Sustainability evaluation of alternative routes for ... - ACS Publications

Subscriber access provided by Kaohsiung Medical University

Process Systems Engineering

Sustainability evaluation of alternative routes for fine chemicals production in an early stage of process design adopting process simulation along with data envelopment analysis Andrea Mio, Phantisa Limleamthong, Gonzalo Guillen Gosalbez, and Maurizio Fermeglia Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.7b05126 • Publication Date (Web): 02 May 2018 Downloaded from http://pubs.acs.org on May 5, 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 92 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

Sustainability evaluation of alternative routes for fine chemicals production in an early stage of process design adopting process simulation along with data envelopment analysis Andrea Mio†, Phantisa Limleamthong‡, Gonzalo Guillén- Gosálbez *‡, Maurizio Fermeglia† †

Molecular Simulation Engineering (MOSE) Laboratory, Department of Engineering and Architecture (DEA), University of Trieste, Piazzale Europa 1, 34127 Trieste, Italy

‡

Centre for Process Systems Engineering, Department of Chemical Engineering, Imperial College London, London, SW7 2AZ, UK

Corresponding author*: Gonzalo Guillén- Gosálbez, e-mail address: [email protected]

Keywords: Sustainable Process Design, Data Envelopment Analysis (DEA), Pioglitazone Hydrochloride, Pharmaceutical, Life Cycle Assessment (LCA), Process Simulator

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 92

Abstract In this work, we propose a framework for the preliminary screening of chemical process designs according to sustainability criteria. Implementing each flowsheet in a process simulator, several indicators based on sustainability pillars, i.e. economy, society and environment, are calculated and then analyzed employing data envelopment analysis (DEA) in order to select the most efficient designs considering multiple criteria simultaneously. Suboptimal alternatives are further investigated adopting a retrofit analysis, aiming to identify the parameters that most contribute to the final indicators scores, therefore highlighting the major sources of deviation from optimal conditions. We applied this framework on a case study based on various patents regarding the production of pioglitazone hydrochloride in order to validate the capabilities of our methodology. Our approach, that embeds process simulation, sustainability indicators and DEA within the same procedure, will support practitioners during the sustainability assessment of promising process designs removing the inefficient ones, as the adoption of well-established LCA methodologies would take much effort in terms of time, money and information to be retrieved for a plethora of alternatives.


Page 3 of 92 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 The technology of modern society has been developed through history to improve life quality and provide a higher satisfaction of human needs, e.g. better hygienic conditions, food stock granted by intensive farming and livestock, effective drugs, etc. Despite bringing significant benefits resulting from improvements in every field of science, the use of these technologies have led to a considerable number of environmental problems that affect globally,1 e.g. global warming, ozone depletion, pollution of air, or reduction of fresh water reserve. In the last two decades an increasing effort has been spent on a smarter exploitation of natural resources, some of which are going to end in the near future.2–4 Therefore, an increasing adoption of renewable energy sources and a trend towards a reduction of the impact of anthropogenic activities on the environment have become essential from a sustainable development viewpoint.5–7 A large number of sustainability indicators and tools for sustainability assessment have been proposed aiming to provide information on the economic, social and environmental impacts of different systems, including transports, manufacturing, supply chain and energy systems.8,9 The industrial sector, in particular, has always played a major role in climate change. Thus, a wide range of different approaches have been recommended to estimate the sustainable development contribution of the entire production process, comprehending extraction of raw material, transportation to manufacturing site, processing to obtain the desired product, utilization by the end user and disposal of wastes.10–12 Furthermore, chemical and process industry specific methodologies13–19 have been published in order to assist decision-makers in the selection of the most sustainable design among a plethora of different alternatives.



Page 4 of 92

One of the most useful and widespread methodology is Life Cycle Assessment (LCA),20 which covers the potential environmental impacts and the resources exploited throughout the entire product’s life-cycle, from cradle, i.e. extraction of raw materials, to grave, i.e. waste disposal.21 The LCA methodology embodies four subsequent steps to perform a full-scale product assessment: Goal and Scope Definition, in which the target of the analysis has to be defined; Life Cycle Inventory (LCI), in which the information about mass and energy balances of materials, equipment, transformation process, transportation and waste treatment need to be retrieved; Life Cycle Impact Assessment (LCIA), which adopts the LCI data to calculate the impact of the entire life-cycle of the product; and Interpretation, in which the results from the LCIA phase have to be analyzed to recognize the most sustainable pathway. Despite the clear benefits of this tool, a fullscale LCA evaluation often requires reliable data, a high level of expertise as well as a considerable amount of time. This is particularly true when applied to process design,22 where many alternatives need to be assessed in terms of a set of economic, environmental and social indicators. In this paper, we introduce a methodology for a preliminary screening of process alternatives that facilitates the selection of flowsheets according to a wide variety of sustainability indicators. The relevance of our approach is enlightened by the peculiarities of the pharmaceutical and fine chemicals fields, i.e. many different routes exist to obtain the same product, data gaps affect the chemicals involved, a substantial amount of resources are spent on R&D, and there is a significant urgency to enter the market with the product under evaluation.23 In essence, our approach is based on the use of data envelopment analysis (DEA), a methodology originally adopted for the efficiency evaluation of economic and social entities,24 to assess the relative efficiency of a set of decision-making units (DMUs). DEA solves primal and dual linear


Page 5 of 92 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


programming models to classify a set of DMUs, each modeled as a single entity consuming multiple inputs to produce multiple outputs, as either efficient or inefficient. The former lie in the Pareto front and show an efficiency score of one, while the latter do not belong to it and feature an efficiency value strictly less than one. For the inefficient units, DEA provides in turn improvement targets that are calculated by projecting them onto the efficient frontier. These targets help to identify sources of inefficiency and assist in turn in the adoption of corrective actions aiming at improving the overall performance of the inefficient units. There is a large body of literature on the use of DEA in a wide variety of applications, including health systems,25 transports,26 academia,27 human development,28 manufacturing industry29 and economy.30 The use of DEA in chemical engineering, on the contrary, has been quite scarce. This is surprising given the fact that DEA is a very useful tool to study trade-offs between objectives like those arising in process design. For instance, Bick et al.31 studied the influence of hydrodynamic conditions on membrane fouling using DEA for the selection of the best trade-off between cross-flow velocity and aeration flow rate. Cabrera-Ríos et al.32 applied DEA to identify the best compromise among various process variables in the polymers manufacturing field. Han et al.33 studied the efficiency of various process designs of an ethylene production plant using DEA. The electricity mix of the European countries have been examined by Ewertowska et al.34 using DEA in order to identify potential improvements attainable by each government, while Limleamthong and Guillén-Gosálbez35 focused their research exclusively on UK, employing DEA principles as a starting point for an accurate bi-level optimization of the UK electricity mix. Chen et al.36 adopted DEA to analyze the efficiency and productivity of biotech industrial sites. Venkata Mohan et al.37 used DEA to investigate the relationship between different pretreatment procedures on anaerobic digestion processes. Finally, Hernández-Sancho



Page 6 of 92

and Sala-Garrido38 evaluated wastewater treatment plants using DEA in order to establish their optimal size and configuration, while Sueyoshi and co-workers39,40 adopted DEA to assess the efficiency of fossil fuel power plants. More recently, DEA was applied to assess facilities producing gases for industrial use.41 Here, we apply DEA to assess chemical flowsheets according to sustainability criteria. Several publications belonging to disparate fields, from electricity generation42,43 to food waste management,44 applied LCA principles coupled with DEA to compare the efficiency of numerous units. The use of DEA in this context shows two main advantages compared to other existing tools. First, DEA can deal with a large number of indicators without the need to rely on any subjective weighting scheme. Second, DEA carries out an analysis of the inefficient (suboptimal) flowsheets, providing insight on how to improve them. We clarify at this point that in process design the emphasis is often placed on generating a large number of configurations and associated operating conditions from which the best ones are selected. In DEA, on the contrary, the alternatives are already given and the focus is on identifying the best according to multiple criteria and, perhaps more importantly, on understanding how suboptimal solutions can be improved. The application of DEA in process design is therefore recommended for cases in which there are already flowsheets at hand, possibly coming from patents, and one wishes to assess them all and benchmark one against the others. DEA does not hence compete with standard superstructure optimization and MINLP methods, but can be rather seen as a complementary tool particularly appealing when one aims to assess multiple indicators simultaneously (like it happens in sustainability evaluations) and when there is a clear interest on identifying sources of inefficiencies and poor performance in suboptimal designs. Particularly, DEA could be applied in the post-optimal analysis of


Page 7 of 92 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


flowhseets identified via superstructure optimization. As an example, we could first generate a set of promising designs by solving an MINLP iteratively with new integer cuts added in each iteration. These designs could then be further analyzed considering multiple criteria using DEA to filter out dominated solutions and also to assess the business as usual flowsheet. This work is organized as follows. An extensive description of the methodology will be presented first, including the characteristics of the indicators chosen as well as the fundamentals of DEA. A case study based on the production of pioglitazone hydrochloride, an active pharmaceutical ingredient (API) employed for diabetes mellitus type 2 treatment, will follow. Finally, some concluding remarks of the work are drawn at the end of the paper.

2. Sustainability evaluation methodology Our DEA-based methodology to screen and assess chemical flowsheets is outlined in Figure 1.



Page 8 of 92

Figure 1. Methodology qualitative decision path.

The methodology consists of five sequential steps that will be described in detail in the ensuing sections: 1. Retrieving different synthesis routes published in the literature or provided by in-house know-how. 2. Develop a flowsheet related to each possible design using process simulators. 3. Choosing and calculating appropriate indicators for the sustainability evaluation of each such alternative. To this end, the type and amount of data available, set of impacts addressed and decision-makers’ preferences will be taken into account. 4. Applying DEA to identify the most sustainable alternatives. 5. Identifying the best unit among the efficient designs and the major sources of inefficiency in the suboptimal designs, providing feedback regarding the parameters that should be modified (if possible) to improve the scores of each design. The adjustments required for improving the sustainability performance can be applied directly on the flowsheet generation step, thereby generating to a virtuous loop for a continuous improvement of the designs under study.

We note that one of the most valuable parameters to consider during a process design phase is time. It is crucial for a company to speed up as much as possible the entrance into the market of a product in order to reduce the R&D costs and take an advantage on the competitors. From this perspective, the methodology proposed in this paper fulfills the need of reducing the time spent


Page 9 of 92 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 performing an extensive analysis on alternative process designs. The reader will find underneath a description of the five main steps to adopt in order to perform the methodology proposed.

2.1.

Retrieving different routes

In order to enter the market with an established product, a company needs to review the existing literature in order to identify as many synthesis routes as possible. Therefore, this step requires the access to patents, literature databases and books. The collection of documents needs to be followed by a check on the expiration date of patents and the replication of routes among different sources. Some trivial changes among different designs are sometimes not worth to assess, while a single paper can provide more than one promising route. During this step, a first guess on the equipment employed and the missing data will result in a homogeneous approach for all designs under investigation, avoiding an imbalance which could end up with misleading evaluations.

2.2.

Modeling routes adopting process simulators

Synthesis routes have to be modelled in a process simulator in order to obtain energy and mass flows that are afterwards required to calculate the emissions and waste generated (and ultimately the impact metrics). Process simulators are a well-established tool adopted to design, develop, analyze and optimize chemical processes using a sequential modular approach.45 In this design phase, the company will take advantage of its know-how about scaling-up, phase equilibria and thermodynamic properties of the compounds involved in the processes. Some quantitative or qualitative data can be usually retrieved from patents, e.g. reaction yields and selectivity, 9 ACS Paragon Plus Environment


Page 10 of 92

byproducts, extraction of solutes, operating conditions, phase separation performances and equipment sizing. Therefore, it is possible to include this information within process simulations in order to achieve a reliable model that should be validated against existing data. In the event that simplifications need to be assumed, similar approximations for each design shall be adopted to produce accurate and meaningful results. Flowsheets should be scaled up to the actual demand of the desired product in order to quantify precisely the mass and energy flows required, while scheduling models might be required to assess batch process designs.

2.3.

Choosing and calculating indicators

It is well known that economy, society and the environment are all connected and dependent upon each other from a sustainable development viewpoint. Thus, it is essential to balance out these criteria when taking decisions. The selection of a suitable set of sustainability indicators is fundamental, while the accuracy of the results of the assessment is particularly related to trustworthy experimental or literature data. Although our methodology can be performed using any set of indicators, for illustrative purposes we adopted a single indicator for each one of the three pillars of sustainability, i.e. economy, society and environment. In order to preserve the simplicity of our approach, we selected indicators for which little information is needed, e.g. mass and energy balances, equipment costs provided by vendors or calculated using process simulators, H-Phrases and chemicals costs sourced from the literature, etc. 2.3.1. Economic – Cost of the Project (CP)


Page 11 of 92 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


CP quantifies the economic performances of each process design accounting for both Capital Expenditures (CapEx) and Operating Expenditures (OpEx) as shown in Equation 1, which is adapted from Schaber et al.46 and evaluates the discounted total cost of the project.

= +

1 1 +

In which is the discount rate and τ is the project lifetime. The CapEx calculations are based on the data reported in Table 1, which is adapted from Schaber’s work.46 The FOB (free on board) cost represents the cost of each process unit, neglecting additional related expenses, e.g. ancillary equipment, delivery, electrical, engineering, or piping. The delivery of each unit is accounted for by increasing its FOB costs by 5%46 and then multiplying by the Wroth factor,47 a heuristics-based coefficient adopted to include the additional expenses formerly excluded by FOB, finally obtaining the battery-limits installed cost (BLIC) of each piece of equipment. Furthermore, it is necessary to include some extra expenditures related to the production site, thereby providing a reliable estimation of equipment unit cost based on heuristics. Therefore, every BLIC needs to be multiplied by some extra coefficients related to each additional plant-related cost, as shown in Table 1, in which we assigned the upper limit value

47

to coefficients from (5) to (8), as we are dealing with

pharmaceutical facilities with stringent hygiene regulations and small scales of production.46



Page 12 of 92

Table 1. CapEx Heuristics47 Item

Cost

(1) FOB cost

sum of processing equipment units, solvent recovery excluded

(2) delivery

5% of FOB cost

(3) installation: ancillary equipment, automation, electrical, piping, and engineering

[(Wroth Factor)-1]∙(delivered equipment cost)

(4) battery-limits installed cost (BLIC)

sum of items (1) to (3)

(5) buildings and structures

10-20% of BLIC

(6) contingency

15-20% of BLIC

(7) offsite capital (for a grass-roots plant)

45-150% of BLIC

(8) service facilities

10-20% of BLIC

(9) waste disposal

not included in CapEx

(10) working capital

30-35% of annual materials costs

(11) total CapEx

sum of items (4) to (10)

The OpEx estimation includes raw materials cost, utilities cost and Labor Expenditures (LabEx), as shown in Equation 2.

=

" − + ! " # + $% 2 "


Page 13 of 92 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 which and are, respectively, the inlet and outlet mass of raw materials; is

the total mass of main product obtained;

is the market price per unit of each chemical;

" is the mass of utility j purchased at " ; and takes into account the estimated percentage of recovery of solvents, unreacted reagents and catalysts (it can vary in the range between zero and one), for which a zero value means a total utilization of chemical i within the process and a value of one indicates a total recovery of the substance for further utilization. The LabEx is evaluated considering the operators’ annual gross salary times the number of operators employed. Both pieces of information can be estimated using well-established methodologies.48 Therefore, in order to reduce the level of uncertainty in the sustainability assessment, the economic evaluation do not account for the expected revenues, since they mainly depend on forecasted product market prices and volumes. 2.3.2. Social – Potential Chemical Risk (PCR)

The Potential Chemical Risk (PCR) evaluates the hazards and risks for humans of the chemicals handled in a chemical plant. We use this index to assess the social dimension of sustainability. This indicator is based on the work by Vincent et al.,49 in which risk classes were related to R-Phrases assigned to each chemical. Aiming to define the correct human risk category that applies to a specific substance, R-Phrases were classified in risk classes considering an increasing risk according to the increasing classes. R-Phrases were replaced by H-Phrases,50, which required updating risk classes to hazard classes using H-Phrases, as reported in S1 of Supporting Information. In particular, hazard class zero is specific for chemicals with a lack of data in the literature, while hazard class one is applied to well-known harmless compound. 13 ACS Paragon Plus Environment


Page 14 of 92

The calculation of PCR needs information from mass balances performed in the process simulator and H-Phrases related to each chemical. The PCR relies on assigning each chemical to specific hazard classes whose contribution define the overall indicator. Whenever more than one H-Phrase is assigned to a substance, its final hazard class is considered to be the highest possible among the classes related to its H-Phrases. The calculation of the indicator is shown in Equation 3, in which '() is the maximum mass flow of component i (inlet flow for raw materials,

catalysts and solvents, outlet flow for products and byproducts); is the total mass of main

product obtained in the process; and *+, -+. is the hazard class assigned to chemical i and related

to risks to human health. We decided to consider the maximum amount of each chemical i within the process to assess the maximum potential risk related to the adoption of a specific substance.

'() 1 526 / = 10 23 4 3

2.3.3. Environmental – Potential Environmental Impact (PEI)

This indicator quantifies the environmental dimension of sustainability in the alternative process designs. It is based on mass balances, transfer coefficients and H-Phrases relying on the work of Vincent et al.,49 although we have modified the hazard classes to replace R-Phrases with H-Phrases in S2 of Supporting Information, using the same procedure as for the PCR. The PEI calculation follows a decision path to define the total potential environmental impact, based on mass flows and H-Phrases of chemicals, as well as transfer coefficients. The latter are driven by physicochemical properties and are related to physical state of substances (such as gas,


Page 15 of 92 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, solid or powder) and the medium (such as air, water or soil) in which a release in the environment is more likely to happen (Figure 2).

Figure 2. PEI Algorithm Flowchart

The introduction of a transfer coefficient provides a more realistic impact evaluation, as defined in the work by Vincent et al.,49 taking into account the relationship between the physical state of a compound and the media in which is more likely to be released. Transfer coefficients are shown in S3 of Supporting Information, which has been adapted from Vincent et al.,49 considering average values. This information could be modified if dispersion or transfer coefficients in various media became available from experimental data or from company expertise. Finally, the PEI score is calculated using Equation 4 from the maximum mass flow of each product, '() , the flowrate of the main product, , the term related to the PEI hazard class of chemical i, *+, -89 , and the transfer coefficient for chemical i, : .



; =

2.4.

'() 1 5

Sustainability evaluation of alternative routes for ... - ACS Publications

Recommend Documents