Review
Clean Products and Processes 1 (1998) 5–19 Q Springer-Verlag 1998
Pollution prevention through process integration Mahmoud M. El-Halwagi
5 Abstract Process integration is a holistic approach to process design and operation. It emphasizes the unity of the process units and objectives. Therefore, it provides a unique framework for integrating environmental issues with other process objectives such as profitability, yield enhancement, debottlenecking and energy reduction. This paper presents a review of recent advances in the area of pollution prevention through process integration. First, the alternative methods for industrial waste reduction are discussed. Then, process integration is defined and categorized into three main components: synthesis, analysis and optimization. Next, mass integration science and methods are reviewed with special emphasis on their critical role in pollution prevention. Throughout the paper, various tools and techniques are described and illustrated.
Introduction With the ever-growing industrial waste problems, the process industries are embarking on major self-regulated as well as legislated pollution-prevention activities. These endeavors are moving in a direction which is fundamentally different than environmental efforts over the past two decades and are gradually shifting from downstream pollution control to a more aggressive practice of trying to prevent pollution at the heart of the process. Pollution control has traditionally been carried out through two alternatives: 1. End-of-pipe treatment refers to the application of chemical, biological, and physical processes to reduce toxicity or magnitude of environmentally undesirable compounds in process waste streams prior to their release to the environment. Treatment options include biological systems, chemical precipitation, flocculation, coagulation, incineration and boilers and industrial furnaces (BIF’s). 2. Disposal involves the use of postprocess activities that can handle waste, such as deep-well injection and off-site Received: 7 July 1998 / Accepted: 8 September 1998 Mahmoud M. El-Halwagi Chemical Engineering Department, Auburn University, Auburn, AL 36849, USA This financial support of the NSF (NYI-CTS-945013) is gratefully acknowledged.
shipment of hazardous materials to waste-management facilities. Therefore, pollution control decisions are made with little or no regard to the process that generates the waste. All that matters is characteristics of terminal streams and environmental goals. This perspective gave rise to prepackaged menus of solutions from which options can be chosen and added to the periphery of the process to address environmental problems. Unfortunately, this approach addresses the symptoms of the problem without dealing with the root cause of the environmental problem which lies at the core of the process. This realization motivated industry to adopt in-plant pollution prevention strategies. Several definitions of pollution prevention can be found in the literature (e.g. El-Halwagi and Petrides 1995; Freeman 1995; Theodore et al. 1994; Noyes 1993). Nonetheless, throughout this paper, the term pollution prevention will be used to designate any activity that reduces the release of undesirable species to the environment. These activities have two main strategies: 1. Source reduction which includes any in-plant actions to reduce the quantity or the toxicity of the waste at the source of generation. Examples include equipment modification, design and operational changes of the process, reformulation or redesign of products, substitution of raw materials, and use of environmentally benign chemical reactions. 2. Recycle/reuse which involves the re-introduction of pollutant-laden streams back into the process. Typically, separation technologies are key elements in a recycle/ reuse system to recover valuable materials such as solvents, metals, inorganic species, and water. The industry-wide move toward in-plant pollution prevention poses the following challenges: – Solutions to the environmental problems can no longer be in the form of simple end-of-pipe or pollution control devices that can be selected based on terminal streams and without delving into the core of the process. – In order to undertake any modifications in the core processing units, it is inevitable to fully understand and appreciate the integrated nature of the process. No longer should we choose from a roster of “environmental technologies” and add them to the periphery of the process. No longer is there a single menu of solutions that fits all. – Changes in a unit or a stream, often propagate
Clean Products and Processes 1 (1998)
6
throughout the process and can have major implications on the operability and profitability of the process. – Environmental objectives cannot be the overriding drivers for process changes. The process has numerous technical, economic, and safety objectives and constraints. All of these must be reconciled without compromising environmental goals. The difficulty lies in determining how far the environmental envelop of the process can be pushed while being reconciled with techno-economic aspects of the plant. – In addition to pollution prevention strategies involving the installation of new equipment, it is surprising that much of the solution lies within existing process equipment. However, it is not intuitively obvious how to identify these opportunities. – Solutions cannot be replicated from one process to another. Subtle differences between one process and another (even if they had common technology) can lead to significantly different solutions. These challenges call for the application of a systematic and generally applicable approach which transcends the specific circumstances of the process and views the environmental problem from a holistic perspective. In this regard, process integration provides a unique framework for addressing the above-mentioned challenges.
Because of the vast number of process alternatives, it is important that the synthesis techniques be able to extract the optimal solution(s) from among the numerous candidates without the need to enumerate these options. In this regard, a key approach is called “targeting.” It is based on tackling the synthesis task via a sequence of stages. Within each stage, a design target can be identified and employed in subsequent stages. Such targets are determined ahead of detailed design and without commitment to the final system configuration. The targeting approach offers two main advantages. First, within each stage, the problem dimensionality is reduced to a manageable size, avoiding the combinatorial problems. Second, this approach offers valuable insights into the system performance and characteristics. The result of process synthesis is a flowsheet which represents the configuration of the various pieces of equipment and their interconnection. Next, it is necessary to analyze the performance of this flowsheet.
Process analysis
While synthesis is aimed at combining the process elements into a coherent whole, analysis involves the decomposition of the whole into its constituent elements for individual study of performance. Hence, once a process is synthesized, its detailed characteristics (e.g., flowrates, compositions, temperature, and pressure) are preProcess integration Process integration is a holistic approach to process dedicted using analysis techniques. These techniques insign, retrofitting, and operation which emphasizes the clude mathematical models, empirical correlations, and unity of the process. It offers a comprehensive framework computer-aided process simulation tools (e.g., ASPEN for fundamentally understanding the global insights of Plus, ChemCAD III, PRO II, HYSIM, EnviroPro). In addithe process, methodically determining its attainable pertion, process analysis may involve predicting and validatformance targets, and systematically making decisions ing performance using experiments at the lab or pilotleading to the realization of these targets. There are three plant scales and even actual runs within existing facilities. key components in any effective process integration Commercially-available simulation tools provide a methodology; synthesis, analysis, and optimization. user-friendly interface to conveniently construct a flowsheet and simulate its operation. Because of their large databases for physical and chemical properties of numerProcess synthesis Process synthesis deals with combining and integrating ous compounds as well as mathematical models for unit process units and streams so as to meet certain objecoperations, they are well suited for the following tasks: tives. Reviews of process synthesis literature are available in literature (e.g. El-Halwagi 1997, 1993; Douglas 1992; Estimation of waste properties Reliable data on dilute and concentrated waste stream Westerberg 1987; Stephanopoulos and Townsend 1986; Nishida et al. 1981). Process synthesis provides an attrac- data are needed. These data include safety factors (e.g. flashpoints and flammability), as well as fate and transtive framework for tackling numerous design problems port of chemical compounds (e.g. evaportation, biodethrough a systemic approach. It guides the designer in the generation and screening of various process technolo- gradation, multi-media partitioning, etc.). gies, alternatives, configurations, and operating conditions. In most applications, the number of process alterPrediction of waste formation and propagation Once the flowsheet structure is determined, it is impornatives is too high (in many cases infinite). Without a tant to account for the formation of various chemical bysystematic approach for process synthesis, an engineer normally synthesizes a few process alternatives based on products that eventually form the process wastes. These formations are typically associated with the reactive sysexperience and corporate preference. The designer then selects the alternative with the most promising economic tems of the process. Once they are formed, it is necessary to account for their distribution and propagation potential and designates it as the “optimum” solution. throughout the process. The simulator models are quite However, by assessing only a limited number of alternaadept in tracking these species throughout the flowsheet. tives one may easily miss the true optimum solution, or even become trapped in a region that is significantly dif- They allow the performance of a pollution balance which is an algorithmic method to account for pollution generaferent from the optimal one. In addition, the likelihood tion. An example of this concept is the WAste Reduction of generating innovative designs is severely reduced by “WAR” algorithm (Hilaly and Sikdar 1994, 1996). It is an exclusive dependence on previous experience.
M.M. El-Halwagi: Pollution prevention through process integration
used as a metric for comparing pollution generation among various process as well as for the same process under different conditions. It uses an index for assessing the extent of generation of pollutants. This algorithm can be used to reduce pollution in new facilities as well as existing processes by focusing on streams with high pollution index. Consequently, units can be systematically modified to reduce the extent of pollution in high-indexed streams. Another example of the use of computer-aided databases for pollution prevention is the Clean Process Advisory System or “CPAS.” This is a cluster of computeraided programs being developed via collaboration among numerous organizations in industry, academia and government. The industrial lead for this endeavor has been the Center for Waste Reduction Technologies “CWRT” of the American Institute of Chemical Engineers “AIChE”. The academic lead has been the National Center for Clean Industrial and Treatment Technologies “CenCITT” through Michigan Technological University. Through these organizations, government co-funding, input, and information sources have been coordinated. CPAS tools provide information on environmental technology, simulation modules, separation methods, material selection, environmental risk, industrial waste reduction case studies, and design options. The various simulation tools available for pollution prevention have been recently surveyed (Hilaly and Sikdar 1996; Samdani 1995).
perature, or flowrate) as well as integer variables (e.g., 0,1,2,...) is called a mixed-integer program (MIP). Depending on the linearity or nonlinearity of MIPs, they are designated as mixed-integer linear programs (MILPs) and mixed-integer nonlinear programs (MINLPs). The principles of optimization theory and algorithms are covered by various books (e.g. Grossmann 1996; Edgar and Himmelblau 1988; Reklaitis et al. 1983; Beveridge and Schechter 1970). It is worth pointing out that most optimization software can efficiently obtain the global solution of LPs and MILPs. On the other hand, no commercial package is guaranteed to identify the global solution of non convex NLPs and MINLPs. Recently, significant research has been undertaken towards developing effective techniques for the global solution of non convex NLP’s and MINLPs (e.g. Vaidyanathan and El-Halwagi 1994a, 1996a; Sahinidis and Grossmann 1991; Visweswaran and Floudas 1990). Within the next few years, these endeavors may indeed lead to practical procedures for globally solving general classes of NLPs and MINLPs. The optimization component of process integration drives the iterations between synthesis and analysis toward an optimal closure. In many cases, optimization is also used within the synthesis activities. For instance, in the targeting approach for synthesis, the various objectives are reconciled using optimization. In the structurebased synthesis approach, optimization is typically the main framework for formulating and solving the synthesis task. The mathematical representation used in this approach is formulated as an MINLP. The objective of the Process optimization Once the process has been synthesized and its performMINLP is to identify two types of variables; integer and ance has been characterized, one can determine whether continuous. The integer variables correspond to the exisor not the design objectives have been met. Therefore, tence or absence of certain technologies and pieces of synthesis and analysis activities are iteratively continued equipment in the solution. For instance, a binary integer until the process objectives are realized. The realization variable can assume a value of one when a unit is seof process objectives implies that we have a solution that lected and zero when it is not chosen as part of the soluworks but not necessarily an optimum one. Therefore, it tion. On the other hand, the continuous variables deteris necessary to include optimization in a comprehensive mine the optimal values of nondiscrete design and operprocess integration methodology. Optimization involves ating parameters such as flowrates, temperatures, presthe selection of the “best” solution from among the set of sures, and unit sizes. candidate solutions. The degree of goodness of the solution is quantified using an objective function (e.g., cost) Motivating example which is to be minimized or maximized. The search process is undertaken subject to the system model and Application of mass-integration to debottleneck an restrictions which are termed constraints. These constraints are in the form of equality and inequality expres- acrylonitrile process by reducing wastewater discharge (El-Halwagi 1997) sions. Examples of equality constraints include material Acrylonitrile (AN, C3H3N) is manufactured via the vaporand energy balances, process modeling equations, and phase ammoxidation of propylene: thermodynamic requirements. On the other hand, the nature of inequality constraints may be environmental (e.g., catalyst * C3H3N c 3 H2O. the quantity of certain pollutants should be below specif- C3H6 c NH3 c 1.5 O2 ic levels), technical (e.g., pressure, temperature or floThe reaction takes place in a fluidized-bed reactor in wrate should not exceed some given values) or thermowhich propylene, ammonia, and oxygen are catalytically dynamic (e.g. the state of the system cannot violate the reacted at 450 7C and 2 atm. The reaction is a single pass second law of thermodynamics). An optimization problem in which the objective function as well as all the con- with almost complete conversion of propylene. The reaction products are cooled using an indirect-contact heat straints are linear is called a linear program (LP); otherwise it is termed a nonlinear program (NLP). The nature exchanger which condenses a fraction of the reactor offof optimization variables also affects the classification of gas. The remaining off-gas is scrubbed with water, then optimization programs. An optimization formulation that decanted into an aqueous layer and an organic layer. The organic layer is fractionated in a distillation column uncontains continuous (real) variables (e.g. pressure, tem-
7
Clean Products and Processes 1 (1998)
8
Fig. 1. Flowsheet of AN production (El-Halwagi 1997)
der slight vacuum which is induced by a steam-jet ejector. Figure 1 shows the process flowsheet along with pertinent material balance data. The wastewater stream of the plant is composed of the off-gas condensate, the aqueous layer of the decanter, the bottom product of the distillation column, and the condensate from the steam-jet ejector. This wastewater stream is fed to the biotreatment facility. Since the biotreatment facility is currently operating at full hydraulic capacity, it constitutes a bottleneck for the plant. Plans for expanding production of AN are contingent upon debottlenecking of the biotreatment facility by reducing its influent or installing an additional treatment unit. The new biotreatment facility will cost about $4 million in capital investment and $360,000/yr in annual operating cost, leading to a total annualized cost “TAC” of $760,000/yr with a 10-yr linear depreciation. The objective of this case study is to use mass integration techniques to devise cost-effective strategies to debottleneck the biotreatment facility. Several constraints are imposed on the quantity and quality of water that can be recycled/ reused in process units (El-Halwagi 1997). In order to identify the optimum solution, one should be able to answer the following challenging questions: – Which phase(s) (gaseous, liquid) should be intercepted to remove the pollutants? – Which process streams should be intercepted? – To what extent should the pollutants be removed from each process stream? – Which separation operations should be used for interception? – Which separating agents should be selected for interception? – What is the optimal flowrate of each separating agent? – How should these separating agents be matched with the pollutant-laden streams (i.e. stream pairings)?
– Which units should be manipulated for source reduction? By what means? – Should any streams be segregated? Which ones? – Which streams should be recycled/reused? To what units? To answer the above-mentioned questions, one can envision so many alternatives they cannot be enumerated. The search for optimum values of continuous variables (e.g. flowrates, temperatures, pressures, extents of separation for the various components) and discrete variables (which units to be used, in what order, etc.) involves numerous (in some cases infinite) alternatives. Typically, an engineer charged with the responsibility of answering these questions examines few process options based on experience and corporate preference. Consequently, the designer develops a simulation model, performs an economic analysis and selects the least expensive alternative from the limited number of examined options. This solution is inappropriately designated as the “optimum.” Normally it is not! Indeed, the true optimum may be an order of magnitude less expensive. The foregoing discussion illustrates that flowsheets (notwithstanding their usefulness) do not readily provide the global insights of the process. The use of repeated analyses to screen a few arbitrarily generated alternatives can be quite misleading. Instead, what is needed is a systematic methodology that can quickly and smoothly guide engineers through the complexities of the flowsheet, allowing them to identify the big picture of mass and energy flows, determine best performance targets of the process, and extract the optimal solution without having to enumerate and analyze the numerous alternatives. Does such a methodology exist? The answer is yes: via mass integration and energy integration.
M.M. El-Halwagi: Pollution prevention through process integration
tailed discussion, the reader is referred to El-Halwagi (1997), El-Halwagi and Spriggs (1998, 1996), and El-Halwagi et al. (1996). Mass integration and energy integration As has been discussed earlier, a fundamental understandMass integration is based on fundamental principles ing of the global flow of mass and energy is instrumental of chemical engineering combined with system analysis in developing optimal design and operating strategies to using graphical and optimization-based tools. The first meet process objectives including cost effectiveness, yield step in conducting mass integration is the development enhancement, energy efficiency, and pollution prevention. of a global mass allocation representation of the whole Over the past two decades, significant contributions have process from a species viewpoint (El-Halwagi et al. 1996; been made in understanding the global flow of mass and Garrison et al. 1995, 1996) as shown in Fig. 2. For each energy within a process. Two key branches of process in- targeted species (e.g., each pollutant), there are sources tegration have been developed: mass integration and en- (streams that carry the species) and process sinks (units ergy integration. Energy integration is a systematic meth- that can accept the species). Process sinks include reacodology that provides a fundamental understanding of tors, heaters/coolers, biotreatment facilities, and disenergy utilization within the process and employs this charge media. Streams leaving the sinks become, in turn, understanding in identifying energy targets and optimiz- sources. Therefore, sinks are also generators of the taring heat-recovery and energy-utility systems.). Of particu- geted species. Each sink/generator may be manipulated lar importance are the thermal-pinch techniques that can via design and/or operating changes to affect the flowrate be used to identify minimum heating and cooling utility and composition of what each sink/generator accepts and requirements for a process. Energy integration is beyond discharges. the scope of this paper but the reader is referred to the In general, sources must be prepared for the sinks numerous articles that have been published on the subthrough segregation and separation via a waste-intercepject (for example see reviews by Shenoy 1995; Linnhoff et tion network (WIN) (Hamad et al. 1996; El-Halwagi et al al. 1994; Linnhoff 1993; Gundersen and Naess 1988). 1996; El-Halwagi and Spriggs 1996; Garrison et al. 1995). On the other hand, mass integration is a systematic Effective pollution prevention can be achieved by a commethodology that provides a fundamental understanding bination of stream of the global flow of mass within the process and em– Segregation, ploys this understanding in identifying performance tar– Mixing, gets and optimizing the generation and routing of species – Recycle, throughout the process. Mass-allocation objectives such – Interception, and as pollution prevention are at the heart of mass integra– Sink/generator manipulation. tion. Mass integration is more general and more involved The following sections summarize these concepts. than energy integration. Because of the overriding mass objectives of most processes, mass integration can poten- Segregation simply refers to avoiding the mixing of tially provide much stronger impact on the process than streams. In many cases, segregating waste streams at the energy integration. Both integration branches are compa- source renders several streams environmentally acceptatible. Mass integration coupled with energy integration ble and hence reduces the pollution-prevention cost. Furprovides a systematic framework for understanding the thermore, segregating streams with different composibig picture of the process, identifying performance tartions avoids unnecessary dilution of streams. This regets, and developing solutions for improving process effi- duces the cost of removing the pollutant from the segreciency including pollution prevention. The next sections gated streams. It may also provide composition levels elaborate on mass-integration techniques. For a more de- that allow the streams to be recycled directly to process units.
Branches of process integration
Fig. 2. Schematic representation of mass-integration strategies for pollution prevention; segregation, mixing, interception, recycle and sink/generator manipulation
Recycle refers to the utilization of a pollutant-laden stream (a source) in a process unit (a sink). Each sink has a number of constraints on the characteristics (e.g. flowrate and composition) of feed that it can process. If a source satisfies these constraints it may be directly recycled to or reused in the sink. However, if the source violates these constraints segregation, mixing, and/or interception may be used to prepare the stream for recycle. A particularly useful tool for identifying recycle strategies is the source-sink mapping diagram. It is a visualization tool that can be used to determine direct-recycling opportunities (Fig. 3). For each pollutant, a diagram is constructed by plotting the pollutant load (flowrate! composition) or flowrate versus composition. On the sourcesink mapping diagram, sources are represented by shaded circles and sinks are represented by hollow circles. Typically, process constraints limit the range of pol-
9
Clean Products and Processes 1 (1998)
10
Interception denotes the utilization of separation unit operations to adjust the composition of the pollutant-laden streams to make them acceptable for sinks. These separations may be induced by the use of mass-separating agents (MSAs) and/or energy separating agents (ESAs). A systematic technique is needed to screen the multitude of separating agents and separation technologies to find the optimal separation system. The synthesis of MSA-induced physical-separation systems is referred to as the synthesis of mass-exchange networks (MENs) (El-Halwagi and Manousiouthakis 1989a). Interception networks using reactive MSAs are termed reactive mass exchange networks (REAMEN) (Srinivas and El-Halwagi 1994a; El-Halwagi and Srinivas 1992). Network synthesis techniques have also been devised for other separation systems that can be used in intercepting pollutants. These systems include pressure-driven membrane separations (e.g., Zhu et al. 1997; El-Halwagi 1992, 1993; Evangelista 1986), heat-induced separation networks (HISENs) (e.g., Dunn and Fig. 3. Identification of recycle opportunities using source-sink Srinivas 1997; Dunn et al. 1995; Dye et al. 1995; Richburg mapping diagram and El-Halwagi 1995; El-Halwagi et al. 1995; Dunn and El-Halwagi 1994a,b, 1996) and distillation sequences (e.g., lutant composition and load that each sink can accept. Malone and Doherty 1995; Wahnschafft et al. 1991). The intersection of these two bands provides a zone of As an example of synthesizing cost-optimum intercepacceptable composition and load for recycle. If a source tion networks, let us consider the problem of designing (e.g., source a) lies within this zone, it can be directly re- an MEN. The problem of synthesizing MENs has been incycled to the sink (e.g., sink S). Moreover, sources b and troduced by El-Halwagi and Manousiothakis (1989a) and c can be mixed using the lever-arm principle (or compo- can be described as follows: Given a number of rich nent material balance) to create a mixed stream that can streams (streams containing targeted species, also called be recycled to sink S. Lever-arm principles can also be sources) and a number of lean streams (MSAs), it is deused to prioritize recycle alternatives (El-Halwagi 1997). sired to synthesize the optimal mass exchange network to Similarly, multiple components can be handled simul- selectively transfer the targeted species from the rich taneously. Figure 4 shows a three-component source-sink streams to the lean streams so as to meet environmental mapping diagram (Parthasarathy and El-Halwagi 1997). or process requirements. Examples of mass exchange opOn this diagram, ternary lever-arm rules can be emerations include stripping, absorption, extraction, washployed to determine the extent of mixing among the var- ing, adsorption, etc. To optimize the cost of mass sepaious streams to satisfy sink constraints. rating agents, the mass-pinch analysis can be utilized (ElThe source-sink mapping diagram can also be used to Halwagi and Manousiouthakis 1989a). As a first step in determine the extent of interception needed. If a source constructing the pinch diagram, each rich (waste) stream lies to the right of a sink, it can be intercepted to bring it is represented as an arrow whose tail and head represent within the band of acceptable recycle. The problem of si- the supply and target compositions of the stream, respecmultaneously intercepting several sources using various tively (Fig. 5). The vertical distance between the head and separation technologies is the focus of the next section. the tail of the arrow represents the mass load to be removed of the targeted species from the stream and the slope of the arrow gives the flowrate of the stream. Next, the “diagonal rule” for superposition is employed to provide a combined representation of both rich streams by adding up mass load in overlapped regions of stream (Fig. 5). This representation results in the rich composite stream (Fig. 5). The same procedure can be adapted for all the rich streams to develop a global rich composite stream. The rich composite stream represents the cumulative mass of targeted species to be removed using MSAs. Having addressed all rich streams, we now turn our attention to the lean streams. A particularly useful concept in screening MSAs is the notion of “corresponding composition scales.” It is a tool for incorporating thermodynamic constraints of mass exchange by establishing a one-to-one correspondence among the compositions of Fig. 4. A ternary source-sink mapping diagram (Parthasarathy all streams for which mass transfer is thermodynamically and El-Halwagi 1997)
M.M. El-Halwagi: Pollution prevention through process integration
lying on the practical-feasibility line, two statements can be made. For a given yi, the value xj corresponds to the maximum composition of the pollutant that is practically achievable in the MSA. Conversely, for a given xj, the value yi corresponds to the minimum composition of the pollutant in the waste stream that is needed to practically transfer the pollutant from the waste stream to the MSA. It is important to derive the mathematical expression relating yi and xj on the practical-feasibility line. To avoid an infinite unit (e.g. column) size, a minimum allowable composition difference (ε) is required. This parameter is used to trade off capital versus operating costs so as to minimize total annualized cost. Details on selecting (ε) are described in literature (e.g. El-Halwagi 1997; El-Halwagi and Manousiouthakis 1990a). For a given yi,the values of xj can be obtained by evaluating xj* that is in equilibrium with yi then subtracting εj, i.e.,
xjpxj*Pεj
(2a)
or
xj*pxjcεj
(2b)
Substituting from (2b) into (1), one obtains
yipmj(xjcεj)cbj
(3a)
or
xj p
yPbj P εj . mj
(3b)
Eq. (3) can be used to establish a one-to-one correspondence among all composition scales for which mass exchange is feasible. We are now in a position to screen the various MSAs. Eq. (3) is employed to generate the correspondence among the rich composition scale, y, and the lean composition scales for all external MSAs. Each external MSA is then represented versus its composition scale as a horizontal arrow extending between its supply and target compositions (Fig. 6). Several useful insights can be gained from this diagram. Let us consider three MSAs; S1, S2 and S3 whose costs ($/kg of recirculating
Fig. 5. a Representing rich streams. b Using superposition to create composite stream. c The rich composite stream
feasible. To demonstrate this concept, let us consider a mass exchanger for which the equilibrium relation governing the transfer of the pollutant from the waste stream, i, to the MSA, j, is given by the linear expression
yipmj xj*cbj,
(1)
which indicates that for a waste stream composition of yi, the maximum theoretically attainable composition of the MSA is xj*. By employing a minimum allowable composition difference of εj, one can draw a “practical-feasibility line” that is parallel to the equilibrium line but offset to its left by a distance εj. In order for an operating line to be practically feasible, it must lie in the region to the left of the practical-feasibility line. Hence, for any pair (yi, xj) Fig. 6. Screening of external MSA’s
11
Clean Products and Processes 1 (1998)
MSA) are c1, c2and c3, respectively. These costs can be converted into $/kg of removed pollutant, c rj, as follows:
12
pulp and paper (Hamad et al. 1995, 1998; Dunn and ElHalwagi 1993), synthetic fuels (Warren et al. 1995), petrochemicals (Stanley and El-Halwagi 1995), polymers cj (Spriggs 1995), and metal finishing (El-Halwagi and Ma(4) c rj p t where jp1, 2, 3. x j Px sj nousiouthakis 1990a). In addition, many examples illustrating the detailed application of MEN’s and mass inteIf arrow S2 lies completely to the left of arrow S1 and r r c 2 is less than c 1, one can eliminate S1 from the problem gration to pollution prevention are published in a recent textbook (El-Halwagi 1997). since it is thermodynamically and economically inferior to S2. On the other hand, if arrow S3 lies completely to the left of arrow S2 but c r3 is greater than c r2, one should Sink/generator manipulation involves design or operating changes that alter the flowrate or composition of polretain both MSAs. In order to minimize the operating lutant-laden streams entering or leaving the process cost of the network, separation should be staged to use units. These measures include temperature/pressure the cheapest MSA where it is feasible. Hence, S2 should changes, unit replacement, catalyst alteration, feedstock be used to remove all the rich load to its left while the substitution, reaction-path changes (e.g., Crabtree and Elremaining rich load is removed by S3 (Fig. 6). The flowrates of S2 and S3 are calculated by simply dividing the Halwagi 1995), reaction system modification (e.g., Gopalakrishnan et al. 1996; Lakshmanan and Biegler 1995), and richload removed by the composition difference for the solvent substitution (e.g. Joback 1994; Joback and StephaMSA. Now that the MSAs have been screened and their nopoulos 1990; Constantinou et al. 1995). optimal flowrates have been determined, one can conIn order to demonstrate how some of mass integration struct the pinch diagram as shown in Fig. 7. tools can be applied, let us revisit the motivating example The subject of physical mass exchange networks has on acrylonitrile production. been addressed extensively in literature. This includes MENs with a single transferable component (El-Halwagi Acrylonitrile example revisited and Manousiouthakis 1989, 1990a), those with multiple transferable components (El-Halwagi and Manousioutha- The first step in the analysis is to identify the target for kis 1989b), those involving regeneration of the MSAs (El- debottlenecking the facility. An overall water balance for the plant (Fig. 8) can be written as follows: Halwagi and Manousiouthakis 1990b) and, mass exchange combined with heat exchange (Srinivas and ElWater in c Water generated by chemical reaction p Halwagi 1994a), removal of fixed loads (Kiperstok and Wastewater out c Water losses Sharratt 1995), variable supply and target compositions Since the wastewater discharge is larger than fresh wa(Garrison et al. 1994), fixed-cost targeting (Hallale and Fraser 1997), MENs providing flexible performance (Zhu ter flowrate, it is possible, in principle, to bring wastewater to a quality that can substitute fresh water using segreand El-Halwagi 1995; Papalexandri and Pistikopoulos 1994), controllable MENs (Huang and Edgar 1995; Huang gation, mixing, recycle and interception. Furthermore, and Fan 1995), and MENs with a single lean stream (wa- sink/generator manipulation can be employed to reduced flowrate of fresh water. Hence, fresh-water usage in this ter) with the objective of minimizing water use (Wang and Smith 1994; Dhole et al. 1996; Kuo and Smith 1998). example can, in principle, be completely eliminated, and for the same reaction conditions and water losses, the Many industrial applications of species interception target for wastewater discharge can be calculated from have also been published including petroleum refining (El-Halwagi and El-Halwagi 1992; El-Halwagi et al. 1992), the overall water balance as follows (Fig. 8):
Fig. 7. Constructing the pinch diagram for external MSA’s
Fig. 8a,b. Establishing targets for biotreatment influent: overall water balance a before and b after mass integration
M.M. El-Halwagi: Pollution prevention through process integration
13
Fig. 9. Segregation, mixing, interception and recycle representation for the AN case study
As can be seen from Fig. 10, the flowrate of recycled gas condensate is 4.1 kg/s and the flowrate of fresh water is 0.9 kg/s (5.8–0.8–4.1). Therefore, direct recycle can reHaving identified this target, let us now determine the duce the fresh-water consumption (and consequently the strategies needed to attain it. To determine segregation, influent to biotreatment) by 5.1 kg/s. mixing, and direct recycle opportunities, the sources and The primary cost of direct recycling is pumping and sinks should be examined as shown in Fig. 9. piping. Assuming that the TAC for pumping and piping Because of the stringent limitation on the BFW (no $80 ammonia or AN), no recycled stream can be used in lieu is and assuming that the total length of piping is m.yr of fresh water (segregation, mixing, recycle and intercep600 m, the TAC for pumping and piping is $48,000/yr. tion can reduce, but not eliminate, ammonia/AN conNext, we include interception in the analysis. In order tent). Hence, the boiler should not be considered as a sink for recycle (with or without interception). Instead, it to eliminate fresh water from the scrubber, the composishould be handled at the stage of sink/generator manipu- tion of ammonia in the off-gas condensate must be relation. This leaves us with the five segregated sources and duced from 14 ppm to 12 ppm. This result may be obone sink (scrubber) for the source-sink mapping diagram tained graphically as shown in Fig. 11. In order to synthesize an optimal MEN for intercept(Fig. 10). The source-sink mapping diagram can be used to determine maximum extent of direct recycle and nec- ing the off-gas condensate, we construct the mass-pinch diagram as shown in Fig. 12. Since the three MSA’s lie essary interception duty. completely to the left of the rich stream, they are all thermodynamically feasible. Hence, we choose the one with the least cost ($/kg NH3 removed); namely the resin. The TAC for removing ammonia using the resin is $119,000/ yr. As a result of segregation, interception, and recycle, we have eliminated the use of fresh water in the scrubber, leading to a reduction in fresh water consumption (and influent to biotreatment) by 6.0 kg/s. Therefore, the target for segregation, interception, and recycle has been realized. Next, we focus our attention on sink/generator manipulation to remove fresh-water consumption in the steam-jet ejector. The challenge here is to alter the design and/or operation of the boiler, the ejector, or the distillation column to reduce or eliminate the use of steam. Several solutions may be proposed including: – Replacing of the steam-jet ejector with a vacuum pump. The distillation operation will not be affected. The operating cost of the ejector and the vacuum pump are comparable. However, a capital investment of $75,000 is needed to purchase the pump. For a fiveyear linear depreciation with negligible salvage value, the TAC of the pump is $15,000/year. Fig. 10. Direct-recycle opportunities for the AN case study Target of minimum discharge to biotreatmentp5.1P0.3 p4.8 kg water/s. (5)
Clean Products and Processes 1 (1998)
14
Fig. 12. The mass-pinch diagram for the AN case study Fig. 11. Interception and recycle opportunities for the AN case study
– Operating the column under atmospheric pressure, thereby eliminating the need for the vacuum pump. Here a simulation study is needed to examine the effect of pressure change. – Relaxing the requirement on BFW quality to a few parts per million of ammonia and AN. In this case, recycle and interception techniques can be used to significantly reduce the fresh water feed to the boiler and, consequently, the net wastewater generated.
Figure 13 illustrates the revised flowsheet with segregation, interception, recycle and sink/generator manipulation. As can be seen from the figure, the flowrate of the terminal wastewater stream has been reduced to 4.8 kg H2O/s. This is exactly the same target predicted in Fig. 8. In order to refine the material balance throughout the plant, a simulation study is needed, as discussed earlier.
The role of chemistry and molecular design in pollution prevention So far, the article has focused on process integration aspects of pollution prevention. Another important element
Fig. 13. Optimal solution to the AN case study with segregation, recycle, interception and sink/ generator manipulation
M.M. El-Halwagi: Pollution prevention through process integration
of pollution prevention is the selection of environmentally-benign chemical reactions, raw materials, solvents and products. Over the past few years, significant progress has been made in this area. This section provides a brief overview of the recent advances in synthesizing “green” reactions and species. For more detailed discussion, the reader is referred to El-Halwagi (1997) Anastas and Williamson (1996), Anastas and Farris (1994), and Chase (1995).
Synthesis of environmentally acceptable reactions For a given desired product, there are typically numerous reaction alternatives that should be identified and screened. The identification of these reactions is not a straightforward task. The problem is further compounded when the search is limited to environmentally-benign cost-effective chemistry. In this context, the following questions should be answered: – What raw materials should be employed? – What is the yield of the desired product? – What is the distribution of byproducts? Are all byproducts environmentally acceptable? – How can undesirable species be removed or minimized? – What are the operating conditions that insure thermodynamic feasibility of these reactions? – Is a catalyst needed to achieve desired kinetics? What catalyst should be employed? – How should the reaction system by sized and configured? In order to address these challenges, a hierarchical approach may be adopted. This approach focuses on the “big picture” first, then adds details to promising solutions. Therefore, preliminary screening ought to be conducted first to identify overall reaction alternatives that meet process requirements in terms of desired product, cost effectiveness, environmental acceptability, and thermodynamic feasibility. At this stage, minimum details are to be invoked. The problem of synthesizing environmentally-acceptable reactions “EAR’s” has been introduced by Crabtree and El-Halwagi (1994) and can be stated as follows: Given a reactor of known size and functionality, and a desired product along with its flowrate, synthesize an overall chemical reaction that features maximum economic potential while complying with all environmental and thermodynamic constraints. In order to generate a candidate EAR, one should consider potential raw materials and by-products, satisfaction of stoichiometric conditions, assurance of thermodynamic feasibility and fulfillment of environmental requirements. These issues can be addressed by employing an optimization formulation to identify an overall reaction that yields the desired product at maximum economic potential while satisfying stoichiometric, thermodynamic and environmental constraints. For a more detailed description of this optimization program, the reader is referred to Crabtree and El-Halwagi (1994). In order to shed more light on synthesizing EARs, it is instructive to discuss the following case study on producing 1-naphthyl-N-methyl carbamate (or carbaryl). This
product was produced in Bhopal (India) by reacting anaphthol and methyl isocyanate “MIC”. In 1984, approximately 50 tons of MIC underwent a chemical reaction and leaked into the atmosphere forming a toxic cloud which killed over 2,000 people and injured more than 250,000. It is possible that this tragic accident could have been averted if carbaryl had been produced from a more environmentally-acceptable reaction. Let us consider the production of 3 tons/hr of carbaryl in a 30 m 3 CSTR. By applying the EARs synthesis procedure of Crabtree and El-Halwagi (1994), the following optimal EAR is identified:
Methyl Formamide
a-Naphthol
Carbaryl
Hydroge (6)
This reaction has an economic potential (gross profit) of $40.05 million per year. An advantage of the optimization-based approach to synthesizing EARs is its ability to generate next-to-optimal solutions. This can be achieved by adding constraints to the optimization program in order to guarantee that the previous solution is not obtained again. Hence, by excluding the optimal EAR and solving the optimization program, we get the following reactions: 0.5 OxygencMethyl Formamideca-Naphtholp CarbarylcWater
(7)
which yields an economic potential of $39.99 million per year, and
Methyl Amine a-Naphthol Chloroformate Carbaryl (8) This reaction, while thermodynamically feasible and environmentally acceptable, shows no sign of commercial profitability. It corresponds to a gross revenue of $–44,300 per hour; a major loss. Hence, no additional effort should be invested in investigating kinetics, catalysis or reactor design for this reaction. On the other hand, reactions (6) and (7) are promising candidates that should be experimentally investigated to validate their feasibility and fine tune their optimal conditions. Once these overall reactions are chosen, it is important to undertake studies on reaction path synthesis to identify the reaction mechanism and the elemental steps involved (e.g. Chase 1995; Douglas 1992; Mavrovouniotis et al. 1990, 1992; Fornari et. al. 1989; Rotstein et al. 1982; Govind and Powers 1981; Rudd and May 1976; Rudd 1976; Rudd et al. 1973, etc.). Next, catalysis and kinetics should
15
Clean Products and Processes 1 (1998)
16
be investigated to manipulate the reaction rates (e.g. Dartt and Davis 1994; Haggin 1994; Sinfelt 1987; Boudart and Djega-Mariadassou 1984; El-Halwagi 1971). Finally, reactor design and optimization is undertaken and linked with separation systems (e.g. Muralikrishnan et al. 1996; Sund and Lien 1996; Friedler et al. 1994; Fogler 1992; Balakrishna and Biegler 1992; Kokossis and Floudas 1991; Hildebrandt et al. 1990; El-Halwagi 1990; El-Halwagi and El-Rifai 1988; Glasser et al. 1987; Levenspiel 1972; Horn 1965; Fan et al. 1965). This hierarchical approach insures that only relevant details are examined at each stage of the analysis and that only promising reaction routes are investigated.
Synthesis of environmentally benign species In many cases, it is possible to replace environmentally hazardous chemicals with more benign species without compromising the technical and economic performance of the process. Examples include alternative solvents, polymers and refrigerants. Group contribution methods have been commonly used in predicting physical and chemical properties of synthesized materials. Two main frameworks have been employed to synthesize alternative materials: knowledge base and computer-aided optimization. Knowledge-based approaches depend on understanding the criteria of the materials to be replaced along with general rules and algorithms that link properties with structure. Examples of this approach can be found in literature (e.g., Joback 1994; Joback and Stephanopoulos 1990; Constantinou et al. 1994). Furthermore, software can be used to screen solvents based on their properties and performance. An example of this approach is the PARIS (Program for Assisting in the Replacement of Industrial Solvents) software (e.g. US EPA 1994; Hilaly and Sikdar 1996; Cabezas and Zhao 1998). Computer-aided optimization approaches are based on formulating the molecular design problem as an optimization program which seeks to maximize a performance function or minimize deviation from desired properties subject to various constraints including structural feasibility, propertystructure correlations and environmental criteria. Examples of this approach include synthesis of solvents (e.g., Brignole et al. 1986; Odele and Macchietto 1990; Naser and Fournier 1991; Dunn et al. 1997, 1995; Hamad and El-Halwagi 1998), polymers (e.g., Vaidyanathan and ElHalwagi 1994 and 1996; Venkatasubramanian et al. 1994) and refrigerants (e.g., Achenie and Duvedi 1996). Process synthesis techniques can be integrated with product synthesis tools. As an illustration, let us revisit the problem of synthesizing mass-exchange networks (MENs) described earlier. A key condition for applying the mass-pinch techniques described earlier is that the designer has a set of candidate mass-separating agents (MSAs). Therefore, it may be necessary to solve a product synthesis problem to generate the candidate MSAs needed for the mass-pinch approach. This problem of simultaneously synthesizing MSAs and MENs has been addressed by Hamad et al. (1998) and can be stated as follows: “Given a set of chemical functional groups, it is desired to synthesize a set of MSAs to selectively transfer certain targeted species (e.g., pollutants) from a set of
Fig. 14. Schematic representation of simultaneous synthesis of MSAs and MEN (Hamad and El-Halwagi 1998)
rich streams at minimum cost. These MSAs must meet technical, environmental and safety constraints. First, candidate chemical groups are chosen as building blocks for the solvents. A common choice is the UNIFAC functional groups. These groups are, in general, classified into three categories: extenders, branches and terminators. Special rules are available for linking these groups in a structurally-feasible manner (e.g., Vaidyanathan et al. 1994b, 1996b). Next, property-structure correlations such as group-contribution methods are selected. The fundamental assumption of most group-contribution methods is that the contribution toward any property made by one group is independent of that made by another group. Mass-exchange equilibrium may be predicted using thermodynamic models such as the UNIFAC method and the regular solution theory. Finally, MSA solvent constraints are integrated with MEN synthesis formulation in a mixed-integer nonlinear program which minimizes the cost of the MSAs and the MEN subject to MSA-synthesis constraints, MEN synthesis constraints, technical requirements as well as safety and environmental restrictions.
Conclusions Process integration provides an attractive framework for cost-effective pollution prevention. In particular, mass-integration science and tools present a systematic and insightful way of developing pollution-prevention strategies that address the root cause of the environmental problems. These strategies are typically simple, implementable but not intuitively obvious. Because of the fundamental nature of this approach, it is generally applicable to processing facilities throughout the chemical process industry. The comprehensive nature of this approach enables the reconciliation of the various process techno-economic process with environmental targets and constraints.
M.M. El-Halwagi: Pollution prevention through process integration
References Achenie LEK, Duvedi AP: (1996) Designing environmentally safe refrigerants using mathematical programming. Chem Eng Sci 51(15):3727–3739 Anastas PT, Williamson TC (eds): (1996) Green chemistry: designing chemistry for the environment. ACS Symp Ser, vol. 626, ACS Pub, Washington, DC Anastas PT, Farris CA (eds): (1994) Benign by design: alternative synthetic design for pollution prevention. ACS Symp Ser, vol. 577, ACS Pub, Washington, DC Balakrishna S, Biegler LT: (1992) Targeting strategies for the synthesis and energy integration of nonisothermal reactor networks. Ind Eng Chem Res 31 : 2152–2164 Beveridge GSG, Schechter R: (1970) Optimization: theory and practice. McGraw Hill, New York Boudart M, Djega-Mariadassou G: (1984) Kinetics of heterogeneous catalytic reactions. Princeton University Press, Princeton, NJ Brignole EA, Bottini S, Gani R: (1986) A strategy for the design and selection of solvents for separation processes. Fluid Phase Equilibria 29 : 125–132 Cabezas H, Zhao R: (1998) Designing environmentally benign solvents: physical property considerations. AIChE Spring Meeting, New Orleans, March Chase V: (1995) Green chemistry: the middle way to a cleaner environment. R & D Magazine 25–26, August Constantinou L, Jacksland C, Bagherpour K, Gani R, Bogle L: (1994) Application of group contribution approach to tackle environmentally-related problems. AIChE Symp Ser 90(303):105–116 Crabtree EW, El-Halwagi MM: (1994) Synthesis of environmentally-acceptable reactions. AIChE Symp Ser 90(303):117–127 Dartt CB, Davis ME: (1994) Catalysis for environmentally benign protection. Ind Eng Chem Res 33 : 2887–2899 Dhole VR, Ramchandani N, Tainsh RA, Wasilewski M: (1996) Make your process water pay for itself. Chem Eng January, pp 100–103 Douglas JM: (1992) Process synthesis for waste minimization. Ind Eng Chem Res 31 : 238–243 Dunn RF, Dobson AM, El-Halwagi MM: (1997) Optimal design of environmentally acceptable solvent blends for coating. Adv Env Res 1(2):243–252 Dunn RF, El-Halwagi MM: (1993) Optimal recycle/reuse policies for minimizing the wastes of pulp and paper plants. Environ Sci Health A28(1):217–234 Dunn RF, El-Halwagi MM: (1996) Design of cost-effective VOC recovery systems. TVA Pub, Muscle Shoals, AL Dunn RF, El-Halwagi MM: (1994a) Optimal design of multicomponent VOC Condensation Systems. J Hazard Mater 38 : 187–206 Dunn RF, El-Halwagi MM: (1994b) Selection of optimal VOC condensation systems. J Waste Manage 14(2):103–113 Dunn RF, Srinivas BK: (1997) Synthesis of heat-induced waste minimization networks (HIWAMINs). Adv Env Res 1 (3), pp 275–301 Dunn RF, Zhu M, Srinivas BK, El-Halwagi MM: (1995) Optimal design of energy induced separation networks for VOC recovery. AIChE Symp Ser 90(303):74–85 Dunn RF, El-Halwagi MM, Lakin J, Serageldin M: (1995) Selection of organic solvent blends for environmental compliance in the coating industries. In: Griffith ED, Kahn H, Cousins MC (eds) Proceedings of the First International Plant Operations and Design Conference, Vol. III, AIChE, New York, pp 83–107 Dye SR, Berry DA, Ng KM: (1995) Synthesis of crytallizationbased separation schemes. AIChE Symp Ser 91(304):238–241 Edgar TF, Himmelblau DM: (1988) Optimization of chemical processes. McGraw Hill, New York
El-Halwagi MM: (1997) Pollution prevention through process integration: systematic design tools. Academic Press, San Diego El-Halwagi MM: (1993) A process synthesis approach to the dilemma of simultaneous heat recovery, waste reduction and cost effectiveness. In: El-Sharkawy AI, Kummler RH (eds) Proceedings of the Third Cairo International Conference on Renewable Energy Sources, Vol. 2, pp 579–594 El-Halwagi MM: (1993) Optimal design of membrane hybrid systems for waste reduction. Sep Sci Technol 28(1–3):283–307 El-Halwagi MM: (1992) Synthesis of reverse osmosis networks for waste reduction. AIChE J 38(8):1185–1198 El-Halwagi MM: (1990) Optimization of bubble column slurry reactors via natural delayed feed addition. Chem Eng Comm 92 : 103–119 El-Halwagi MM: (1971) An engineering concept of reaction rate. Chem Eng 75–78, May El-Halwagi AM, El-Halwagi MM: (1992) Waste minimization via computer aided chemical process synthesis – A new design philosophy. TESCE J 18(2):155–187 El-Halwagi MM, El-Halwagi AM, Manousiouthakis V: (1992) Optimal design of dephenolization networks for petroleumrefinery wastes. Trans Inst Chem Eng 70 : Part B, 131–139 El-Halwagi MM, El-Rifai MA: (1988) Mathematical modeling of fluidized-bed reactors – I. The multistage three phase model. Chem Eng Sci 43(9):2477–2486 El-Halwagi MM, Hamad AA, Garrison GW: (1996) Synthesis of waste interception and allocation networks. AIChE J 42(11):3087–3101 El-Halwagi MM, Manousiouthakis V: (1989a) Synthesis of massexchange networks. AIChE J 35(8):1233–1244 El-Halwagi MM, Manousiouthakis V: (1989b) Design and analysis of mass exchange networks with multicomponent targets. AIChE Annu Meet, San Francisco, November El-Halwagi MM, Manousiouthakis V: (1990a) Automatic synthesis of mass exchange networks with single-component targets. Chem Eng Sci 45(9):2813–2831 El-Halwagi MM, Manousiouthakis V: (1990b) Simultaneous synthesis of mass exchange and regeneration networks. AIChE J 36(8):1209–1219 El-Halwagi MM, Petrides DM: (1995) Pollution prevention via process and product modifications. AIChE Symp Ser 90(303), AIChE, New York El-Halwagi MM, Spriggs HD: (1998) Employ mass integration to achieve truly integrated process design. Chem Eng Prog, August El-Halwagi MM, Spriggs HD: (1996) An integrated approach to cost and energy efficient pollution prevention. Proceedings of Fifth World Congr of Chem Eng Vol. III, San Diego, pp 344–349 El-Halwagi MM, Srinivas BK: (1992) Synthesis of reactive mass exchange networks. Chem Eng Sci 47(8):2113–2119 El-Halwagi MM, Srinivas BK, Dunn RF: (1995) Synthesis of optimal heat-induced separation networks. Chem Eng Sci 50(1):81–97 Evangelista F: (1986) Improved graphical analytical method for the design of reverse osmosis desalination plants. Ind Eng Chem Process Des Dev 25(2):366–375 Fan LT, Erickson LE, Sucher RW, Mathad GS: (1965) Optimal design of a sequence of continuous-flow stirred-tank reactors with product recycle. Ind Eng Chem 4 : 432–440 Fornari T, Rotstein E, Stephanopoulos G: (1989) Studies on the synthesis of chemical reaction paths – II. Reaction schemes with two degrees of freedom. Chem Eng Sci 44(7):1569–1579 Fogler SH: (1992) Elements of chemical reaction engineering. Prentice Hall, Englewood Cliffs, New Jersey Freeman H: (1995) Industrial pollution prevention handbook. McGraw Hill, New York Friedler F, Varga JB, Fan LT: (1995) Algorithmic approach to
17
Clean Products and Processes 1 (1998)
18
the integration of total flowsheet synthesis and waste minimization. AIChE Symp Ser 90(303):86–97, AIChE, NY Garrison GW, Spriggs HD, El-Halwagi MM: (1996) A global approach to integrating environmental, energy, economic and technological objectives. Proceedings of Fifth World Congr of Chem Eng Vol. I, San Diego, pp 675–680 Garrison GW, Hamad AA, El-Halwagi MM: (1995a) Synthesis of waste interception networks. AIChE Annu Meet, Miami Garrison GW, Cooley BL, El-Halwagi MM: (1995b) Synthesis of mass exchange networks with multiple target mass separating agents. Dev Chem Eng Miner Proc 3(1) : 31–49 Gopalakrishnan M, Ramdoss P, El-Halwagi M: (1996) Integrated design of reaction and separation systems for waste minimization. AIChE Annu Meet, Chicago Glasser D, Hildebrandt D, Crowe C: (1987) A geometric approach to steady flow reactors: the attainable region and optimization in concentration space. Ind Eng Chem Res 26 : 1803–1810 Govind R, Powers GJ: (1981) Studies in reaction path synthesis. AIChE J 27(3):429–442 Grossmann IE: (1996) Global optimization in engineering design. Kluwer Academic Pub, Dordrecht, The Netherlands Gundersen T, Naess L: (1988) The synthesis of cost optimal heat exchanger networks: an industrial review of the state of the art. Comput Chem Eng 12(6) : 503–530 Haggin J: (1994) Catalysis gains widening role in environmental protection. Chem Eng News 72 : 22–30 Hallale N, Fraser DM: (1997) Synthesis of cost optimum gas treating process using pinch analysis. In: Ho WS, Luo RG (eds) Proceedings, Top Conf on Sep Sci and Techs, Part II, AIChE, New York, pp 1,708–1,713 Hamad AA, Varma V, Krishnagopalan G, El-Halwagi MM: (1998) Application of mass integration to reduce methanol and effluent discharge in pulp mills. TAPPI J 81(10) : 171–180 Hamad AA, El-Halwagi MM: (1998) Simultaneous synthesis of mass separating agents and interception networks. Trans I.Chem.E, 76, Part A, pp 376–388 Hamad AA, Garrison GW, Crabtree EW, El-Halwagi MM: (1996) Optimal design of hybrid separation systems for waste reduction. Proceedings of Fifth World Congr of Chem Eng Vol. III, San Diego, pp 453–458 Hamad AA, Varma V, El-Halwagi MM, Krishnagopalan G: (1995) Systematic integration of source reduction and recycle reuse for the cost-effective compliance with the cluster rules. AIChE Annu Meet, Miami Hilaly A, Sikdar S: (1994) Pollution balance: a new methodology for minimizing waste in manufacturing process. J Air Waste Manage Assoc 44 : 1303–1308 Hilaly A, Sikdar S: (1996) Process simulation toos for pollution prevention: new methods reduce the magnitude of waste streams. Chem Eng, pp 98–105, Feb Hildebrandt D, Glasser D, Crowe C: (1990) Geometry of the attainable region generated by reaction and mixing: with and without constraints. Ind Eng Chem Res 29 : 49–58 Horn F: (1965) Attainable and non-attainable regions in chemical reaction technique. Third European Symposium on Chemical Reaction Engineering, Pergammon Press, London, pp 293–302 Huang YL, Fan LT: (1995) Intelligent process design and control for in-plant waste minimization. In: Rossiter AP (eds) Waste minimization through process design. McGraw Hill, New York, pp 165–180 Huang YL, Edgar TF: (1995) Knowledge based design approach for the simultaneous minimization of waste generation and energy consumption in a petroleum refinery. In: Rossiter AP (eds) Waste minimization through process design. McGraw Hill, New York, pp 181–196 Joback KG: (1994) Solvent substitution for pollution prevention. AIChE Symp Ser Vol. 90, Number 303 : 98–104
Joback KG, Stephanopoulos G: (1990) Designing molecules possessing desired physical property values. In: Siirola JJ, Grossmann I, Stephanopoulos G (eds) Foundations of Computer Aided Process Design ’FOCAPD’ III, CACHE/Elsevier, New York, pp 363–387 Kiperstok A, Sharratt PN: (1995) On the optimization of mass exchange networks for removal of pollutants. Trans Inst Chem Eng 73 : Part B, 271–277 Kokossis AC, Floudas CA: (1994) Optimization of complex reactor networks: nonisothermal operation. Chem Eng Sci 49 : 1037–1051 Kuo WCJ, R Smith: (1998) Designing for the interactions between water use and effluent treatment. Trans Inst Chem Eng 76 : Part A, 287–301 Lakshmanan A, Biegler LT: (1995) Reactor network targeting for waste minimization. AIChE Symp Ser 90(303):128–138 Levenspiel O: (1972) Chemical reaction engineering. Second Edition, John Wiley and Sons, New York Linnhoff B, Townsend DW, Boland D, Hewitt GF, Thomas BEA, Guy AR, Marsland RH: (1994) A user guide on process integration for the efficient use of energy. Revised 1st Ed, Institution of Chemical Engineers, Rugby, UK Linnhoff B: (1993) Pinch analysis – A state of the art overview. Trans Inst Chem Eng Chem Eng Res Des 71 : Part A5, 503–522 Malone MF, Doherty MF: (1995) Separation system synthesis for nonideal liquid mixtures. AIChE Symp Ser 91(304):9–18 Mavrovouniotis ML, Stephanopoulos G: (1990) Computer-aided synthesis of biochemical pathways. Biotechnology and Bioengineering 36 : 1119–1132 Mavrovouniotis ML, Stephanopoulos G: (1992) Synthesis of reaction mechanisms consisting of reversible and irreversible steps. Ind Eng Chem Res 31 : 1625–1637 Muralikrishnan G, Ramdoss P, El-Halwagi MM: (1996) Integrated design of reaction and separation systems for waste minimization. AIChE Annu Meet, Chicago, November Naser SF, Fournier RL: (1991) A system for the design of an optimum liquid-liquid extractant molecule. Comp Chem Eng 15(6):397–414 Nishida N, Stephanopoulos G, Westerberg A: (1981) A review of process synthesis. AIChE J 27(3):321–351 Noyes R: (1993) Pollution prevention technology handbook. Noyes Publications, Park Ridge, NJ Odele O, Macchietto S: (1993) Computer aided molecular design: A novel method for optimal solvent selection. Fluid Phase Equilibria 82 : 47–54 Parthasarathy G, El-Halwagi MM: (1997) Mass integration for multicomponent nonideal systems. Presented at AIChE Annual Meeting (Los Angeles), AIChE, New York (Nov. 1997) Papalexandri KP, Pistikopoulos EN: (1994) A multiperiod MINLP model for the synthesis of heat and mass exchange networks. Comput Chem Eng 18(12):1125–1139 Reklaitis GV, Ravindran A, Ragsdell KM: (1983) Engineering optimization. Wiley, New York Richburg A, El-Halwagi MM: (1995) A graphical approach to the optimal design of heat-induced separation networks for VOC recovery. AIChE Symp Ser 91(304):256–259 Rotstein E, Resasco D, Stephanopoulos G: (1982) Studies on the synthesis of chemical reaction paths – I. Reaction characteristics in the (DG, T) space and a primitive synthesis procedure. Chem Eng Sci 37 (9):1337–1352 Rudd DF, Powers GJ, Siirola JJ: (1973) Process synthesis. Prentice Hall, Inc, New Jersey Rudd DF, May D: (1976) Development of solvay clusters of chemical reactions. Chem Eng Sci 31 : 59–69 Rudd DF: (1976) Accessible designs in solvay cluster synthesis. Chem Eng Sci 31 : 701–703 Sahinidis NV, Grossmann IE: (1991) Convergence properties of generalized Benders decomposition. Comput Chem Eng 15(7):481–491
M.M. El-Halwagi: Pollution prevention through process integration
Samdani S: (1995) Cleaner by design. Chem Eng, July 1995, pp 32–37 Shenoy UV: (1995) Heat exchange network synthesis: process optimization by energy and resource analysis. Gulf Pub Co, Houston, TX Sinfelt JH: (1987) Catalyst design: selected topics and examples. In: Liu YA, McGee HA, Epperly WR (eds) Recent developments in chemical process and plant design. John Wiley and Sons, New York, pp 1–41 Spriggs HD: (1995) Design for pollution prevention. AIChE Symp Ser 90 (303), pp 1–11 Srinivas BK, El-Halwagi MM: (1994a) Synthesis of reactive mass-exchange networks with general nonlinear equilibrium functions. AIChE J 40(3):463–472 Srinivas BK, El-Halwagi MM: (1994b) Synthesis of combined heat reactive mass-exchange networks. Chem Eng Sci 49(13):2059–2074 Srinivas BK, El-Halwagi MM: (1993) Optimal design of pervaporation systems for waste reduction. Comp Chem Eng 17 (10), pp 957–970 Stanley C, El-Halwagi MM: (1995) Synthesis of mass-exchange networks using linear programming techniques. In: Rossiter AP (eds) Waste minimization through process design. McGraw Hill, New York, pp 209–224 Stephanopoulos G, Townsend D: (1986) Synthesis in process development. Chem Eng Res Des 64(3):160–174 Sund EB, Lien K: (1996) An optimal control formulation of the reaction-mixing problem. AIChE Annu Meet, Chicago, November Theodore L, Dupont RR, Reynolds J: (1994) Pollution prevention: problems and solutions. Gordon & Breach, Amsterdam US EPA: (1994) Paris Manual. EPA, Cincinnati, Ohio Vaidyanathan R, El-Halwagi MM: (1994a) Global optimization of nonconvex nonlinear programs via interval analysis. Comput Chem Eng 18(10):889–897
Vaidyanathan R, El-Halwagi MM: (1994b) Computer-aided design of high performance polymers. J Elastomers Plastics 26 : 277–293 Vaidyanathan R, El-Halwagi MM: (1996a) Global optimization of nonconvex MINLP’s by interval analysis. In: Grossmann IE (ed) Global optimization in engineering design. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp 175–194 Vaidyanathan R, El-Halwagi MM: (1996b) Computer-aided synthesis of polymers and blends with target properties. Ind Eng Chem Res 35 : 627–634 Venkatasubramanian V, Chan K, Cauthers JM: (1994) Computer-aided molecular design using genetic algorithms. Comp Chem Eng 18 : 833–844 Visweswaran V, Floudas CA: (1990) A global optimization procedure for certain classes of nonconvex NLP’s – II. Application of theory and test problems. Comput Chem Eng 14(2):1419–1434 Wahnschafft OM, Jurian TP, Westerberg AW: (1991) SPLIT: a separation process designer. Comput Chem Eng 15 : 565–581 Wang YP, Smith R: (1994) Wastewater minimization. Chem Eng Sci 49(7):981–1006 Warren A, Srinivas BK, El-Halwagi MM: (1995) Design of costeffective waste-reduction systems for synthetic fuel plants. J Environ Eng 121(10):742–747 Westerberg AW: (1987) Process synthesis: A morphological view. In: Liu YA, McGee HA, Epperly WR (eds) Recent developments in chemical process and plant design. John Wiley and Sons, New York, pp 127–145 Zhu M, El-Halwagi MM: (1995) Synthesis of flexible mass exchange networks. Chem Eng Commun 138 : 193–211 Zhu M, El-Halwagi MM, M Alahmad: (1997) Optimal design and scheduling of flexible reverse-osmosis networks. J Membrane Sci 129 : 161–174
19