Integrated Waste Management in Multiproduct ... - ACS Publications

Subscriber access provided by La Trobe University Library

Article

#ntegrated waste management in multi-product bio-refineries: systems optimization and analysis of a real-life industrial plant. Aikaterini Mountraki, Marinella Tsakalova, Anna Panteli, Aikaterini Papoutsi, and Antonis Constantinos Kokossis Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.5b03431 • Publication Date (Web): 01 Mar 2016 Downloaded from http://pubs.acs.org on March 2, 2016

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 free 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 accessible to all readers and 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.

Industrial & Engineering Chemistry Research 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 43

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

Integrated waste management in multi-product bio-refineries: systems optimization and analysis of a real-life industrial plant

A. MOUNTRAKI1, M. TSAKALOVA1, A. PANTELI1, A.I. PAPOUTSI1 and A.C. KOKOSSIS1,* 1

School of Chemical Engineering, National Technical University, Athens, Greece. *Corresponding author: [email protected], +30-2107724275

Keywords: Biorefinery, systems engineering, waste management, process integration, optimization

Abstract

The paper presents a methodology for the integrated treatment of biorefinery effluents using a systems engineering approach. The methodology uses generic bipartite graphs to integrate biorefinery units and to compose superstructures. The graph representation accounts for biorefinery processes, waste treatment technologies, raw materials, intermediates, and products. Graphs are applied in conjunction with allocation maps that link treatment technologies with biorefinery liquid, solid, and gas streams. The superstructures integrate process technologies, include options for central and distributed treatment, and are applied both in grassroots and retrofit applications. The mathematical optimization requires the solution of mixed-integer nonlinear programming models, and the methodology is illustrated by using a real-life lignocellulosic biorefinery featuring 49 streams and 22 treatment technologies (6 for liquids, 4 for solids, 7 for gas pollutants, 2 for water reuse, and 2 for catalyst regeneration). When first generation plants are retrofitted into second generation biorefineries, systems integration proves capable to identify cost-effective alternatives that restrict cost. Results generally demonstrate that integration is exceptionally important, often leading to significant savings and cost reductions, even able to turn treatment costs into profits.

ACS Paragon Plus Environment


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

1- INTRODUCTION Strong bio-based economies are expected to create, both directly and indirectly, significant revenues and jobs. They are also expected to increase farmers’ income and to improve economic activity in developing rural regions. Βiorefinery is the most powerful concept towards a Bio-based Economy1 and is founded by the innovative and cost-efficient use of biomass for the production of food, feed, energy, and chemicals. Lignocellulosic biorefineries process the most abundant type of biomass through intermediates, which include cellulose, hemicelluloses, and lignin.2,3 Biorefineries account for the most efficient valorization of biomass, but, in order to remain competitive in the chemical industry, they should build efficiency with process integration. The latter relates to the efficient use of energy4-6 water7-11, and process flowsheeting12,13. Integrated design of waste treatment has been largely underestimated even though the biorefinery waste streams can be numerous and different in nature. The integrated technology could not only save cost but also produce profit. Industrial waste is distinguished to gas, liquid, and solids with each class regulated by separate and different terms. Waste treatment ensures discharge under environmental regulation standards, but there is a multitude of treatment technologies to use.14,15 First, solid waste treatment can be subdivided to biological and thermal treatment. Thermal processing can be combined with energy recovery. The majority of the thermal treatment refers to incineration with some limited use of pyrolysis and/or gasification.16 Second, air pollution control methods largely depend on the pollutants: choices differ between suspended particles, volatile organic compounds (VOCs), acid gas (SO2), or CO2 emissions.17,18 Finally, wastewater technologies are classified as chemical, physical, and biological. The main benefit of biological treatment is the recovery of stabilized organic matter and nutrients. Physical pretreatment requires less energy but is less effective in pollutant reduction. On the other hand, biological treatment demands higher energy but yields much higher reductions, so chemical processes stand as intermediate choices. Wastewater treatment is laid out across primary, secondary, and tertiary systems that are benefitted by combining different technologies. Treatment of wastewater includes the removal of specific contaminants as well as the removal and control of nutrients.14,19 The abundance of technologies complicates the analysis whereas their complementary merits highlight the potential to compound benefits by integrating each other. Instead, common practice applies intuition and is pre-occupied on popular choices. The paper introduces an approach that copes with the challenge and enables a high-throughput evaluation of options in large-scale applications.


Page 2 of 43

Page 3 of 43

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


Systems technology has already proved its potential to address waste treatment. The Pinch Technology offers the most popular branch of methods in Oil & Gas.4,8,20,21 Model-based methods, which use computer-aided synthesis to identify tasks by examining waste properties against sets of environmental targets, have been applied to pharmaceutical industries.25 Databases offer access to different options, like nonlinear models to predict residues and costs, or superstructure models to evaluate the better options. Extensions of the work can address uncertainties26 and retrofitting problems.27 A separate class of technologies includes systems based on ontology models, developed to account for treatment options and embedding rules to support decisions.28,29 In the case of biorefineries, it is important to holistically address both, the waste streams as well as the waste treatment technologies. The proposed methodology highlights a systematic and generic approach. The potential of the work is illustrated with a real-life lignocellulosic biorefinery consisted by 11 valorization paths. Waste streams have been first mapped to treatment options. Next, the proposed mapping is structured as a synthesis problem that is formulated and optimized in the form of an MINLP. The optimization selects technologies, and the optimal solution determines the technology to use.

2- INDUSTRIAL CASE The work addresses the holistic valorization and the integrated treatment of by-products (biowaste) in multi-product biorefinery plants. The methodology can be scaled up to industrial problems, and its application is illustrated by using a real-life biorefinery30, which valorizes the biomass feedstock through the organosolv technology as developed by the French company CIMV. CIMV ProcessTM allows the separation without degradation of all the constituents of the vegetable matter with exceptional valorization of cellulose, lignins, and xylose under mild reaction conditions by a cost-effective process in which solvents are recovered and recycled at the end of the process.31 These three main products (cellulose, hemi-cellulose, and lignin) can be converted into fuels and chemicals through a number of valorization pathways, as shown in Figure 1.

FIGURE 1 Figure 1: Perimeters of environmental assessment.



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

Other than the core pretreatment process (organosolv) and the hydrolysis of sugars (cellulose to glucose and hemi-cellulose to xylose), the biorefinery includes eleven valorization paths. Hydrolyzed hemicelluloses (xylose) can be converted to xylitol either by biological or catalytic process. Hydrolyzed cellulose (glucose) can be converted to itaconic acid. Lignin can be converted either to Poly-Urethanes or Phenol Formaldehyde resins. Ethanol can be produced from celluloses, hemicelluloses, or from a combination of those two. Ethanol can be further converted to ethylene and PVC. These processes produce 49 waste streams which need to be classified and properly treated. Using the biorefinery process described above, the work considers portfolios that include: a) xylitol, cellulosic ethanol, and lignin-based poly-urethanes (Porfolio I), and b) xylitol, cellulosic PVC, and lignin based polyurethanes (Portfolio II) The plant operates at a fixed capacity as dictated by the availability of biomass in the area of plant construction.32 Higher capacities are not possible whereas lower capacities will be less economical. The waste streams produced are first classified and mapped onto lists of eligible technologies. The treatment technologies cover all the needs of the selected plant. Depending on their water content, waste streams are classified as liquids (higher than 50% water content) or solids (less than 50% H2O). Streams with water content close to 50% can be treated as liquids, as well as solids, and, in this paper, they are mentioned as mixed streams. Liquid streams, with water content higher than 99.5%, are considered recyclable, but if water content is between 97-99.5%, they can be reused after treatment. If the effluent streams contain substantive quantities of catalysts, the catalyst is recovered and regenerated. The concentration threshold to regeneration apparently depends on the catalyst type and its formulation. A more detailed classification of the liquid streams depends on their C/N content, their Lower Heating Value (LHV), and their BOD concentration. Similarly, gaseous emissions are classified as waste streams if they are rich in carbon dioxide. Those lower than 4% carbon dioxide content may be released into the environment. The effluent gas streams are rich in carbon dioxide and can be considered "green" as a product of fermentation processes. In this paper, the scenarios of CO2 capture examine the commercial applications of utilizing this product. The proposed classification resulted in 10 gaseous effluents, 4 solid streams, 1 mixed, 14 liquids; 2 streams are driven for regeneration, 1 is reusable, and 17 are recyclables. Sections 3, 4, and 5 illustrate the methodology as applied to the industrial problem described.


Page 4 of 43

Page 5 of 43

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


3- THE SYSTEMS APPROACH A systems approach is presented by means of a general methodology that is applicable to the general problem. The problem is typical to biorefineries where one has to match several waste streams with several treatment technologies. Other than selecting isolated treatment technologies, one is faced with options to either deploy a particular technology on several streams or, on the contrary, treat a single stream using several technologies. The biorefinery capacity and/or its value chain paths are often degrees of freedom, thus, a systems approach holds the potential to assess the extent to which the biorefinery is able to ‘integrate’ its production. The problem can be stated as one where given is a set of •

Biorefinery processes,

•

waste streams and waste treatment technologies,

•

engineering economics and costing models for each technology, and

•

design specifications, including environmental regulations and constraints.

Then the systems approach is required to determine •

treatment technologies appropriate to deploy

•

whether each waste stream should be treated by single or multiple technologies,

•

whether waste streams should be treated jointly or separately.

The paper advocates a model-based approach that capitalizes on a bipartite graphical representation (BBR) as presented by Kokossis et al..33 The BBR is used to access and evaluate the biorefinery value chain. The approach is capable to review and integrate available product portfolios searching exhaustively and systematically the options. The networks produced by the BBR can be translated into a superstructure of synthesis units. In Section 4 the synthesis units are described in detail alongside the allocation maps that define their possible connectivities. The network produced by the BBR yields a superstructure that is modeled as an optimization problem; conceptual models are used to match the level of granularity dictated by the problem data. The purpose of the study has been to point the major trade-offs in the design and to assess the scope for investment. The conceptual models are short-cut models that include mass and energy balances, costing, and the economics are presented in Section 5. The models are developed using simulation and regressing design and sizing parameters. The problem is formulated and solved as a



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

mathematical programming problem. The optimization includes integer and continuous variables and is formulated as a mixed-integer nonlinear programming problem (MINLP). Binaries are assigned to the type of treatment technologies and to type of model that is appropriate to use as different models are valid as the conditions vary. The mathematical formulation is presented in Section 6; the application of the approach to a real-life problem is presented in Section 7. Alternative treatment models can be used to replace the ones selected by the current approach. By a similar token, the types of technologies are possible to introduce. The paper employs a set of treatment technologies, which is different for liquid, gas, and solid streams. In the case of liquid effluents, treatment options include anaerobic digestion (AD), activated sludge (AS), trickling filter (TF), rotating biological contactors (RBC), aerated lagoon (AL), and stabilization ponds (SP). In the case of gaseous emissions, the treatments studied are calcinations-loop (Ca-loop), membrane reactor (MR), cryogenic methods (CR), chemical absorption, with either MEA or MDEA solvent, and physical absorption with either rectisol (RCT) or selexol (SLX). Solid treatments include combustion (COM), incineration (INC), gasification (GSF), torrefaction (TOR), and slow (SPY) and fast pyrolysis (FPY) technologies.

4. SYNTHESIS MODELS Synthesis blocks are based on the bipartite biomass graph representation (BBR) introduced by Kokossis and co-workers.33 The BBR has been introduced as a weighted graph that takes the form: , , , ) = ( Where •

accounts for graph places (graph nodes) and includes raw materials, intermediates, and products.

•

accounts for transitions (graph bars) and consists of technologies/processing units

•

accounts for arcs (connections) and stands for weights that may be assigned to binary variables in the formulation


Page 6 of 43

Page 7 of 43

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


(graph places) is extended to include additional lists of For the purpose of waste treatment,

end-products, such as biogas and bioenergy. (transitions and graph bars) are similarly extended to account for treatment technologies and their preceding mixers. The set

(connections) requires the use of direct and indirect allocation. In direct allocations the connections follow the single rules of assignment introduced by Kokossis and co-workers.33 However, the allocation of waste streams to technologies requires the use of specific properties of the waste streams (e.g. content, composition, flows etc.). Figure 2 (left) explains the graph places (biorefinery processes, treatment processes) and their connections (arcs). Graph properties dictate connections, and they integrate to networks. A set of basic connections relates the raw materials, the intermediates, and the products. In other words, raw materials/intermediates are feeding/produced by processing technologies; connections process outputs to process inputs (simple allocation). Let us consider 2 biorefinery processes ( - ), 3 mixers (m1, m2, m3), 3 treatment processes ( - - ), and 2 co-products (p1, p2). The network of Figure 2 (right) connects the waste streams flowing out of the biorefinery processes as feed to the treatment technologies. The stream allocation (essentially the extension of ) constitutes the indirect allocation that is noted earlier and which is discussed in detail at the end of the section. FIGURE 2 Figure 2: A BBR graph representation

As explained in Kokossis and co-workers,33 BBRs do not correspond to the complete bipartite graphs (all-to-all connections) that are common to the majority of superstructure-based methods of the systems literature. Instead, the BBR holds all the properties of a place/transition net (Petri net),50 offering a notation that includes choice, iteration, and concurrent execution. Still, the BBR is equivalent to a superstructure by translating arcs into flows. States include nodes with single/multiple outlets (raw materials), single/ multiple inlets (products), or combined single and outlet nodes (intermediates). Hence, the BBR can set the basis of a conventional superstructure. Figure 3 is accordingly the superstructure representation that corresponds to the BBR network of Figure 2 (right hand). Figure 4 represents a real-life treatment network following the connections of Figure 3. FIGURE 3



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 3: Superstructure corresponding to Figure 2 (right hand)

FIGURE 4 Figure 4: Treatment configuration By activating and deactivating appropriate streams and units in the network, different scenarios of decentralized (Figure 5a) and/or centralized (Figure 5b) configurations can be produced. As a result, the proposed approach is, among others, capable to evaluate the potential in deviating from centralized treatment to a combined use of technologies at different places in the plant. Whereas different technologies are assigned to different streams, in order to curtail the complexity of the treatment network, the work assumes that each waste stream is treated by a single, not multiple, technology.

FIGURE 5 Figure 5: Decentralized (a) and centralized (b) liquid manufacture processes Indirect allocation Streams are allocated to technologies following established literature27-29 and a basic classification of streams between liquids, solids, and gases. Different allocation rules are applied for each case. (a) Liquids - Allocation related to capacity, the organic content (expressed as BOD), and the nitrogen content. The allocation map is sketched in Figure 6. Reuse technologies are restricted by the amount of suspended solids (TSS) that they hold. Regeneration depends on the substance regenerated, following Figure 7.

FIGURE 6 Figure 6: Allocation maps – Liquids (a)

FIGURE 7 Figure 7: Allocation maps – Liquids Reuse/ Regeneration (b)


Page 8 of 43

Page 9 of 43

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


(b) Solids - The allocation of solid streams follows the diagram of Figure 8. Streams are candidates as fuels, provided that they contain H2O less than 40%; otherwise, they are desiccated. If nitrogen, cells, sugars, or yeasts are present, then, the streams can be used for animal feed or fertilizer.

FIGURE 8 Figure 8: Allocation maps – Solids.

(c) Gas emissions - The allocation diagram for gas emissions is shown on Figure 9.

FIGURE 9 Figure 9: Allocation maps – Gas emissions.

Allocation of streams: biorefinery example The biorefinery includes 54 streams, 14 production processes, and 12 products (cellulosic pulp, glucose, lignin powder, C5 sugar syrup, xylose, itaconic acid, xylitol, PF resin, polyurethanes, ethylene, PVC, ethanol). Table 1 summarizes the results of the allocation maps, as applied to this problem. Treatment technologies are allocated to processes following the algorithms presented on Table 2. The co-products consist of biogas, electric and/or thermal energy, sludge, cleaned water, compressed CO2, and the regenerated catalyst.

Table 1: Waste streams from each production processes TABLE 1

Table 2: Treatment technologies options TABLE 2 Adopting the BBR representation, the total network along with the waste streams and the treatment technologies are shown in Figure 10. Based on the graph representation the superstructure produced is illustrated on Figure 11.



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 10 Figure 10: BBR graph of multiproduct biorefinery FIGURE 11 Figure 11: Superstructure of multiproduct biorefinery

5- WASTE TREATMENT MODELS For each waste treatment process j, models are developed to relate inlet flows and conditions with product outflows (Pj), energy use (Ej), solvents’ requirements (Rsj), capital (Cjf), and operating costs (Cjop). Different treatment technologies are studied separately for gas, liquid and solid waste. The treatment models made extensive use of literature34,35 in order to tune process and cost parameters at accuracy levels compatible with those in the plant data. All model parameters are presented in Appendix I, II, and III, in the Supporting Material section. Nonlinearities are accounted with the use of polynomial expressions as discussed in the following section. The regression settled with deviations less than 15% (e.g. generally lower than the 15-20% of the plant data). Liquid Waste Figure 12 illustrates the presumed input-output structure of the models. The output has been modeled based on total inlet flows (Sj), total suspended solids (TSSj), and the BOD of the inlet stream (BODj). FIGURE 12 Figure 12: Input-output structure: liquid treatment Anaerobic digestion (AD) Biodegradable material is biologically converted into compost, water, and biogas. Biogas is assumed to contain 70% methane.34,36,37 The regression ranged from 25-1100 kg/m3. Pj is modeled as a function of Sj, while Cjf and Cjop are modeled as functions of BODj. Accordingly,

= f ≅ = ⋅ + , j = AD (1)


Page 10 of 43

Page 11 of 43

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


C$% = f &' ≅ ( ) = * ⋅ &' , j = AD (2)

,-

C$

= f &' ≅ ( ./ = 0 ⋅ &' , j = AD (3)

k denotes linearization intervals. The regression parameters of the models are presented in the Supporting Material (Table S1).

Activated Sludge (AS) Micro-organisms metabolize the suspended and soluble organic matter in aeration basins. Sludge is the main product. The regression ranged from 120-380 kg/m3. Pj is modeled as a function of Sj following Eq (1) (j=AS). The energy required for aeration, Ej and the costs Cjf and Cjop are modeled as functions of Sj following, 2 = f ≅ 2 = 3 / ⋅ + 3 / , j = AS (4)

C$% = f ≅ ( ) = f ⋅ + f , j = AS (5)

,-

C$

= f ≅ ( ./ = 0 ⋅ + 0 , j = AS (6)

k denotes linearization intervals. The regression parameters of the models are summarized in the Supporting Material (Table S2).

Trickling Filters (TF) TFs are used to remove organic matter from wastewater. The regression ranged from 120520 kg/m3. Pj is modeled as a function of Sj following Eq. (1) (j=TF). Energy is required for the sludge aeration and the rotation of distributor arms. Ej is modeled as a function of Sj and is based on Eq. (4). Cjf is modeled as a polynomial function of Sj and the operating cost Cjop is modeled as a power function of Sj.



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

C$% = f ≅ ( ) = 8f ⋅ + * ⋅ + f 9 , j = TF (7)

,-

C$

=> ?@

= f ≅ ( ./ = 0 ⋅

, j = TF (8)

k denotes linearization intervals. The regression parameters of the models are presented in the Supporting Material (Table S3).

Rotating Biological Contactors (RBC) RBC involves a number of rotating discs attached on a shaft that is submerged in a tank and is partially or completely filled with the treated liquid. In many respects, RBC is similar to TF above. The regression ranged from 100-380 kg/m3. Costs Cjf and Cjop, are modeled as functions of Sj following respectively Eqs. (5) and (6) (j=RBC). Pj is modeled as a function of TSSj.

= f TS ≅ = ⋅ + , j = RBC (9)

k denotes linearization intervals. The regression parameters of the models are summarized in the Supporting Material (Table S4).

Aerated Lagoon (AL) & Stabilization Pond (SP) The aerated lagoons (ALs) are ponds with artificial aeration to promote biological oxidation. Capable to perform at shock (peak) loads, they feature low maintenance cost. The regression ranged from 4000-150000 kg/day. Pj is modeled as a function of TSSj following Eq. (9). Ej, Cjf and Cjop, are modeled as functions of Sj following Eqs. (4), (5), and (6) respectively (j=AL). SPs are large shallow excavations that drain sewage from various systems. The regression ranged from 80-290 kg/m3. Pj is modeled as a functions of TSSj using Eq. (9) (j=SP). Cjf is modeled as a function of Sj following Eq. (6) (j=SP). As SPs do not need aeration the operating cost is assumed negligible. The regression parameters of the models are presented in the Supporting Material (Table S5 and S6).


Page 12 of 43

Page 13 of 43

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


Solid Waste The main products include ash and thermal energy. The selected technologies include combustion, incineration, gasification, fast pyrolysis, slow pyrolysis, and torrefaction.35,37 Each treatment output is modeled as a function of the inlet flow (Sj) and the energy required (Ej). Figure 13 illustrates the input-output structure of the models of solid treatment. FIGURE 13 Figure 13: Input-output structure: solid treatment

Combustion (COM), incineration (INC), and gasification (GSF) Combustion produces energy by turning solids into energy and ash (assumed 15% of the solids). The energy produced depends on the total inlet flow, its LHV, and the combustion technology. Pj is modeled as a function of Sj, based on Eq. (1). The regression considered LHV values from 2.5 - 40 MJ/kg. Costs Cjf and Cjop are modeled as functions of Sj, following Eqs. (5) and (6) respectively. Gasification operates at high temperatures (>700°C), converting organics into carbon monoxide, hydrogen, and carbon dioxide. Pj, Cjf, and Ej are expressed as functions of Sj respectively, following Eqs. (10), (11), and (12) (j= GSF). Incineration uses excess of oxygen and produces ash, flue gas, and heat. Pj and Cjf are modeled as functions of Sj, based on Eqs. (10) and (11) respectively.

= ⋅ + , j = INC, GSF (10) ( ) = * ⋅ + * , j = INC, GSF (11) 2 = 3 / ⋅ + 3 / , j = GSF (12) The regression parameters are presented in the Supporting Material (Table S7, S8, and S9).

Fast pyrolysis (FPY), slow pyrolysis (SPY), and torrefaction (TOR) Fast pyrolysis operates at 650-1000oC, while slow pyrolysis requires lower temperatures (~400oC) and longer residence times. Products include oil (50%), char (20%), and gas (30%).36-38 Pj, Ej, and Cjop depend on Sj and are modeled following Eqs. (10), (12), and (13).



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

In fast pyrolysis Cjf is calculated as power function of Ej, following Eq. (14), while slow pyrolysis models for cost follow Eq. (11). ( ./ = 0 ⋅ + 0 , j = FPY, SPY (13) ( ) = * ⋅ 2 )> ? , j = FPY, SPY (14) Torrefaction is a thermo-chemical process, which operates under anaerobic conditions at 250300°C. The end product is a stable, homogeneous, high quality biofuel with upgraded energy density and calorific value. The models of Pj and Ej, follow those in fast pyrolysis (j=SPY). Cjop is modeled as a linear function of Ej and Cjf following Eq. (16) ( ./ = 0 ⋅ 2 , j = TOR (15) ( ) = * ⋅

)> ?

, j = TOR (16)

The regression parameters of the models are presented in the Supporting Material (Table S10, S11 and S12).

Gas emissions Gases are treated to reduce COx, NOx, and SOx at acceptable levels. Selected treatment technologies include chemical and physical absorption, carbonation-calcination loop, and cryogenic methods.34 Byproducts include gas streams of acceptable environmental limits. The output is calculated with respect to the inlet stream flow (Sj) and its CO2 content, RCO2j, expressed in kg/day. The model calculates the products (Pj), the energy required (Ej), the capital investment (Cjf), the operating cost (Cjop), and the quantity of the solvent (Rs j) as required by the treatment technologies. Figure 14 illustrates the input-output structure of the models of gas treatment. FIGURE 14 Figure 14: Input-output structure: gas treatment

Chemical Absorption (MEA & MDEA)


Page 14 of 43

Page 15 of 43

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


Two different solvents are considered, namely mono-ethanolamine (MEA) and N-methyl diethanolamine (MDEA). Regression models are produced based on established literature.39,40 Pj and Cjf are functions of Sj based on Eqs. (10) and (13) respectively. Rsj, Ej and Cjop depend on RCO2j as follows. L = M ⋅ =.N , j = MEA/MDEA (17) 2 = 3 / ⋅ =.N , j = MEA/MDEA (18) ( ./ = 0 ⋅ =.N , j = MEA/MDEA (19) The regression parameters are shown in the Supporting Material (Table S13 and S14).

Physical Absorption (RCT & SLX) The methods address two different solvents: Rectisol and Selexol. The analysis is based on detailed designs available in the literature that process 30% CO2 in the feed.39 Rsj is a function of RCO2 j and follows Eq. (17). Ej is required to regenerate the solvent. For practical reasons, it is calculated as a function of RCO2j using Eq. (18). Cjf is a function of Sj, while Cjop is modeled with respect to RCO2 j, following chemical absorption Eqs. 13 and 19. Regression parameters are presented in the Supporting Material (Table S15 and S16).

Carbonation-calcination loop (Ca-loop) & Cryogenic methods (CR) The Ca-loop carbonates CO2 from the flue gas using CaO at high temperature. The model is based on available literature39 that assumes 30% or less CO2 content in the incoming flow. Cryogenic purification separates gas mixtures using fractional distillation and condensation at low temperature. The CO2 is liquefied and condensed from high concentration (>90%) streams, offering advantages as the captured CO2 can be directly transported and/or sold.41 Both models (Ca-loop and cryogenic) follow the same equations as in physical absorption. Regression parameters are summarized in the Supporting Material (Table S17 and S18).

6- MATHEMATICAL FORMULATION



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

The superstructure of Figure 15 is formulated as an optimization model. The model includes basic mass and energy balances, expressions that account for the economic flows, and a set of logical constraints. Objective function: The objective is the minimization of the total annualized cost with the latter expressed by: C$R = CR% + CRS − RV R (20) ∈W

CR% and CRS are fixed capital and energy costs associated with each treatment technology

Negative energy costs indicate operational profits in processing waste. RV R denotes revenues in selling valuable streams from treatment. Mass Balances around mixers [[[[[[ &'N denote the BOD composition at the entrance of treatment i. The composition is made up by the mixing of upstream flows and is calculated using Eq 21. By a similar token, TSS_i denotes the solid composition as it develops at the entrance of unit i. The composition is made up by the mixing of upstream flows using Eq 22. [[[[[[N = ∙ &' , ∀ ` ∈ a (21) N ∙ &' ∈^

N∈]

[[[[[N = ∙ , ∀ ` ∈ a (22) N ∙ N∈]

∈^

N = , ∀ ` ∈ a (23) N∈]

∈^

Fb = N , ∀ c ∈ d (24) ∈W

Design Constraints as presented in Section 5 Equations 25 and 26 calculate the amount of useful products, as they are produced at the exit of the treatment units. They depend on the process efficiency (design parameters a1 and a2) and the selection of the treatment units (zjk).


Page 16 of 43

Page 17 of 43

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


,fR Pr$= (g / ∙ + g / ∙ h ), ∀ i ∈ , ∈^

,fR = (g / ∙ j + g / ∙ h ), ∀ i ∈ , Pr$∈^

j = AD, AS, TF (25) j = RBC, AL, SP (26)

,fR Pr-R = Pr$, ∀ i ∈ (27) ∈W

Energy requirements and balances The energy needs of the treatment technologies can be estimated by Eqs 4, 12, and 18, E$ = (3 / ∙ + 3 / ∙ h ), ∀ ` ∈ a (28) ∈^

ER = E$ (29) ∈W

Costs and revenues These include the cost models presented by Eqs 2, 5, 7, 11, 14, and 16. The fixed cost are functions of the inlet flowrate (Eq 30) and/or the BOD composition of the stream entering treatment (Eq 31), as is the case of anaerobic digestion (AD). C$% = (f $ ∙ M$l + f $ ∙ z$l ), j = AS, TF, RBC, AL, SP (30) l∈n

C$% = (f $ ∙ BOD$l + f $ ∙ z$l ), j = AD (31) l∈n

CR% = C$% (32) $∈o

C$S = E$- ∙ pS , j ∈ J, p ∈ P (33) CRS = C$S (34) $∈o

,fR RV$ = Pr$∙ p- , j ∈ J, p ∈ P (35)

RV R = RV$ (36) $∈o



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

Logical Constraints: A first set of constraints relates to binary and continuous variables. The logical constraints eliminate the continuous variables associated with the unit once the unit is not selected. M$l − z$l ∙ LU s ≤ 0, ∀ j ∈ J, k ∈ K (37) R $l − y$l ∙ LU x ≤ 0, ∀ i ∈ I, k ∈ K (38) A second set of constraints is used to manipulate the maximum and minimum BOD compositions. For existing units the composition is set to the maximum levels required by the technology. The minimum composition is relaxed when upstream flows are eliminated. BOD$l − z$l ∙ BODz{| ≤ 0 , ∀ j ∈ J, k ∈ K (39) l ≥ 0 , ∀ j ∈ J, k ∈ K (40) BOD$l − z$l ∙ BODzb} l Additional logical constraints include exclusivity constraints on the selection of treatment technologies and constraints to enforce appropriate linearization intervals, yb$ = 1, ∀ i ∈ I (41) $∈o

z$l ≤ 1, ∀ j ∈ J (42)

l∈n

They also include contingency constraints and bounds on the number of technologies to use. yb$ ≥ z$l , ∀ j ∈ J , ∀ k ∈ K (43) b∈

z$l ≥ yb$ , ∀ i ∈ I , ∀ j ∈ J (44)

l∈n

yb$ ≤ N, ∀ i ∈ I , ∀ j ∈ J (45) b∈ $∈o

The mathematical formulation yields a MINLP problem with non-linearities mostly in the form of bilinear terms. The bilinear terms have all resulted from the regression models presented in Section 5. On average, they included 400-500 equations, 350-420 continuous


Page 18 of 43

Page 19 of 43

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


variables, and 100-150 discrete variables. The MINLPs were solved using GAMS/BARON (integrality gap 10-8) and results required 0.5-2.5min on a small machine (Intel i5 2.80 GHz processor) to converge. The quality of the optimal solutions has been tested by initiating optimization from 10 different initial points for each case.

7. RESULTS AND DISCUSSION Results are obtained in the background of the biorefinery portfolios presented in Section 2. Biorefinery products in Portfolio I include xylitol, cellulosic ethanol, and lignin-based polyurethanes; Portfolio II includes xylitol, cellulosic PVC, and lignin based polyurethanes. The value chain analysis, as this is required to produce the product portfolios, follow Kokossis and co-workers.33 On the basis of the two portfolios, three different examples are presented: •

Example 1: Evaluation of trade-offs between centralized and decentralized treatment (liquid streams)

•

Example 2: Integrated treatment of solid and gas byproducts

•

Example 3: Retrofitting existing installations

Example 1: Centralized vs decentralized treatment Portfolio I is selected as a basis. The problem involves 5 liquid streams, whose treatment technologies include anaerobic digestion, activated sludge, tricking filters, rotating biological contractors, aerated lagoons, and stabilization ponds. Figure 15 illustrates the liquid streams following the BBR representation discussed in Section 4. The graph properties, in conjunction with the allocation maps presented in Section 4, give rise to the superstructure of Figure 16. Centralized treatment corresponds to the case where the number of units (N) is set to 1. As N increases, options for decentralized treatment are possible. The solutions produced by the optimization models are presented and discussed separately for each case.

FIGURE 15 Figure 15: BBR representation Example 1



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 16 Figure 16: Superstructure Example 1

i) Centralized management - By restricting the approach to the use of a single technology (N=1), the optimization selected the anaerobic digestion (AD). The total installation cost is 15 $M/yr and the revenues from the production of biogas is 1.5 $M/ yr. The selection of AD is essentially forced as AD is the only technology suitable for all streams; no other technology can stand as an alternative option. ii) Decentralized management – As N increases, the optimization is able to introduce options for decentralized management. Results are presented for N=2 and N=3. The optimization effort is similar for higher values of N; still, higher values for N are considered less realistic. For N=2, the selected technologies include AD and RBC. The installation cost is 4.2 $M/yr. In comparison with the previous case, distributed treatment saves 33% of the cost. Table 3 presents the allocation of streams across treatment technologies and the cost distribution in the selected technologies. The BBR with the optimal solution is illustrated in Figure 17.

Table 3: Results summary – Example 1 (N=2) TABLE 3

FIGURE 17 Figure 17: BBR of Example 1, DC - optimal solution for N=2

For N=3 the selected treatment includes AD, RBC, and SP. The installation cost is 0.42 $M/yr, namely a tenfold decrease as compared with the previous case. Table 4 presents the allocation of streams across treatment technologies and the cost distribution. Figure 18 illustrates the optimal solution. Integration options are important, and Liquid 8 is forced to split between into SP and AD.


Page 20 of 43

Page 21 of 43

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


Table 4: Results summary Example (N=3) TABLE 4

According to the results, the integrated approach and the use of distributed treatment lead to a significant decrease in cost, and, essentially, it reduces the burden by one or two orders of magnitude.

FIGURE 18 Figure 18: Optimal Solution - Example 1 (N=3)

Example 2: Integrated treatment of solid and gas byproducts Two different biorefineries are presented for the treatment of solid and gas waste. The treatment of solids is based on Portfolio II whereas the treatment of gases is using Portfolio I. Solids - The superstructure integrates solid streams, according to Figure 19. Results

(i)

are summarized in Table 5. Unlike Example 1, solid streams can be treated by all 6 technologies. Still, the optimization selects a single treatment (e.g. combustion) as the better choice, even when N is relaxed at higher values. A single treatment is the optimal solution and central management emerges as the better choice to handle solids. The optimization produces negative values for the objective (see Table 5), essentially indicating that the treatment of solids is economically profitable.

FIGURE 19 Figure 19: Superstructure - Example 2 (solids)

Table 5: Results summary - Example 2 (solids) TABLE 5

(ii)

Gases - The superstructure integrates gas streams according to Figure 20. Results are summarized in Table 6. Unlike the case for solids, the optimization selected 2



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

different technologies, indicating benefits in the distributed management of waste and the use of Ca-loops (even though MEA offers 80% lower costs than Ca-loop).

Table 6: Results summary - Example 2 (gases)

TABLE 6

FIGURE 20 Figure 20: Superstructure - Example 2 (gas)

Example 3: Retrofitting The example assumes a bioethanol plant that already uses a small anaerobic digestion (AD) unit to treat its waste. The plant is upgraded into a second generation biorefinery using the products of Portfolio I. Options stand to (a) upgrade treatment adhering to anaerobic technology, now using a larger unit, and (b) combine the existing unit with one or more new units (e.g. new additional units, not excluding options to purchase other small AD units). The fixed cost of Case (a) is estimated at 14.9 $M/yr; the annualized fixed costs account for 16.4 $M/yr and revenue of 1.474 $M/yr available from the biogas that is produced. Let us assume that in Case (b) (i) the set of treatment options remains as discussed in Example 1, and (ii) the existing unit does not incur payback costs (e.g. it only contributes to energy balances/costs yielding revenues in producing biogas). The purpose of the optimization remains to identify treatment technologies that could be combined with the existing unit to improve the economic performance of Case (a).

FIGURE 21 Figure 21: Optimal Solution – Example 3 (N=1) In the case where a single technology (N=1) is allowed to be selected, the best solution is shown in Figure 21. The additional unit is, also, an AD process. Table 7 presents the cost distribution and the allocation of streams between the existing and the new unit.


Page 22 of 43

Page 23 of 43

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


Table 7: Costing features comparison. TABLE 7

FIGURE 22 Figure 22: Optimal Solution – Example 3 (N=2) For N=2 the technologies selected include RBC and SP, none involves an anaerobic unit. The result is counter-intuitive since, in all previous cases studied, the design options have been dominated by AD units. Figure 22 illustrates the optimal solution. Table 8 summarizes results, explaining the stream and the cost allocation between the 3 technologies available. Moreover, the overall objective has now turned negative (-0.104 $M/yr), indicating a profitable process in operation. This is a significant improvement over

the results of the nominal case (purchase of new and/or complementary AD units) that required 14.9 and 4.234$M/yr, respectively (Table 7).

Table 8: Costing features comparison. TABLE 8 8. CONCLUSIONS The paper outlines a structured methodology towards the systematic valorization of bio-waste in a multi-product biorefinery. The methodology deploys a graph representation (BBR) to synthesize and develop superstructures that are subsequently formulated into mathematical programming models. The integration of technologies is systematic leading to the holistic evaluation of options and the development of designs that are economic and efficient. Results are presented against a background of a real-life plant featuring 49 streams, 22 treatment technologies, and 17 pollutants. The mathematical optimization has performed robustly producing solutions that justify both the use of optimization and the drive to integrate. An important conclusion is that the integrated approach may yield significant savings in using decentralized treatment and non-conventional solutions. The methodology can be adjusted to restrict the level of integration by constraints on the number of technologies (either per plant



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

or per stream) that are allowed to integrate. The savings from decentralization was found to be as high as 97%. In the context of the selected example, savings are pronounced in the case of liquid streams. The approach is further illustrated in retrofitting applications where the integrated designs essentially convert operating costs into clean profits for the plant. 9. SUPPORTING INFORMATION FOR PUBLICATION Supporting material includes adopted reference equations used from the literature and material to explain the development of design parameters used in the mathematical models (Appendix I, II, and III). This information is available free of charge via the Internet.

10. ACKNOWLEDGMENTS Financial support is gratefully acknowledged from the EU Research Program BIOCORE (FP7-241566) and RENESENG (FP-607415).

Nomenclature Sets ` c i

treatment technology waste streams waste products discretization intervals

Parameters g / , / 3 / , 3 / * , * 0 , 0 M /

p- , pS jN j/N N [[[[[[ &'N N [[[[[N

&' ,&'N j , j

discretization parameters for product p produced by technology j discretization parameters for energy needs produced by technology j discretization parameters for fixed cost of technology j discretization parameters for operating cost of technology j parameter for the estimation of solvent p (component) required by technology j product/energy price lower heatig value of waste liquid stream i latent heat of component p in waste liquid stream i number of technologies BOD of waste liquid stream i [kg/yr] flow of waste liquid stream i [kg/day] TSS of waste liquid stream i [kg/day] maximum / minimum BOD going to discretization k [kg/day] upper bounds for logical constraints


Page 24 of 43

Page 25 of 43

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


Continuous variables N /N &' j

(

waste stream i in technology j component p of waste stream i in technology j total inlet to technology j total waste from technology j going to discretization k total BOD from technology j going to discretization k total TSS to technology j total TSS from technology j going to discretization k product p from technology j / from all the waste management processes energy needs of technology j / total energy needs of treatment processes installation cost of technology process j / of biorefinery [M$ per year] operational cost of technology process j [M$ per year] energy cost of technology process j / total energy cost of biorefinery [M$ per year] revenue from technology j / total revenue from waste management [M$ per year] total cost from waste management treatment [M$ per year]

Binary variables h N

modeling variable of technology j and interval k modeling variable of waste stream i and the technology j

.

/ , /

2 , 2 C$% , CR% C$,

C$S , CRS ,

REFERENCES [1] International Energy Agency (IEA), www.ieabioenergy.com (accessed May 29, 2014). [2] Ding S.Y.; Himmel M.E. The maize primary cell wall microfibril: A new model derived from direct visualization. J. Agric. Food Chem. 2006 54 (3), 597–606. [3] Zhang Y.H.P.; Ding S.Y.; Mielenz J.R.. Cui J.B.; Elander R.T.; Laser M.; Himmel M.E.; McMillan James R.; Lynd L.R.; Fractionating recalcitrant lignocellulose at modest reaction conditions. Biotechnol. Bioeng. 2007 97 (2), 214–223. [4] Smith R. Chemical Process Design and Integration. J. Willey & son, 2005 Chichester, UK. [5] Smith R.; Jobson M.; Chen L. Recent development in the retrofit of heat exchanger networks, Appl. Therm. Eng., 2010 30, 2281-2289. [6] Gadalla M.; Jobson M.; Smith R. Optimization of existing heat-integrated refinery distillation systems, Trans IChemE, 2003 81, Part A, 147-152. [7] Gunaratnam M.; Alva-Argez A.; Kokossis A.C.; Kim J.-K.; Smith R. Automated Design of Total Water Systems, Ind. Eng. Chem. Res., 2005 44, 588-599.



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

[8] Nikolakopoulos A.; Karagiannakis P.; Galanis A.; Kokossis A. A water saving methodology for the efficient development of biorefineries, ESCAPE 22 on Comp. Aided Process Eng., 2011 30, 6-10. [9] Wang, Y.P.; Smith, R. Wastewater Minimisation, Chem. Eng. Sci., 1994 49, 981-1006. [10] Alva-Argaez A.; Kokossis A.C.; Smith R. Wastewater minimization of industrial systems using an integrated approach, Comp. Chem. Eng., 1998 22, S741-S744. [11] Alva-Argaez A.; Vallianatos A.; Kokossis A.C. A multi-contaminant transshipment model for mass exchange networks and wastewater minimization problems. Comp. Chem. Eng. 1999 23, 1439–1453. [12] Stefanakis M.E.; Pyrgakis K.; Mountraki A.D.; Kokossis A.C. The Total Site Approach as a synthesis tool for the selection of valorization paths in lignocellulosic biorefineries, ESCAPE 23, 2013 June 9-12, Finland. [13] El-Halwagi M.M. Process integration, Academic Press 2006, San Diego. [14] Woodard F. Industrial Waste Treatment Handbook, 2001 Butterworth–Heinemann. [15] Chakraborty A.; Linninger A.A; Plant-Wide Waste Management. 1. Synthesis and Multiobjective Design, Ind. Eng. Chem. Res. 2002 41, 4591-4604. [16] Arnold M.E.; Merta E.S. Towards energy self-sufficiency in wastewater treatment by optimized sludge treatment, Water Practice & Technology 2011 6 (4), IWA Publishing. [17] Koppmann R.; von Czapiewski K.; Reid J.S. A review of biomass burning emissions, part I: gaseous emissions of carbon monoxide, methane, volatile organic compounds, and nitrogen containing compounds, Atmos. Chem. Phys. Discuss. 2005 5, 10455–10516. [18] Brinckerhoff P. Accelerating the uptake of CCS: Industrial use of captured carbon dioxide, 2011 Global CCS Institute. [19] Economic and Social Commission for Western Asia (ESCWA) Waste-Water Treatment Technologies: A General Review, 2003 United Nations. [20] Strauss K.J. Application of pinch technology in water resource management to reduce water use and wastewater generation for an area, WRC, 2006 Report No. 1241/1/06. [21] El-Halwagi, M.M. Pollution prevention through process integration, Academic Press, 2003 San Diego. [22] Kokossis A.C.; Yang, A. On the use of systems technologies and a systematic approach for the synthesis and the design of future biorefineries. Comp. Chem. Eng. 2010 34, 1397–1405.


Page 26 of 43

Page 27 of 43

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


[23] Kokossis A.C.; Yang A.; Tsakalova M.; Lin T.-C. A systems platform for the optimal synthesis of biomass based manufacturing systems. Comp. Aided Chem. Eng., 2010 28,1105-1110. [24] Kokossis A.C.; Yang A.; Tsakalova M.; Lin T-C. Systematic Screening of Multiple Processing Paths in Biorefineries: The ABC (Accessing Biomass to Chemicals) Project and Its Potential to BuildProcess Synthesis Capabilities, Integrated Biorefineries: Design, Analysis, and Optimization, 2012 Taylor and Francis. [25] Linninger A.A.; Chakraborty A. Synthesis and optimization of waste treatment flowsheets. Comp. Chem. Eng. 1999 23, 1415-1425. [26] Linninger A.A.; Chakraborty A.; Colberg R.D. Planning of waste reduction strategies under uncertainty. Comp. Chem. Eng. 2000 24, 1043-1048. [27] Cavin L.; Dimmer P.; Fischer U.; Hungerbu1hler K. A Model for Waste Treatment Selection and Costing under Uncertainty, Ind. Eng. Chem. Res, 2001 40, 2252-2259. [28] Muñoz E.; Capón-García E.; Hungerbühler K.; Espuña A.; Puigjaner L. Decision Making Support Based on a Process Engineering Ontology for Waste Treatment Plant Optimization, AIDIC Conference Series, 2013 11, 261-270. [29] Muñoz E.; Capón-García E.; Hungerbühler K.; Espuña A.; Puigjaner L. Decision Making Support based on a Process Engineering Ontology for Waste Treatment Plant Optimization, Chem. Eng. Trans., 2013 32, 277-282. [30] Mountraki A.D.; Nikolakopoulosα A.; Mlayah B.B.; Kokossis A.C. BIOCORE– A systems integration paradigm in the real-life development of a lignocellulosic biorefinery, ESCAPE21 on Comp. Aided Process Eng. 2011. [31] Delmas M. Vegetal Refining and Agrichemistry, Chem. Eng. Techn., 2008 31 (5), 792797. [32] Tsakalova M.; Yang A.; Kokossis A. Implementation and initial evaluation of a decision support platform for selecting production routes of biomass-derived chemicals. Comp. Aided Chem. Eng., 2011 29, 1470-1474. [33] Kokossis A.; Tsakalova M.; Pyrgakis K. Design of integrated biorefineries, Comp. Chem. Eng., 2015 81, 40-56. [34] Panteli A. Mathematical modeling and integrated design of treatment technologies for liquid waste and gaseous emissions originated from 2nd generation biorefineries, National Technical University of Athens, 2014 Greece, Dissertation – in Greek



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

[35] Papoutsi I.A. Mathematical modeling and integrated design of treatment technologies for solid waste and reusable effluents originated from 2nd generation biorefineries, National Technical University of Athens, 2014 Greece, Dissertation – in Greek [36] Appels L.; Baeyens J.; Degrève J.; Dewil R. Principles and potential of the anaerobic digestion of waste-activated sludge. Elsevier Ltd., Progr. Energy Comb. Sci., 2008 34, 755-781. [37] Van Haandel A.; van der Lubbe J. Handbook Biological Waste Water Treatment: Design and optimization of activated sludge systems. Leidschendam: Quist Publishing, 2007 Chapter 8, 389-396. [38] Bezanson A. Pyrolysis and Torrefaction of Biomass, Mech 1999 4840. [39] Atsonios K.; Koumanakos A.; Panopoulos K.D.; Doukelis A.; Kakaras E. TechnoEconomic Comparison of CO2 Capture Technologies Employed With Natural Gas Derived GTCC. ASME Turbo Expo 2013: Turbine Technical Conference and Exposition, San Antonio, Texas, USA, 2013 2, No. GT2013-95117, pp. V002T07A018. [40] Yu C.H.; Huang C.H.; Tan C.S. A Review of CO2 Capture by Absorption and Adsorption. Aerosol and Air Quality Research, 2012 12, 745-769. [41] Atsonios K.; Panopoulos K.D.; Doukelis A.; Koumanakos A.; Kakaras E. Cryogenic method for H2 and CH4 recovery from a rich CO2 stream in pre-combustion carbon capture and storage schemes. Elsevier Ltd., Energy, 2013 53, 106-113. [42] Economic and Social Commission for Western Asia (ESCWA) Waste-Water Treatment Technologies: A General Review. 2003 United Nations. [43] Qiu Y.; Shi H.C.; He M. Nitrogen and Phosphorous Removal in Municipal Wastewater Treatment Plants in China: A Review. Int. J. Chem. Eng., 2010, Article ID 914159, pp10. [44] NPTEL, National Program on Technology Enhanced Learning, http://nptel.ac.in/ (accessed 13/11/2013). [45] Daigger G.T.; Boltz J.P. Trickling Filter and Trickling Filter – Suspended Growth Process Design and Operation: A State-of-the-Art Review. J. Water Env. Res., 2011 38 (5), 388-404. [46] U.S. Environmental Protection Agency (USEPA) Process Design Manual for Sludge Treatment and Disposal. EPA 625/1-79-011 U.S. Environmental Protection Agency, Cincinnaty, Ohio. EPA Office of Research and Development, 1979 Washington, D.C. [47] Swedish College of Engineering & Technology, Wah Cantt, http://scetcivil.weebly.com/ (accessed 01/11/2013).


Page 28 of 43

Page 29 of 43

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


[48] Spellman F.R. Mathematics Manual for Water and Wastewater Treatment Plant Operators. United States: CRC Press, 2004 Chapter 25, pp. 251-256. [49] Patwardhan A.W. Rotating Biological Contactors: A Review. Ind. Eng. Chem. Res., 2003 42 (10), 2035-2051. [50] Peterson, J.L. Petri Net Theory and the Modeling of Systems. 1981 Prentice-Hall.



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

For Table Contents only


Page 30 of 43

Page 31 of 43

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


FIGURES

Figure 1: Perimeters of environmental assessment.

Figure 2: A BBR graph representation



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 3: Superstructure corresponding to Figure 2b

Figure 4: Treatment configuration

Figure 5: Decentralized (a) and centralized (b) liquid manufacture processes


Page 32 of 43

Page 33 of 43

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


Liquid s Trickling Filter

Aerated Lagoon

OR

BOD: 120 - 310

BOD: 120 - 520

YES

Waste stream 50-97% H2O

H2O

Biogas

Anaerobic Digestion

NO

N2

Sludge Rotating Biological Contactors

BOD: 25 - 1100

BOD: 100 - 380

Activated Sludge

Stabilization Pond

BOD: 120 - 380

BOD: 80 - 290

Figure 6: Allocation maps – Liquids (a)

H2O

Catalyst

Chemical Precipitation

OR

Ion Exchange

YES

YES

Waste stream

Catalyst > tr

H2O

NO

H2O>97%

Chemical Precipitation NO OR

Ion Exchange

Reuse/ Regeneration

Figure 7: Allocation maps – Liquids Reuse/ Regeneration (b)


H2O


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 34 of 43

Feed

Solids

YES

Liquid

Zymo Sugars Cell

Gas

Ash

Fertilize r

YES

Combustion N2

NO

Burnable NO

Drying

Incineration

YES OR

Pyrolysis

Gasification

Figure 8: Allocation maps – Solids.

Gas

CO2

Gas

Membrane Reactor

Cryogenic Methods

Ca -Loop

YES

OR

Air Pollutant Waste stream 50 < H2O

CO

NO

Discharge

COx, SOx, NOx NO

YES

Chemical Absorption MEA Chemical Absorption MDEA

OR

Physical Absorption Rectisol Physical Absorption Selexol

Figure 9: Allocation maps – Gas emissions.


Page 35 of 43

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 10: BBR graph of multiproduct biorefinery

Figure 11: Superstructure of multiproduct biorefinery



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 12: Input-output structure: liquid treatment

Figure 13: Input-output structure: solid treatment

Figure 14: Input-output structure: gas treatment


Page 36 of 43

Page 37 of 43

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 15: BBR representation Example 1

Figure 16: Superstructure Example 1



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 17: BBR of Example 1, DC, optimal solution for N=2

Figure 18: Optimal Solution - Example 1 (N=3)


Page 38 of 43

Page 39 of 43

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 19: Superstructure - Example 2 (solids)

Figure 20: Superstructure - Example 2 (gas)



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 21: Optimal Solution – Example 3 (N=1)

Figure 22: Optimal Solution – Example 3 (N=2)


Page 40 of 43

Page 41 of 43

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


TABLES

Table 1: Waste streams from each production processes Production Process

Gas

CIMV C5 to xylose

Liquid

for Reuse

Solid

Catalyst regeneration

1 1

to xylitol bio

1

to xylitol cat to ethanol

1

1

1+1*

1*

1

1

1

1

C6 to glucose

1

to Itaconic Acid

3

to ethanol

1

1

1

Lignin to PF to PU

1

SSH ethanol

1

1

SHF ethanol

1

1

Ethanol to ethylene

1

Ethylene to PVC

3

2

Table 2: Treatment technologies options Treatment Technologies Chemical absorption

Physical absorption

Gas MEA

MDEA

Rectisol

Selexol

Liquid

Anaerobic Digestion

Activated Sludge

Trickling Filters

Rotating Biological Contactors

Reuse

Reverse Osmosis

Micro filtration

Solid

Combustion

Catalyst Regeneration

Chemical precipitation

Incineration Gasificatio n

Torrefaction

Ca-loop

Membrane reactors

Aerated Lagoon

Stabilization Pond

Slow Fast Pyrolysis Pyrolysis

Ion exchange

Table 3: Results summary – Example 1 (N=2)

Technology

Liquid 1 RBC

Liquid 2 RBC

Fixed Cost

0.185

Energy Cost

0.172

Liquid 3 RBC

Liquid 8 AD

4.722

Total M$/yr

4.907 0.172

Revenue Total Cost M$/yr

Liquid 11 AD

0.845 4.234


0.845

Cryogenic methods


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 42 of 43

Table 4: Results summary Example (N=3) Liquid 1

Liquid 2

Liquid 3

Technology

RBC

SP

SP

Fixed Cost

0.139

Liquid 8 SP

Liquid 11

AD

0.029

AD 0.136

0.129

Energy Cost

0.013

Total Cost

0.420

(M$/yr)

Table 5: Results summary - Example 2 (solids) Solid 1

Solid 2

Combustion

Combustion

Fixed Cost

0.784

Revenue

2.577

Total Cost (M$/yr)

-1.809

Table 6: Results summary - Example 2 (gases) Gas 1

Gas 2

Technology

MEA

Ca-loop

Fixed Cost

0.006

0.031

Total Cost (M$/yr)

0.304 0.129

Revenue

Technology

Total $M/yr

0.038


0.013

Page 43 of 43

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


Table 7: Costing features comparison. AD costing

AD+ ADs costing

(all liquid streams)

Liquid

Liquid

1+8

1+2+3+8+11

Technology

AD

AD

ADs

Fixed Cost

16,404

5.712

0

Revenue

1,474

0.479

0.998

Total Cost (M$/yr)

14,930

4,234

Table 8: Costing features comparison. AD costing (all liquid streams)

RBC

SP

ADs

Liquid

Liquid

Liquid

1

8

1+2+3+8+11

Fixed Cost

16,404

0.082

0.025

0

Revenue

1,474

0.077

0

1.225

Total Cost (M$/yr)

14,930


-0.1040

Integrated Waste Management in Multiproduct ... - ACS Publications

Recommend Documents