C., Meier, R.G., Vane, L.M., 2007. Stability of MFI zeolite-filled PDMS
membranes during pervaporative ethanol recovery from aqueous mixtures
containing acetic acid. J. Membr. Sci. 298, 117–125.
Clarke, W.P., Radnidge, P., Lai, T.E., Jensen, P.D., Hardin, M.T., 2008. Digestion of
waste bananas to generate energy in Australia. Waste Manage. 28, 527–533.
Chovau, S., Degrauwe, D., Van der Bruggen, B., 2013. Critical analysis of techno-
economic estimates for the production cost of lignocellulosic bio-ethanol.
Renewable Sustainable Energy Rev. 26, 307–321.
Chovau, S., Gaykawad, S., Straathof, A.J.J., Van der Bruggen, B., 2011. Influence of
fermentation by-products on the purification of ethanol from water using
pervaporation. Bioresour. Technol. 102, 1669–1674.
Dermibas, A., 2011. Competitive liquid biofuels from biomass. Appl. Energy 88, 17–
28.
Dobrak, A., Figoli, A., Chovau, S., Galiano, F., Simone, S., Vankelecom, I.F.J., Drioli, E.,
Van der Bruggen, B., 2010. Performance of PDMS membranes in pervaporation:
effect of silicalite fillers and comparison with SBS membranes. J. Colloid
Interface Sci. 346, 254–264.
EMBRAPA (2006) Empresa Brasileira de Pesquisa Agropecuária. Embrapa fruta
<http://www.jornalentreposto.com.br/anteriores/janeiro_2006/
transporte.htm> June 2008. (accessed 13.08.12).
España-Gamboa, E., Mijangos-Cortes, J., Barahona-Perez, L., Dominguez-Maldonado,
J., Hernández-Zarate, G., Alzate-Gaviria, L., 2011. Vinasses: characterization and
treatments. Waste Manage. Res. 29, 1235–1250.
Fernandes, E.R.K., Marangoni, C., Souza, O., Sellin, N., 2013. Thermochemical
characterization of banana leaves as a potential energy source. Energy
Convers. Manage. 75, 603–608.
Gaykawad, S.S., van der Wielen, L.A.M., Straathof, A.J.J., 2013. Effects of yeast-
originating polymeric compounds on ethanol pervaporation. Bioresour.
Technol. 116, 9–14.
Graefe, S., Dufour, D., Giraldo, A., Muñoz, L.A., Mora, P., Solís, H., Garcés, H.,
Gonzalez, A., 2011. Energy and carbon footprints of ethanol production using
banana and cooking banana discard: a case study from Costa Rica and Ecuador.
Biomass Bioenergy 35, 2640–2649.
Gonçalves Filho, L.C., Fischer, G.A.A., Sellin, N., Marangoni, C., Souza, O., 2013.
Hydrolysis of banana tree pseudostem and second-generation ethanol
production by saccharomyces cerevisae. Bioresour. Technol. 2, 65–69.
Hammond, J.B., Egg, R., Diggins, D., Coble, C.G., 1996. Alcohol from bananas.
Bioresour. Technol. 56, 125–130.
Hossain, A.B.M.S., Ahmed, S.A., Ahmed, M.A., Faris, M.A.A., Annuar, M.S.M., Hadeel,
M., Norah, H., 2011. Bioethanol fuel production from rotten banana as an
environmental waste management and sustainable energy. Afr. J. Microbiol.
Res. 5, 586–598.
 R.H. Bello et al. / Waste Management 34 (2014) 1501–1509
Jiraratananon, R., Chanachai, A., Huang, R.Y.M., Uttapap, D., 2002. Pervaporation
dehydration of ethanol-water mixtures with chitosan/hydroxyethylcellulose
(CS/HEC) composite membranes: I. Effect of operating conditions. J. Membr. Sci.
195, 143–151.
Kazi, F.K., Fortman, J.A., Anex, R.P., Hsu, D.D., Aden, A., Dutta, A., Kothandaraman, G.,
2010. Techno-economic comparison of process technologies for biochemical
ethanol production from corn stover. Fuel 89, S20–S28.
Korelskiy, D., Leppäajärvi, T., Zhou, H., Grahn, M., Tanskanen, J., Hedlund, J., 2013.
High flux MFI membranes for pervaporation. J. Membr. Sci. 427, 381–389.
Lai, C.-L., Fu, Y.-J., Chen, J.-T., An, Q.-F., Liao, K.-S., Fang, S.-C., Hu, C.-C., Lee, K.-R.,
2012. Pervaporation separation of ethanol/water mixture by UV/O3-modified
PDMS membranes. Sep. Purif. Technol. 100, 15–21.
Lee, H.-J., Cho, E.J., Kim, Y.-G., Choi, I.S., Bae, H.-J., 2012. Pervaporative separation of
bioethanol using a polydimethylsiloxane/polyetherimide composite hollow-
fiber membrane. Bioresour. Technol. 109, 110–115.
Lewandowska, M., Kujawski, W., 2007. Ethanol production from lactose in a
fermentation/pervaporation system. J. Membr. Sci. 79, 430–437.
Li, L., Xiao, Z., Tana, S., Pua, L., Zhang, Z., 2004. Composite PDMS membrane with
high flux for the separation of organics from water by pervaporation. J. Membr.
Sci. 243, 177–187.
Liang, L., Dickson, J.M., Jiang, J., Brook, M.A., 2004. Effect of low flow rate on
pervaporation of 1,2-dichloroethane with novel polydimethylsiloxane
composite membranes. J. Membr. Sci. 231, 71–79.
Luis, P., Degrève, J., Van der Bruggen, B., 2013. Separation of metanol-n-butyl
acetate mixtures by pervaporation: potential of 10 commercial membranes. J.
Membr. Sci. 429, 1–12.
Mabee, W.E., McFarlane, P.N., Saddler, J.N., 2011. Biomass availability for
lignocellulosic ethanol production. Biomass Bioenergy 35, 4519–4529.
Mohammadi, T., Aroujalian, A., Bakhshia, A., 2005. Pervaporation of dilute alcoholic
mixtures using PDMS membrane. Chem. Eng. Sci. 60, 1875–1880.
Molina, J.M., Vatai, G., Bekassy-Molnar, E., 2002. Comparison of pervaporation of
different alcohols from water on CMG-OM-010 and 1060-SULZER membranes.
Desalination 149, 89–94.
Nomura, M., Bin, T., Nakao, S., 2002. Selective ethanol extraction from fermentation
broth using a silicalite membrane. Sep. Purif. Technol. 27, 56–69.
Oberoi, H.S., Vadlani, P.V., Saida, L., Bansal, S., Hughes, J.D., 2011. Ethanol production
from banana peels using statistically optimized simultaneous saccharification
and fermentation process. Waste Manage. 31, 1576–1584.
O’Brien, D.J., Senske, G.E., Kurantz, M.J., Craig Jr., J.C., 2004. Ethanol recovery from
corn fiber hydrolysate fermentations by pervaporation. Bioresour. Technol. 92,
15–19.
Ortiz, I., Alonso, P., Urtiaga, A., 2002. Pervaporation of azeotropic mixtures ethanol/
ethyl tert-butyl ether: Influence of membrane conditioning and operation
variables on pervaporation flux. Desalination 49, 67–72.
Pereira, C.C., Ribeiro Jr., C.P., Nobrega, R., Borges, C.P., 2006. Pervaporative recovery
of volatile aroma compounds from fruit juices. J. Membr. Sci. 274, 1–23.
Rivas-Cantu, R.C., Jones, K.D., Mills, P., 2013. A citrus waste-based biorefinery as a
source of renewable energy: technical advances and analysis of engineering
challenges. Waste Manage. Res. 31, 413–420.
Sarkar, N., Ghosh, S.K., Bannerjee, S., Aikat, K., 2012. Bioethanol production from
agricultural wastes: an overview. Renewable Energy 37, 19–27.
1509
Schulz, M.A. 2010 Bioethanol production from banana cultivation waste: banana
peels and pulp. Dissertation. MSc in Process Engineering – UNIVILLE, Joinville,
SC. (In Portuguese).
Shepherd, A., Habert, A.C., Borges, C.P., 2002. Hollow fibre modules for orange juice
aroma recovery using pervaporation. Desalination 148, 111–114.
Schnabel, S., Moulin, P., Nguyen, Q.T., Roizard, D., Aptel, P., 1998. Removal of volatile
organic components (VOCs) from water by pervaporation: separation
improvement by Dean vortices. J. Membr. Sci. 142, 129–141.
Soccol, C.S., Vandenberghe, L.P.S., Medeiros, A.B.P., Karp, S.G., Buckeridge, M.,
Ramos, L.P., Pitarelo, A.P., Leitão, V.F., Gottschalk, L.M.F., Ferrara, M.A., Bon,
E.P.S., Moraes, L.M.P., Araújo, J.A., Torres, F.A.G., 2010. Bioethanol from
lignocelluloses: status and perspectives in Brazil. Bioresour. Technol. 101,
4820–4825.
Sommer, S., Melin, T., 2005. Influence of operation parameters on the separation of
mixtures by pervaporation and vapor permeation with inorganic membranes.
Part 1: dehydration of solvents. Chem. Eng. Sci. 60, 4509–4523.
Souza, O., Federizzi, M., Coelho, B., Wagner, T.M., Wisbeck, E., 2010. Biodegradação
de resíduos lignocelulósicos gerados na bananicultura e sua valorização para a
produção de biogas. Revista Brasileira de Engenharia Agrícola e Ambiental 14,
438–443 (in Portuguese).
Sun, D., Li, B.-B., Xu, Z.-L., 2013. Pervaporation of ethanol/water mixture by
organophilic nano-silica filled PDMS composite membranes. Desalination 322,
159–166.
Swana, J., Yang, Y., Behnam, M., Thompson, R., 2011. An analysis of net energy
production and feedstock availability for biobutanol and bioethanol. Bioresour.
Technol. 102, 2112–2119.
Tewari, H.K., Marwaha, S.S., Rupal, K., 1986. Ethanol from banana peels. Agric.
Wastes 16, 135–146.
Tock, J.Y., Lai, C.L., Lee, K.T., Tan, K.T., Bhatia, S., 2010. Banana biomass as potential
renewable energy resource: a Malaysian case study. Renewable Sustainable
Energy Rev. 14, 798–805.
van Baelen, D., Reyniers, A., Van der Bruggen, B., Vandecasteele, C., Degreve, J., 2004.
Pervaporation of binary and ternary mixtures of water with methanol and/or
ethanol. Separ. Sci. Technol. 39, 559–576.
Vane, L.M., 2008. Separation technologies for the recovery and dehydration of
alcohols from fermentation broths. Biofuels Bioprod. Biorefin. 2, 553–588.
Velásquez-Arredondo, H.I., Ruiz-Colorado, A.A., De Oliveira Jr, S., 2010. Ethanol
production process from banana fruit and its lignocellulosic residues: energy
analysis. Energy 35, 3081–3087.
Wu, Y., Xiao, Z., Huang, W., Zhong, Y., 2005. Mass transfer in pervaporation of active
fermentation broth with a composite PDMS membrane. Sep. Purif. Technol. 42,
47–53.
Zereshki, S., Figoli, A., Madaeni, S.S., Simone, S., Esmailinezhad, M., Drioli, E., 2011.
Pervaporation separation of MeOH/MTBE mixtures with modified PEEK
membrane: effect of operating conditions. J. Membr. Sci. 371, 1–9.
Zhang, W., Ma, H., Wang, Q., Zhao, F., Xiao, Z., 2012. Pretreatment technology for
suspended solids and oil removal in an ethanol fermentation broth from food
waste separated by pervaporation process. Desalination 293, 112–117.
Zhou, H., Su, Y., Chen, X., Wan, Y., 2011. Separation of acetone, butanol and ethanol
(ABE) from dilute aqueous solutions by silicalite-1/PDMS hybrid pervaporation
membranes. Sep. Purif. Technol. 79, 375–384.