Computer Aided Design for Sustainable Industrial Processes

Computer Aided Design for Sustainable Industrial Processes: Specific Tools and Applications Maurizio Fermeglia ... There are several advantages derivin...

0 downloads 114 Views 391KB Size
Computer Aided Design for Sustainable Industrial Processes: Specific Tools and Applications Maurizio Fermeglia Molecular Simulation Engineering Laboratory, Department of Chemical, Environmental and Raw Material Engineering, University of Trieste, Piazzale Europa 1, Trieste 34127, Italy

Gennaro Longo and Letitia Toma International Centre for Science and High Technology, ICS-UNIDO, AREA Science Park, Trieste 34012, Italy DOI 10.1002/aic.11730 Published online March 11, 2009 in Wiley InterScience (www.interscience.wiley.com).

Chemical Process Sustainability can be estimated using different sustainability indicators. The quantitative estimation of those indicators is necessary (i) for evaluating the environmental impact of a chemical process and (ii) for choosing the best design among different available alternatives. To accomplish these goals, the computerized calculation of sustainability indicators requires the use of at least three computer tools: (i) process simulation, (ii) molecular modeling and a (iii) sustainability indicators software code. In this work, a complete software platform, Process Sustainability Prediction Framework, integrated with process simulation programs, which support the CAPE-OPEN interfaces, is presented and discussed. The article contains also description and application of molecular modeling techniques to estimate different toxicological data, which are used in the calculation of sustainability indicators. A representative example of one chemical process and thermo-physical properties used in the toxicological data calculation, are reported to demonstrate the applicability of the software to C 2009 American Institute of Chemical Engineers AIChE J, 55: 1065–1078, 2009 real cases. V Keywords: sustainability, process simulation, molecular modeling, toxicity evaluation, CAPE-OPEN, WAR algorithm

Introduction Sustainable development has been defined as the creation of goods and services using processes and systems that are non-polluting; conserving energy and natural resources; economically viable; safe and healthy for employees, communities and consumers; socially and creatively rewarding for all working people.1 Correspondence concerning this article should be addressed to M. Fermeglia at [email protected] C 2009 American Institute of Chemical Engineers V

AIChE Journal

In the sustainable development context, an important role is played by chemical engineering, basically for two main reasons: (i) the production of chemical process plants contributes greatly to national income and is absolutely essential to modern society; thus, the society development depends on the development of the chemical process industry and vice versa; (ii) many critical environmental problems can be associated directly with chemical processes or to the use of chemical products and by-products that result in waste streams to soil, water, and air.2 Chemical industry produces materials for multiple consumer markets, which need to be produced, used and recycled by manufacturing processes that

April 2009 Vol. 55, No. 4

1065

Figure 1. Potential Chemical Risk Evaluation.

are clean, safe and economical.3 To eliminate or reduce the environmental impact, the environmental performance of chemical processes should be identified and quantified at an early stage of process design. Taking into account the environmental aspects at design time, an answer to ‘‘which design is more environmentally friendly, and implicitly more sustainable, for a specific chemical process?’’ can be provided. The present work deals with the development, implementation and application of sustainable industrial development concept in the design stage of a chemical plant. In particular, the development and implementation of a specific software tool, Process Sustainability Prediction (PSP) Framework, is presented. The framework is intended to be used in the design stage of a chemical process, for selecting the best solution in terms of environmental impact, among different design alternatives. The PSP software framework developed in this work contains also a toxicological data estimation methodology, using molecular modeling techniques, and the implementation of a toxicological database. Applications of the PSP Framework are discussed and an example is presented. The aforementioned listed issues are discussed in detail in the next sections. There are several advantages deriving from the application of the PSP Framework. The most important advantage is the possibility of including the potential environmental impact criteria in the optimization of a chemical process and not only the technical and economical criteria. Modern tools such as molecular modeling and process simulation, both used in the present work, are of crucial importance because they can evaluate possible technical solutions faster and with much lower costs then experimental activity. Benefits from the coupling of molecular modeling with process simulation and design include but are not limited to: (i) shortened product-process development cycles, (ii) optimization of existing processes to improve energy efficiency and minimize production of waste, (iii) efficient design of new products and processes; and (iv) improvements in health, safety, and environment.4

Sustainability Evaluation Economy, society and environment are three basic elements of sustainability. Each of these elements is connected to and dependent upon the others.5 It is generally acknowledged that sustainability results from a balance among the 1066

DOI 10.1002/aic

three aspects. The selection of an adequate set of indicators for sustainability evaluation is very important, for comparative analysis between different versions of a process.6 With the objective of providing an easy-to-implement method for applying indicators for the purpose of analyzing industrial systems for sustainability, Sikdar6 proposed a typology of indicators, considering the three dimensions of sustainability in three distinct hierarchical groups: (i) one dimensional (1D) indicators, which provide information about only one dimension of sustainability: economical, ecological, or social; (ii) two dimensional (2D) indicators, which provide information simultaneously about two dimensions of sustainability: socio-ecological, socio-economical, or economicecological; (iii) three dimensional (3D) indicators, which provide information about all three dimensions of sustainability.7 Four 3D and four 1D indicators have been studied, implemented and analyzed in the present work. A brief description of those is presented later to understand the significance of each factor and their relation to process data on one side and to toxicological data on the other side.

Three dimensional indicators The Material Intensity (MI) Indicator measures the amount of non-renewable resources required to obtain a unit mass of products. It includes raw materials, solvents, and other ingredients.7 Details regarding the calculation of this indicator are given by Schwarz et al.8 The Energy Intensity (EI) Indicator measures the energy demands of the process. It is calculated per unit mass of products and focuses primarily on the use of non-renewable energy.7 Details about the calculation of this indicator are given by Tanzil et al.9 The Potential Chemical Risk Evaluation (PCRE) Indicator regards the risk associated with the manipulation, storage and use of hazardous chemical substances in the process. It is schematically represented in Figure 1. The chemicals used in the process under study are classified depending on their relative quantity, frequency of use, danger characteristics and potential exposure. Thus, for each chemical, the designated ‘‘frequency class,’’ ‘‘quantity class,’’ ‘‘hazard class,’’ and ‘‘potential exposure class’’ are determined according to Vincent et al.10 The Potential Environmental Impact (PEI) Indicator is due to the emissions and discharge of hazardous chemicals to the environment. It is similar to the one for the potential

Published on behalf of the AIChE

April 2009 Vol. 55, No. 4

AIChE Journal

Figure 2. Potential Environmental Impact Evaluation.

chemical risk, as represented in Figure 2. The chemicals are classified according to their relative quantity and hazard characteristics. These lead to the calculation of the total potential environmental impact. Furthermore, the potential environmental impact of interest is dependent on the transfer to a receiver medium, such as air, water or soil, which is governed by physicochemical characteristics and transfer coefficients as shown in Figure 2.7

One dimensional indicator: waste reduction algorithm The Waste Reduction (WAR) Algorithm, based on the concept of environmental impact balance, analogous to mass or energy balances, uses data from the process (stream flowrates and mass fractions) and toxicological data to calculate the environmental impact of a chemical plant.11–13 According to the algorithm, toxicological data are grouped in different environmental impact categories. Eight environmental impact categories: Global Warming Potential (GWP), Acidification Potential (AP), Ozone Depletion Potential (ODP), Photochemical Oxidation or Smog Formation Potential (PCOP), Human Toxicity Potential by Ingestion (HTPI), Human Toxicity Potential by Inhalation or Dermal Exposure (HTPE), Aquatic Toxicity Potential (ATP), Terrestrial Toxicity Potential (TTP) are used in evaluation of potential environmental impact using the WAR Algorithm.12 A brief description of each category is presented later. GWP is determined by comparing the extent to which a unit mass of a chemical absorbs infrared radiation over its atmospheric lifetime to the extent the CO2 absorbs infrared radiation over its respective lifetimes. AP is determined by comparing the rate of release of Hþ in the atmosphere as promoted by a chemical to the rate of release of Hþ in the atmosphere as promoted by SO2. ODP is determined by comparing the rate at which a unit mass of chemical reacts with ozone to form molecular oxygen to the rate at which a unit mass of CFC-11 (trichlorofluoromethane) reacts with ozone to form molecular oxygen. PCOP is determined by comparing the rate at which a unit mass of chemical reacts with a hydroxyl radical OH to the rate at which a unit mass of ethylene reacts with OH. HTPI is calculated for a chemical if it existed as a liquid or solid at a temperature of 0 C AIChE Journal

April 2009 Vol. 55, No. 4

and atmospheric pressure. As a first approximation, the lethal-dose that produced death in 50% of rats by oral ingestion (LD50) has been used as an estimate for the HTPI. HTPE was determined for that chemical if it existed as a gas a temperature of 0 C and atmospheric pressure. ATP was estimated by using toxicological data for a single, representative species of fish, Pimephales Promelas (Fathead Minnows). The data for this assay comes in the form of a LC50, a lethal concentration which causes death in 50% of the test specimens.11–13 Different molecular modeling methods will be presented for the evaluation of the thermo-physical properties which lead to environmental impact categories. The four 1D environmental impact indexes used in WAR Algorithm and in the PSP Framework are: the total rate of ðtÞ impact output—I_out , the total impact output per mass of ðtÞ ðtÞ ^ products—Iout , the total rate of impact generation—I_gen and ðtÞ the total impact generated per mass of product—I^gen . Those indexes have been defined by Cabezas and Young11–13 and are referred in the PSP Framework as Iout, Iout_mp, Igen and Igen_mp. The comparison of the aforementioned presented indexes permits to choose the most environmentally friendly design between different process alternatives.

PSP framework: design and development The PSP Framework implementation has been made according to Martins and Vincent for 3D Indicators7,10 and to Young and Cabezas for 1D Indicators.11–13 A general schema illustrating the main steps of the development is given in Figure 3. The two starting points for calculating the sustainability indicators are the process simulator and the database containing toxicity and risk phrases data. One of the main goals of the present work is to incorporate the toxicology data calculation issue into process simulation. For this aim, toxicological effects of the substances, present in the process flowsheet, and their quantities are processed in special unit operations, implemented using the CAPE OPEN (CO) methodology. The choice of the CO technology has been made according to a precise goal: to develop a module able to interact with all the process simulator software programs which support the CO interfaces. Once the module is created it is ready to be used anywhere, being independent of the

Published on behalf of the AIChE

DOI 10.1002/aic

1067

Figure 3. General schema of the PSP Framework.

process simulator software used. This is one of the greatest advantages offered by the CO technology. Four CO Modules have been developed (two CO Modules for 3D Indicators and two CO Modules for 1D Indicators). These modules are inserted in the process flow-sheet, as any other unit operation (e.g. a heat exchanger, a reactor, a pump, etc.), with the aim of extracting all the necessary data requested for further calculations (flow-rate quantities, flow-rate composition, energy demand of the process, toxicological data, risk phrases data). It is important to mention that the two CO Modules used in the 1D calculation are directly linked with the toxicological database (TDB), containing the environmental impact categories, and one of the CO Module used in the 3D calculation is linked with the same toxicological database which contains the risk-phrases (R-Phrases) for several chemical compounds. The modules have been tested successfully using different process simulators such as: PROII, Aspen Plus, 1068

DOI 10.1002/aic

COCO/COFE. The connection between the CO Modules and the database is done using ActiveX Data Objects (ADO) technology. The environmental impact categories and the RPhrases are obtained from the TDB as a function of the Chemical Abstract Service (CAS) Number. The TDB contains about 4900 compounds. The process data and toxicological data obtained through the CO Modules are processed in the PSP software. Other information such as H2O, O2, N2 quantities formed in the reaction; energy demand of the process; frequency of use and physical state of each substance involved in the process should be supplied by the user on the PSP graphical user interface. The software has been implemented in Microsoft Visual Basic (version 6.0) and generates the 3D and 1D sustainability indicators. After the calculation, the indicators will be used in a comparative manner to choose the best environmental design between different design alternatives.

Published on behalf of the AIChE

April 2009 Vol. 55, No. 4

AIChE Journal

Figure 4. Thermo-physical properties and software programs used in the calculation of the Environmental Impact Categories.

Details regarding the implementation of the CO Modules, the development of the toxicological relationship database, the PSP software and its applications are given by Toma.14

Estimation of Toxicological Properties by Molecular Modeling The rapid increase in the world’s human population and the development of human society have been accompanied by an equally rapid increase in the utilization of chemical compounds and the subsequent deterioration of the environment by the release of these compounds.15 Toxicological data are usually necessary to assess the impact of such compounds on the environment and on human health. However, because of the time and financial resources required, it is impractical to perform toxicity tests for all the chemical compounds in use.16 In this context, molecular simulation methods become an important tool for toxicological data prediction. As it has been mentioned in the previous sections, GWP, ODP, PCOP, AP, HTPE, HTPI, TTP, and ATP are the necessary data on which the WAR Algorithm is based. In view of developing a strategy for the estimation of these eight environmental impact categories from the molecular structure of the compound under investigation, it is interesting to map such indexes on some reference intermeAIChE Journal

April 2009 Vol. 55, No. 4

diate physical properties that could be in turn estimated by some molecular simulation method. Figure 4 summarizes such intermediate physical properties and the relevant methods for their calculation starting from the molecular structure. Each physical property is then mapped to the related environmental impact category (top of the figure) and to the molecular simulation method allowing its calculation (bottom of the figure). The Atmospheric Oxidation Program (AOP), based on Structure Activity Relationship (SAR) calculation, was developed by U.S.EPA. The program has been used in the present work for the estimation of the lifetime of a compound in the atmosphere, for the reaction rate with ozone and reaction rate with hydroxyl. For the calculation of the octanol-water partition coefficient (Kow), different molecular modeling methods can be used. For example KOWWIN, the Log Octanol-Water Partition Coefficient Program, is software developed by U.S.EPA which is based on SAR. ClogP is a program, based on group-contribution methods, which can be also used for the octanol-water partition coefficient. ALOGPS is software based on neural networks developed by the Virtual Computational Chemistry Laboratory. Conductor like Screening Model Thermodynamics (COSMOTherm) is a software tool based on quantum mechanical methods, which calculates

Published on behalf of the AIChE

DOI 10.1002/aic

1069

different thermo-physical properties such as: vapor pressures, boiling points, activity coefficients, excess enthalpy and entropy, Henry’s Law constants, solubility, octanol-water partition coefficient, vapor-liquid equilibrium, liquid-liquid equilibrium, solid-liquid equilibrium, density and viscosity for pure compounds.17 ECOSAR, developed by the Risk Assessment Division of the U.S.EPA, is a computer program based on SAR, used to predict the aquatic toxicity of chemicals based on their similarity of structure to chemicals for which the aquatic toxicity has been previously measured. TOPKAT, a computer program developed by Accelrys, employs robust and cross-validated Quantitative Structure Toxicity Relationship (QSTR) models for estimating 16 toxicity endpoints (e.g. Mutagenicity, Rat Oral LD50, Carcinogencity, Developmental Toxicity, etc.).18 It is evident, form Figure 4, that some properties are of fundamental importance for the calculation of the eight environmental impact categories. As it can be noticed, HTPE, HTPI, ATP and TTP impact categories are related to a phase equilibrium property, namely the Kow. The other four impact categories: GWP, ODP, PCOP and AP are related to transport properties such as the release rate of Hþ or to reactivity properties such as reaction rate with hydroxyl, reaction rate with ozone (lifetime) and therefore can be calculated using quantum mechanics (QM) or (Q)SAR methods. Based on these considerations, the octanol-water partition coefficient, the lifetime of a chemical compound in the atmosphere, the reaction rate with ozone and the reaction rate with hydroxyl are discussed in the next section.

Thermo-physical properties estimation Octanol-Water Partition Coefficient. Numerous properties, including bioavailability, skin permeability, toxicity and environmental fate, have all been successfully related to the differential solubility of solutes in aqueous and organic solvents.19 Octanol-water partition coefficient is one of the most effective parameters used in the toxicity evaluation of four Environmental Impact Categories: HTPI, HTPE, ATP, TTP (see Figure 4). The Kow is defined as the ratio of the concentration the component of interest has in octanol phase to its concentration in the aqueous phase in a two-phase octanol-water system.20 In recent years, the Kow has become a key parameter in studies of the environmental fate of organic chemicals. It has been found to be related to water solubility, soil/sediment adsorption coefficients, and bio concentration factors for aquatic life. Because of its increasing use in the estimation of these properties, Kow is considered a required property in studies of new or problematic chemicals.16,20 The successfulness of Kow in predicting bioaccumulation and toxicity has been attributed to the fact that the octanolwater partitioning is a good representative of a compound going from more aqueous like-phases (extra cellular phase) to organic-like phases (cellular phase) inside the body, which is the rate controlling step during the interaction of a compound and a biological system.21 Because Kow is a free energy function, it is directly related to the energy of transfer between two phases. There are different methods, available in literature, to calculate Kow using molecular simulations. Some of these methods are: Free Energy Perturbation, Monte Carlo and Molecular Dynamics Method19,21; Free 1070

DOI 10.1002/aic

Energy Calculation—Expanded Ensemble Method21,22; the configurational-bias Monte Carlo (CBMC)22; Continuum Solvation Model (CSM); Quantitative Structure Activity Relationship (QSAR) Method.23–27 The last two approaches have been applied in the present work. Lifetime. To evaluate the environmental impact of a given gas molecule, one needs to know, first, how long it remains in the atmosphere.28 The environmental damage caused by atmospheric pollutants is proportional to the duration of their effects. The global impacts of greenhouse gases, as measured by global warming potential, and ozone depleting substances, as measured by ozone depletion potential, have traditionally been calculated using the atmospheric lifetime of the source gas as a quantitative measure of the impact’s duration, assuming that the gas quickly reaches a steady state pattern which decays exponentially according to the lifetime.29 Consequently, lifetime is another important parameter used in the calculation of two environmental impact categories: GWP and ODP (see Figure 4). Lifetime of chemical compounds in the atmosphere can be estimated using Structure Activity Relationship (SAR) methods. Reaction Rate with Ozone. Ozone plays a central role in tropospheric chemistry. Not only is it a highly reactive and toxic species, but it absorbs both infrared and ultraviolet light, contributing to the ‘‘greenhouse effect’’ and providing protection from exposure to damaging UV.30 The reaction with ozone represents one of the most common reactions which take place in the atmosphere and the reaction rate with ozone is used to calculate the ODP environmental impact category. It should be also specified that ozone concentrations are very difficult to model because of the different interactions between pollutants and meteorological variables.31 Reaction rate of chemical compounds with ozone in the atmosphere can be estimated using the QSAR method.32 Reaction Rate with Hydroxyl. Organic chemicals emitted into the troposphere are degraded by several important transformation processes that include reaction with hydroxyl OH radicals or other photo chemically-produced radicals, reaction with ozone or direct photolysis. The dominant transformation process for most compounds that occur in the troposphere is the daylight reaction with OH radicals. The reaction rate with hydroxyl is used to calculate the PCOP environmental impact category. The rates at which organic compounds react with OH radicals are a direct measure of their atmospheric persistence.33 As experimental measurements can be difficult, time-consuming, and expensive, the ability to estimate rate constants has become very important. The estimation methods have become an important tool to reduce the number of compounds for which measurements are necessary.34 Several methods are used to calculate the reaction rate with hydroxyl, such as: Empirical Models; QSAR models (QSAR models with measured physicochemical descriptors and QSAR models with structural descriptors); ab initio molecular orbital calculations. A complete overview of the methods, found in the scientific literature, that calculate the reaction rate with hydroxyl was published by Sablijic and Peijnenburg in the IUPAC Technical Report for 200135 and by Gu¨sten.36 In the present work, the reaction rate with hydroxyl radical has been evaluated using group contribution (GC) and QSAR methods.

Published on behalf of the AIChE

April 2009 Vol. 55, No. 4

AIChE Journal

Results and Discussions on the Estimation of Thermo-Physical Properties Octanol-water partition coefficient Our attention was focused on two methods: QSAR and CSM. The choice of these methods as well as the advantages and disadvantages correspondent to each of them is briefly summarized later. The advantages of QSAR methods are: these methods are very fast, allowing them to be applied to large databases of structures, in many cases they require only 2-D molecular connectivity, and can be very accurate for molecules with welldefined group equivalents. The disadvantages of QSAR methods are that they require large numbers of empirically derived parameters; they cannot be used to examine conformational effects in the solute and, many times, no parameters exist for calculating the logKow for the new chemical classes.19 Another problem with such models is their extremely limited scope. In other words, they perform well, but only for very narrow classes of compounds.24 Direct calculation of octanolwater partition coefficients, using the CSM methods, offers the promise of calculating accurate partition coefficients for a wide variety of molecules with a degree of generality not available from the fragment-based methods.19 The CSM used in this work are based on the original theory of COSMO, based on computational quantum mechanics that allows us to predict different thermo-physical properties without any experimental data. The COSMO method has been implemented in the literature in two models: conductor like screening model-realistic solvation (COSMO-RS) and conductor like screening modelssegment activity coefficient (COSMO-SAC). The two methods are very similar and they predict intermolecular interactions based only on molecular structure and a few adjustable parameters. COSMO-RS is the first extension of dielectric continuum solvation model to liquid-phase thermodynamics, and COSMO-SAC is a variation of COSMO-RS.37 Kow is calculated by the COSMO-RS by considering the activity coefficients for the systems of interest. The theory is based on considering the ideally screened molecules as starting points for the description of molecules in solution. The deviations from ideal screening, which occur in any real solvent, are described as pair wise misfit interactions of the ideal screening charges on parts of the molecules in contact in the fluid. Because COSMO-RS does not depend on experimental data or any parameterization for the solvent, it efficiently enables the calculation of the chemical potential of almost any solute in almost any solvent. Basically, COSMO-RS method assumes that the molecule is made up of small surface elements. The charge density of each of these elements is evaluated by means of quantum mechanics calculations. An ensemble of interacting molecules is replaced by the corresponding ensemble of interacting surface pieces characterized by a screening charge density, namely, the contribution of all of the differences in polarity obtained when the molecules are not isolated. The characteristic distribution of charges obtained by considering all of these surface elements is the sigma profile (r-profile) of the molecule. The sigma profile describes the amount of surface in the ensemble having a screening charge density between r and r þ Dr.38 Sigma profiles depict the surface charge density distribution over the entire molecule. It AIChE Journal

April 2009 Vol. 55, No. 4

has to be specified that each molecule’s profile is unique and the profiles are sensitive to conformation. The sigma profiles provide a rich and detailed quantitative information about the polarity of molecules.17,39 The difference in the r-potentials between the two phases can be considered as the local contribution to the partition coefficient, and the partition coefficient can be visualized as a surface property.40 Many are the advantages of the method. First, because each molecule is characterized by a sigma profile, the method can treat proximity effects, can distinguish among isomers, and can also take into account the result of a conformational analysis.38 The above-presented methods for Kow calculation have been implemented in different software programs: COSMORS method has been implemented in COSMOTherm software and QSAR has been implemented in Estimation Programs Interface (EPI) Suite software developed by EPA. The results of the calculations performed are presented in Table 1. Analyzing the results shown in Table 1, it can be noticed that the logKow relative absolute deviation (RAD) between the experimental and the calculated value, has lower values for nine substances when the COSMO-RS method is used. The nine substances considered are: methanol, phenol, chloroform, toluene, methyl ethyl chloride, benzene, methane, ethane and propane. Also the mean relative absolute deviation (MRAD) has lower value (0.00765 vs. 0.10715) when the COSMO-RS method is used. These results confirm what stated earlier about the reliability of the methods and the good accuracy of the QM-COSMO method.

Lifetime As it has been mentioned earlier, SAR method has been used for the lifetime estimation. The Atmospheric Oxidation Program (AOPWIN) is a computer program, based on SAR, developed by US EPA, used to calculate atmospheric halflives for organic compounds. The program belongs to the EPI suite software. The only data required for AOPWIN to estimate the half-life of a chemical substance is the chemical structure of the compound. The chemical structure is entered into the program as a SMILES notation.41 The procedure has been applied to calculate the lifetime for several substances. Table 2 contains the experimental lifetime values for some representative chemical compounds as well as the results obtained using SAR. The experimental values have been taken from the World Meteorological Organization Global Atmosphere Watch Report.42 By analyzing the results shown in Table 2, it can be noticed that the lifetime relative absolute deviation (RAD) has a variation range of 0.0235–0.4514. The mean relative absolute deviation (MRAD) is 0.04843. Taking into account the complexity of the problem and the limited number of methods, software programs and experimental data available in the literature, the results obtained using the SAR method are considered satisfactory.

Reaction rate with ozone As it has been mentioned earlier, SAR method has been used for the reaction rate with ozone estimation. The AOPWIN has been used to calculate the reaction rate with ozone. It should be specified that the program is able to estimate the reaction rate with ozone only for two classes of

Published on behalf of the AIChE

DOI 10.1002/aic

1071

Table 1. Kow Results Using Different Methods LogKow Calculated

LogKow RAD

Substance Name

Substance Formula

Substance CAS

LogKow (exp.)

Using QSAR method(1) KOWWIN-software

Using CSM Method(2) COSMOTherm software

RAD for (1)

RAD for (2)

Methanol Naphthalene Phenol Chloroform Toluene Anisole Methyl ethyl chloride Benzene Methane Ethane Propane Benzoic acid

CH4O C10H8 C6H6O CHCl3 C7H8 C7H8O CH2Cl2 C6H6 CH4 C2H6 C3H8 C7H6O2

67-56-1 91-20-3 108-95-2 67-66-3 108-88-3 100-66-3 75-09-2 71-43-2 74-82-8 74-84-0 74-98-6 65-85-0

0.77 3.30 1.46 1.97 2.73 2.11 1.25 2.13 1.09 1.81 2.36 1.87

0.63 3.17 1.51 1.52 2.54 2.07 1.34 1.99 0.78 1.32 1.81 1.87

0.7292 3.0939 1.4319 2.1001 2.6549 2.2271 1.2591 2.1043 1.0162 1.6319 2.1782 2.2379

0.1818 0.0393 0.0342 0.2284 0.0696 0.0189 0.0720 0.0657 0.2844 0.2707 0.2330 0

0.0529 0.0625 0.0192 0.0660 0.0275 0.0555 0.0073 0.0121 0.0677 0.0984 0.0770 0.1967 MRAD 0.00765

0.10715

substances: acetylene and olefins. The procedure has been applied to calculate the reaction rate with ozone for several substances. Table 3 contains the experimental reaction rate with ozone values for some representative chemical compounds as well as the results obtained using SAR. The results shown in Table 3, shows also the relative absolute deviation (RAD) of the reaction rate with ozone. As it can be noticed the method is able to calculate the reaction rate for isomers too (for example 2-butene cis and trans, 1-3 dichlorpropene cis and trans). The mean relative absolute deviation (MRAD) is 0.271. The results obtained using the SAR method are reasonable taking into account the complexity of the problem and the limited number of methods, software programs and experimental data available in the literature.

Reaction rate with hydroxyl As it has been mentioned earlier, the reaction rate with hydroxyl radical has been evaluated using GCM and QSAR methods. The QSAR method has been implemented in the Atmospheric Oxidation Program (AOP). The MOOH-GCM method has been implemented in AM1-MOOH code developed by Klamt. The code runs under MOPAC software. The results obtained are presented in Table 4. Analyzing the

results shown in Table 4, it can be noticed that the relative absolute deviation (RAD) between the experimental and the estimated values (see formula) is lower for eight substances using the QSAR method. Investigating the results, we can conclude that the QSAR method is more accurate for the estimation of reaction rate with hydroxyl than the GCM.

Process Sustainability Prediction Framework Application Ten chemical processes have been simulated, using PROII and/or Aspen Plus process simulators, and analyzed using the PSP Framework. The processes can be divided, in the following four categories: (i) processes where one or more parameters have been modified: Acrylic Acid Production Process, Sweetening Natural Gas by DGA Absorption, Formaldehyde Production Process, Ethanol Production Process from Sugar Cane Molasses; (ii) processes where the same product is obtained starting from two different raw materials: Phthalic Anhydride Production Process, Maleic Anhydride Production Process, (iii) processes with heat integration: Dimethylether Production Process, R-134a Production Process, Multiple Effect of Sugar Cane Juice Process, (iv) processes with different streams recovery: Electroplating Wastewater Discharge Process. Details of all the ten processes14

Table 2. Lifetime of Different Chemical Substances Lifetime Substance Name

Substance Formula

Substance CAS

Experimental Value (days)

Methods Used to Measure Experimental Values

SAR AOP Value (days)

Lifetime RAD

Methanol Toluene Benzene Ethane Propane n-Butane Formaldehyde Acetone Ethanol

CH4O C7H8 C6H6 C2H6 C3H8 C4H10 H2CO C3H6CO C2H5OH

67-56-1 108-88-3 71-43-2 74-84-0 74-98-6 106-97-8 50-00-0 67-64-1 64-17-5

12 2 10 45 11 5 1 51 4

GC/FID PTR-MS GC/FID GC/MS GC/FID GC/MS GC/FID GC/FID GC/FID GC/MS DOAS GC/FID PTR-MS GC/FID PTR-MS

17.364 2.047 5.486 39.323 8.432 4.064 1.316 52.431 2.991

0.4470 0.0235 0.4514 0.1262 0.2335 0.1872 0.3160 0.0280 0.2523 MRAD 0.04843

GC-FID, gas chromatography-flame ionization detection; GC-MS, gas chromatography-mass spectrometry; DOAS, differential optical absorption spectroscopy; PTR-MS, proton transfer reaction mass spectrometry.

1072

DOI 10.1002/aic

Published on behalf of the AIChE

April 2009 Vol. 55, No. 4

AIChE Journal

Table 3. Reaction Rate with Ozone for Different Chemical Substances Reaction Rate with Ozone Substance Name

Substance Formula

Substance CAS

Experimental Values (cm3/molecule-sec)

AOP Values (cm3/molecule-sec)

Reaction Rate with Ozone RAD

Acetylene Hexene 1-pentene 1,3 dichlorpropene

C2H2 C6H12 C5H10 C3H4Cl2

74-86-2 592-41-6 109-67-1 10061-01-5

4.8e 20 1.17e 17 1e 17 1.5e 19

2-butene

C4H8

107-01-7

2e 16

1-propene Ethene 2-methyl 1-pentene

C3H6 C2H4 C6H12

115-07-1 74-85-1 763-29-1

1.13e 17 1.75e 18 1.31e 17

3e 20 1.2e 17 1.2e 17 1.126e 19 (cis) 2.25e 19 (trans) 1.3e 16 (cis) 2e 16 (trans) 1.2e 17 1.75e 18 1.2e 17

0.375 0.0256 0.2 0.249 (cis) 0.5 (trans) 0.35 (cis) 0 (trans) 0.062 0 0.084 MRAD 0.271

are available from the authors upon request. In this article, one representative example taken from the list above is reported in details. In the example, we show: (i) that the PSP Framework is applicable to the process of interest and that could be used in direct connection with a process simulator for evaluating the sustainability of a process, (ii) the sensitivity of the process to the predicted values of toxicological data from molecular modeling and (iii) the use of the predictive methods to enhance the evaluation when the toxicological data are missing.

Production of maleic anhydride from butane (CASE 1). The described chemical plant is capable to produce 40 million pounds of maleic anhydride per year from n-butane. The following reactions occur when butane is mixed with oxygen: C4 H10 þ 3:5O2 ! C4 H2 O3 þ 4H2 O butane maleic anhydride C4 H10 þ 5:5O2 ! 2CO þ 2CO2 þ 5H2 O butane C4 H10 þ 3:5O2 ! C3 H4 O2 þ CO2 þ 3H2 O butane acrylic acid

Maleic anhydride production process Maleic anhydride is a versatile chemical intermediate used to make unsaturated polyester resins, lube oil additives, alkyd resins, and a variety of other products. There are different ways to obtain maleic anhydride. The most common ways are: (i) production of maleic anhydride from butane and (ii) production of maleic anhydride from benzene. Both process have been considered and simulated in the following section. The goal of this study is to analyze the processes, to evaluate the toxicological data of the substances involved in the processes and identify the best alternative in terms of environmental impact using the PSP methodology described in this work.

C4 H10 þ 6O2 ! CH2 O2 þ 3CO2 þ 4H2 O butane formic acid Process schema is presented in Figure 5. Pure butane, S2, and air, S1, are mixed and fed to the first rector R-1. The reaction is exothermic, therefore, one could consider either a fluidized bed or a packed bed reactor with heat removal. The reactor effluent, S5, is cooled and sent to the absorber F-1, where it is in contact with water, S7, to remove the light gases, S8. All maleic anhydride reacts with water, in R-2, to form maleic acid. The output reactor stream, S16, is flashed. The bottom flash stream, S12, is sent to R-3 where maleic acid is broken down to maleic anhydride and water. The reactor

Table 4. Reaction Rate with Hydroxyl for Different Chemical Substances Reaction Rate with Hydroxyl RAD

Reaction Rate with Hydroxyl Substance Name

Substance Formula

Substance CAS

Methanol Toluene Benzene Ethane Propane n-Butane Formaldehyde Acetaldehyde Acetone Ethanol Dimethyl ether

CH4O C7H8 C6H6 C2H6 C3H8 C4H10 H2CO CH3CHO C3H6CO C2H5OH CH3OCH3

67-56-1 108-88-3 71-43-2 74-84-0 74-98-6 106-97-8 50-00-0 75-07-0 67-64-1 64-17-5 115-10-6

AIChE Journal

Experimental Value (cm3/molecule-sec) 9.44e 5.96e 1.23e 2.68e 1.15e 2.54e 9.37e 15.8e 2.19e 3.27e 2.98e

13 12 12 13 12 12 12 12 13 12 12

April 2009 Vol. 55, No. 4

GCM AOP Value (cm3/molecule-sec)

QSAR AM1-MOOH Value

GCM-AOP RAD

QSAR AM1-MOOH RAD

6.16e 13 5.226e 12 1.9498e 12 2.7e 13 1.2686e 12 2.6322e 12 8.13e 12 16.98e 12 2.04e 13 3.5763e 12 1.6592e 12

9.3e 13 4.79e 12 1.51e 12 3.51e 13 1.276e 12 2.519e 12 9.77e 12 15.5e 12 2.3e 13 3.27e 12 4.177e 12

0.3470 0.1231 0.5852 0.0075 0.1031 0.0363 0.1323 0.0747 0.0685 0.0937 0.4430

0.0148 0.1963 0.2280 0.3097 0.1095 0.0083 0.0427 0.0189 0.0502 0 0.4017

Published on behalf of the AIChE

DOI 10.1002/aic

1073

Figure 5. Maleic anhydride production process from butane.

effluent, S13, is sent to the distillation column T-1 where maleic anhydride and water are separated. The distillate, S14, is sent to waste treatment. The bottom stream, S15, consists of 99-wt% maleic anhydride. Production of maleic anhydride from benzene (CASE 2). The following reactions occur during the reaction of benzene with oxygen: C6 H6 þ 4:5O2 ! C4 H2 O3 þ 2CO2 þ 4H2 O benzene maleic anhydride C6 H16 þ 7:5O2 ! 6CO2 þ 3H2 O benzene C4 H2 O3 þ 3O2 ! 4CO2 þ H2 O maleic anhydride C6 H6 þ 1:5O2 ! C4 H4 O2 þ 4H2 O benzene

quinone

mixed stream, S6, is heated in E-2, prior to being sent to a packed catalytic reactor, R-1. The reactions described earlier take place in R-1. The reactor has been simulated as a plug flow reactor with 10% of input stream by-passing the reactor. The reactor effluent, S12, containing small amounts of un-reacted benzene, maleic anhydride, quinone and combustion products are cooled in E-3. Water can react with some maleic anhydride to form maleic acid. This phenomenon takes place in R-2. The R-2 output stream, S17, is sent to an absorber where it is contacted with heavy organic solvent (dibutyl phtalate), S16. This solvent absorbs the maleic anhydride, quinone, and small amounts of water. The F-1 top output stream, S18, contains off gasses, which are released into the atmosphere. The bottom stream, S19, is sent to T-1 where dibutyl phtahlate is recovered as the bottom product and recycled back to the absorber, S21. The product process,maleic anhydride, S20, is obtained in the top of T-1 column.43

PSP framework application

Process schema is shown in Figure 6. Benzene, S1, is vaporized in E-1 and mixed with compressed air, S4. The

Both processes have been simulated using PROII process simulator and their environmental impact has been estimated. The goal is to choose, from these two cases (production

Figure 6. Maleic anhydride production process from benzene. 1074

DOI 10.1002/aic

Published on behalf of the AIChE

April 2009 Vol. 55, No. 4

AIChE Journal

thermo-physical properties which lead to the environmental impact categories calculation. The molecular modeling results are presented in Table 8. These thermo-physical properties lead to the environmental impact categories presented in Table 9. To validate the proposed methodology, the values of one environmental impact category—HTPE for two substances (butane and benzene) have been replaced in the EPA’ DB with the values calculated using MM techniques to evaluate the sensitivity of the final calculation to the estimation method of the thermo-physical properties. The results, in terms of Environmental Impact Indexes, are presented in Table 10. As it can be noticed from Table 10 the results in terms of Environmental Impact Indexes are very similar, showing that the experimental values can be safely substituted by the predicted ones. These results allow as moving further more in this direction and completing some missing data of the original database with values calculated using MM techniques (from Table 9). The results obtained in this case are presented in Table 11. It can be noticed that the missing environmental impact categories inserted in the DB contribute to increase the values of the environmental impact indexes. The index reported in Table 11 should be considered more reliable of the ones reported in Table 10 because all the data for all the substances are now taken into account. Comparing the results obtained using MM calculation the best environmental solution still remain the second alternative design: production of maleic anhydride from benzene.

Table 5. 3D Indicators Obtained in the Maleic Anhydride Production Process 3D Indicators Case Study

MI

EI

PCR

PEI

CASE 1 CASE 2

29,734.45 7416.67

677.55 62.9629

20,070 10,110

8776.601 16,289.101

from butane and production from benzene) the best process in terms of environmental impact. For this purpose, 3D and 1D indicators have been calculated using the mass flowrates, the input and output streams composition (mass fraction) and toxicological and risk information data. The 3D indicators calculated using data from the process simulators and from the database are presented in Table 5. From Table 5, it can be noticed, MI, EI and PCR are lower in the second case while PEI is lower in the first alternative design. To decide which design is more environmentally friendly, 1D indicators must be considered. Table 6 reports the environmental impact categories, as extracted from the EPA toxicological data base, that are directly used for the calculation of the 1D indicators according to the WAR Algorithm. Table 7 contains the Environmental Impact Indexes of the WAR Algorithm which are calculated in the PSP Framework by coupling the toxicological data reported in Table 6 with the mass flow-rates and composition of the input and output streams of the processes of interest. Taking into account the 1D indicators it can be noticed that the best alternative design in terms of environmental impact is the second one, CASE 2. As it can be noticed from Table 6 not all the environmental impact categories for all substances are present in EPA’s database: this is where molecular modeling comes into play. In fact, the molecular modeling techniques, described in the previous sections, have been used here to predict some

Conclusions This article presents and discusses a methodology to quantify the environmental impact of a chemical process. The methodology, used in the early stages of process design,

Table 6. Environmental Impact Categories of the Substances Involved in the Maleic Anhydride Production Process (data from EPA’s DB) GWP

ODP

PCOP

HTPE

– – – – – – –

0.8425 – – – – – –

0.0001553 0.00003278 0.0053647 0.03278 0.05 0.295 –

– – – – – –

– – 0.38837 – – –

0.295 0.05901 0.092207 0.73765 0.00003278 –

Production from butane Butane – CO2 0.0003174 CO – Formic acid – Acrylic acid – Maleic anhydride – Maleic acid – Production from benzene Maleic anhydride – Dibutyl phtalate – Benzene – Quinone – CO2 0.000317 Maleic acid –

HTPI

TTP

0.2867 – – 0.31982 10.5017 0.8795 0.4969

0.2867 – – 0.31982 10.5017 0.8795 0.4969

0.8795 0.04397 0.1064147 2.706208 – 0.4969

0.8795 0.04397 0.1064147 2.706208 – 0.4969

AP

ATP

– – – – – – –

0.2867 – – 0.31982 10.5017 0.8795 0.4969

– – – – – –

0.8795 0.04397 0.106415 2.706208 – 0.4969

Table 7. Environmental Impact Indexes Obtained in the Maleic Anhydride Production using WAR Algorithm Enivoronmental Indexes Name Case Study Name

Iout (PEI/hr)

Iout_mp (PEI/kg)

Igen (PEI/hr)

Igen_mp (PEI/kg)

Case 1 Case 2

4056.57 1095.02

1.54 0.44

1643.38 988.00

0.625 0.398

AIChE Journal

April 2009 Vol. 55, No. 4

Published on behalf of the AIChE

DOI 10.1002/aic

1075

Table 8. Thermo-Physical Properties Obtained Using MM

Production from butane Butane CO2 CO Formic acid Acrylic acid Maleic anhydride Maleic acid Production from benzene Maleic anhydride Dibutyl phtalate Benzene Quinone CO2 Maleic acid

Log Kow (QSAR)

Log Kow (QM)

lifetimeOH (SAR) (hr)

lifetimeO3 (SAR) (hr)

2.31 0.83 1.78 0.46 0.44 1.62 0.05

3.078 – 1.71 0.17 0.642 1.29 0.66

48.762 – 18710.184 246.828 13.198 56.894 1.041

– – – – 78.588 78.588 24.179

2.63 E – 6.90 E 5.20 E 9.73 E 2.26 E 7.95 E

15 13 12 12 12

56.894 13.836 65.827 28.447 – 1.041

78.588 – – 78.583 – 24.179

2.26 9.28 1.95 4.51

12 12 12 12

1.62 4.61 2.1043 0.25 0.83 0.05

1.29 6.49 1.99 0.5788 – 0.66

rOH (SAR) (cm3/molecule-sec)

E E E E – 7.95 E

rO3 (SAR) (cm3/molecule-sec)

12

– – – – 1.75 E 18 1.75 E 18 8.75 E 19 1.75 E 18 – – 3.50 E 18 – 8.75 E 19

12

Table 9. Environmental Impact Categories of the Substances Involved in the Maleic Anhydride Production Process (Data Calculated Using MM Techniques)

Production from butane Butane CO2 CO Formic acid Acrylic acid Maleic anhydride Maleic acid Production from benzene Maleic anhydride Dibutyl phtalate Benzene Quinone CO2 Maleic acid

GWP

ODP

PCOP

HTPE

HTPI

TTP

AP

ATP

– – – – – – –

– – – – – – –

0.6349 – 0.0017 0.1254 2.3455 0.5441 1.9172

0.000145811 0.00003401 0.005584308 0.038486484 0.040917836 0.370544293 0.001412429

0.2691888 – – 0.3754409 8.5923 1.1044884 0.3613818

0.2691888 – – 0.3754409 8.5923 1.1044884 0.3613818

– – – – – – –

0.0454 0.0024 0.0289 0.0667 0.0370 0.0060 0.2000

– – – – – –

– – – – – –

0.5441 2.2375 0.4703 1.0882 – 1.9172

0.370544293 0.06042072 0.093333335 0.708151296 0.00003401 0.001412429

1.1044884 0.0450192 0.1077144 2.1649664 – 0.3613818

1.1044884 0.0450192 0.1077144 2.1649664 – 0.3613818

– – – – – –

0.0060 0.5914 0.0167 0.0455 0.0024 0.2000

Table 10. Environmental Impact Indexes Obtained in the Maleic Anhydride Production: Comparison Between Indexes Obtained Using Initial HTPE Value (EPA’s DB) and HTPE Value Calculated Using MM Enivoronmental Indexes Name Case Study Name Case Case Case Case

1 1 2 2

(initial) (MM calculation) (initial) (MM calculation)

Iout (PEI/hr)

Iout_mp (PEI/kg)

Igen (PEI/hr)

Igen_mp (PEI/kg)

4056.57 4056.56 1095.02 1095.37

1.54 1.54 0.44 0.44

1643.38 1643.39 988.00 991.374

0.625 0.625 0.398 0.399

Table 11. Environmental Impact Indexes Obtained in the Maleic Anhydride Production: Comparison Between Indexes Obtained Using Initial DB (EPA’s DB) and Final DB (Containing Data Estimated with MM) Enivoronmental Indexes Name Case Study Name Case Case Case Case

1 1 2 2

(initial) (MM calculation) (initial) (MM calculation)

Iout (PEI/hr)

Iout_mp (PEI/kg)

Igen (PEI/hr)

Igen_mp (PEI/kg)

4056.57 4123.51 1095.02 1329.60

1.54 1.57 0.44 0.53

1643.38 1710.31 988.00 815.71

0.625 0.651 0.398 0.32

allows choosing the most environmentally friendly design among different process alternatives. The present work has considered several issues for the definition of the PSP Framework developed: Sustainable development concept 1076

DOI 10.1002/aic

analysis; Identification of different indicators for sustainability evaluation; Development and implementation of a specific software, PSP Framework, for industrial sustainability evaluation; Development of a toxicological database;

Published on behalf of the AIChE

April 2009 Vol. 55, No. 4

AIChE Journal

Analysis and application of different molecular modeling techniques for evaluating the toxicity of chemical compounds; PSP Framework application in various case studies. The software developed contains interesting elements such as: (i) CAPE OPEN Modules ready to be used in any process simulator which provides CO interfaces, (ii) a toxicological database, (iii) various support program codes developed in Visual Basic and (iv) a methodology for calculating missing toxicological data from molecular modeling. The CO modules developed in this work can be directly used in any process simulator and they automatically transfer the necessary data from process simulator and from the toxicological data base for calculating the environmental indexes. The modules have been tested successfully using different process simulators such as: PROII, Aspen Plus, COCO/COFE. A relational database containing about 4900 substances has been created and implemented. Different thermo-physical properties correspondent to some toxicological data, such as the environmental impact categories HTPI, HTPE, TTP, ATP, GWP, ODP, PCOP, AP for various chemical substances have been estimated using molecular modeling techniques. The molecular modeling techniques used are: QSAR, GCM, QM, and COSMO. For each estimated thermo-physical: octanol-water partition coefficient, lifetime, reaction rate with ozone, reaction rate with hydroxyl, representative results, discussions, and interpretations are reported. Some support programs are also included in the PSP Framework. The programs are able: (i) to set the weighting factors for the environmental impact categories according to local and political needs; (ii) to calculate the 3D indicators: MI, EI, PCR, and PEI; (iii) to calculate the 1D indicators: Iout, Iout_mp, Igen and Igen_mp used in WAR Algorithm. Ten processes, taken from the literature developed ad hoc starting from real processes located in developing countries, have been simulated and analyzed from the environmental point of view using the PSP Framework. The maleic anhydride production process is reported here as a case study where the application of the entire PSP Framework is shown (including the effect of the prediction of the environmental indexes from molecular modeling). In conclusion, for the sustainable development of the chemical industry, it is essential to consider environmental issues at design time. The PSP Framework presented in this article provides a complete and reliable tool that can quantify the environmental impact of the process under study.

Acknowledgments The authors would like to thank everyone who provide support and suggestions to this work: PROII and Aspen Plus Support teams, COLaN (especially to Michel Pons), Marco Carone (University of Trieste), Ioan Toma (University of Innsbruck), Branislav Opacic and Circe la Placa (ICS-UNIDO).

Literature Cited 1. Veleva V, Ellenbecker M. Indicators of sustainable production: framework and methodology. J Cleaner Prod. 2001;9:519–549. 2. Korevaar G. Sustainable Chemical Processes and Products—New Design Methodology and Design Tools, PhD Thesis, Technical University of Delft, 2004. 3. Jenck JF, Agterberg F, Droescher M. Products and processes for a sustainable chemical indutry: a review of achievements and prospects. Green Chem. 2004;6:544–556.

AIChE Journal

April 2009 Vol. 55, No. 4

4. Fermeglia M, Pricl S, Longo G. Molecular modeling and process simulation: real possibilities and challenges. Chem Biochem Eng Q. 2003;17:69–79. 5. ‘‘NJSSI’’ (2007). New Jersey Sustainable State Institute. Available at: http://njssi.org/about.asp?Level2ItemID¼55. Last accessed September 2008. 6. Sikdar SK. Sustainable development and sustainable metrics. AIChE J. 2003;49:1928–1932. 7. Martins AA, Mata TM, Costa CAV, Sikdar SK. Framework for sustainability metrics. Ind Eng Chem Res. 2007;46:2962–2973. 8. Schwarz J, Beloff B, Beaver E. use sustainability metrics to guide decision-making. Environmental protection. Available at: http:// www.bridgestos.org/CEP-July.pdf. Last accessed September 2008. 9. Tanzil D, Ma G, Beloff B. Automating the Sustainability Metrics Approach. AIChESpring Meeting, April 25–29, 2004, New Orleans. Available at: http://www.bridgestos.org/AIChESpring2004Metrics PPT.pdf (BRIDGES to Sustainability Institute). Last accessed September 2008. 10. Vincent R, Bonthoux F, Mallet G, Iparraguirre JF, Rio S. Me´thodologie D’ Nospace Evalutionsimplifie´e du Risque Chimique: Un Outil d’Aide a` De´cision. INRS Hyg Secur Travail. 2005;195 (Second Quarter), 7 (INRS Ref. No. ND 2233). 11. Cabezas H, Bare JC, Mallick SK. Pollution prevention with chemical process simulators: the generalized waste reduction (WAR) algorithm. Comput Chem Eng. 1997;21:305–310. 12. Young DM, Cabezas H. Designing sustainable processes with simulation: the waste reduction (WAR) algorithm. Comput Chem Eng. 1999;23:1477–1491. 13. Young D, Scharp R, Cabezas H. The waste reduction (WAR) algorithm: environmental impacts, energy consumption, and engineering economics. Waste Manage. 2000;20:605–615. 14. Toma L. Computer Aided Design of Sustainable Industrial Processes, PhD Thesis, University of Padua, Italy, 2008. 15. Hansen BG, Van Haelst AG, Van Leeuwen K, Van der Zandt P. Priority setting for existing chemicals: European union risk ranking method. Environ Toxicol Chem. 1999;18:772–779. 16. Ren S, Frymier PD, Schultz TW. An exploratory study of the use of multivariate techniques to determine mechanisms of toxic action. Ecotoxicol Environ Saf. 2003;55:86–97. 17. COSMOlogic GmbH & Co. KG. COSMOTherm versionC21–0106, A Graphical User Interface to the COSMOtherm Program Tutorial. 18. ‘‘Accelrys’’ (2007). Available at: http://www.accelrys.com/products/ topkat/index.html. Last accessed September 2008. 19. Best SA, Merz KM, Reynolds CH. Free energy perturbation study of octanol/water partition coefficients: comparison with continuum GB/SA calculations. J Phys Chem. 1999;103:714–726. 20. Lin Z, Shi P, Gao S, Wang L, Yu H. Use of partition coefficients to predict mixture toxicity. Water Res. 2003;37:2223–2227. 21. Ulas S, Diwekar UM. Efficient molecular simulations for environmentally benign processes. Mol Simul. 2006;32:315–329. 22. Lyubartsev AP, Jacobsson SP, Sundholm G, Laaksonen A. Solubility of organic compounds in water/octanol systems. A expanded ensemble molecular dynamics simulation study of logP parameters. J Chem Phys. 2001;105:7775–7782. 23. Barratt MD, Rodford RA. The computational prediction of toxicity. Curr Opin Chem Biol. 2001;5:383–388. 24. Kaiser KLE. The use of neural networks in QSARs for acute aquatic toxicological endpoints. J Mol Struct (Theochem). 2003;622:85–95. 25. Hansch C, Muir R, Fujita T, Peyton P, Maloney P, Geiger F, Streich M. The correlation of biological activity of plant growth regulators and chloromycetin derivates with Hammett constants and partition coefficients. J Am Chem Soc. 1963;85:2817–2824. 26. Di Marzio W, Saenz ME. Quantitative structure-activity relationship for aromatic hydrocarbons on freshwater fish. Ecotoxicol Environ Saf. 2004;59:256–262. 27. Schultz TW, Cronin MTD, Netzeva TI. The present status of QSAR in toxicology. J Mol Struct. 2003;622:23–38. 28. Velders GJM, Madronich S, Clerbaux C, Derwent R, Grutter M, Hauglustaine D, Incecik S, Ko M, Libre JM, Nielsen OJ, Stordal F, Zhu T. Chemical and radiative effects of halocarbons and their replacement compounds. IPCC/TEAP special report: safeguarding the ozone layer and the global climate system. Available at: http:// arch.rivm.nl/env/int/ipcc/pages_media/SROC-final/SROC02.pdf. Last accessed september 2008.

Published on behalf of the AIChE

DOI 10.1002/aic

1077

29. Prather M. Lifetimes of atmospheric species: integrating environmental impacts. Geophys Res Lett. 2002,22:2063–2065. 30. Finlayson-Pitts BJ, Pitts JN Jr. Chemistry of the Upper and Lower Atmosphere. London: Academic Press, 2000. 31. Borrego C, Tchepel O, Costa AM, Amorim JH, Miranda AI. Emission and dispersion modeling of Lisbon air quality at local scale. Atmos Environ. 2003;37:5197–5205. 32. Pompe M, Veber M. Prediction of rate constants for the reaction of O3 with different organic compounds. Atmos Environ. 2001;35: 3781–3788. 33. Kwok ESC, Atkinson R. Estimation of hydroxyl radical reaction rate constants for gas-phase organic compounds using a structurereactivity relationship: an update. Atmos Environ. 1995;29:1685– 1695. 34. Klamt A. Estimation of gas-phase hydroxyl radical rate constants of organic compounds based on molecular orbital calculations. Chemosphere. 1993;32:171–726. 35. Sablijic A, Peijnenburg W. Modeling lifetime and degradability of organic compounds in air, soil, and water (IUPAC Technical Report). Pure Appl Chem. 2001;73:1331–1348. 36. Gu¨sten H. Predicting the abiotic degradability of organic pollutants in the troposphere. Chemosphere. 1999;38:1361–1370.

1078

DOI 10.1002/aic

37. Mullins E, Oldland R, Liu YA, Wang S, Sandler SI, Chen CC, Zwolak M, Seavey KC. Sigma-profile database for using COSMO-based thermodynamic methods. Ind Eng Chem Res. 2006;45:4389–4415. 38. Fermeglia M, Braiuca P, Gardossi L, Pricl S, Halling PJ. In silico prediction of medium effects on esterification equilibrium using the COSMO-RS method. Biotehcnol Prog. 2006;22:1146–1152. 39. Klamt A, Eckert F. COSMO-RS: a novel and efficient method for the a priori prediction of thermophysical data of liquids. Fluid Phase Equilib. 2002;172:43–72. 40. Klamt A. COSMO-RS: From Quantum Chemistry to Fluid Phase Thermodynamics and Drug Design. Netherlands: Elsevier. 2005. 41. AOP User’s Guide. AOP Program version v. 1.9.2. US EPA. 2002. 42. GAW Report No. 171. A WMO/GAW Expert Workshop on Global Long-Term Measurements of Volatile Organic Compounds (VOCs), Geneva, Switzerland, January 30 – February 1, 2006. Available at: http://instaar.colorado.edu/arl/pdf/gaw171final.pdf. Last accessed November 2007. 43. ChE (2007). Chemical Engineering Department, West Virginia University. Available at: http://www.che.cemr.wvu.edu/publications/projects. Last accessed October 2007. Manuscript received Apr. 17, 2008, and revision received Sept. 3, 2008.

Published on behalf of the AIChE

April 2009 Vol. 55, No. 4

AIChE Journal