Screening of process alternatives based on sustainability metrics: comparison of two decision-making approaches

Abstract

Importance of sustainability concerns demands consideration of sustainability design criteria in early stages of process design. Most often there are several alternatives for a particular process which must be compared in terms of sustainability metrics to select the most sustainable design. There are several tools available for evaluation of process alternatives in terms of sustainability concerns. However, there is a need for a screening method which enables engineers to compare and select several alternatives in terms of a multitude of metrics and also incorporate their preferences if required. This study utilizes a recently developed sustainability evaluator for process evaluation and compares overall impact factor method for process screening with a novel fuzzy-based approach. The applicability of the proposed method is demonstrated though application of the methodology on ibuprofen case study to select the most sustainable design.

Keywords::

• decision-making tool
• sustainability
• sustainable process engineering

1. Introduction and background

Our world has been constantly striving towards a green revolution for decades now, and scientists have been trying to introduce technologies more benign to the environment and society which are yet lucrative. Chemical industries are no exception in this regard and everyday new technologies and design alternatives are introduced to help more sustainable productions. The current challenges in sustainable design of chemical processes fall in two main categories: one is to evaluate processes quantitatively in terms of the concerns of sustainability using a generic applicable methodology and another is to develop applicable tools which enable final decision-making and process screening in the presence of a multitude of often conflicting criteria imposed by sustainability concerns. During the past decade most researchers have been trying to address the mentioned issues and develop frameworks which address two main challenges encountered in incorporating sustainability concerns: sustainability evaluation and multi-criteria decision-making.

Incorporating the sustainability criteria into process design requires the evaluation of the sustainability concerns quantitatively using a generic and unified approach. Minimizing waste generation, environmental penalties (global emission) (Cabezas, Bare, and Mallick 1999), mass of pollutant of concern and total mass of waste (Ahmad and Barton 1995) have been often used as representative of environmental concerns of sustainability (Li et al. 2009). However, due to the involvement of energy consumption and the effect of recycle flow along with materials in a process, the use of traditional methods are not sufficiently comprehensive. In addition, social impacts and safety of processes are very important to be considered in process synthesis.

To overcome the challenge of incorporation of sustainability into process design, several metrics have been proposed to guide sustainable decision-making. Examples of metrics proposed by researchers include material intensity, energy intensity, water consumption, toxic emissions, pollutants emission and potential chemical risk (Tugnoli, Santarelli, and Cozzani et al. 2008; Azapagic, Perdan, and Clift 2004; Beloff, Lines, and Tanzil 2005; Beloff, Beaver, and Schwartz 2002; Martins et al. 2007; Sikdar 2009). Recently, some methodologies also have been presented for evaluating the social and environmental impact of processes. Examples of these methods are the environmental fate and risk assessment tool (EFRAT) (Achour et al. 2005), the atmospheric hazard index (Gunasekera and Edwards 2003) and thermodynamic analysis method (emergy and exergy) (Hau, Yi, and Bakshi 2007; Yi and Bakshi 2007; Bakshi 2000).

Three categories of sustainability indicators exist based on dimensions of sustainability presented in Figure 1. The most common metrics of sustainability according to these categories are shown in Table 1. Several tools and metrics have been devised by researchers in the past decade. Some well-known examples of such tools and metrics are TRACI developed by the Environmental Protection Agency (EPA) for the reduction and assessment of environmental impacts, waste reduction algorithm developed by the EPA, the EFRAT developed by Shonnard and Hiew (2000), the Methodology for Environmental Impact Minimization developed by Pistikopoulos, Stefanis, and Livingston (1994), etc. However, none of those is comprehensive to encompass all the metrics and categories of sustainability or they just consider the impacts limited to the battery of the process. Many other systems of sustainability metrics and index have been developed by other researchers and organizations, described in more detail by Shadiya (2010). A summary of these metrics and sustainability indices are summarized in Table 2. Shadiya (2010) have developed a computational tool for evaluating all sustainability concerns shown in Table 1 referred as the SUSTAINABILITY EVALUATOR in this work. This tool takes inputs from a process simulation and outputs a breakdown of 41 environmental, social and economic sustainability metrics. The novel contribution to the sustainability tool was the introduction of health metrics and the selection of the metrics that apply to process design. This tool enables process designers to comprehensively and quantitatively evaluate processes in terms of sustainability.

Figure 1 Graphical presentation of the sustainability aspects.

Table 1 Breakdown of all metrics analysed by the SUSTAINABILITY EVALUATOR.

	Environmental index		Social index
Economic index	Environmental burden	Resource usage	Total inherent safety index	Health risks
Product revenue	Acidification	E-factor	Heat of Reaction	Carcinogenic risk
Energy costs	Global warming	Mass productivity	Flammability index	Immune system damage
Raw material costs	Ozone depletion	Reaction mass efficiency	Toxic exposure index	Skeletal system damage
Annualized capital costs	Smog	Energy Intensity	Explosisivity index	Developmental damage
Waste treatment costs	Ecotoxicity to aquatic life	Water intensity	Temperature index	Reproductive system damage
Material value added	Eutrophication		Pressure index	Respiratory system damage
Profit	Aquatic oxygen demand		Corrosivity index	Cardiovascular system damage
			Equipment safety index	Endocrine system damage
			Safety level of process structure	Liver damage
				Nervous system damage
				Sensory system damage
				Kidney damage

Source: Adapted from Shadiya (2010).

Table 2 Summary of the proposed metrics, indicators and index systems.

Name of the system	Application	Drawbacks
Sustainable Process Index (Krotscheck and Narodoslawsky 1996)	Applicable to detailed process design	Lack of incorporating reaction efficiencyInapplicable to early stages of designNot addressing the social impacts
Inherent Process Safety Index (Heikkila 1999)	Applicable for assessing the safety of a chemical process	Limited to safety concerns
Sustainability Indicators (Afgan, Carvalho, and Hovano 2000)	Assessment of sustainability at early stages	Limited to energy indicators
BASF Socio-Eco-efficiency Metrics (Saling et al. 2002)	Useful in evaluate the impact of products and process during detailed design	Limited to early stage of design due to extensive data requirementThe social metrics impose problems in correlating process design parameters
GreenMetrics (Constable, Curzons, and Cunningham 2002)	Evaluation of reaction efficiency	Addressing only resource efficiency but not sustainability
IChemE Sustainability Metrics (IChemE Metrics 2002)	Assessment of the sustainability of production processes	Difficulty in correlating process parameter to the presented social metrics
Indicators of sustainable production (Krajnc and GlaviČ 2003)	Assessment of the sustainability of an operating unit	Not all of the suggested metrics are applicable for early stages of design
Global Environmental Risk Assessment Index (Achour et al. 2005)	Applicable for addressing health and safety risks of an operating unit	Limited to risk and safety assessment
Bridges to Sustainability Metrics (Tanzil and Beloff 2006)	Comparison of environmental impact of chemical processes	Limited to environmental aspectMetrics are accumulated into one metric
Three Dimensional Sustainability Metrics (Martins et al. 2007)	Evaluation of the sustainability of an industrial process	Lack of direct correlation between operating conditions, chemical process risk and environmental impact in the two presented metrics
Sustainability Indices (Tugnoli, Santarelli, and Cozzani 2008)	Evaluation of the sustainability of process alternatives	Not applicable to early stage of design
AIChE Sustainability Index (AIChE, SI^a: “Strategic Commitment to Sustainability,” 2008)	Comparison of different companies' performance	Lack of quantitative metricsNot applicable to early stages of design

Source: Adapted from Shadiya (2010).

a Sustainability index.

As Finkbeiner et al. (2010) concluded, besides the need for developing more quantitative sustainability measures in all three categories, a major challenge is the trade-off between alternatives and decision-making on the selection of a final sustainable solution. Due to the existence of conflicting objectives in a design for sustainability, usually there are more than one optimum alternative that among all only one needs to be selected as the final solution. Al-Sharrah, Elkamel, and Almanssoor (2010) have formulated sustainability indicators as objectives of a multi-criteria optimization to find the most sustainable solution of a petrochemical process. They converted the multi-objective problem to a single-objective optimization by normalizing the objectives according to their maximum values. Shadiya (2010) proposed an overall sustainability impact for comparing process alternatives which is calculated as a weighted sum of the impact of each category. However, the aggregation of the three kinds of metrics was done using weighting factors. The justification for using those weighting factors presented in the work by Shadiya (2010) was based on the author's preferences. This approach suffers from lack of generality and in addition it is unable to provide sufficient information necessary for final decision-making and screening of process alternatives. Some of the more recent efforts to improve on this field are the works by Kasivisvanathan et al. (2012) and Tay et al. (2011) based on fuzzy optimization approach for process retrofit and synthesis. Gerber, Fazlollahi, and Maréchal (2013) also presented a systematic method for sustainable process design considering an additional objective for environmental concerns.

Most of the existing approaches either require priori-knowledge and experience about the case of study which is not readily available to decision-makers or are limited in terms of sustainability metrics accounted for in decision-making. Therefore, there is still a great demand for development of methods that conveniently enables decision-makers to select the most desirable sustainable process alternative among several options based on decision-maker's preferences and information obtained from the analysis of problems.

This article specifically presents a simple fuzzy-based method to address the challenge of decision-making among several conflicting process alternatives while reducing computational burden and the cognition load on decision-makers. The sustainability metrics proposed by Shadiya (2010) are used to evaluate process alternatives in terms of sustainability. In addition, the overall impact factor approach proposed by Shadiya (2010) and the proposed fuzzy-based decision-making methodology are applied on two process alternatives of ibuprofen production to compare the performance of these methods in selection of the most favourable alternative considering the multitude of sustainability concerns. Ibuprofen process was selected as it presents a case that shows some drawbacks of overall impact factor methodology.

2. Overall methodology

The proposed methodology consists of three main steps (presented in Figure 2); (1) simulation of the process alternatives, (2) evaluation of the process alternatives using the SUSTAINABILITY EVALUATOR and comparison of the processes and (3) selection of a final process design based on the calculated sustainability metrics.

Figure 2 Steps of the proposed methodology.

Commercial process simulators are an integral part of the chemical process industry, as they are used for analysis and design. In this study, Aspen Plus was used to simulate the ibuprofen process rigorously. Aspen Plus is very comprehensive in terms of thermodynamic package compared with most of the other commercialized process simulators. Economic and cost evaluation module available in Aspen Plus was used to evaluate the capital and operating cost of the simulated process. Basic steps of a process simulation in Aspen Plus can be listed as follows:

• chemical component selection,

• selection of thermodynamic models,

• feed stream selection with compositions,

• specification of operating conditions such as temperature and pressure,

• configuration of reactor systems and other processing equipment,

• configuration of separators to separate products and un-reacted raw materials.

Subsequently, the outputs of the simulation are entered in the SUSTAINABILITY EVALUATOR to evaluate the metrics of sustainability in each category. The SUSTAINABILITY EVALUATOR is a Microsoft Excel-based tool which uses mass and energy rates of a process as the inputs to evaluate sustainability of that process. This tool utilizes a set of metrics and indices to indicate economic, environmental, health and safety impacts (Shadiya 2010). Some of the concerns addressed by this tool are as follows.

• Economic concerns: profit, energy costs, waste treatment costs etc.

• Environmental concerns: atmospheric acidification, global warming, environmental burdens, ozone depletion, photochemical smog, resource usage etc.

• Health and safety impact: health and safety risks such as risk of exposure, explosion and flammability.

The input/output structure of the SUSTAINABILITY EVALUATOR is shown in Figure 3. It should be mentioned that the first and second steps can be done using an optimization algorithm integrated with process models and sustainability metrics. Regardless of the method used for generating the alternatives, all alternatives need to be evaluated in terms of sustainability criteria in order to select a final sustainable alternative at the final step of decision-making.

Figure 3 Sustainability evaluator structure.

The final step of this methodology is the screening of process alternatives and selection of a final design alternative. Two strategies are discussed and compared in this study for screening and selecting a final solution. An aggregated fuzzy-based method is introduced and compared with the overall impact factor method.

3. The SUSTAINABILITY EVALUATOR

Assessment of a process in terms of economy is one of the most important tasks of process design, and it is also one of the three main aspects of sustainability. There are several methods for assessing the economy of a process: some of them are presented by Dantus (1999), Seider, Seader, and Lewin (2008) and Turton et al. (2009). In this work, the set of economic metrics used for economic evaluation are as follows:

• Product revenue: It is the measure of the total income produced by selling main products and by-products.

• Raw material costs: This is the end cost of buying and transporting the raw material into the process boundary.

• Waste treatment costs: All the costs imposed for treating and disposing wastes are presented by this metric.

• Operating costs: The cost of all utilities consumed in the process is considered as operating cost.

• Material value added: It is the difference between the price of the products and the raw material (Carvalho, Gani, and Matos 2008).

• Annualized capital costs: This is the total fixed capital cost and working capital cost multiplied by capital recovery factor, CRF.

  (1)
  where n is the nmber of years and i is the interest rate.

• Profit: It is defined as the annual revenue minus all the annualized costs:

  (2)

Another important aspect of sustainability is the environmental aspect including two categories: environmental burden and resource usage. Metrics and indices for the assessment of environmental impacts are presented in IChemE Metrics (2002), GreenMetrics (Constable, Curzons, and Cunningham 2002) and Bridges to Sustainability (Tanzil and Beloff 2006). The metrics suggested for this purpose are global warming, stratospheric ozone depletion, photochemical smog, aquatic oxygen demand, atmospheric acidification, aquatic acidification, eco-toxicity to aquatic life, eutrophication for environmental burden and E-factor, reaction mass efficiency, mass productivity, mass intensity, energy intensity and water intensity for resource usage (Shadiya 2010). These metrics are calculated by multiplying the mass flow rate and potency factor of chemical components. The chemicals causing environmental burden are converted to an equivalent chemical component using potency factors. For example, all components contributing to global warming are converted into CO₂ equivalent using specific potency factors. The list of chemicals and their corresponding potency factors is presented in Shadiya (2010).

Resource usage consists of several metrics each addressing a different type of resource usage. For more information, the reader is referred to the works of Constable, Curzons, and Cunningham (2002) and Tanzil and Beloff (2006). The sub-metrics of this category are listed with their calculation method (Constable, Curzons, and Cunningham 2002; Tanzil and Beloff 2006). A short description of each metrics in this category is as follows:

• E-factor: total waste/mass of product,

• reaction mass efficiency: mass of product/mass of reactants,

• mass productivity: (1/mass intensity) × 100,

• mass intensity: total mass used in a process step/mass of the products,

• energy intensity: energy consumed/mass of product,

• water intensity: water consumed/mass of product.

Except for reaction mass efficiency and mass productivity, the lower value of other environmental metrics indicates better sustainability.

Social aspect of sustainability includes a variety of societal impacts. Many social metrics have been developed each assessing a different impact on societies. For the scope of this work, health and safety metrics of social aspect are selected. Due to the importance of safety on process design and also health impact of processes, these metrics have been studied by many researchers such as Heikkila (1999) and Tugnoli, Santarelli, and Cozzani (2008). They presented quantitative metrics for social impacts. The safety risk metrics and health risk metrics, respectively, developed by Heikkila (1999) and “as of August 14, 2005 Score Card listed on its http://scorecard.goodguide.com/health-effects/” are used for the purpose of this work. These two categories of social impacts, process safety risks and health risks, are explained in the later sections.

Safety metrics used for this work which include heat of main and side reactions index, flammability index, explosivity index, corrosive index, toxic exposure index, temperature index, pressure index, equipment process safety index and process structure safety index were collected from various sources such as American Conference of Governmental Industrial Hygienists (2009) and American Institute of Chemical Engineers (AIChE) by Shadiya (2010). Heikkila (1999) proposed to use a cumulative safety index which is calculated as the sum of all mentioned safety indices. A chemical process with a score of 100 is considered as an extremely unsafe process (Shadiya 2010).

In order to address the health risk associated with processes, the following indices are utilized: carcinogenic health risk, developmental health risk, reproductive health risk, cardiovascular health risk, endocrine system health risk, liver damage health risk, immune system damage health risk, kidney damage health risk, skeletal system damage health risk, neurological damage health risk and respiratory system health risk. The risk imposed due to exposure to toxic chemical components is indicated by an index corresponding to each category. The chemicals suspected as toxicants are embedded in the SUSTAINABILITY EVALUATOR. A specific score for each category of toxic chemicals was suggested by Shadiya (2010). In order to calculate the indices for each chemical, its amount emitted into the environment is multiplied by its corresponding score. The list of toxic chemicals for each index can be found in Shadiya (2010).

4. Process screening and selection methodologies

Sustainability characteristics of the process can be assessed with the previously introduced metrics. More often a process with a better metrics in a specific category has a worse metrics in at least another category. Consequently, considering all sustainability metrics, shown in Table 1, for designing a process is often a complicate multi-criteria decision-making problem. In order to compare process alternatives considering all sustainability metrics, there is a need for a method that properly addresses the relativeness of sustainability concepts and enables assessment of process alternatives compared with one another. The method must be general such that it is not limited to a specific application and must be applicable for screening of any number of process alternatives even with very close metrics.

4.1. Overall impact factor approach

This method was suggested by Shadiya (2010) to calculate an overall sustainability factor which can be used for the comparison of process alternatives in terms of sustainability. According to this method, for each category of sustainability a normalized impact factor is calculated. In order to enable decision-making based on impact factors of several sustainability categories, weighted sum of normalized impact factors are used as the overall sustainability impact factor. The overall sustainability impact factor is calculated using Equation (3). The justification of weights is based on the preferences presented in Shadiya (2010):

(3)

where EI is economic impact, ENVI is environmental impact and SCI is social impact.

This method suffers from several drawbacks. In this method, weighting factors for calculating the overall sustainability impact must be given by decision-makers. Weighting factors can be justified based on the problem specifications or decision-maker's preferences, however; in many practical cases, there is no priori-knowledge to define these factors. The method proposed by Shadiya (2010) results in the same values for the overall sustainability indicators for alternatives with close metrics values. This does not provide sufficient information for comparison of the alternatives. Last, the overall impact calculated using this method implies absolute sustainability which does not exist. In addition, this method implies absoluteness of sustainability in different categories such as economic aspect considering that every profitable process is absolutely sustainable. This prevents a relative comparison between process alternatives that all may be profitable but with different profitability. Next section introduces a novel ranking methodology proposed by the author for relative comparison of process alternatives based on sustainability metrics.

4.2. Aggregated fuzzy rank approach

In order to enable decision-makers to select final, “most sustainable” alternatives out of process alternatives with several conflicting metrics independent of any priori-knowledge, a novel decision-making method is utilized. In this method, decision-makers interactively incorporate their preferences which is defined based on the metric values obtained by the assessment of all process alternatives.

The fuzzy logic concept was adopted in order to incorporate decision-makers' preferences. Based on fuzzy preferences an aggregated rank is assigned to each process alternative. The alternative with the highest rank is selected as the most desirable solution. The values of sustainability metrics obtained by the assessment of process alternatives define an acceptable range on each metric. Minimum and maximum values of sustainability metric are utilized for defining decision-makers' preferences. Depending on the type, either maximum or minimum of metrics could be desirable. For the purpose of explanation, it was assumed that the minimum values of metrics are desirable.

After presenting the range of a particular sustainability metric to decision-makers, they are asked to define their preferred range for each metric. However, there is no need to define a compromised or preferred range meaning that the preferred range can be the same as the range of objective values. Preferred range of each metric can also be considered as a compromised range since its limits may fall within the limits of that particular metric obtained from the evaluation of process alternatives. In order to select one single final solution, decision-makers sometimes must compromise on the minimum value of some metrics by selecting a compromised minimum value within the limits of those metric values. Therefore, the lower limit of a compromised range for a particular metric is the value that decision-makers can accept as the compromised minimum of that metric in order to find a solution which meets their desired criteria in terms of the other metrics. The upper limit of a compromised range is the maximum value of a particular metric that can be accepted for that metric. Thus, any value less than the specified upper limit is acceptable and has a rank according to decision-makers' preferred (compromised) range of a particular metric. After obtaining the compromised range for all metrics from decision-makers, these values are used for fuzzification of the metric ranges and ranking each process alternative. The value of each metric corresponding to a particular alternative, i, is normalized as:

(4)

In order to define the rank (membership) for each process alternative, a strictly monotonic membership function, μ_i,k, is used:

(5)

(6)

where μ_i,k is the membership function for alternative i in terms of metric k, is the compromised upper limit on metric k, is the compromised lower limit of metric k and f_i,k is the value of metric k for alternative i. A form of the membership function based on Equation (6) is shown in Figure 4. It should be mentioned that using factors a, b and c in Equation (6); the form of membership function can be adjusted to asymptotically approach to one or zero so as to allow slow change in the membership function close to one of the objective function extremes.

Figure 4 Membership function versus metric values in the range of [0, 1].

However, it is also possible not to limit the upper bound of membership function to one because if the metric value of a solution is less than the compromised value, it is more desirable. Thus, it is maybe preferable in some cases to assign a higher rank to it.

An aggregated rank, R_i, is assigned for each process alternative. Indeed, the aggregated rank indicates the intersection of all fuzzy memberships for a particular solution. Two methods for calculating aggregated fuzzy ranks, presented in the textbook of Sakawa (1993), are as follows:

• Convex Aggregated Fuzzy Rank:

  (7)
  where μ_i,k and α_k are, respectively, the membership values of alternative in terms of objective function k and weighting factor for objective k.

  This method is used to find the most desirable alternative according to decision-makers' preferences. In this work, the lower limit of the membership function is not limited to zero and can become negative. However, the fuzzy rank for all alternatives with a lower metric value than the lower compromised metric limit is restricted to one. This helps in the selection of alternatives for which their metrics are closer to the boundaries of compromised ranges even though some of them are outside the compromised ranges and some are better than lower compromised limits. Convex Aggregated Fuzzy Rank is more informative in the cases that there is no alternative with all metrics values less than the corresponding upper limits of compromised ranges. This situation may arise when there are many process alternatives. However, in that case, Convex Aggregated Fuzzy Rank method does not strictly ensure selection of the alternatives with all metrics exactly within the compromised ranges.

  In order to strictly select alternatives with all metrics within the compromised ranges, product aggregated fuzzy rank can be used.

• Product Aggregated Fuzzy Rank:

  (8)
  where μ_i,k is the membership value of solution i in terms of objective function k.

The product fuzzy method is used for the strict selection of alternatives for which all of their metric values are in the ranges specified by decision-makers. If only one metric of a solution is greater than the upper limit of its corresponding compromised range, the fuzzy product becomes zero due to the limitation on the upper bound for the membership function presented by Equation (5). Both methods will find exactly a similar solution in the cases where all objectives of all alternatives fall in the specified compromised ranges.

The advantages of the fuzzy approach over using decision weights are as follows:

• The fuzzy approach is independent of decision weights which are biased and based on decision-makers' concerns. Decision weights do not incorporate the information available about relative sustainability of design alternatives. Decision weight approach may not result in the selection of the most sustainable alternative.

• The fuzzy approach does not require involvement of decision-makers' preferences to select the most sustainable solutions.

• The fuzzy approach also gives the flexibility for the incorporation of decision-makers' preferences by either defining preferred ranges or applying weighting factors in the calculation of Convex Aggregated Fuzzy Rank.

• The fuzzy approach incorporates the sustainability information of design alternatives and conserves the relativeness of their sustainability even if decision-maker's preferences are used by applying preferred ranges or weighting factors for calculation of Convex Aggregated Fuzzy Rank.

• Non-linearity of the membership function has an advantage as it can be adjusted to asymptotically approach to one or zero so as to allow slow change in the membership function close to one of objective function extremes and bias the membership value towards one of the extremes.

5. Case study

Ibuprofen is identified as a core analgesic by the World Health Organization (WHO) and, along with other painkillers, it accounts for a large portion of the $500 Billion worldwide pharmaceutical industry (as of 2012, WHO listed on its website www.who.int/about/copyright/en/). Due to the importance and large demand for this product, sustainability and efficiency of its production process has been of interest. In this study, the proposed methodology is applied to evaluate two process alternatives for the production of ibuprofen: semi-batch process and continuous process.

5.1. Brief description of ibuprofen semi-batch process

Typically, ibuprofen is produced using a semi-batch process based on the main reaction presented by Equation (9):

(9)

In this process, isobutyl benzene and acetic anhydride, liquid raw materials, are loaded in a batch reactor and gas raw materials, hydrogen and carbon monoxide, are continuously fed to the reactor to maintain a pressure of 1.4 e7 Pa. Both streams are fed as an equimolar mixture to the reactor. The desirable conversion for this reaction is 90% and the operating temperature is kept constant at 410 K. The raw materials are heated and pressurized using compressors and heat exchangers before feeding to a 2 m³ semi-batch reactor. The temperature of the reactor is also maintained constant using a heating jacket. The kinetic of the reaction for reactor design is modelled approximately using the rate model presented in Equation (10). Wilson activity model was selected for thermodynamic property calculation. The following estimated kinetic rate is used for the reactor design of this process.

(10)

where k = 0.349 m¹²/mol³ s.

After the required time to convert 90% of the key component, isobutyl benzene, the contents of the reactor are cooled to 313 K and sent to the separation units to separate the produced ibuprofen with desirable purity. The separation of ibuprofen is carried out in a crystallization unit to crystallize and separate 95% of the produced ibuprofen from the reactor output mixture. The process was simulated to calculate the time required for the reaction, component distribution and mass and energy flow rates of the streams.

Figure 5 shows the process flow diagram of the process. The key information of the streams obtained from the simulation is presented in Table 3. The reaction time was calculated as 89.8 min, and 10 min is considered for reloading and preparing the reactor for the next batch. The information required by the SUSTAINABILITY EVALUATOR was extracted and entered into the evaluator for further analysis.

Figure 5 Process flow diagram of the ibuprofen semi-batch process simulated in Aspen Plus.

Table 3 Key information of process streams of the semi-batch process.

	GAS1	GAS2	LIQ1	LIQ2	6	7	Product
Temperature (K)	298.15	298.15	298.15	298.15	410.01	313.15	313.15
Pressure (atm)	123.37	148.04	1.00	1.00	138.17	1.50	1.50
Vapour (fraction)	1.00	1.00	0.00	0.00	0.07	0.10	0.00
Mole flow (kmol/h)	4.80	4.80	425.46	425.46	10.54	10.54	4.05
Mass flow (kg/h)	134.34	9.67	43435.02	57105.57	1263.52	1263.52	835.57
Mass flow: vapour (kg/h)	134.34	9.67			8.75	16.76	0.00
Mass flow: liquid (kg/h)			43435.02	57105.57	1254.77	1246.76	0.00
Enthalpy (Gcal/h)	− 0.13	0.00	− 63.11	− 7.01	− 1.04	− 1.09	− 0.51
Mole flow (kmol/h)
Isobutyl benzene	0.00	0.00	0.00	425.46	0.47	0.47	0.00
Acetic anhydride	0.00	0.00	425.46	0.00	0.47	0.47	0.00
Carbon monoxide	4.80	0.00	0.00	0.00	0.53	0.53	0.00
Hydrogen	0.00	4.80	0.00	0.00	0.53	0.53	0.00
Ibuprofen	0.00	0.00	0.00	0.00	4.26	4.26	4.05
Acetic acid	0.00	0.00	0.00	0.00	4.26	4.26	0.00

The data from Aspen Plus were imported to Aspen Economic Evaluation module to calculate the capital cost and operating cost of the process using the given prices in Table 4. The values for capital cost and operating cost, shown in Table 5, were then entered to the SUSTAINABILITY EVALUATOR to evaluate the economical metrics.

Table 4 Prices used in economic calculations.

Isobutyl benzene ($/kg)	132
Acetic anhydride ($/kg)	1.1
CO ($/l)	4.55
H2 ($/l)	3.46
Ibuprofen ($/kg)	320
Waste treatment ($/kg)	0.026
Cooling water ($/m^3)	0.03
Electricity (kWh)	0.05

Table 5 Economical inputs calculated by Aspen for semi-batch process.

Inputs for economic metrics	Value	Units
Enter utility costs from Aspen	9.39E+04	$/yr
Enter capital costs from Aspen Plus	2.85E+06	$

5.2. Brief description of ibuprofen continuous process

The alternative design path involves a novel approach developed by Bogdan et al. (2009). Using their methods, a set of micro-tubular reactors are used for the synthesis of ibuprofen from the same raw material presented in Section 5.1. The set of reactors was optimized to reach 90% conversion and produce the same amount of ibuprofen per year to first meet the market demand and second to enable comparisons between process alternatives. The micro-reactor consists of 120 parallel tubes with a length of 1.5 m and diameter of 50 mm. The operating condition of the reactor for the continuous process is exactly the same as semi-batch process. The raw materials are pre-treated and fed to the reactor as shown in the flow diagram for this process, presented in Figure 6. The continuous process was simulated to obtain all mass and energy balances. The results are presented in Table 6. The information from Aspen Plus simulation was imported to Aspen process evaluator to calculate capital and operating costs presented in Table 7. The required inputs were then entered into SUSTAINABILITY EVALUATOR where all the 41 sustainability metrics were evaluated for further analysis.

Figure 6 Process flow diagram of the ibuprofen continuous process simulated in Aspen Plus.

Table 6 Key information of process streams of the ibuprofen continuous process.

	LIQ1	LIQ2	GAS1	GAS2	6	7	Product
Temperature (K)	298.15	298.15	298.15	298.15	409.39	313.15	313.15
Pressure (atm)	1.00	1.00	123.37	148.04	138.17	1.50	1.50
Vapour (fraction)	0.00	0.00	1.00	1.00	0.07	0.10	0.00
Mole flow (kmol/h)	4.80	4.80	4.80	4.80	10.68	10.68	4.04
Mass flow (kg/h)	490.03	644.26	134.34	9.67	1278.30	1278.30	834.34
Mass flow: vapour (kg/h)			134.34	9.67	8.83	16.96
Mass flow: liquid (kg/h)	490.03	644.26			1269.47	1261.34	834.34
Enthalpy (Gcal/h)	− 2.83	− 0.31	− 0.50	0.00	− 4.18	− 4.36	− 2.01
Mole flow (kmol/h)
Isobutyl benzene	0.00	4.80	0.00	0.00	0.54	0.54	0.00
Acetic anhydride	4.80	0.00	0.00	0.00	0.54	0.54	0.00
Carbon monoxide	0.00	0.00	4.80	0.00	0.54	0.54	0.00
Hydrogen	0.00	0.00	0.00	4.80	0.54	0.54	0.00
Ibuprofen	0.00	0.00	0.00	0.00	4.26	4.26	4.04
Acetic acid	0.00	0.00	0.00	0.00	4.26	4.26	0.00

Table 7 Economical inputs calculated by Aspen for continuous process.

Inputs for economic metrics	Value	Units
Enter utility costs from Aspen	2.29E+05	$/yr
Enter capital costs from Aspen Plus	2.46E+05	$

For both processes, only the amount of undesirable materials sent beyond the battery of the process was considered for the evaluation of environmental and social metrics since the rest are considered to be either desirable products or to be dealt appropriately within the process.

6. Results

SUSTAINABILITY EVALUATOR was applied on both the alternatives for ibuprofen processing to evaluate all the sustainability metrics for this process. The flow rates and economic information were entered in the evaluator and the outputs were calculated. Calculated metrics for both processes are presented in Tables 8910 for comparison. It should be mentioned that the total hours of production for continuous process was considered as 8150 h/year and for the semi-batch process considering the cycle time and reloading time the total hours of production was calculated as 7333 h/year.

Table 8 Economic metrics calculated by the SUSTAINABILITY EVALUATOR.

OUTPUTS for economic evaluation
Metric	Semi-batch process	Continuous process
Revenue	$1,960,813,456	$2,175,948,572
Operating costs	$93,851	$228,542
Waste treatment costs	$81,596	$94,075
Raw material costs	$978,986,863	$755,102,147
Capital costs	$2,845,999	$246,469
Annualized capital cost	$334,405	$28,960
Material value added	$981,826,594	$1,420,846,424
Profit	$981,316,742	$1,420,494,846
Overall normalized impact factor	0.0	0.0

Table 9 Environmental burden metrics calculated by the SUSTAINABILITY EVALUATOR.

OUTPUTS for environmental burden evaluation
Impact category	Impact value for the semi-batch process (tonnes/year)	Impact value for the continuous process (tonnes/year)	Chemicals present
Global warming	2.6E+04	3.0E+04	Isomers of butylbenzene, acetic acid, carbon monoxide
Photochemical smog formation	3.0E+02	3.3E+02	Acetic acid, carbon monoxide
Aquatic acidification	3.8E+01	4.2E+01	Acetic acid
Aquatic oxygen demand	2.0E+03	2.2E+03	Acetic acid

Table 10 Resource usage metrics calculated by the SUSTAINABILITY EVALUATOR.

OUTPUTS for resource usage evaluation
	Metric value for the semi-batch process	Metric value for the continuous process	Units
Effective mass yield	50%	50%	%
E-factor	0.5	0.5	kg/kg
Atom economy	71%	71%	%
Mass intensity	2.18	2.18	kg/kg
Mass productivity	46%	46%	%
Reaction mass efficiency	46%	46%	%
Material intensity	1.2	1.2	kg/kg
Energy intensity/fossil fuel usage	0.19763	0.19763	kW/kg
Water intensity	0.0	0.0	kg/kg

As shown in Table 8, both process alternatives are lucrative and therefore are economically promising. However, as sustainability is a relative concept, the question is which process is to be chosen as a more sustainable alternative. The economic impact calculated using the SUSTAINABILITY EVALUATOR is 0.0 for both alternatives which indicates that both processes are perfectly economically sustainable. The calculated overall impact factor provides no information to compare the alternatives in terms of economic aspect of sustainability and implies an absolute sustainability for both alternatives.

Environmental burden metrics were also calculated using this tool by entering the amount of harmful material emitted into the environment. The contribution of materials in ibuprofen processing to each environmental category is presented in Table 9.

Resource usage metrics, as one of the subcategories of environmental impacts, were evaluated for these processes as shown in Table 10. The overall environmental impact factor calculated by the SUSTAINABILITY EVALUATOR for both alternatives is 0.15 that shows the same sustainability impact for both processes although they have different metrics. Again, the overall impact factor for the environmental burden aspect cannot be used for the screening of processes.

Health evaluation metrics and safety metrics as two subcategories of social aspect of sustainability also were calculated and are presented in Tables 11 and 12. The chemicals present in this process contributing to each health impact category are presented in Table 11. Safety metrics are presented in Table 12 with the percentage of each metrics' maximum possible value. The lesser is the percentage, the safer is the process. The same overall social impact of 0.31 was calculated for both processes.

Table 11 Health evaluation metrics calculated by the SUSTAINABILITY EVALUATOR.

OUTPUTS for health evaluation
Metrics	Semi-batch process	Continuous process	Units	Chemicals
Developmental damage	1.1E+02	1.2E+02	Tons/yr	Carbon monoxide
Reproductive system damage	6.6E+01	7.4E+01	Tons/yr	Carbon monoxide
Respiratory system damage	1.4E+03	1.6E+03	Tons/yr	Acetic anhydride, acetic acid, hydrogen, carbon monoxide
Cardiovascular system damage	1.2E+03	1.3E+03	Tons/yr	Acetic acid, carbon monoxide
Liver damage	1.1E+03	1.3E+03	Tons/yr	Acetic acid
Nervous system damage	6.6E+01	7.4E+01	Tons/yr	Carbon monoxide
Sensory system damage	1.1E+03	1.3E+03	Tons/yr	Acetic anhydride, acetic acid

Table 12 Safety evaluation metrics calculated by the SUSTAINABILITY EVALUATOR.

OUTPUTS for process safety evaluation
	Results for the semi-batch process	Results for the continuous process	% of maximum value
Heat of main reaction index	2	2	25
Heat of side reaction index	0	0	0
Flammability index	4	4	50
Explosiveness index	6	6	75
Toxic exposure index	20	20	67
Corrosiveness index	2	2	50
Temperature index	6	6	75
Pressure index	2	2	25
Equipment safety index	6	6	75
Safety level of process structure index	4	4	40

Table 13 summarizes the impact factors using the method proposed by Shadiya (2010) based on the sustainability metrics calculated by the SUSTAINABILITY EVALUATOR. As both processes are profitable, same impact factor is calculated for both alternatives. There are some waste streams leading to environmental burdens; therefore, there are environmental impacts associated with the process alternatives. The overall impact for environmental burden is 0.15. It should be mentioned that the environmental impacts are affected by the size of plant as the amount of its waste stream increases. The safety index for the process is 52, and also there are several health concerns associated with these processes which results in an overall social impact of 0.31. The overall sustainability impact of both process alternatives is 0.19 according to the overall impact method. This case study demonstrates a case in which overall impact methodology fails to provide sufficient information for the comparison of processes although there are differences in their sustainability metrics. According to the overall impact factor method, both processes are absolutely sustainable and one cannot distinguish them in terms of sustainability.

Table 13 Summary of sustainability impacts for ibuprofen process alternatives.

Summary of results	Semi-batch	Continuous
Profit	$981,316,742	$1,420,494,846
Economic impact	0.0	0.0
Environmental concerns	Global warming, stratospheric ozone depletion, photochemical smog formation, aquatic acidification, aquatic oxygen demand	Global warming, stratospheric ozone depletion, photochemical smog formation, aquatic acidification, aquatic oxygen demand
Environmental impact	0.15	0.15
Safety index	52	52
Health concerns	Developmental damage, reproductive system damage, respiratory system damage, cardiovascular system damage, liver damage, nervous system damage, sensory system damage	Developmental damage, reproductive system damage, respiratory system damage, cardiovascular system damage, liver damage, nervous system damage, sensory system damage
Social impact	0.31	0.31
Overall sustainability impact	0.19	0.19

Convex Aggregated Fuzzy Rank approach with no weighting factor was applied for selecting the best alternative. In this case, the metric values of the only two alternatives, presented in Tables 89101112, define the upper and lower limits. Thus, compromised ranges become the same as the whole range for all metrics and all metrics of both alternatives fall within the compromised ranges.

Table 14 presents the aggregated ranks calculated for each alternative. Although there are very small differences in sustainability metrics of the two alternatives, aggregated fuzzy rank approach calculates different ranks for each alternative which has a better overall sustainability impact. Decision-makers may incorporate their preferences by assigning different ranges than those used for this study to overlook the importance of one alternative over the other. For example, global warming metric of semi-batch process is 2.6E+04 which is better than that of the continuous process. However, one can favour the continuous alternative by defining both lower and upper limits on this metrics, same as the global warming metric of the continuous process. This could be the case in process screening involved with more alternatives and a range of values for each metric.

Table 14 Aggregated fuzzy rank calculated for each alternative.

	Semi–batch	Continuous
Aggregated fuzzy rank	5	1

Although the continuous process is slightly more profitable, the semi-batch process is better in terms of other sustainability metrics such as environmental metrics which results in a higher rank for semi-batch process. It should be mentioned that in order to show generality and independency of the fuzzy rank method from preferences, the calculated ranks in this study are unbiased ranks and no preferences were incorporated for the final decision-making. One can readily use compromised limits on the metrics for membership calculations or incorporate weighting factors for the calculation of final ranks using Equation (7).

7. Conclusion

As more alternatives and technologies are introduced for chemical processes, there is a demand for a systematic methodology to enable engineers to screen process alternatives and select the most desirable process. Due to the importance of sustainability concerns, process screening is often involved with many metrics considered as process design criteria. The methodology proposed in this work enables screening of process alternatives without a need for assuming any decision weights and reduces the recognition load on decision-makers imposed by a multitude of objectives while conserving the relativity of sustainability concept. This method is based on three main steps: design and generation of process alternatives, evaluation of process alternatives using a sustainability tool and finally comparison of all alternatives using a fuzzy-based approach to select the most sustainable process.

This study also discussed the drawbacks of the overall impact method for selection and comparison of process alternatives. Both overall impact method and the proposed fuzzy-based method were applied on a case study to demonstrate capabilities and generality of the methods. The presented case study is an example of cases for which the overall impact factor method cannot provide sufficient information for comparison of alternatives. In addition, as mentioned previously the overall impact method is dependent on weighting factors requiring justification for every application. However, application of aggregated fuzzy rank approach demonstrated applicability of this method for comparison of any number of process alternatives with a multitude of sustainability metrics. This method is universal as it can be applied on any application only based on the value of metrics evaluated for process alternatives using any sustainability evaluator tool. While this method accounts for the relativity of sustainability concerns, decision-makers can incorporate their preferences by defining new desired ranges for the metrics based on the values of metrics of the evaluated process alternatives.

The significance of this methodology is that it enables engineers to incorporate economic, environmental and social concerns of sustainability into early stages of chemical process design and decision-making. This methodology involves the use of the SUSTAINABILITY EVALUATOR, a newly developed impact assessment tool, and a screening method based on fuzzy ranking which was introduced in this study. Application of this methodology on two process alternatives of ibuprofen production, which were very similar in terms of overall sustainability impact, proved the applicability of the aggregated fuzzy ranking method in process design decision-making problems. Semi-batch process was chosen as the more sustainable alternative without considering any bias towards a specific metric.

Notes

^1. Email: [email protected]

^2. Email: [email protected]