Optimization of multi-product biorefinery processes under

Subscriber access provided by Kaohsiung Medical University

Process Systems Engineering

Optimization of multi-product biorefinery processes under consideration of biomass supply chain management and market developments Kirsten Ulonska, Andrea Koenig, Marten Klatt, Alexander Mitsos, and Joern Viell Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.8b00245 • Publication Date (Web): 03 May 2018 Downloaded from http://pubs.acs.org on May 4, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Optimization of multi-product biorenery processes under consideration of biomass supply chain management and market developments Kirsten Ulonska,

†,‡

Andrea König,

†,‡

Marten Klatt,

Viell

†

†

Alexander Mitsos,

and Jörn

∗,†

†RWTH Aachen University, Aachener Verfahrenstechnik,Process Systems Engineering,

Forckenbeckstraÿe 51, 52074 Aachen, Germany ‡Both authors contributed equally to this work. E-mail: [email protected]

Abstract Even though a shift from conventional to renewable resources is often envisaged, lignocellulosic biorenery concepts struggle with economic viability and sustainability. In order to overcome these barriers, a full analysis from biomass supply chain, process performance optimization and product-portfolio selection is targeted. Addressing the economic viability and sustainability already at an early process development stage when only limited knowledge is available, Process Network Flux Analysis (PNFA) [Ulonska et al., AIChE J. 2017], is capable of systematically identifying the most valuable processing pathways. This enables a rst performance ranking based on the prot or global warming potential of pathways, thereby accelerating process development. While so far only processing networks have been considered, the methodology is herein extended 1

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 41

to consider biomass supply chain optimization and market-dependent price developments such that all main inuencing factors are considered simultaneously. The extended methodology is validated identifying reasonable plant locations in North RhineWestphalia, Germany. Enhancing economic viability of the best performing biofuel ethanol, a multi-product biorenery is targeted co-producing value-added chemicals. Herein, a co-production of iso-butanol raises the prot signicantly: a mass ratio of at most 1.9 (ethanol:iso-butanol) is required to break even.

1. Introduction

In times of crude oil depletion and an increasing awareness of global warming, the development of sustainable and ecient processes for the production of fuels and chemicals becomes more and more important. In particular, reducing the use of fossil fuels in engines by renewables is envisioned, considering that fuel combustion in transportation is responsible for 23% of global carbon dioxide emissions. 1 However, a limited availability of biomass, high capital investment, as well as uncertainties concerning supply chains limit the global development of biorenery processes. 2,3 Besides economic viability, the identication of optimal biorenery processes poses a complex design problem with a variety of degrees of freedom, including supply chain, feedstocks, and plant location, reaction and separation processes as well as product-portfolio selection. Consequently, for the development of sustainable and protable solutions, there is an urgent need for systematic approaches that address these questions. 48

Compared to heating values of 42

MJ kg

for crude oil 9,10 and 40-44

els 11 , the heating value of biomass is low, approximately 15-20

MJ kg

MJ kg

for petroleum fu-

depending on the biomass

type and composition. 9,10 In addition, biomass is spread over large areas, such that biomass transportation can contribute signicantly to the production cost. 11,12 In fact, Angus-Hankin et al. 13 , Kumar et al. 14 and Ek³io§lu et al. 15 determined that transportation of forestry biomass accounts for 20-40% of the total delivered fuel cost. This is considered one of the 2


Page 3 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


main reasons preventing lignocellulosic ethanol plants from commercialization. 15 To reduce transportation costs, compared to the fossil feedstock-based industry, smaller plant sizes and thus distributed production are often envisaged. 10 However, as a consequence of the economy of scale, the relative investment cost contribution is reduced at larger plant sizes thus targeting for centralized production. These contradictory eects can be addressed using the overall production costs or greenhouse gas emissions to determine optimal plant capacities and locations 16 . Examples are given by You and Wang 16 and Lara and Grossmann 17 , who describe an optimal combination of centralized and distributed production by simultaneous supply chain and capacity analysis. This illustrates the fact that an integrated process and supply chain design already at early-stage is required to screen for the most viable solutions. 10,16 Protability can be further increased by selling value-added chemicals in a multi-product biorenery. For this purpose,an early-stage product-portfolio selection is important, which depends on attainable prices and corresponding market sizes. 18,19 Thus, a profound model covering the biomass supply chain, processing and product portfolio as well as market developments is required.

The identication of an optimal process design and thereby product portfolio is challenging due to a high number of possible reactions in the context of bioreneries. 9,20,21 For these variety of conversions, which are often proven at laboratory scale only, a detailed process design and benchmarking is infeasible such that early-stage screening approaches are required, based on proof-of-concept laboratory data. 22 To that extend, a number of superstructurebased approaches has been proposed in literature, which address the identication of optimal processing concepts. 2328 These models can be further rened considering heat 8,29,30 and water integration 31 , additionally. However, all of these models rely on process design data extracted from literature or require a high number of process simulations with common preparation times of more than a year. 8 Furthermore, the available data is often restricted to a specic process or unit operation and specic feed compositions. The dierent operat3



ing conditions thus render the analysis of a number of integrated unit operations dicult. By a systematic ontology-based data storage including consistency checks and a subsequent automated network formulation and solution, the Super-O implementation proposed by 32 ensures feasible connections between individual processing steps as well as between supply chain constraints and processing pathways or pathways and product selection. Thereby the eort to setup the required superstructures is reduced and available data is eciently stored for future re-uses. However, only existing processing concepts are taking into account. In order to further accelerate process development and also consider novel processes, Reaction Network Flux Analyis (RNFA) and PNFA have recently been introduced as early-stage design methods for the identication of optimal reaction 33 or processing pathways. 22 While the RNFA considers reaction pathways only assuming ideal separations, 33 the PNFA systematically identies separation tasks and estimates the energy and cost thereof. Thus, the PNFA is capable of evaluating full processing pathways based on laboratory data to allow for a rst analysis and insight into novel biorenery processing pathways. The result is a quantitative evaluation of process eciency and an identication of possible process bottlenecks. While it has already been demonstrated for a complex case study based on a large processing network, 22 neither the biomass supply itself nor market considerations have been part of the design approach yet.

In order to analyze the eect of biomass transportation on the processes viability, the PNFA needs to be extended to cover the biomass supply chain. For this purpose, a profound list of supply chain models already exist in literature (cf. Shah 34 , Papageorgiou 35 , Garcia and You 36 for general reviews and Cambero and Sowlati 12 as well as Ghaderi et al. 37 for forest biomass supply chain) varying in the number of analyzed aspects and thus in their complexity. The availability of dierent biomass compositions 16,38 and their seasonality 15,16,39,40 highly inuences the determination of an optimal supply chain. However, detailed biomass availability and distribution data is scarce. The biomass transportation includes dierent 4


Page 4 of 41

Page 5 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


transportation modes (rail, ship, truck) 41 and biomass storage. 15,4244 The identication of optimal plant locations 16,39,42,43 depends on the decision regarding distributed or centralized production or a combination thereof to obtain more exibility. 16,17,38 Since data is scarce and scope of this paper is to integrate the biomass supply chain into the PNFA to analyze process viability, a pragmatic supply chain is chosen in a rst step considering biomass availability and transportation to identify optimal plant locations for a distributed production. Similarly to existing work on supply chains, an economic 38,39,41,4548 and environmental 16 objective is utilized in the optimization. Thus, the eect of social aspects as proposed by You et al. 49 and Cambero and Sowlati 12 is not included. Finally, the analysis is complemented by a sensitivity analysis to address the eect of uncertainties as proposed by Kim et al. 48 ,Sharifzadeh et al. 38 and Gao and You 50 . All of the aforementioned models are connected either to a biochemical ethanol production 15,44,45,49,51 , a thermochemical conversion plant 16,38,46,47 or a combination thereof. 39 Thus, the number of products and hence processing alternatives is limited. Furthermore, the products are already pre-selected. In this work, supply chain design is only utilized to determine biomass transportation cost and global warming. The number of product and processing alternatives is enhanced and a multi-product biorenery with product-portfolio selection is targeted.

A product-portfolio determination requires market analysis on obtainable prices. Furthermore, a biorenery is limited by the economically accessible biomass feed. Thus, price and market uncertainties need to be considered for both the biomass feedstock as well as for the nal products 3,8,5254 , which are subject to supply and demand principles. Thorough market models and resilient forecasts for bio-based chemicals are not available in the open literature and publication is often restricted from companies. 53 Although economic forecasts are generally dicult, they are particularly challenging in the eld of biorenery markets, which have not been established. Therefore, market models are often simplied and can only be used for a rst insight. For this purpose, sensitivity analyses are the most simple 5



approach to consider price uncertainties analyzing the eect of uctuating prices on the product selection. 47,55 Alternative approaches derive and analyze potential future scenarios to account for market uncertainties 56 , correlate prices of bio-products to an oil-price model or adjust them based on fossil replacement or substitution products. 57,58 This is restricted to fossil replacement products such as biopolyethylene or substitution products exhibiting the same functionality as the fossil counterpart. Furthermore, these scenarios or oil-price correlation models come along with a high degree of nonlinearities as well as a high number of assumptions, like estimates of the replaced fossil processes or feedstock margins. 57 These models omit the inuence of the market size on attainable product prices. Preventing an excess production, a constraint might be added guaranteeing a certain customer demand 38 or restricting the maximum product demand. 47,59 While an excess production is prohibited thereby, the dependency of prices on the market exploitation is not fully depicted. Existing biomass market models by Sorda and Madlener 18 and Voll 19 take these dependencies into account but rely on a high number of uncertain parameters, like price elasticities, which vary between dierent biomass types by more than one order of magnitude and are highly specic for the analyzed countries. 60 Due to large market uncertainties in the context of bioreneries and a signicant increase in nonlinearity, complex market models are not suitable for an early-stage design. Instead, a pragmatic approach is targeted, which considers simplied price - market dependencies.

In the current article, a twofold extension of the PNFA methodology is proposed to consider feedstock availability and transportation as well as the impact of price variations. Therefore, a biomass supply chain design is introduced as additional sub-model in the PNFA tool, in order to perform a simultaneous screening of a larger number of existing and novel pathways, while taking into account the biomass availability, transportation as well as the identication of an optimal plant location. Furthermore, a simple, but meaningful early-stage supply and demand model is added to consider the biomass and product-price development 6


Page 6 of 41

Page 7 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


depending on the specic market exploitation. With these extensions, PNFA can be used for the identication of an optimal plant location depending on local biomass availability, processing concept, and product-portfolio selection. The following Section 2 provides at rst a brief review of the PNFA methodology, before introducing the model extensions for the biomass supply chain in Section 2.2 and for the market developments in Section 2.3. Finally, Section 2.4 presents a description of the overall optimization problem formulation. The applicability of the extended model is illustrated in Section 3 for a biorenery converting lignocellulosic biomass, which has previously been investigated with the non-extended version of PNFA. 22 Optimal concepts for single (cf. Section 3.1) and multi-product bioreneries (cf. Section 3.2) are determined.

2. Methodology

PNFA is an optimization-based screening method for early process design by consideration of the economic performance and sustainability of a large number of reaction and processing pathways. The PNFA represents an extension of the Reaction Network Flux Analysis (RNFA) 61 , which is a screening method based on reaction mole balances, providing a rst insight into the performance of biorenery pathways. While the RNFA considers only reactions, which implicitly assumes feasible and ideal separations, PNFA covers additional aspects like the inuence of reaction solvents, as well as the feasibility and choice of separations. The PNFA implementation is based on a network consisting of components (nodes) and reaction or processing steps (arcs), hence connecting reagents, like dierent types of biomass, to platform chemicals and nally products. The major objective of PNFA is the identication of optimal processing pathways as well as the detection of bottlenecks in the very early-stage of process design, when only little information is available.

By coupling PNFA with supply chain design and market modeling, a full picture from

7



biomass transportation to process design and product-portfolio selection is established, which is illustrated in Figure 1. In the left part of the gure the biomass supply chain is visualized connecting the dierent harvest sites to the various production sites by biomass transportation. At each production site, dierent feedstocks types might be processed at a certain tonnage. Converting the lignocellulosic biomass via its constituents to platform chemicals and nal products, the best performing pathways need to be identied. The right part of the gure presents the product-portfolio selection, which depends on the applied objective functions such as highest protability or lowest environmental impact. A more detailed review of the PNFA is subsequently provided, while the extensions for supply chain design and market and price models are described afterwards.

Figure 1: Model scheme for biomass supply chain a), which is coupled to process network analysis b) and product-portfolio selection c). Herein, biomass transportation uxes connect the various biomass harvesting to the potential plant sites. The resulting biomass supply chain is specic for each production site. For every site, a biomass conversion concept is determined as result of an optimal process network ux analysis b). For a multi-product biorenery, the process network is further constrained by the determination of an optimal product-portfolio selection c).

2.1. Review Process Network Flux Analysis Prior to the application of the PNFA tool, a reaction network is set up containing all conceivable reaction pathways from biomass and its constituents to the desired products. Necessary separations between the reactions need to be addressed systematically. For this purpose, the 8


Page 8 of 41

Page 9 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


approach of thermodynamic insights according to Jaksland et al. 62 is utilized, which uses property ratios for a driving force determination, thereby identifying potential separations. Finally, the separation tasks are accessed using thermodynamically sound separation models. 6365 The energy or solvent demand required for the separations is then introduced into the PNFA model and the reaction network is extended towards a processing network containing all the dierent alternatives for the reactions and separations. Thus, PNFA model uses a linear ux balance:



A · f = b, A = A A A A

1

2

3

4

,

f

    =   

f f f f

 1 2 3

    ,   

(1)

4

with the stoichiometric matrix A, the ux vector f and the product vector b. The submatrices

A

1

and

A

2

are dened identically to the RNFA as raw material supply matrix

A ) and reaction matrix (A ), while additional sub-matrices are introduced to account for mixing (A ) and separation (A ). Balancing the process uxes, the pathways can then be

(

1

2

3

4

evaluated for instance based on their economic eciency and sustainability in a bi-objective optimization. Herein, the economic eciency is expressed as total annualized cost (TAC). The TAC are dened as

T AC =

Ns X j=1

fs,j Ms Ps +

Nw X i=1

Nutility

bw,i Mi Pw +

X

Eutility,k Putility +

k=1

ir IC, 1 − (1 + ir)−n

(2)

consisting of the raw material, waste disposal and utility cost as well as the annualized investment cost. The raw material and waste disposal cost are deduced from the ux balance, the respective molar masses M and prices P for supply s and waste w. The investment costs IC are derived based on an empirical method of El-Halwagi 66 and are annualized for a plant lifetime n and an interest rate ir. Herein, the heuristic step-counting approach by

9



Page 10 of 41

El-Halwagi 66 relies on the number of processing steps and the plant's capacity. For this purpose, the number of active processing steps is counted. For more details regarding the IC determination, refer to Ulonska et al. 22 . The utility cost are derived based on utility prices and the specic energy duties determined using aforementioned thermodynamic sound separation models. These models are capable of considering non-idealities like azeotropes, which often occur in the context of bioreneries. Bausa et al. 63 and Skiborowski et al. 67 report only small deviations of the energy demand to tedious rigorous calculations of 1-5% even for complex separations. Therefore, these models are sucient for a rst pathway screening and can also be used to initialize rigorous calculations subsequently. 67,68 Further information on the inclusion of the utility demand into the PNFA is given in Section 2 of the supplementary information.

While minimizing the TAC leads to economically ecient processes, the sustainability of these pathways is addressed considering the cumulative energy demand CEDprocess ,

CEDprocess =

eheat Eheat Nproducts P

+ eelec Eelec

.

(3)

bproduct,i ∆Hcomb,product,i

i=1

In addition, a direct assessment of the global warming potential, GW Pprocess is feasible,

GW Pprocess =

gwpheat Eheat Nproducts P

+ gwpelec Eelec

,

(4)

bproduct,i ∆Hcomb,product

i=1

enabling a comparison of a global warming reduction potential compared to conventional fuels. Herein, the overall energy duties E for electricity (elec) and steam (heat) are summed up using the respective primary energy e or global warming factors gwp. Furthermore, a normalization on the heating value ∆Hcomb,product of the product uxes bproduct is carried out, such that both former equations are bilinear. A biomass combustion for excess energy production is prohibited as the overall goal is the production of transportation fuels and 10


Page 11 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


chemicals, such that (5)

CEDprocess ≥ 0.

2.2 Supply chain design The PNFA methodology is extended to cover the biomass supply chain with the scope of analyzing the eect of transportation on the process viability. Since the case study presented later on covers a comparably small region and transportation by ship or railway is associated with high non-distance-related cost 41 and limited accessibility, truck transportation only is considered. The method can be easily extended to identify dierences between distributed and centralized production as described by Lara and Grossmann 17 and You and Wang 16 .

Figure 1 presents a scheme of the linear biomass supply chain model, in which nodes represent harvest sites as origins (index o) and production sites as destinations (index d) whereas the arcs indicate the biomass transport from the harvest site (o) to the production site (d). In a rst step, the amount of biomass collected at each harvest site as well as the total collected biomass need to be identied. The biomass transportation ux fBT,o,d from one harvest site to all destinations is limited by the maximal biomass capacity available at the origin Cbiomass,o , Nd X

fBT,o,d Mbiomass ≤ Cbiomass,o

∀ o.

(6)

d=1

The underlying assumption herein is that biomass, i.e., residual and small wood, is a homogeneous good, meaning that it has the same characteristics and quality regardless of the time and location it was delivered from. The utilized data by Dieter et al. 69 does not include information on seasonal biomass composition. Therefore, seasonality is not yet included in the method. So far, the same molar mass Mbiomass is utilized for all origins o. In a next step, the total amount of biomass collected, fS1 , equals the amount from all origins and delivered

11



to all destinations

fS1 =

Nd X No X

fBT,o,d .

Page 12 of 41

(7)

d=1 o=1

Identifying production sites, the big M formulation is used such that the binary variables

ysite,d equal one if the site d is chosen No X

fBT,o,d Mbiomass ≤ fBT,max,d · ysite,d ,

(8)

o=1

with fBT,max,d representing the maximal amount, which can be delivered to a plant. Summing up the binary variables for the production sites Nd X

ysite,d ≤ λ

(9)

d=1

the maximal allowable number of plants λ species an upper boundary of production sites. Based on the transport ux and the distance D between harvest and production site, the transportation cost can be determined. The biomass transportation cost consists of a distance-independent price PBT,f ix and a distance-dependent price PBT,var which covers the expenses for fuel and wages. 41,70 These prices are specic for each country and depend on the biomass type and water content, wH2O . In Section 4 of the supplementary information, an overview is given with literature-known price parameters for the various regions in the world and dierent biomass types. In general, residual wood causes higher variable prices compared to wood chips due to a lower packing density. Furthermore, the variable prices strongly depend on the region. In this work, the price parameters of Schwaderer 41 are utilized, which are specic for residual wood in Germany and consider a water content wH2O of 0.5. Herein, the reason to address biomass composition and water content separately is the data on biomass availability reported by Dieter et al. 69 , which is given in dry tons and neither species the biomass nor its water content. Considering forest residues specically, a water content of 50% agrees well with a typical water content of 44% proposed by Mantau 71 .

12


Page 13 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Finally, the transportation price is updated by a factor for logistics prices in Germany called "Verkehrsrundschau" (VR) index from the reference year 2011 to the recent year. The biomass transport cost (BTC) from one harvest to one production site can then be written as

BT Co,d = (PBT,f ix + PBT,var · Do,d ) ·

VR 1 · fBT,o,d Mbiomass . 1 − wH2O V R2011

(10)

The yearly total biomass transportation cost (TBTC) are obtained by summing up the BTC for all harvest and production sites

T BT C =

Nd X No X

(11)

BT CBT,o,d .

d=1 o=1

As biomass transportation contributes to the overall global warming potential, GWP is utilized as second objective function. The GWP for biomass transport (GW PBT ) is based on a diesel truck specic factor (GW Ptruck ) which correlates with the amount of dry biomass transported and the total distance

GW PBT = GW Ptruck ·

Nd X No X

(12)

fBT,o,d Mbiomass Do,d .

d=1 o=1

In literature, dierent GW Ptruck factors are given depending on the truck size, reference year, biomass type, density, and moisture content. Börjesson 72 estimate 0.073 2015, Kumar et al. 14 consider 0.095 0.18

kg CO2 t km ,

kg CO2 t km ,

kg CO2 t km

for the year

whereas Garcia and You 73 propose a value of

all without further specifying the underlying assumptions, truck size or biomass.

Based on the transport of wood chips in Germany, Kappler 74 determines a GWPtruck factor of 0.257

kg CO2 t km

for a water content of 50 wt%. Since this work aims at a supply chain

optimization in Germany as well, the GWPtruck factor of Kappler 74 is applied. The values for the biomass capacities Cbiomass,o are based on dry biomass, hence the GW Ptruck factor is divided by a factor of two (wH2O of 0.5), leading to a GW Ptruck factor of 0.129 which is applied in the following. The supply chain sub-model is fully linear.

13


kg CO2 dry t km ,


Page 14 of 41

2.3 Market and price model Considering the impact of a new biorenery process, a simplied supply and demand model is introduced to account for biomass and chemicals price variations depending on the market sizes. A simple model with a linear price dependency on the market quantity is proposed. Although the model is accompanied by uncertainties, the degree of nonlinearity as well as a need for a vast amount of market data is kept low. 75,76 Since the output of small-scale bioreneries are orders of magnitudes lower compared to the fuel demand, no additional price - market dependency is considered for fuel production. With this rst rough market assessment, an integration into early-stage process design is enabled. The price of chemicals is determined based on a current initial price Pinitial,chem , the fraction of current market volume qchem and a price scenario parameter Pchem,scenario :

Pchem = Pinitial,chem − Pchem,scenario · qchem .

(13)

In literature, market sizes and prices of chemicals are often given for a specic year. To allow a fair comparison of the dierent chemicals, the market sizes herein are calculated for a reference year using the specic component annual growth rates (CAGR) and the prices are updated applying the producer price index for industrial chemicals. A more detailed description is given in Section 1 of the supplementary information. The biomass price function is built analogously:

Pbiomass = Pinitial,biomass + Pbiomass,scenario · qbiomass .

(14)

The fraction of chemicals produced

qchem =

bchem , marketchem

14


(15)

Page 15 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


and biomass consumed by the biorenery

qbiomass =

fS1 , marketbiomass

(16)

are calculated based on the market sizes marketchem and marketbiomass , the output uxes of the chemicals bchem or the biomass supply ux fS1 , respectively. For the value-added chemicals, access to the global market is assumed, while the biomass market is restricted to the local borders of the supply chain design space, without any in- or export of biomass in the current study. By means of the introduced price functions, the plant's total annual revenues TR,

TR =

NX chem

Nf uel

Pchem bchem Mchem +

chem=1

X

Pf uel

f uel=1

bf uel Mf uel . ρf uel

(17)

can be determined based on the output ux b, the molar mass, and in case of fuel production, the density ρf uel . Due to a variable price for chemicals Pchem , this equation is bilinear. A plant's total prot, TP, is then calculated based on the dierence between total annualized revenues and cost, as well as biomass transportation cost

T P = T R − T AC − T BT C.

(18)

In order to prevent an overproduction of chemicals, the global market demand for every chemical is set as upper limit 47,59 :

bchem ≤ marketchem .

(19)

A similar constraint is given in the supply chain design for the biomass market preventing an excessive use of biomass (Eq. 6).

15



Page 16 of 41

2.4 Optimization problem Taking into account the extensions of the PNFA, an updated formulation of the optimization problems is presented. The PNFA is complemented by a biomass supply chain design under the restriction of the biomass market. To incorporate the inuence of transportation on the overall sustainability, the global warming potential is now addressed as environmental objective function combining the process energy demand and the transportation into one single metric. This leads to an extended bilinear equation for the GWP determination:

GW P = GW Pprocess +

GW PBT Nproducts P

.

(20)

bproduct,i ∆Hcomb,product

i=1

Allowing a fair comparison of single fuel and multi-product plants, GWP is normalized in both cases on the heat of combustion. Determining the viability of the concepts, the prot as dened in Equation 18 is applied as second objective function. The viability and economic eciency of a biorenery can be increased considering biofuels and value-added chemicals. These value-added products might be either inevitable side products or added value products, actively selected by parallel pathways. Although the latter results in higher IC and utility cost, it increases the biomass exploitation thus reducing the waste disposal cost. Furthermore, the protability gained by selling value-added chemicals is often signicantly higher compared to fuel production. To analyze the inuence of value-added chemicals on the economics, the optimization problem is further extended such that the price and market constraints for a multi-product biorenery are included. The overall optimization problem

16


Page 17 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


is formulated as

   −T P min f ,b ,y   GW P s.t.

    

A · f = b, yield constraint, biomass composition, prof it calculation (Eqn. 2, 10 − 11, 17 − 18), GW P calculation (Eqn. 4, 12, 20), supply chain design (Eqn. 6 − 9)

(21)

biomass price constraints (Eqn. 14 − 16) market and price constraints (Eqn. 13 − 19) Nf uels

X

bf uel,i ∆Hcomb,f uel ≥ α,

i=1

CEDprocess ≥ 0, λ = 1,

f , b ≥ 0, y ∈ {0, 1}. Compared to the original PNFA for process design, the problem complexity is increased due to the additional supply chain model, as well as the market and protability analysis. The nonlinear terms in the resulting mixed-integer nonlinear programming problem are mainly bilinearities, which are present in Eqn. (2), (3), (4), (17) and (20). While the general ux balance is linear, the nonlinearities are introduced for the computation of the objective functions, namely the economic performance in terms of the prot and the global warming potential, which is nonlinear for a multi-product analysis. These objective functions are minimized subject to yield constraints of the processing steps and the biomass composition. 17



Page 18 of 41

In addition, a heating value equivalent is set as design target α allowing a fair comparison of fuel production processes. 22 The binary variables y indicate active processing steps and are required for the IC determination (cf. Section 2 of the supplementary information), which also holds a bilinear term with a xed exponential factor. Additional binary variables are introduced for every potential plant site. While specialized solution methods are conceivably more ecient we use the general-purpose solver BARON. All required price, sustainability, and model parameters are given in Section 1 of the supplementary information. In order to address the questions of parametric uncertainties and dierentiability of pathways under uncertainty, a one at a time analysis is performed for the subsequent case study, similar to our previous work. 22

For the following case study, specic parameters are applied. A fuel design target of

α = 2.77 · 1012

kJ year

is set, which is the equivalent of an annual production of 100,000

t ethanol year .

For both, biomass and chemicals, a weak (optimistic) and a strong (pessimistic) market response scenario are analyzed. The price parameters Pscenario for the optimistic and pessimistic scenarios are determined such that the prices change by a factor of only two in the optimistic scenario and by a factor of ten in the pessimistic scenario for a full market exploitation. All calculations have been performed on an Intel(R) Core(TM) i5-6200 CPU (3.10 GHz), with the model being implemented in the modeling environment GAMS. The model consists of 1752 equations, 2871 continuous and 107 binary variables and is solved by means of BARON version 16.3 77 in GAMS 24.7 78 with a relative optimality tolerance (optcr) of 0.01. For single-product bioreneries, solvability is not an issue even though the CPU time increases, when decreasing the grid length and thus improve the resolution (cf. Supplementary Information Section 4.1. for more detail). The market model increases the model non-linearity in particularly for multi-product bioreneries compared to single-product bioreneries, which leads to extended CPU times by a factor of 75 (comparison based on ethanol fuel production). Thus, a more sucient solution strategy might be mandatory when 18


Page 19 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


introducing more value-added products to the network.

3. Case study

The following case study demonstrates the potential and applicability of the extensions towards multi-product biorenery processes under consideration of biomass supply chain management and market developments. The processing network (cf. Figure 2) is given in our previous publication. 22 Instead of TAC and CED in Ulonska et al. 22 , TP and GWP are formulated as objective function as described above.

The process network includes the production of ethanol, iso-butanol, 2-butanone, ethyllevulinate, γ -valerolactone, and 2-butanol from lignocellulosic biomass. The reaction network is shown in Figure 2 visualizing pathways starting from cellulose and hemicellulose, while lignin can be used for an internal energy supply by combustion. This leads to a lower GWP as well as lower utility and waste disposal cost, but increases IC, raw material, and transportation cost. Therefore, combustion is only activated if the processes are energy-intensive, which renders a simple GWP comparison challenging. In addition, process bottlenecks are dicult to detect as the eect of introducing residue combustion superimposes any energy intensive process step. To overcome this issue, herein we exclude combustion. Results including combustion show the same product ranking and are given in Section 6 of the supplementary information.

In the following, rst a validation of the extended model and the inuence of biomass transportation on the economics and sustainability of biorenery processes are presented. In order to focus on the process viability, a pragmatic supply chain using truck transportation only is integrated as rst step. Similar to our previous publication, 22 an energetic product usage as transportation fuels

19



Figure 2: Reaction network for the production of ethanol, iso-butanol, 2-butanone, ethyllevulinate, γ -valerolactone, and 2-butanol from lignocellulosic biomass, which is adapted from a previous work. 22 The adaptation includes the addition of formic acid as by-product of levulinic acid formation. is analyzed for this purpose. The biomass availability as well as potential plant sites in the region of North Rhine-Westphalia (NRW), Germany, are considered. Even though, ethanol is an established biofuel, an economic production from lignocellulosic residues is challenging such that viability is improved by selling value-added chemicals in addition. In the supplementary material (Section 3), the procedure on biomass availability data acquisition is described. The resulting data on biomass potential per harvest site as well as a list of potential plant locations is given. In addition, an analysis of dierent grid sizes is presented (Section 4.1), which is important as a sucient resolution is required that allows for appropriate computation times. Here, a cell length of 20 km is deemed suciently accu20


Page 20 of 41

Page 21 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


rate and computational favorable compared to a grid length of 4 and 10 km.

3.1. Single-product scenario Figure 3 shows the forest area of NRW along with ten chemical site locations, which are analyzed in the following. This selection is based on the decision criteria by Clausen et al. 79 who propose to consider the distance to federal roads and freeways, the availability of full service oers, a harbour, a rail connection, a similar production eld and research and development facilities as criteria for the selection. The optimization results show that process data as well as cost other than transportation costs are equal for all sites. The analysis identies the chemical sites in Leverkusen, Cologne and Hürth as slightly superior in location. The dierence in transportation cost is only 3% such that all three sites should be considered equally good. The results are reasonable as all three sites are located close to the largest area of biomass availability. A large discrepancy is shown for Heinsberg, which is expected, since Heinsberg is located close to the border in an area with only sparse biomass availability. The transportation cost contribute to the overall cost with 9-12%, which is low compared to 20-40% of other literature examples. 1315 Compared to a GWPprocess of 16.01 0.5-0.8

g CO2 MJ ethanol

g CO2 MJ ethanol ,

the GWPBT with values ranging between

is negligible. The analysis proves the applicability of the extended PNFA

model since (i) reasonable plant locations are identied and (ii) transportation cost are only slightly lower compared to literature ranges. Figure 4 illustrates the optimization results for all products and their processing pathways considering either a weak (a-b) or a strong (c-d) market scenario. The arrows indicate the optimization goal of low GWP and less negative prot, meaning that no single product process achieves protability. For 2-butanol, 2-butanone, and ethyllevulinate, Pareto curves are shown in the weak market scenario (cf. Figure 4 a) and b), while for the remaining three candidates these collapse into a single point. The reasons for this are versatile. For 21



Figure 3: For the region of NRW, Germany, the distribution of available forest area (grey) as well as ten potential chemical site locations (N) are shown. In addition, biomass transportation cost and GWP for an annual production of 100,000 t ethanol are determined for every plant location. iso-butanol there exists only one feasible pathway (R5-R38). For γ -valerolactone the pathway via the reactions R5 and R33 outperforms all alternatives considering a weak market response (cf. Figure 4 a) and b). Minimizing GWP or maximizing TP leads to a similar ethanol production from celluloses and hemicelluloses. For process optimization only, a TP maximization leads only to a cellulose conversion (cf. Ulonska et al. 22 ). Thus, due to additional biomass transportation and increased raw material cost an improved biomass conversion is achieved. Similarly, due to the increase in cost pressure, short pathways are preferred for ethyllevulinate and γ -valerolactone, which directly reduces the number of separation steps and hence the energy demand. This illustrates the general eect of a higher cost pressure, which leads to a higher resource conversion or reduces the number of active processing steps. A more detailed discussion of the results and a comparison to literature values is given in Section 7 of the Supporting Information.

22


Page 22 of 41

Page 23 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Compared to process optimization only (cf. Ulonska et al. 22 ), the annual cost rise by 10% for ethanol to 38% for 2-butanol considering biomass supply and price development. Main reason is the high impact of biomass transportation, while roughly one third of the cost increase is caused by a higher raw material price. However, the eect of biomass transportation on GWP is rather small with only 3% for ethanol to 7% for 2-butanol. Hence, process design is key for future GWP reductions. The utilized fraction of available biomass varies strongly between 16% for ethanol and 91% for 2-butanol. While Leverkusen is identied optimal for all Pareto optimal solutions, dierent supply radii are required for the products. A maximal distance of 67 km is sucient for ethanol, but for 2-butanol a maximal (average) distance of 127 (83) km is determined at the point of maximum prot and 193 (110) km at the point of minimum GWP. With the latter being the only exception, all other processes require distances of less than 150 km. According to Kappler 74 , Hamelinck et al. 80 and Mahmudi and Flynn 70 , this conrms an economic transportation by truck only.

Despite an increase in cost pressure resulting from biomass transportation, which enforces a higher resource conversion, the product ranking is similar to that obtained by process pathway optimization only. 22 The ranking is once more conrmed by a complementary sensitivity analysis of the main cost and GWP parameters. For the weak market scenario, an applied parameter uncertainty of ± 5% results in the same processing pathways and thus ranking. For brevity reasons, a detailed description is given in Section 5 of the supporting information.

Major dierence between the weak and strong market scenario, is the (negative) profitability of the processes, while the GWP as well as the pathways remain almost always similar. The only exception is the production of γ -valerolactone. Here, an additional pathway is activated in the strong market scenario (cf. Figure 4 c) and d), which is again caused by a higher cost pressure, requiring a higher feedstock conversion. Any novel biorenery drastically impacts the local biomass market. While the fuel output is small compared to 23



Figure 4: Pareto curves for a weak (a-b) and strong (c-d) market response on biomass market exploitation. While in the left part of the gure all results are shown, the right part presents a more detailed view for the top performing products of gures a) and c), respectively. the overall fuel market, the production of value-added chemicals strongly alters these low volume markets. Being conservative, only the strong market scenario is pursued in the following section, which presents a solution for economically attractive multi-product bioreneries.

3.2. Multi-product biorenery The viability of a biorenery can be signicantly improved by selling value-added chemicals. Therefore, an economically ecient biorenery is targeted with the constraint of producing ethanol as fuel. Herein, ethanol is chosen since it constitutes the best performing product and is established in today's fuel market. In the following, the most promising co-products available in the network are identied under consideration of biomass supply chain management and market response. Furthermore, the systematic analysis allows for a break even 24


Page 24 of 41

Page 25 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


analysis to dene a minimum quantity of required value-added co-products. In Figure 5 the Pareto curves for a co-production of fuel, i.e., ethanol and value-added chemicals, are shown. At the point of minimal GWP, only ethanol is produced such that these results are equal to those of the single product analysis. Since ethanol production has already been identied as process exhibiting the lowest GWP, any additional active step or pathway adds greenhouse gas emissions. This enables however higher (positive) protability. A co-production of iso-butanol with a mass ratio of 1.9 (ethanol:iso-butanol) only is required to break even. If 2,3-butanediol and furfurylalcohol are produced in addition, protability is maximized. For this point, the cost structure, mass and revenue distribution are shown in Figure 5 as well. The cost structure exhibits a high fraction of raw material and annualized investment costs. Compared to conventional crude oil processes, a higher investment cost contribution is determined. This results from multiple parallel processing steps. Furthermore, the economy of scale eect is lower than in conventional industry, as the positive scaling eect is oset by signicantly increasing raw material and transportation cost. Thus, only 38% of the available biomass is utilized. A further biomass exploitation is not economically attractive. In addition to higher costs, the chemical prices decrease due to a higher market occupancy. This demonstrates the full potential of the extended methodology, which is capable of determining the optimal plant capacity under current market constraints for the chemicals and biomass. In order to produce the specied amount of fuel and thus fulll the design constraint

α, ethanol is produced. Even though 45% of the mass output is associated with ethanol, it has a low contribution of 5% to the overall revenues, which is caused by a low Pf uel . The residual glucose fraction is converted into iso-butanol and 2,3-butanediol. The latter is responsible for 62% of the revenues, with only a share of 15% of the total mass output. However, a further 2,3-butanediol production would lead to a market oversaturation and therefore drastic price decline. Even though it increases the investment cost, this is the reason for a parallel iso-butanol production. A signicant larger iso-butanol market, is also 25



Figure 5: Multi-product biorenery results for a strong market response along with the cost structure, product mass and revenue distribution at the point of maximal protability. For comparison reasons, single-product results for ethanol are given as well. responsible for a more stable price. Since the ethanol design constraint is already achieved using the cellulose fraction, the hemicellulose fraction can be fully utilized for a value-added production of furfurylalcohol. In addition, the multi-product biorenery benets from the same logistics and biomass pretreatment for all products. Overall, the results systematically prove a positive protability of multi-product bioreneries based on a full model of biomass supply chain, process network and market analysis.

4. Conclusion

In this work, the PNFA methodology is extended incorporating biomass supply chain and market developments. The integration of a linear supply chain model enables the impact analysis of a biomass supply radius and transportation on the production cost and GWP besides identifying an optimal plant location. The simplied market model oers the advantage of requiring only few data such as an initial price, market size and annual growth rate of every component. A bi-objective optimization is formulated maximizing the prot

26


Page 26 of 41

Page 27 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


on the one hand and minimizing the GWP on the other hand. With these extensions, PNFA is capable of evaluating novel pathways and determining a product portfolio depending on local biomass availability and supply chain as well as global chemical market developments. Hence, a full analysis from feedstock supply to process optimization and product analysis is enabled. The applicability and potential of the extended PNFA is shown for a complex case study for the region of North-Rhine Westphalia, Germany. In this study, Leverkusen is identied as optimal plant location with small dierences compared to Cologne and Hürth. An analysis of grid lengths revealed a suciently high accuracy for a length of 20 km compared to 10 and 4 km, while simultaneously reducing the CPU time signicantly. Based on the chosen grid size, no deviations from process analysis only is determined including a transportation and market model. A sensitivity analysis proves that these product performances as well as their identied pathways are still distinguishable under uncertainty. Although the inuence of transportation on GWP is rather low, it aects the overall production cost by up to 14%. Cost contribution from biomass transportation is higher compared to conventional processes, thus a simultaneous biomass supply chain and process model is mandatory for a proper determination of bioreneries viability. Since fuel production only is not protable, multi-product bioreneries are envisaged. With the novel extensions, an optimal product portfolio is identied. A co-production of ethanol as fuel and iso-butanol, 2,3-butanediol, and furfurylalcohol as value-added chemicals reveals a large potential for protability increase. To break even, a mass ratio of 1.9 (fuel: iso-butanol) is required assuming a conservative market scenario. Hence, already a simple biorenery with two product streams can be protable such that an ecient fuel production from lignocellulosic residues is enabled.

Acknowledgement

This work was performed as part of the Cluster of Excellence "Tailor-Made Fuels from

27



Biomass", which is funded by the Excellence Initiative by the German federal and state governments to promote science and research at German universities.

Supporting Information Process and supply chain input data, elaboration on PNFA model and procedure to determine the biomass availability, analysis of grid size, sensitivity analysis, single-product results including residue combustion, comparison of results to literature values.

Nomenclature

Abbreviations CAGR component annuals growth rate CEPCI chemical engineering plant cost index MINLP mixed-integer nonlinear programm NFU number of unit operations NRW North-Rhine Westphalia PNFA Process Network Flux Analysis RNFA Reaction Network Flux Analysis HMF 2,5-hydroymethylfurfural

Symbols A stoichiometric matrix, [-] b product vector,[ kmol year ] 28


Page 28 of 41

Page 29 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


$ BTC specic biomass transportation cost,[ US year ] $ TBTC total biomass transportation cost,[ US year ]

C capacity MJ ] CED cumulative energy demand,[ kmol

CED* cumulative energy demand fuel production,[ MJ MJ ] $ P prices,[ kg ]

Pf uel prices,[ L$ ] D distance,[km] kJ ] E energy demand,[ year kJ Espec specic energy demand,[ kmol ]

e primary energy factors,[-] f ux vector,[ kmol year ] GWP global warming potential,[ gCO2 MJ ] gwp global warming potential factors,[-] kJ h enthalpy,[ kmol ]

IC investment cost,[US $] Inv1 pre-factor investment cost calculation,[-] Inv2 exponential factor investment cost calculation,[-] ir interest rate,[%] i control variable,[-] 29



j control variable,[-] k control variable,[-] kg ] M molar mass,[ kmol

N number,[-] n runtime,[years] $ TAC total annualized cost,[ US year ] $ TP total prot,[ US year ] $ TR total revenues,[ US year ]

VR logistic factor "Verkehrsrundschau",[-] w weight fraction,[-] y integer variable,[-]

Greek Letters kJ α design target,[ year ]

λ number of plant sites ζ split fraction boiler/turbine η eciency ν stoichiometric coecient

Subscripts 30


Page 30 of 41

Page 31 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


BT biomass transportation chem. chemical cool cooling water comb combustion d destinations elec electricity excess surplus energy m mixing o origins r reaction s supply t separation w waste residues

References

(1) International

Energy

Agency,

CO2

emissions

from

fuel

combustion,

http://www.iea.org/publications/freepublications/publication/CO2EmissionsfromFuel Combustion_Highlights_2016.pdf. accessed November 22, 2016. (2) Balat, M. Production of bioethanol from lignocellulosic materials via the biochemical pathway: A review. Energy Conversion and Management 2011, 52, 858875.

31



(3) Giuliano, A.; Poletto, M.; Barletta, D. Process optimization of a multi-product biorenery: The eect of biomass seasonality. Chemical Engineering Research and Design

2016, 107, 236252. (4) Marquardt, W.; Harwardt, A.; Hechinger, M.; Kraemer, K.; Viell, J.; Voll, A. The biorenewables opportunity - toward next generation process and product systems. AIChE

Journal 2010, 56, 22282235. (5) Upadhye, A. A.; Qi, W.; Huber, G. W. Conceptual process design: A systematic method to evaluate and develop renewable energy technologies. AIChE Journal 2011, 57, 2292 2301. (6) Daoutidis, P.; Marvin, W. A.; Rangarajan, S.; Torres, A. I. Engineering Biomass Conversion Processes: A Systems Perspective. AIChE Journal 2013, 59, 318. (7) Yuan, Z.; Chen, B.; Gani, R. Applications of process synthesis: Moving from conventional chemical processes towards biorenery processes. Computers & Chemical Engi-

neering 2013, 49, 217229. (8) Kokossis, A.; Tsakalova, M.; Pyrgakis, K. Design of integrated bioreneries. Computers

& Chemical Engineering 2015, 81, 4056. (9) Huber, G. W.; Iborra, S.; Corma, A. Synthesis of transportation fuels from biomass: chemistry, catalysts, and engineering. Chemical Reviews 2006, 106, 40444098. (10) Serrano-Ruiz, J. C.; Dumesic, J. A. Catalytic routes for the conversion of biomass into liquid hydrocarbon transportation fuels. Energy & Environmental Science 2011, 4, 8399. (11) Demirbas, A. Competitive liquid biofuels from biomass. Applied Energy 2011, 88, 17 28.

32


Page 32 of 41

Page 33 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(12) Cambero, C.; Sowlati, T. Assessment and optimization of forest biomass supply chains from economic, social and environmental perspectives A review of literature. Renew-

able and Sustainable Energy Reviews 2014, 36, 6273. (13) Angus-Hankin, C.; Stokes, B.; Twaddle, A. The transportation of fuelwood from forest to facility. Biomass and Bioenergy 1995, 9, 191203. (14) Kumar, A.; Sokhansanj, S.; Flynn, P. C. In Twenty-Seventh Symposium on Biotech-

nology for Fuels and Chemicals ; McMillan, J. D., Adney, W. S., Mielenz, J. R., Klasson, K. T., Eds.; Humana Press: Totowa, NJ, 2006; pp 7187. (15) Ek³io§lu, S. D.; Acharya, A.; Leightley, L. E.; Arora, S. Analyzing the design and management of biomass-to-biorenery supply chain. Computers & Industrial Engineering

2009, 57, 13421352. (16) You, F.; Wang, B. Life Cycle Optimization of Biomass-to-Liquid Supply Chains with DistributedCentralized Processing Networks. Industrial & Engineering Chemistry Re-

search 2011, 50, 1010210127. (17) Lara, C. L.; Grossmann, I. E. Global Optimization for a Continuous Location-Allocation Model for Centralized and Distributed Manufacturing. Computer Aided Chemical En-

gineering 2016, 38, 10091014. (18) Sorda, G.; Madlener, R. Cost-Eectiveness of Lignocellulose Bioreneries and their Impact on the Deciduous Wood Markets in Germany, http://www.fcn.eonerc.rwthaachen.de/go/id/dnfj/le/206264. 2012. (19) Voll, A. Model-based screening of reaction pathways for processing of biorenewables. Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, Germany, 2014. (20) Alonso, D. M.; Bond, J. Q.; Dumesic, J. A. Catalytic conversion of biomass to biofuels.

Green Chemistry 2010, 12, 1493. 33



(21) Climent, M. J.; Corma, A.; Iborra, S. Conversion of biomass platform molecules into fuel additives and liquid hydrocarbon fuels. Green Chemistry 2014, 16, 516547. (22) Ulonska, K.; Skiborowski, M.; Mitsos, A.; Viell, J. Early-stage evaluation of biorenery processing pathways using process network ux analysis. AIChE Journal 2016, 62, 30963108. (23) Bao, B.; Ng, D. K.; Tay, D. H.; Jiménez-Gutiérrez, A.; El-Halwagi, M. M. A shortcut method for the preliminary synthesis of process-technology pathways: An optimization approach and application for the conceptual design of integrated bioreneries. Comput-

ers & Chemical Engineering 2011, 35, 13741383. (24) Santibanez-Aguilar, J. E.; Gonzalez-Campos, J. B.; Ponce-Ortega, J. M.; SernaGonzalez, M.; El-Halwagi, M. M. Optimal Planning of a Biomass Conversion System Considering Economic and Environmental Aspects. Industrial & Engineering Chemistry

Research 2011, 50, 85588570. (25) Rangarajan, S.; Bhan, A.; Daoutidis, P. Identication and analysis of synthesis routes in complex catalytic reaction networks for biomass upgrading. Applied Catalysis B:

Environmental 2014, 145, 149160. (26) Cheali, P.; Gernaey, K. V.; Sin, G. Toward a computer-aided synthesis and design of biorenery networks: Data collection and management using a generic modeling approach. ACS Sustainable Chemistry & Engineering 2014, 2, 1929. (27) Kelloway, A.; Daoutidis, P. Process synthesis of bioreneries: optimization of biomass conversion to fuels and chemicals. Industrial & Engineering Chemistry Research 2014,

53, 52615273. (28) Zondervan, E.; Nawaz, M.; de Haan, A. B.; Woodley, J. M.; Gani, R. Optimal design of a multi-product biorenery system. Computers & Chemical Engineering 2011, 35, 17521766. 34


Page 34 of 41

Page 35 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(29) Niziolek, A. M.; Onel, O.; Elia, J. A.; Baliban, R. C.; Floudas, C. A. Coproduction of liquid transportation fuels and C 6 _C 8 aromatics from biomass and natural gas.

AIChE Journal 2015, 61, 831856. (30) Tock, L.; Maréchal, F. Co-production of hydrogen and electricity from lignocellulosic biomass: Process design and thermo-economic optimization. Energy 2012, 45, 339349. (31) Martín, M.; Grossmann, I. In Integrated bioreneries ; Stuart, P. R., El-Halwagi, M. M., Eds.; Green Chemistry and Chemical Engineering; CRC Press: Boca Raton, FL, 2013; pp 325348. (32) Bertran, M.-O.; Frauzem, R.; Sanchez-Arcilla, A.-S.; Zhang, L.; Woodley, J. M.; Gani, R. A generic methodology for processing route synthesis and design based on superstructure optimization. Computers & Chemical Engineering 2017, (33) Voll, A.; Marquardt, W. Reaction network ux analysis: Optimization-based evaluation of reaction pathways for biorenewables processing. AIChE Journal 2012, 58, 1788 1801. (34) Shah, N. Process industry supply chains: Advances and challenges. Computers & Chem-

ical Engineering 2005, 29, 12251235. (35) Papageorgiou, L. G. Supply chain optimisation for the process industries: Advances and opportunities. Computers & Chemical Engineering 2009, 33, 19311938. (36) Garcia, D. J.; You, F. Supply chain design and optimization: Challenges and opportunities. Computers & Chemical Engineering 2015, 81, 153170. (37) Ghaderi, H.; Pishvaee, M. S.; Moini, A. Biomass supply chain network design: An optimization-oriented review and analysis. Industrial Crops and Products 2016, 94, 9721000.

35



(38) Sharifzadeh, M.; Garcia, M. C.; Shah, N. Supply chain network design and operation: Systematic decision-making for centralized, distributed, and mobile biofuel production using mixed integer linear programming (MILP) under uncertainty. Biomass and

Bioenergy 2015, 81, 401414. (39) u£ek, L.; Martín, M.; Grossmann, I. E.; Kravanja, Z. Multi-period synthesis of optimally integrated biomass and bioenergy supply network. Computers & Chemical Engi-

neering 2014, 66, 5770. (40) Avraamidou, S.; Pistikopoulos, E. N. A Multiparametric Mixed-integer Bi-level Optimization Strategy for Supply Chain Planning Under Demand Uncertainty. IFAC-

PapersOnLine 2017, 50, 1017810183. (41) Schwaderer, F. Integrierte Standort-, Kapazitäts- und Technologieplanung von Wertschöpfungsnetzwerken zur stoichen und energetischen Biomassenutzung. Dissertation, Karlsruher Institut für Technologie, Germany, 2012. (42) Ng, R. T.; Maravelias, C. T. Design of biofuel supply chains with variable regional depot and biorenery locations. Renewable Energy 2017, 100, 90102. (43) Ng, R. T.; Maravelias, C. T. Economic and energetic analysis of biofuel supply chains.

Applied Energy 2017, 205, 15711582. (44) Ng, R. T.; Kurniawan, D.; Wang, H.; Mariska, B.; Wu, W.; Maravelias, C. T. Integrated framework for designing spatially explicit biofuel supply chains. Applied Energy 2018,

216, 116131. (45) Marvin, W. A.; Schmidt, L. D.; Daoutidis, P. Biorenery Location and Technology Selection Through Supply Chain Optimization. Industrial & Engineering Chemistry

Research 2013, 52, 31923208.

36


Page 36 of 41

Page 37 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(46) Elia, J. A.; Baliban, R. C.; Floudas, C. A.; Gurau, B.; Weingarten, M. B.; Klotz, S. D. Hardwood Biomass to Gasoline, Diesel, and Jet Fuel: 2. Supply Chain Optimization Framework for a Network of Thermochemical Reneries. Energy & Fuels 2013, 27, 43254352. (47) Kim, J.; Real, M. J.; Lee, J. H. Optimal design and global sensitivity analysis of biomass supply chain networks for biofuels under uncertainty. Computers & Chemical

Engineering 2011, 35, 17381751. (48) Kim, J.; Real, M. J.; Lee, J. H.; Whittaker, C.; Furtner, L. Design of biomass processing network for biofuel production using an MILP model. Biomass and Bioenergy

2011, 35, 853871. (49) You, F.; Tao, L.; Graziano, D. J.; Snyder, S. W. Optimal design of sustainable cellulosic biofuel supply chains: Multiobjective optimization coupled with life cycle assessment and input-output analysis. AIChE Journal 2012, 58, 11571180. (50) Gao, J.; You, F. Modeling framework and computational algorithm for hedging against uncertainty in sustainable supply chain design using functional-unit-based life cycle optimization. Computers & Chemical Engineering 2017, 107, 221236. (51) Huang, Y.; Chen, C.-W.; Fan, Y. Multistage optimization of the supply chains of biofuels. Transportation Research Part E: Logistics and Transportation Review 2010,

46, 820830. (52) Black, M.; Sadhukhan, J.; Day, K.; Drage, G.; Murphy, R. Developing database criteria for the assessment of biomass supply chains for biorenery development. Chemical

Engineering Research and Design 2016, 107, 253262. (53) Mansoornejad, B.; Pistikopoulos, E. N.; Stuart, P. Metrics for evaluating the forest biorenery supply chain performance. Computers & Chemical Engineering 2013, 54, 125139. 37



(54) Shabani, N.; Akhtari, S.; Sowlati, T. Value chain optimization of forest biomass for bioenergy production: A review. Renewable and Sustainable Energy Reviews 2013, 23, 299311. (55) Mansoornejad, B.; Chambost, V.; Stuart, P. Integrating product portfolio design and supply chain design for the forest biorenery. Computers & Chemical Engineering 2010,

34, 14971506. (56) Sepiacci, P.; Depetri, V.; Manca, D. A systematic approach to the optimal design of chemical plants with waste reduction and market uncertainty. Computers & Chemical

Engineering 2016, (57) Sharma, P.; Romagnoli, J.; Vlosky, R. Options analysis for long-term capacity design and operation of a lignocellulosic biomass renery. Computers & Chemical Engineering

2013, 58, 178202. (58) Geraili, A.; Romagnoli, J. A. A multiobjective optimization framework for design of integrated bioreneries under uncertainty. AIChE Journal 2015, 61, 32083222. (59) Cambero, C.; Sowlati, T.; Pavel, M. Economic and life cycle environmental optimization of forest-based biorenery supply chains for bioenergy and biofuel production. Chemical

Engineering Research and Design 2016, 107, 218235. (60) Simangunsong, B. C. H.; Buongiorno, J. International Demand Equations for Forest Products: A Comparison of Methods. Scandinavian Journal of Forest Research 2001,

16, 155172. (61) Voll, A.; Marquardt, W. Benchmarking of next-generation biofuels from a process perspective. Biofuels Bioproducts & Biorening 2012, 6, 292301. (62) Jaksland, C. A.; Gani, R.; Lien, K. M. Separation process design and synthesis based on thermodynamic insights. Chemical Engineering Science 1995, 50, 511530. 38


Page 38 of 41

Page 39 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(63) Bausa, J.; Watzdorf, R. v.; Marquardt, W. Shortcut methods for nonideal multicomponent distillation: I. Simple columns. AIChE Journal 1998, 44, 21812198. (64) Skiborowski, M.; Harwardt, A.; Marquardt, W. Conceptual Design of Distillation-Based Hybrid Separation Processes. Annual Review of Chemical and Biomolecular Engineer-

ing 2013, 4, 4568. (65) Redepenning, C.; Marquardt, W. Pinch-based shortcut method for the conceptual design of adiabatic absorption columns. AIChE Journal 2017, 63, 12131225. (66) El-Halwagi, M. M., Ed. Sustainable Design Through Process Integration: Fundamen-

tals and Applications to Industrial Pollution Prevention, Resource Conservation, and Protability Enhancement, 1st ed.; Butterworth-Heinemann: Boston, MA, 2012. (67) Skiborowski, M.; Harwardt, A.; Marquardt, W. Ecient optimization-based design for the separation of heterogeneous azeotropic mixtures. Computers & Chemical Engineer-

ing 2015, 72, 3451. (68) Recker, S.; Skiborowski, M.; Redepenning, C.; Marquardt, W. A unifying framework for optimization-based design of integrated reactionseparation processes. Computers

& Chemical Engineering 2015, 81, 260271. (69) Dieter, M.;

Englert, H.;

Klein, M. Abschätzung des Rohholzpotentials für

dieenergetische Nutzung in der Bundesrepublik Deutschland, Release date 2001, http://literatur.thuenen.de/digbib_extern/bitv/dk040192.pdf. accessed May 05,2016. (70) Mahmudi, H.; Flynn, P. C. Rail vs Truck Transport of Biomass. Applied Biochemistry

and Biotechnology 2006, 129-132, 88103. (71) Mantau, ien

des

U.

Holzrohstobilanz

Holzaufkommens

und

Deutschland der

39

-

Entwicklungen

Holzverwendung


von

und

1987

bis

Szenar2015,


http://www.saegeindustrie.de/tmp_uploads/00_holzrohstobilanz_2012.pdf:

Page 40 of 41

Tech-

nical Report. accessed July 6, 2016. (72) Börjesson, P. Energy analysis of biomass production and transportation. Biomass and

Bioenergy 1996, 11, 305318. (73) Garcia, D. J.; You, F. Multiobjective optimization of product and process networks: General modeling framework, ecient global optimization algorithm, and case studies on bioconversion. AIChE Journal 2015, 61, 530554. (74) Kappler, G. O. Systemanalytische Untersuchung zum Aufkommen und zur Bereitstellung von energetisch nutzbarem Reststroh und Waldrestholz in Baden-Württemberg. Dissertation, Albert-Ludwigs-Universität Freiburg, Germany, 2008. (75) Mansoornejad, B.; Stuart, P.; Pistikopolous, E. N. Incorporating Flexibility Design into Supply Chain Design for Forest Biorenery. Journal of Science & Technology for Forest

Products and Processes 2011, 1, 5466. (76) Malueg, D. A. Monopoly Output and Welfare: The Role of Curvature of the Demand Function. The Journal of Economic Education 1994, 25, 235250. (77) Tawarmalani, M.; Sahinidis, N. V. A polyhedral branch-and-cut approach to global optimization. Mathematical Programming 2005, 103, 225249. (78) GAMS Development Corporation, GAMS − General Algebraic Modeling System, http://www.gams.com. Accessed August, 10, 2012. (79) Clausen, U.; Langkau, S.; Goedicke, I.; Lier, S. Standort- und Netzwerkplanung für modulare Containeranlagen in der Prozessindustrie. Chemie Ingenieur Technik 2015,

87, 13651372. (80) Hamelinck, C. N.; Suurs, R. A.; Faaij, A. P. International bioenergy transport costs and energy balance. Biomass and Bioenergy 2005, 29, 114134. 40


Page 41 of 41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6: For Table of Content Only.

41


Optimization of multi-product biorefinery processes under

Recommend Documents