Accepted Manuscript

Methodology for the optimal design of an integrated first and second generation
ethanol production plant combined with power cogeneration

Rami Bechara, Adrien Gomez, Valérie Saint-Antonin, Jean-Marc Schweitzer,

François Maréchal

PII: S0960-8524(16)30626-5
DOI: http://dx.doi.org/10.1016/j.biortech.2016.04.130
Reference: BITE 16489

To appear in: Bioresource Technology

Received Date: 19 February 2016

Revised Date: 27 April 2016
Accepted Date: 28 April 2016

Please cite this article as: Bechara, R., Gomez, A., Saint-Antonin, V., Schweitzer, J-M., Maréchal, F., Methodology
for the optimal design of an integrated first and second generation ethanol production plant combined with power
cogeneration, Bioresource Technology (2016), doi: http://dx.doi.org/10.1016/j.biortech.2016.04.130

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Methodology for the optimal design of
an integrated first and second
generation ethanol production plant
combined with power cogeneration
Rami Bechara1*, Adrien Gomez1, Valérie Saint-Antonin2, Jean-Marc Schweitzer1, François

Maréchal3

1 Process Modeling and Design, IFPEN, Insitut Français du Pétrole et des

Energies Nouvelles Rond Point de l'Echangeur de Solaize, BP3, 69360 Solaize,
France
2
Economics and Information Watch and Management, IFPEN, 1-4 Avenue du
Bois Préau, 92852 Rueil-Malmaison
3
Industrial Process and Energy Systems Engineering, Ecole Polytechnique
Fédérale de Lausanne, EPFL Valais Wallis , Rue de l'Industrie 17, CH-1951 Sion,
Switzerland
rami.bechara@etu.univ-lyon1.fr, adrien.gomez@ifpen.fr; valerie.saint-
antonin@ifpen.fr; jean-marc.schweitzer@ifpen.fr; francois.marechal@epfl.ch
* Corresponding Author : Rami Bechara rami.bechara@etu.univ-lyon1.fr

Abstract

The application of methodologies for the optimal design of integrated processes has seen

increased interest in literature. This article builds on previous activities and applies a systematic

methodology to an integrated first and second generation ethanol production plant with power

cogeneration. The methodology breaks into process simulation, heat integration, thermo-

economic evaluation, exergy efficiency vs. capital costs, multi-variable evolutionary optimization,

and process selection via profitability maximization. Optimization generated Pareto solutions with

exergy efficiency ranging between 39.2% and 44.4% and capital costs from 210 M$ to 390 M$.

The Net Present Value was positive for only two scenarios and for low efficiency, low hydrolysis

points. The minimum cellulosic ethanol selling price was sought to obtain a maximum NPV of zero

for high efficiency, high hydrolysis alternatives. The obtained optimal configuration presented

1
maximum exergy efficiency, hydrolyzed bagasse fraction, capital costs and ethanol production

rate, and minimum cooling water consumption and power production rates.

Keywords

First and second generation ethanol production; Process modeling; Evolutionary bi-objective

multi-variable optimization; Selection by Profitability maximization: Minimum Cellulosic Ethanol

Selling Price

Introduction

The application of systematic methodologies for the optimal design of integrated processes has

seen increased interest in literature. (Morandin et al., 2010) applied a systematic methodology to

the optimization of a combined sugar and ethanol production process integrated with a CHP

system with the objective of maximizing power production. (Bechara et al., 2014) on the other

hand applied a similar methodology for the minimization of the utility consumption of a stand-

alone second generation ethanol production process. Finally, (Albarelli et al., 2015) used such a

methodology for the optimization of the joint production of ethanol and methanol from

sugarcane with energy efficiency and capital costs as chosen objective functions.

In the context of biomass valorization and renewable biofuels production, the integrated first and

second generation ethanol from sugarcane production process combined with power

cogeneration has seen increased interest in literature. (Dias et al., 2012b) investigated the

improvement of the integrated process by modifying the operating conditions of the biomass

combustion section, namely the boiler pressure and superheating temperature. (Macrelli et al.,

2012) performed a thermo-economic evaluation of several process configurations integrating

different second generation ethanol producing technologies with various first generation

schemes. (Dias et al., 2013) on the other hand studied the impact of varying hydrolysis solids

2
loading and conversion yield on process specific steam consumption (kg steam/ton sugarcane),

specific ethanol production (l ethanol/ton sugarcane) and specific power production (kWh/ ton

sugarcane). (Furlan et al., 2012) went a step further by coupling process simulation with a global

optimization algorithm with the goal of determining the optimal fraction of bagasse to be

diverted to second generation ethanol production with regards to revenue maximization.

Moreover, (Ensinas et al., 2013) made use of a similar tool, but with a bi-objective optimization:

maximizing electricity production versus maximizing ethanol production. Furthermore, this work

incorporated heat integration into the optimization problem. Likewise, (Costa et al., 2015)

performed multiple bi-objective optimization runs to a variant of the study process, with the

possibility of distillation waste, vinasse, concentration. All these works stressed the importance of

using a systematic methodology for optimal process design, and highlighted the predicament

posed by diverting large quantities of bagasse to hydrolysis. Finally, (Macrelli et al., 2014) studied

the additional effect of varying market factors on the choice of process alternatives, which are

also a key factor in profitability and minimum selling price.

Considering the previous, this present article expands on these works and applies a systematic

process design methodology for an integrated first and second generation ethanol production

plant coupled with electricity cogeneration. This article starts by describing and applying the

chosen methodology and its constitutive steps to the studied process. It is then followed by a

visualization, assessment and discussion of the obtained results, before finishing off with

conclusions.

1 Materials and Methods

The used methodology, highlighted and employed in (Gassner and Maréchal, 2012),can be broken

down into three main steps as highlighted in Figure 1. The first step consists in generating the

process model with the ultimate goal of enabling thermo-economic evaluation. Its key sub-steps

3
are process simulation, heat integration and ultimately thermo-economic evaluation. The second

step consists in global process optimization. The chosen technique is multi-variable, bi-objective

optimization using evolutionary algorithms as highlighted in (Leyland, 2002). This step leads to

the generation of a Pareto Optimal Frontier (POF) for the optimization problem. This frontier

highlights the optimal compromise between the chosen objective functions. Considering this, the

last step consists in the selection of the most interesting process configuration from the

previously obtained Pareto Optimal Frontier. This selection step makes use of decision making

techniques which guides the decision maker towards the most interesting solutions.

The application of this methodology and its constitutive components for the optimal design of an

integrated first and second generation ethanol production plant combined with power

cogeneration is highlighted in this section

1.1 Step I: Generate Process Model

As indicated in Figure 1, the process model breaks down into simulation, heat integration and

thermo-economic evaluation. The application of each component to the studied process will be

detailed herein.

1.1.1 Process simulation Model

The process simulation model is described for the studied process in this section where process

capacity and block flow diagram are indicated.

1.1.1.1 Process capacity

Process capacity was set to 500 tons of sugarcane (TC)/h, and to 33 tons of leaves/h (70 kg leaves

/TC). Input sugarcane is composed of: water (71.57 wt.%), sugars (13.92 wt.%), dirt (0.6 wt.%),

impurities (1.99 wt.%) and bagasse fibers (11.92 wt.%). This bagasse is composed of: cellulose (43

wt.%), hemicellulose (26 wt.%), lignin (24 wt.%) and ashes (7 wt.%). Input leaves is on the other

composed of: water (15 wt. %), ash (2 wt.%), and biomass fibers (83 wt.%). The process has

4
moreover a third input material, enzymes, whose mass flow rate is set to 0.1 g/g hydrolyzed

bagasse cellulose. It is thus directly dependent on the problem’s optimization variables. Finally,

The NREL database was chosen for modeling the various components. These information are in

line with (Dias et al., 2009; Ensinas et al., 2013).

1.1.1.2 Process Block Flow Diagram

The studied process consists in an integrated first and second generation ethanol production

distillery, combined with a heat and power production plant, and a cold utility system. The block

flow diagram for this process, as inspired from previous literature works (Ensinas et al., 2013;

Macrelli et al., 2012), is provided in Figure 2. Three important sections can be identified therein.

The first deals with the distillery which encompasses (1) the sugarcane mill, (2) sugarcane juice

treatment, (3) bagasse hydrolysis, (4) concentration and (5) fermentation, (6) wine concentration

and (7) ethanol dehydration, for anhydrous ethanol production. The concentration sub-step takes

in two juice streams. The first relates to the sugarcane juice, mainly composed of sucrose. The

second stream on the other hand is made of glucose C5 sugars produced by the hydrolysis of the

bagasse cellulose fraction. The second section deals with the cogeneration system which provides

steam and electricity to power the process, with the possibility of surplus electricity production.

This cogeneration system takes four inputs: sugarcane bagasse, leaves, unhydrolyzed biomass

and biogas obtained from xylose biodigestion. The last section constitutes the cold utilities used

to cool process streams. The process has also three input streams: sugarcane, enzymes and

leaves.

The chosen technologies for the distillery separation sections are as follows multiple-effect

evaporation for juice concentration (Ensinas et al., 2007; Morandin et al., 2010; Urbaniec et al.,

2000), the Melle-Boinot process for glucose fermentation (Basso et al., 2011), two-column

distillation for distillation and rectification (Bessa et al., 2013, 2012; Dias et al., 2011b; Palacios-

5
Bereche et al., 2015), and extractive distillation via Methyl-Ethyl Glycol for ethanol dehydration

(Errico et al., 2013; GarciÌa-Herreros et al., 2011).

Moreover, the block flow diagram for the hydrolysis section breaks down as follows. The chosen

pretreatment technology is
  catalyzed steam explosion (Ramos et al., 2014), motivated by

its wide application both in experimentation (Aguiar et al., 2013; Ramos et al., 2014), simulation

(Dias et al., 2013; Ensinas et al., 2013; Macrelli et al., 2012) and pilot scale (Agbor et al., 2011).

This technology leads to the autohydrolysis of hemicellulose to soluble xylose. For this reason,

this step is followed by a solid/liquid separation step which yields a liquid stream, namely made of

xylose from hemicellulose hydrolysis, and a solid stream composed of the remaining lignin-

cellulose fraction. The first stream is sent to biodigestion for the production of biogas, sent to

cogeneration. The second is diluted with water and sent to the hydrolysis reactor where the

cellulose fraction of pretreated biomass is converted into glucose by the action of added

enzymes. This step is also followed by a solid/liquid separation step, which produces a soluble

glucose juice, sent to fermentation, and hydrolysis biomass, later dried in a pres filter to reach the

humidity level needed for its usage in cogeneration.

The block flow diagram for the cogeneration section breaks down as follows. The biogas is burned

in a gas turbine producing electricity along with hot flue gases. The three solid biomass streams

are: leaves, subjected to washing, and bagasse and hydrolysis biomass, subjected to drying.

Hydrolysis biomass is mainly composed of the unreacted biomass fraction combined with the

cellulose hydrolyzing enzymes. The solid biomass streams are used as solids in burners producing

hot flue gases. Burner heat and hot flue gases are then available for a variety of purposes as

highlighted in (Ensinas et al., 2010). This includes: the multi-level Rankine cycle for the combined

production of heat (by condensation of low pressure steam) and power as highlighted in

(Morandin et al., 2010) and the preheating of boiler water and combustion air and finally biomass

drying through flue gas recirculation. This technique is in line with the works of (Ensinas et al.,

2010), namely due to its utilization of the waste heat present in the flue gases. Finally, the cold

6
utility section is made of cooling water and refrigerant as indicated in (Dias et al., 2012b;

Morandin et al., 2010).

This model was developed and run using the Aspen Plus V7.2™platform. The goal of the present

article is to find a configuration of interest for this fixed process scheme.

1.1.2 Heat integration Model

The heat integration model, as detailed in (Gassner and Maréchal, 2012), uses the heat cascade

method to represent the heat exchanger network and calculate the optimal utility system, and

this by applying the following sequence. Heat and power streams are first extracted from the

process simulation model Process composite curves are then generated by the use of the

temperature interval heat cascade method (Linnhoff and Flower, 1978). Third, the optimal utility

consumption rate is obtained by the use of a Mixed Integer Linear Programming (MILP) model as

highlighted in (Gassner and Maréchal, 2012). This utility rate meets the hot and cold process

Minimum Energy Requirements, whilst maximizing power production in the utility system. This

system contains the boiler, the multi-level steam turbine along with cold water and refrigeration.

The 8 controlled variables pertain to the mass flow rates in these systems, whereas the selection

criterion pertains to their respective nominal power production rate. These rates are dependent

on the choice of operating conditions. Certain concepts can however be summarized: the cold

water system has a lower power consumption rate than the refrigeration system. It is thus

preferred except when temperatures drop below its operational range. The power production

rate decreases with decreasing pressure in the turbines. This is however offset by the need for

higher temperature heat by the process. Finally, the process Heat Exchange Network area is

calculated via the Enthalpy-Interval method as provided in (Linnhoff and Ahmad, 1990). This area

encompasses all heat exchanging equipment, evaporators, reboilers, condensers, heaters,

coolers. After its convergence, this model yields values for utility consumption rates, power

production in the utility system and the total required heat exchange area. This model was

7
developed using the Matlab ® based OSMOSE platform, whereas the MILP algorithm was solved

using GLPK.

1.1.3 Thermo-Economic Evaluation Model

This section deals with the evaluation of key process performance indicators. These indicators

include: the ethanol production rate (t/h), the power production rate (MW), exergy efficiency (%)

and capital cost (M$). This step takes place once the process simulation and heat integration

models have converged, and extracts information from both these models. The elements for this

calculation are provided in this section

1.1.3.1 The ethanol production rate

The ethanol production rate is a result of the simulation model. This indicator has always been

measured and reported in the various literature works related to the studied process (Dias et al.,

2013, 2012b; Ensinas et al., 2013; Furlan et al., 2012; Macrelli et al., 2012). In this context,

increases in this rate are associated with increases in the extent of hydrolysis.

1.1.3.2 The electricity production rate

The electricity production rate (

 ) is also a key indicator measured in all literature works

(Dias et al., 2013, 2012b; Ensinas et al., 2013; Furlan et al., 2012; Macrelli et al., 2012). Contrary to

ethanol, the electricity production rate can be calculated only after the convergence of the

process heat integration model, yielding the electricity production in the utility system. The

power production in the gas turbine is added to this value, whereas the process power

consumption is subtracted. Such consumption is related to the sugarcane mill (16 kWh/ton

sugarcane), centrifuges (1 kWh/ m3 of input juice), hydrolysis reactors (0.3 kWh / m3 of reactor),

and pres filters (11 kWh / ton hydrolysis biomass). In this context, increases in this rate are a

direct indicator of increased process integration. Decreases in this rate relate however to greater

hydrolysis rates.

8
1.1.3.3 Exergy efficiency

Exergy efficiency is an indicator of global process efficiency. It is measured as the ratio of the

exergy contents of output streams to that of input streams. Its use has been condoned by (Rosen

et al., 2008). An exergy analysis was performed for a variant of the studied process (Palacios-

Bereche et al., 2013). Moreover, (Ensinas et al., 2013) evaluated the exergy efficiency of the

optimization results. Process output streams include ethanol and net power whereas input

streams include sugarcane, leaves and enzymes. The formula for exergy efficiency is provided in

Equation 1.

Equation 1 Formula for exergy efficiency of investigated process

+
  +   #$ × ex()
* + 
 ×  
  = = + + +
  +   ! +  " #$ × ex)
 + #$ ! × ex),- + #$" × ex
.

In this equation #$ , #$ #$ ! and #$" represent the mass flow rates of ethanol,

cane and leaves respectively, evaluated in t/h. They are set design specifications for cane and

+ +
leaves, and varying in the case of ethanol and enzymes. ex()
* (8.2 MWh/t), ex)
 (1.4

+ +
MWh/t), ex),- (4.4 MWh/t) and ex
. (6.1 MWh/t) represent the nominal chemical exergy

content of the corresponding streams. These components were calculated for the studied process

based on literature data (Palacios-Bereche et al., 2013). 

 on the other hand presents the

net power produced by the process. Its nominal exergy content   is equal to 1 MW/MW,

since electricity represents pure work.

1.1.3.4 Capital Cost

Capital cost is a key performance indicator measured in almost all related works (Dias et al., 2013,

2012b; Ensinas et al., 2013; Furlan et al., 2012; Macrelli et al., 2012). The construction of a grass

root facility was considered in this work. Considering this, process capital cost was calculated

based on the methodology of (Chauvel et al., 2003). The fixed capital cost is chosen equal to the

process Grass Root (GR) Cost, calculated given equipment purchase cost along with installation

and erection factors. The Marshal and Swift index was chosen equal to 1597.7 relative to Q4

9
(2011) in line with (Ensinas et al., 2013). Equipment sizing is realized based on simulation results,

except for heat exchanging equipment, whose dimensions are included in the global heat

exchange network, constructed as a result of the heat integration step. Nonetheless, heat

integration parameters ultimately help calculate the cost of utility equipment, namely turbines.

Both the investment costs and exergy efficiency were calculated via the Matlab ® based OSMOSE

platform.

1.1.3.5 Profitability indicator:

Profitability was also evaluated in the works of (Dias et al., 2012a; Ensinas et al., 2013; Furlan et

al., 2013; Macrelli et al., 2014, 2012). This evaluation was realized by calculating: the Net Present

Value (Furlan et al., 2013), the internal rate of return or the simple rate of return (Dias et al.,

2012a; Macrelli et al., 2012) and the minimum ethanol and cellulosic ethanol selling prices

(Ensinas et al., 2013; Macrelli et al., 2012). Such indicators were also investigated in other

literature works namely (Pintarič and Kravanja, 2015)who vindicated the use of the Net Present

Value as a profitability indicator, namely for its suitability in selecting between mutually exclusive

process alternatives. With this in mind, the Net Present Value will be the selected profitability

indicator. Nonetheless, the remaining indicators will be kept for use in case of need, as will be

later highlighted. Considering this, the NPV of a given configuration under a chosen economic

scenario was calculated, via a dedicated Excel ® flowsheet, as indicated in (Chauvel et al., 2003),

taking into account operating costs, revenue, taxes, depreciation, construction and start-up

period along with cash flow actualization. Finally, profitability indicators depend greatly on

economic parameters and choices made by the decision maker, as highlighted in (Macrelli et al.,

2014). This dependence will be exploited in this work.

1.2 Step II: Perform process optimization

The chosen optimization problem is highlighted in this section.

10
1.2.1 Choice of objective functions

Two objective functions were selected for the optimization problem: maximize exergy efficiency

(  ) and minimize capital cost (/0123 ), as highlighted in Table 1. This choice is motivated by

the following reasons. First of all, exergy efficiency is a suitable indicator for polygeneration, as is

the case for this process. Second, exergy efficiency and capital cost are conflictive: a higher

efficiency requires a higher cost (Rosen et al., 2008). Third, these functions do not depend on

market conditions, as opposed to profitability indicators. Finally; these functions constitute the

core elements of any subsequent selection step.

1.2.2 Choice of optimization variables

Optimization variables are process design variables whose values are controlled by the

optimization algorithm to give better results for the objective functions. These variables were

chosen based on related literature works as highlighted in Table 1 They pertain to all process

sections with three variables related to hydrolysis. Certain key ideas related to the optimization

interactions are highlighted below. The variation of distillery and cogeneration parameters, lead

to a greater heat integration, a smaller heat demand, a greater power production and a greater

bagasse availability (Dias et al., 2012b; Macrelli et al., 2012). This was partially exploited by

(Ensinas et al., 2013) who included the sugar concentrations at the output of the evaporators as

optimization variables. The variation of hydrolysis parameters can be broken down as follows.

First, a greater fraction of bagasse diverted to hydrolysis leads to greater ethanol production but

to a higher investment cost, and a lower heat production, with the possibility of not meeting

process heat demands, configurations that are rejected in this case. Second, a higher solid loading

leads to a lower conversion rate but to lower energy demands and investment costs. Third, a

higher hydrolysis rate leads to greater ethanol production rate at the expense of a higher power

demand and investment cost. With this in mind, only the hydrolyzed bagasse fraction was an

optimization variable in the works of (Ensinas et al., 2013).

11
Finally, Equation 2 provides the mathematical model for the hydrolysis reaction, considered as a

first order reaction. This model highlights the linear dependence of the glucose concentration in

the hydrolysates (/45,73 ) on the hydrolysis residence and the solids loading. This choice is

supported by the works of (Ramos et al., 2014) from which the values of the various parameters

were deduced. Values for kinetic parameters were as follows (8 = 257.33, > = 5.73, =

−0.00398, C = 0.0069, EF73,+ = 24ℎ). Finally, this data is inputted into the process simulation

model which evaluates the conversion yield required to obtain this concentration.

Equation 2 Model for hydrolysis reaction – dependence on reaction parameters

/45,73 Jres(MN , sd*)N P = (a × sd*)N + b) × exp ((c × sd*)N + d) × (res(MN − res(MN,+ )

1.2.3 Measured parameters

Table 1 further contains measured parameters, highlighted by their direct link to the chosen

objective functions. This work differs from that of (Ensinas et al. 2013) (Ensinas et al., 2013) who

chose ethanol production and power production as the objective functions with capital costs and

exergy efficiency evaluated after the optimization. Moreover, no profitability criterion is included

at this level. This is because such criteria will be employed in the subsequent selection step as

highlighted previously.

1.2.4 Chosen optimization algorithm

In continuity with previous research works (Albarelli et al., 2015; Bechara et al., 2014; Ensinas et

al., 2013; Morandin et al., 2010), the application of the QMOO algorithm, an elitist multi-objective

evolutionary algorithm, developed and run in Matlab ®, to the previously constructed process

model was considered in this article. QMOO breaks down as follows (Leyland, 2002):

• Random sampling to generate an initial number, set to 100, of configurations.

• Add the number of evaluated configurations to the total number of evaluated points

• Removal of a chosen number, set to 20, of bad ranking configurations, others are kept “alive”

12
 • Crossover and mutation to c the corresponding number of new configurations

• Return to point b, unless a chosen maximum number of evaluations is reached, set to 1000

• The Pareto set contains the best ranking points in the last set of configurations.

Finally, the analysis of optimization results was realized via a dedicated Excel ® flowsheet.

1.3 Step III: Select most interesting process configuration

As evidenced in Figure 1, the selection of the most interesting process configuration is the last

step of the methodology. The chosen method in this article is the evaluation of profitability,

expressed by the Net Present Value along with other parameters. This profitability is highly

dependent on market and financial conditions (Macrelli et al., 2014). This is why several economic

scenarios deduced from literature works (Dias et al., 2011a; Ensinas et al., 2013; Furlan et al.,

2013; Macrelli et al., 2012) were chosen for this goal. Values for certain parameters are

highlighted for these scenarios in Table 2. Other non specified parameters were constant for all

scenarios and include tax rate (34%), discount rate (10%), depreciation strategy (10 year-linear),

maintenance and overhead costs (10% of investment cost), among others.

2 Results & Discussion

The results obtained for both bi-objective multi-variable evolutionary optimization and the

subsequent process selection step are highlighted in this section

2.1 Results corresponding to process optimization

Figure 3 highlights the resulting Pareto set containing a total of 42 points with exergy efficiency

ranging from 39.2% to 44.4% and capital costs from 210 M$ to 390M$. These results indicate that

capital costs increase with exergy efficiency, albeit at a far greater rate. As a matter of fact, the

capital cost of the most expensive, most efficient system is 85% greater than the capital cost of

13
the least expensive, least efficient system, whereas its exergy efficiency is but 13% greater. Figure

3 also highlights the existence of knee points, for which capital costs witness a greater increase

than exergy efficiency, along with the two extreme points corresponding to low efficiency and

low cost and high efficiency and high cost. Figure 3 also provides a comparison with the results of

(Ensinas et al., 2013). This comparison is possible because this work adopts a similar process

scheme, with bagasse hydrolysis and leaves combustion, along with similar input sugarcane

content. It also makes use of bi-objective optimization with process simulation and heat

integration. This optimization has however a reduced number of variables, and different objective

functions consisting in maximizing ethanol and electricity production, with exergy and capital

costs evaluated after optimization. Considering this, 4 points relative to this study are plotted on

Figure 3 in comparison to the obtained curve. Moreover, points 1 and 2 coincide with the

obtained Pareto curve, whereas points 3 and 4 have largely negative results, marked by low

efficiencies and high costs. Optimization results as well as their comparison to literature can be

further understood by evaluating the values for measured parameters, namely ethanol and

power production rates.

Consequently, the net power production rate is plotted against the ethanol production rate in

Figure 4, with exergy efficiency highlighted in color codes. The curve obtained as a result of

optimization can be split into two domains. The first has a nearly constant ethanol production

rate for an increasing power production rate, and an exergy efficiency ranging from 39 to 42%.

The second section, with efficiencies from c.a. 42% to 44%, is highlighted by a nearly linear

(VW = 0.992) increase in ethanol production and decrease in electricity production. The first

section is characteristic of heat integration and electricity cogeneration with limited hydrolysis,

whereas the second is characterized by a combination between heat integration, electricity

cogeneration and hydrolysis. Hydrolysis thus leads to higher efficiencies than heat integration and

electricity cogeneration under current configurations. Figure 4 also contains a comparison with

14
the works of) (Ensinas et al., 2013). This curve is characterized by a continuous decrease in net

power production faced by an increase in ethanol production. Moreover, certain points intersect

with the present optimization results. These points are comparable to points 1 and 2. For the

remaining points, different results are obtained, characteristic of the associated optimization

problems. First, the first section of the current optimization curve is absent from the works of

(Ensinas et al., 2013), because of their choice of objective functions. This choice prohibits the

inclusion of alternatives with lower ethanol production rates yet similar power production rates

into the Pareto curve, and this despite their lower costs. Second, the higher electricity and

ethanol production rates are obtained in this work by virtue of its more complex optimization

problem. In fact, the current optimization problem contains a total of 28 variables, whereas that

of (Ensinas et al., 2013) contains but 5. More importantly, this work contains hydrolysis related

parameters that are absent from the works of (Ensinas et al., 2013). The authors’ choice of values

for these parameters was also the main reason for the low efficiency high cost alternatives

highlighted in Figure 3. With this being said, the current results highlight the advantages of the

chosen optimization problem. Moreover, this problem shifted configurations with high hydrolysis

contribution towards higher efficiencies, when lower efficiencies were encountered for these

case in (Ensinas et al., 2013).

2.2 Results for NPV analysis

The Net Present Value was computed for each Pareto configuration under each economic

scenario. The obtained results are highlighted in Figure 5, where curves present the evolution of

the Net Present Value with respect to exergy efficiency across the Pareto population for all

scenarios. The obtained curves follow a similar trend albeit with different NPV values, which vary

depending on the given economic scenario. More importantly, all hydrolysis configurations

(section 2,   > 41.8%)) have rather negative NPV values, except for the first hydrolysis

points (  = 42%) who have slightly positive values for the 2nd and 3rd scenarios. That can be

15
namely attributed to the great investment cost that cannot be overcome by the current

production rates at the given economic conditions.

2.3 Control of economic conditions to ensure profitability

The negative trends for the NPV of the hydrolysis configurations highlighted in Figure 5 were

overcome by seeking a minimum second generation ethanol selling price ([\ − 2], expressed

in $/L second generation (2G) ethanol). This parameter renders the maximum NPV value for the

hydrolysis section of the curve equal to zero, under the various defined economic conditions. This

price has been investigated namely in the works of (Ensinas et al., 2013; Macrelli et al., 2012).

This method was applied for the 1st and 4th economic scenarios, namely due to their all-negative

results in Figure 5. This led to calculate the MESP-2G and the identification of one corresponding

selected configuration for the two cases. This configuration had an exergy efficiency of 44.4% and

a fixed capital cost of 393 M$. Values for [\ − 2] were then sought for this configuration

under the 2nd and 3rd scenario. Considering this, Figure 6 highlights the evolution of the Net

Present Value with the use of the previously obtained values of MESP-2G for the four different

economic scenarios. The selected configuration (  = 44.4%) is encircled in black for all the

economic scenarios. Moreover, the corresponding [\ − 2] values for the various scenarios

are highlighted in the legend. Scenarios 2 and 3 had 30% smaller minimum second generation

ethanol selling prices, depicting the positive effects imparted by prosperous economic conditions.

The fourth scenario provided also slightly better results, attributed to its greater number of

operating days (210 days vs. 167 days for scenario 1, ref. Table 2).

A quick comparison between Figure 5 and Figure 6 shows that the NPV is not modified for the

first section and this because of its low cellulosic ethanol production rate. The NPV of second

section points witness a net increase. A clear difference can however be seen between the 1st and

4th and the 2nd and 3rd scenarios. In fact, whereas the obtained optimal point is a maximum for the

1st and 4th scenarios, it is a minimum for the 2nd and 3rd scenarios. This can be related to the

16
initially better performance of the latter scenarios, due to the high electricity price for the 2nd

(maximum at 86 MW ref. Table 2) and the high ethanol price for the 3rd.(ref. Table 2) scenarios.

2.4 Characteristics of selected hydrolysis configuration

Analysis of values of parameters of the selected configuration obtained via the [\ − 2]

criterion indicated that the optimal configuration has the highest value for efficiency (44.42%),

cost (393 M$), hydrolyzed bagasse fraction (0.95), and ethanol production (42 t/h), with the

lowest cold water consumption (1130 t/h) and electricity production rates (46 MW) and the

second lowest boiler water flow rate (367 t/h). It also presented a great heat integration rate

between the various process steps namely juice concentration and evaporation, along with high

pressure (94 atm) high superheating boiler (291 °C) operation. Hydrolysis was performed at great

residence times (96h) and mild solids loadings (13 wt.%). Finally, the humidity content of

hydrolysis biomass and bagasse streams was reduced to 30% via drying. All of these results

supplement and validate the previous research works. Moreover, the obtained value for the

hydrolyzed bagasse fraction represents a new frontier for the studied process, further

highlighting the benefits of the employed methodology.

Conclusions

The application of a methodology for the design of an optimal first and second generation

ethanol production process with power cogeneration was investigated. Bi-objective optimization

yielded Pareto points with exergy efficiency ranging from 39.2% to 44.4% and capital costs from

210 M$ to 390M. The Net Present Value was positive for two scenarios only for low efficiency

points. The minimum cellulosic ethanol selling price was sought to obtain a maximum NPV of zero

for high efficiency alternatives. The optimal configuration presented a maximum hydrolyzed

bagasse fraction and ethanol production rate, and minimum cooling water consumption and

power production rates.

17
List of abbreviations

Exergy of produced ethanol

  Exergy efficiency (%)  
(MW)
Exergy of net electricity
    Exergy of sugarcane (MW)
produced (MW)
  ! Exergy of leaves (MW)  " Exergy of enzymes (MW)
Specific chemical exergy of Specific electricity exergy
ex+()
*  
ethanol (MWh/t) content (MW/MW)
Mass flow rate of sugarcane + Specific chemical exergy of
#$ ex)

(t/h) cane(MWh/t)
Mass flow rate of leaves + Specific chemical exergy of
#$ ! ex),-
(t/h) leaves (MWh/t)
Mass flow rate of enzymes + Specific chemical exergy of
#$" ex
.
(t/h) enzymes (MWh/t)
/0123 Fixed capital cost (M$) ^_ Net Present Value (M$)
Solids loading in hydrolysis Residence time in hydrolysis
sd*)N res(MN
reactor (wt. %) reactor (h)
Kinetic parameters for
Glucose concentration in a, b, c, calculating /45,73 as a
/45,73
hydrolysates (g/l) d, res(MN,+ function of hydrolysis
parameters
Minimum selling price for
[\ − 2] second generation
ethanol ($/l 2G ethanol)

Acknowledgements

This work was funded and supervised by the IFP Energies Nouvelles (IFPEN), Solaize, France, and

more specifically by the process design department and the process modeling and conception

direction.

The authors recognize the contributions of Adriano V. Ensinas, of Universidade Federal do ABC,

Brazil, Juliana Q. Albarelli, of UNICAMP, Brazil, and Victor Codina, EPFL, for their collaboration

which was necessary for the correct modeling and investigation of the studied process.

References

1. Agbor, V.B., Cicek, N., Sparling, R., Berlin, A., Levin, D.B., 2011. Biomass pretreatment:

Fundamentals toward application. Biotechnol. Adv. 29, 675–685.

2. Aguiar, R.S., Silveira, M.H.L., Pitarelo, A.P., Corazza, M.L., Ramos, L.P., 2013. Kinetics of

18
 enzyme-catalyzed hydrolysis of steam-exploded sugarcane bagasse. Bioresour. Technol.

147, 416–

3. Albarelli, J.Q., Onorati, S., Caliandro, P., Peduzzi, E., Meireles, M.A.A., Marechal, F.,

Ensinas, A. V, 2015. Multi-objective optimization of a sugarcane biorefinery for integrated

ethanol and methanol production. Energy -.

doi:http://dx.doi.org/10.1016/j.energy.2015.06.104

4. Basso, L.C., Rocha, S.N., Basso, T.O., 2011. Ethanol production in Brazil: the industrial

process and its impact on yeast fermentation. INTECH Open Access Publisher.

5. Bechara, R., Viana Ensinas, A., Queiroz Albarelli, J., Maréchal, F., Gomez, A., Saint-

Antonin, V., Schweitzer, J., 2014. Methodology for Minimising the Utility Consumption of

a 2g Ethanol Process.

6. Bessa, L.C.B. a., Batista, F.R.M., Meirelles, A.J. a., 2012. Double-effect integration of

multicomponent alcoholic distillation columns. Energy 45, 603–612.

7. Bessa, L.C.B. a., Ferreira, M.C., Batista, E. a. C., Meirelles, A.J. a., 2013. Performance and

cost evaluation of a new double-effect integration of multicomponent bioethanol

distillation. Energy 63, 1–9.

8. Chauvel, A., Fournier, G., Raimbault, C., 2003. Manual of process economic evaluation.

Editions Technip.

9. Costa, C.B.B., Potrich, E., Cruz, A.J.G., 2015. Multiobjective optimization of a sugarcane

biorefinery involving process and environmental aspects. Renew. Energy.

doi:http://dx.doi.org/10.1016/j.renene.2015.10.043

10. Dias, M.O.S., Cunha, M.P., Jesus, C.D.F., Rocha, G.J.M., Pradella, J.G.C., Rossell, C.E.V.,

Maciel Filho, R., Bonomi, A., 2011a. Second generation ethanol in Brazil: Can it compete

with electricity production? Bioresour. Technol. 102, 8964–8971.

11. Dias, M.O.S., Ensinas, A. V., Nebra, S. a., Maciel Filho, R., Rossell, C.E. V, Maciel, M.R.W.,

2009. Production of bioethanol and other bio-based materials from sugarcane bagasse:

19
 Integration to conventional bioethanol production process. Chem. Eng. Res. Des. 87,

1206–1216.

12. Dias, M.O.S., Junqueira, T.L., Cavalett, O., Cunha, M.P., Jesus, C.D.F., Rossell, C.E.V.,

Maciel Filho, R., Bonomi, A., 2012a. Integrated versus stand-alone second generation

ethanol production from sugarcane bagasse and trash. Bioresour. Technol. 103, 152–161.

13. Dias, M.O.S., Junqueira, T.L., Jesus, C.D.F., Rossell, C.E.V., Maciel Filho, R., Bonomi, A.,

2012b. Improving second generation ethanol production through optimization of first

generation production process from sugarcane. Energy 43, 246–252.

14. Dias, M.O.S., Junqueira, T.L., Rossell, C.E. V., Maciel Filho, R., Bonomi, A., 2013. Evaluation

of process configurations for second generation integrated with first generation

bioethanol production from sugarcane. Fuel Process. Technol. 109, 84–89.

15. Dias, M.O.S., Modesto, M., Ensinas, A. V., Nebra, S. a., Filho, R.M., Rossell, C.E.V., 2011b.

Improving bioethanol production from sugarcane: evaluation of distillation, thermal

integration and cogeneration systems. Energy 36, 3691–3703.

16. Ensinas, A. V., Nebra, S. a., Lozano, M. a., Serra, L.M., 2007. Analysis of process steam

demand reduction and electricity generation in sugar and ethanol production from

sugarcane. Energy Convers. Manag. 48, 2978–2987.

17. Ensinas, A. V, Arnao, J.H.S., Nebra, S.A., 2010. Increasing energetic efficiency in sugar,

ethanol, and electricity producing plants, in: Luis Augusto Barbosa Cortez (Ed.), Sugarcane

Bioethanol: R&D for Productivity and Sustainability. Blucher Sao Paulo, pp. 583–600.

18. Ensinas, A. V, Codina, V., Marechal, F., Albarelli, J., Aparecida, M., 2013. Thermo-

Economic Optimization of Integrated First and Second Generation Sugarcane Ethanol

Plant. Chem. Eng. Trans. 35, 523–528.

19. Errico, M., Rong, B.-G., Tola, G., Spano, M., 2013. Optimal synthesis of distillation systems

for bioethanol separation. Part 1: Extractive distillation with simple columns. Ind. Eng.

Chem. Res. 52, 1612–1619.

20
 20. Furlan, F.F., Costa, C.B.B., Fonseca, G.D.C., Soares, R.D.P., Secchi, A.R., Cruz, A.J.G. Da,

Giordano, R.D.C., 2012. Assessing the production of first and second generation

bioethanol from sugarcane through the integration of global optimization and process

detailed modeling. Comput. Chem. Eng. 43, 1–9.

21. Furlan, F.F., Filho, R.T., Pinto, F.H., Costa, C.B., Cruz, A.J., Giordano, R.L., Giordano, R.C.,

2013. Bioelectricity versus bioethanol from sugarcane bagasse: is it worth being flexible?

Biotechnol. Biofuels 6, 142.

22. GarciÌa-Herreros, P., GoÌmez, J.M., Gil, I.D., RodriÌguez, G., 2011. Optimization of the

design and operation of an extractive distillation system for the production of fuel grade

ethanol using glycerol as entrainer. Ind. Eng. Chem. Res. 50, 3977–3985.

23. Gassner, M., Maréchal, F., 2012. Thermo-economic optimisation of the polygeneration of

synthetic natural gas (SNG), power and heat from lignocellulosic biomass by gasification

and methanation. Energy Environ. Sci. 5, 5768.

24. Leyland, G.B., 2002. Multi-objective optimisation applied to industrial energy problems.

25. Linnhoff, B., Ahmad, S., 1990. Cost optimum heat exchanger networks—1. Minimum

energy and capital using simple models for capital cost. Comput. Chem. Eng. 14, 729–750.

26. Linnhoff, B., Flower, J.R., 1978. Synthesis of heat exchanger networks: I. Systematic

generation of energy optimal networks. AIChE J. 24, 633–642.

27. Macrelli, S., Galbe, M., Wallberg, O., 2014. Effects of production and market factors on

ethanol profitability for an integrated first and second generation ethanol plant using the

whole sugarcane as feedstock. Biotechnol. Biofuels 7, 26.

28. Macrelli, S., Mogensen, J., Zacchi, G., 2012. Techno-economic evaluation of 2nd

generation bioethanol production from sugar cane bagasse and leaves integrated with

the sugar-based ethanol process. Biotechnol. Biofuels 5, 22.

29. Morandin, M., Toffolo, A., Lazzaretto, A., Maréchal, F., Ensinas, A. V., Nebra, S. a., 2010.

Synthesis and parameter optimization of a combined sugar and ethanol production

21
 process integrated with a CHP system. Energy 36, 3675–3690.

30. Palacios-Bereche, R., Ensinas, A. V., Modesto, M., Nebra, S. a., 2015. Double-effect

distillation and thermal integration applied to the ethanol production process. Energy 82,

512–523.

31. Palacios-Bereche, R., Mosqueira-Salazar, K.J., Modesto, M., Ensinas, A. V., Nebra, S. a.,

Serra, L.M., Lozano, M.-A., 2013. Exergetic analysis of the integrated first- and second-

generation ethanol production from sugarcane. Energy 62, 46–61.

32. Pintarič, Z.N., Kravanja, Z., 2015. The importance of proper economic criteria and process

modeling for single- and multi-objective optimizations. Comput. Chem. Eng. 83, 35–47.

33. Ramos, L.P., da Silva, L., Ballem, A.C., Pitarelo, A.P., Chiarello, L.M., Silveira, M.H.L., 2014.

Enzymatic hydrolysis of steam-exploded sugarcane bagasse using high total solids and low

enzyme loadings. Bioresour. Technol. 175, 195–202.

34. Rosen, M. a., Dincer, I., Kanoglu, M., 2008. Role of exergy in increasing efficiency and

sustainability and reducing environmental impact. Energy Policy 36, 128–137.

35. Urbaniec, K., Zalewski, P., Zhu, X.X., 2000. A decomposition approach for retrofit design of

energy systems in the sugar industry. Appl. Therm. Eng. 20, 1431–1442.

22
Figure Captions

Figure 1 Breakdown of methodology into constitutive steps and sub-steps

Figure 2 Block Flow Diagram for studied process

Figure 3 Obtained Pareto curve & comparison with results of (Ensinas et al., 2013)

Figure 4 Pareto Ethanol vs. Electricity: exergy efficiency variation and comparison with (Ensinas et

al., 2013)

Figure 5 NPV values for the various Pareto points under the different economic scenarios

Figure 6 Evolution of NPV with MESP-2G under the fourth chosen economic scenarios

23
Tables

Table 1 Optimization problem: objective functions, variables and parameters

Objective functions: Maximize Exergy Efficiency – Minimize Fixed Capital Cost

Block Unit operation Optimization Variable Reference

Hydrolyzed bagasse fraction (Ensinas et al.,

(%) 2013)

(Dias et al., 2013;

Solids loading in hydrolysis
Ramos et al.,
reactor (wt. %)
Hydrolysis Hydrolysis 2014)

(Aguiar et al.,

Residence time in hydrolysis 2013; Macrelli et

reactor (h) al., 2012; Ramos

et al., 2014)

Vaporization rates (6 (Urbaniec et al.,

variables) (-) 2000)

Juice Temperature decrements (5 (Morandin et al.,

Evaporation
concentration variables) (°C) 2010)

Sugar content in (Ensinas et al.,

concentrated juice (kg/kg) 2007)

Stripping column Operating pressure (atm) (Bessa et al.,

Distillation 2013, 2012; Dias

Rectifying column Operating pressure (atm)
et al., 2011b;

24
 Palacios-Bereche

et al., 2015)

Feed Vapor fraction (Errico et al.,

(mol/mol) 2013)

(GarciÌa-Herreros
Dehydration Extraction column Solvent to feed ratio (kg/kg)
et al., 2011)

Solvent input temperature

(°C)

(Macrelli et al.,

Bagasse drying Dryer Cake humidity (mol/mol) 2012; Morandin

et al., 2010)

Boiler pressure (atm)

(Dias et al.,
Boiler Superheating temperature
2012b)
(°C)

Boiler and

steam turbine
Operating pressures (5 (Morandin et al.,
Multi-level Turbine
variables) (atm) 2010)

Measured parameter: Net Electricity Production rate / Ethanol production

25
Table 2 Values for economic parameters based on economic scenarios deduced from literature

Scenario 1 2 3 4

(Dias et al., (Macrelli et (Ensinas et (Furlan et al.,

Source
2011a) al., 2012) al., 2013) 2013)

Ethanol price ($/l) 0.5 0.53 0.72 0.51

Electricity price ($/MWh) 71 86 51 69

Sugarcane price ($/t) 19.41 19.5 31.17 28.76

Leaves price ($/t) 15 13 15 17

Enzymes price ($/t) 1680 1250 1680

Operating days (d/y) 167 200 210

Project life time (y) 25 20 25

Construction and start-up period (y) 2 1 2

26
Figures

27
28
29
30
31
32
Graphical abstract

33
Highlights :

• Method for optimal design of 1st and 2nd generation ethanol and power plant
• Process simulation, integration and evaluation model
• Bi-objective multi-variable evolutionary optimization run
• Profitability analysis for choosing point under different economic scenarios
• Analysis of Pareto curve and characterization of optimal point

34