ASIA-PACIFIC JOURNAL OF CHEMICAL ENGINEERING Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 Published online 18 June 2009 in Wiley InterScience (www.interscience.wiley.com) DOI:10.1002/apj.292 Research Article Design and synthesis of separation process based on a hybrid method Chunshan Li,1 * Gu?nter Wozny2 and Kenzi Suzuki1 1 2 EcoTopia Science Institute, Nagoya Unviersity, Nagoya, Japan Technische Universita?t Berlin, Berlin, Germany Received 3 September 2008; Revised 15 February 2009; Accepted 16 February 2009 ABSTRACT: A new general hybrid methodology for separation process synthesis and design is proposed, which considers different separation technologies by integrating mathematical modeling ? Analytical Hierarchy Process (AHP) with heuristic approaches and thermodynamic insights. The methodology can provide suitable guidance for the initial separation process design and energy saving. Firstly, a general separation synthesis system based on thermodynamic insights is developed to select suitable separation techniques before sequencing, which reduces the complexity and size of synthesis search space. Then, the pseudo-component concept is proposed and used to deal with the azeotrope contained in the mixture, which widens the scope of the application of the proposed methodology. The AHP method is used to make a separation sequence by pairwise comparison matrices. Lastly, the separation of the pseudo-component will be considered, and it performs energy integration and a detailed process design. Application of the proposed methodology is highlighted through two industrial examples: one is the separation synthesis of a light-end refinery mixture. The other is azeotrope system, the mixture of phenol, o-cresol, p-cresol, and m-cresol. ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. KEYWORDS: separation synthesis; AHP; thermodynamic insight; mathematical method INTRODUCTION In order to obtain products of purity, the separation parts are crucial for every chemical process, whose capital and operating costs will occupy from 40 to 70% of that of the whole chemical plants. So finding suitable synthesis methodology for separation process is urgent. For each of the resulting separation tasks, more than one separation technique may be feasible. Furthermore, since more than one separation task will frequently be required, the best sequence of separation tasks must be established. Finally, a suitable and consistent set of operating conditions must be determined; thus the synthesis of separation process is a complex problem in chemical engineering.[1] In principle, separation process synthesis is based on either a heuristic, thermodynamic insight or mathematical programming approaches. Heuristic method uses the rule of thumb resulting from an engineering experience and insights from the physics and chemistry of the separation method, which does not require any mathematical background and computational skill for the users; *Correspondence to: Chunshan Li, EcoTopia Science Institute, Nagoya Unviersity, Nagoya, 464-8603, Japan. E-mail: chunshanli@gmail.com ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. thus it is easy for application.[2 ? 6] Thompson and King (1972) proposed the use of heuristics to generate process alternatives, followed by eliminating the infeasible separations with known constraints.[2] Nadgir (1983) developed a simple-ordered heuristic method for the systematic synthesis of initial sequences for multicomponent separations.[3] Barnicki and Fair (1990,1993) provided a comprehensive set of rules for liquid and gas/vapor separations and demonstrated how to arrive at a design.[6] Douglas (1995) described how to consider a wide range of separation techniques in a hierarchical manner, emphasizing the interactions among the subsystems for gas/vapor, liquid, and solid separations.[5] Shi(1997) put up a relative cost function replacing the coefficient of ease of separation (CES), but which does not fundamentally differ from the ordinary rules.[6] Although heuristic rules to guide the order of separation sequencing have been available, many of the known heuristic contradict or overlap others; many procedures to resolve these conflicts have not been adequately developed.[7,8] Recently, Bek-Pedersen(2000, 2004) proposed a driving-force-based approach to design and synthesize distillation systems that were mainly based on thermodynamic insights by graphic method,[9,10] which provide a measure of the differences in chemical/physical 906 C. LI, G. WOZNY AND K. SUZUKI properties between two co-existing phases in a separation unit. Approach based on thermodynamic insight is often adopted by experience engineers and is easy to use. But it is often trapped in local optimum, overlooks global optimization, and also a little difficult to be applied in a complex system.[11] Mathematical approach is another totally different way, which bases on rigorous modeling; in principle it should be rigorous and infallible. For example, Grossmann (1996, 2003) put up some rigorous algorithms for the separation synthesis,[12,13] but the defects of these approaches are also obvious, cumbersome and overwhelming in terms of computational time and much effort required for algorithms, especially for a complex separation system. Thus, efforts were made by some researches to develop combined methods, for example, Shah (2002) developed a new synthesis framework screens and examined complex distillation sequences based on knowledge and mathematical methods;[14] Zhao(2002) applied Analytical Hierarchy Process (AHP) method to determine the optimal sequence for multicomponent distillation separation.[15] The mathematic algorithmic methods are adopted for modeling; the well-accepted heuristics are used to reduce the size of the search space of the algorithmic methods for simplifying solution, which exploit the advantages of mathematical approach and heuristic or thermodynamic method. But in these studies, only one separation technique ? distillation was considered, although it is the most frequently used separation technique, in many industrial situations; other techniques, e.g. extraction, absorption etc. are more economic. Therefore, more than one type of separation technique may have to be used to separate a given mixture economically. A ?general? separation synthesis system should also be able to select from multiple separation techniques before sequencing. And also, most of product mixtures contain azeotrope especially for fine chemicals, which could not be separated by the simple common techniques, such as atmosphere distillation and so on. From the summary of the previous research, it can be concluded that these problems are not well solved yet. In the study, one general methodology for separation process synthesis and design is proposed, which is based on a mathematical approach, and combines the thermodynamic insights and heuristic rules. Firstly, the pseudo-component concept was proposed and was used to tackle the azeotrope contained in the mixture, which widened the scope of application of the conventional heuristic method. Then the AHP was adopted to construct separation model and determine the optimal separation sequence. The heuristic rules and thermodynamic insights were applied for the selection of separation techniques and operations to reduce the complexity and size of synthesis search space. The hybrid method is applied to two illustrative examples which have been ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pacific Journal of Chemical Engineering carefully selected by taking into account the importance of practical relevance, The first example deals with a light-end refinery mixture;[14] the second example deals with phenol, o-cresol, p-cresol, m-cresol system containing azeotrope. METHODOLOGY OF THE HYBRID METHOD The analytic hierarchy process (AHP) AHP is an approach developed by mathematician Thomas Saaty,[16] and applied in different area,[17] which can provide decision making that involves structuring multiple choice criteria into a hierarchy, assessing the relative importance of these criteria, comparing alternatives for each criterion, and determining an overall ranking of the alternatives. In principle, the separation synthesis such as the selection of separation techniques and decision of separation sequence is one complex decision-making process, in which the final results are usually influenced by the subjective factors because of different designer, angle, or experience. So AHP can be used to help capture both subjective and objective evaluation measures for this process, provide a useful mechanism for checking the consistency of the evaluation measures and alternatives suggested by the team, thus reducing bias in decision making. AHP includes four main steps. Decomposing The goal is to structure the complex problem into humanly manageable subproblems. To do so, iterating from top (the more general) to bottom (the more specific), split the problem, which is unstructured at this step, into submodules that will become subhierarchies. Navigating through the hierarchy from top to bottom, the AHP structure comprises goals (systematic branches and nodes), criteria (evaluation parameters), and alternative ratings (measuring the adequacy of the solution for the criterion). Each branch is then further divided into an appropriate level of detail. At the end, the iteration process transforms the unstructured problem into a manageable problem organized both vertically and horizontally under the form of a hierarchy of weighted criteria. By increasing the number of criteria, the importance of each criterion is thus diluted, which is compensated by assigning a weight to each criterion. Weighing Assign a relative weight to each criterion, based on its importance within the node to which it belongs. The sum of all the criteria belonging to a common direct parent criterion in the same hierarchy level must equal to 100% or 1. A global priority is computed, which quantifies the relative importance of a criterion Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj Asia-Pacific Journal of Chemical Engineering SEPARATION PROCESS BY HYBRID METHOD within the overall decision model. The detail process for weighing will be explained in the next section. Evaluating Score alternatives and compare each one with others. Using AHP, a relative score for each alternative is assigned to each leaf within the hierarchy, then to the branch the leaf belongs to, and so on, up to the top of the hierarchy, where an overall score is computed. Selecting Compare alternatives and select the one that best fits the requirements. The above is the explanation in theory; details of practical application in separation design will be shown in the next section. The procedure of the methodology The hybrid method procedure is presented in Fig. 1, which includes six main steps. Steps 1?3, thermodynamic calculation, handling azeotrope, simplifying the scale of the mathematical problem, eliminating the unsuitable separation technique, and selecting the suitable separation techniques based on thermodynamics, experience, and heuristic rules. Step 4, the most important separation technique distillation, constructs the optimal separation sequence based on the AHP method. Step 5 solves the pseudo-component (azeotrope) separation. Step 6 optimizes the final separation sequence. For the final separation sequence, we only use common column structure (one feed, two products) for the distillation separation method, not consider thermally coupled schemes such as the side-stripper, the side-rectifier, mixture List possible separation techniques Step1 Calculate pure and mixture properties Step2 containing azeotrope No Yes consider the azeotrope mixture as one pseudo-component Step3 seletion suitable method according thermodynamic analysis distillation separation method No other separation methods Yes list all separation heurstic rules pure product construct the model of AHP as Fig.3 Step4 No Yes construct the matrix by pairwise comparisons calculate the ordering vectors and overall ordering vectors normalization of the ordering vectors Step5 the optimal separation sequence (containing pseudo-component) select suitable separation methods for pseudo-component(azeotrope) Step6 Optimal flowsheet Figure 1. The scheme of the proposed hybrid method. ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj 907 C. LI, G. WOZNY AND K. SUZUKI Asia-Pacific Journal of Chemical Engineering Table 1. Relationship between separation techniques and pure component properties. Separation type Distillation Absorption Filtration Crystallization Vapor?liquid separation Gas separation Extractive distillation Vapor?liquid separation Azeotropic distillation Liquid?Liquid extraction Liquid membranes Vapor?liquid?liquid separation Liquid?liquid separation Liquid separation Drying Gas separation Solid?vapor separation Important pure component properties Differences in volatilities(vapor pressure) Preferential solubility Size of solid greater than pore size of filter medium Difference in freezing tendencies; preferential participation in crystal structure Vapor pressure, heat of vaporization, boiling point, solubility parameter Difference in volatilities Difference in volatilities Solubility parameter, molar volume, radius of gyration, affinity to carrier Solid?liquid separation Step 1: thermodynamic data calculation The commonly known separation techniques (distillation, extraction, crystallization, membrane separation, adsorption, filtration, and so on) suitable for the multicomponents separation are listed in Table 1. Some qualitative and quantitative thermodynamic properties of pure and mixture components are calculated, which are the base for the selection of the separation techniques. For example, the calculation results of relative volatility and the difference of boiling point between the two components will be the base for the decision of distillation. If there is azeotrope in the mixture, then go to Step 2, otherwise go to Step 3. Step 2: pseudo-component If two components form azeotrope (if the relative volatility between the two components is less than 1.05, then we assume them to be azeotrope), then they are considered and handled as one pseudo-component until the last step, and the contents, volatility, or other thermodynamic data will be calculated again. Step 3: selection of separation techniques According to the calculation result of the thermodynamic properties from Step 1, the existing experience and heuristic, a series of tests and selection are made to evaluate the feasibility of application of the separation techniques. In the past decades, a variety of new separation methods emerged and have been applied in industry, but the energy-driven (based on the heats of vaporization of the components) techniques such as distillation and evaporation are still employed by the majority The level of application of technology and its maturity ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. are summarized by Keller,[18] and shown in Fig. 2; it was the status 20 years ago, but it can still prove helpful in the selection of different techniques and also emphasized the importance of distillation separation techniques. Some experiences from chemical engineering are summarized in the following. The advantages of distillation are in its simple flowsheet, low capital investment, and low risk. If components to be separated have a relative volatility of 1.2 or more and are thermally stable, distillation is typically the separation method of choice.[19] The disadvantages of distillation are its low energy efficiency and that it requires thermal stability of compounds at their boiling points. Large energy savings could be obtained by replacing distillation with lowenergy intensity operations. The most promising technologies for replacement of distillation include membranes, extraction, sorption (absorption and adsorption), and hybrid systems. Maximum or Petlyuk column and so on, but in natural circumstances, the thermally coupled schemes are formed by the common column structure; thus, after the optimal separation sequence is available, by improving, a last optimal separation sequence can also be available. Vapor pressure, diffusivity Industry Application Level Separation technique 1 2 3 4 6 5 7 8 First Application 908 13 12 9 10 11 14 15 Invention Maximum Technical Maturity Level Figure 2. The technology application and maturity level. Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj Asia-Pacific Journal of Chemical Engineering SEPARATION PROCESS BY HYBRID METHOD Membrane permeation can generally be used only for dilute liquid mixtures, else should be eliminated. No bulk liquid separations are done commercially. Also, a high-purity product will not result from membrane permeation, and with a correspondingly low recovery rate. Thus, if high level of purity and recovery rate are essential, membrane separation can be eliminated as a potential separation method. Feasible crystallization processes typically require 20?30 ? C differences in the freezing points of pure components. In addition, the freezing points should be at or above ambient temperatures if the added expense of refrigeration is to be avoided, else the crystallization process should be eliminated. The choice of other separation techniques is also based on thermodynamic insight and experience. Jaksland recommended some key binary parameters standard for the selection of separation techniques,[11] which are shown in Table 2. If no information of a specific external medium is found (or available through a knowledge base), other separation techniques (such as adsorption, filtration, prevaporation using membranes, and so on explained in Table1) are usually eliminated and distillation will be the best choice. If the distillation technique cannot applied for this mixture, then go to Step 5, otherwise Step 4. Step 4: distillation methods AHP in distillation. For multicomponents separation, distillation is by far the most widely used and the first choice of separation process for mixtures. For example, in the United States, distillation processes constitute 90?95% of all separations in the chemicals and petroleum refining industries and account for over 99% of the total energy associated with all industrial separation technologies, offering the most promising energy reduction opportunities.[19] Thu, in Table 2. Recommended values for separation feasibility indices. Separation process Partial condensation Flash operation Distillation Cryogenic distillation Per-vaporation Micro-filtration Property (pj ) Boiling point Boiling point Vapor pressure Boiling point Vapor pressure Boiling point Vapor pressure Molar volume Solubility Molecular diameter Molecular weight Binary ratio value of pure property j for binary pair i (ri ,j ) this study, distillation technique of separation is the first choice. As explained the Section on The Analytic Hierarchy Process, by reducing complex decisions to a series of one-on-one comparisons, then synthesizing the results, AHP not only helps decision makers arrive at the best decision but also provides a clear rationale that it is the best. For the multicomponents separation, the separation sequence in nature is the decision problem, by the allside trade-off between the different separation point and separation technique. So the separation synthesis can be solved by the AHP. The model of AHP is shown as Fig. 3. A hierarchy structure shown in Fig. 3 is established by decomposing the complex problem (mixture separation) into a hierarchy of inter-related decision elements (the sequence of different points). This structure is the key to interrelate and chain all decision elements of the hierarchy from the top level down to the bottom. So the top of the hierarchical structure is the optimal result of all the alternative flowsheet; the lowest level of the hierarchical structure is the feasible separation points. For example, for the mixture ABC three-component, it includes two possible flowsheets, A/(BC), then A/B/C or (AB)/C, then A/B/C. So after the separation technique is decided, different thermodynamic insights or experience rules (i.e. Rules in Fig. 3) can be calculated based on the AHP method to decide the first separation point (i.e. S in Fig. 3), Prioritization and construction of the pairwise comparison matrices. Once the hierarchical structure is established, the relative importance (weights) of all decision elements is explicitly captured and revealed through ratio scale approach. Pairwise comparisons of these elements within the same hierarchical level with respect to the parent elements in the next higher level are established. The numerical scales ranging from 1 (equal importance) to 9 (absolute importance) (Saaty, 1980) as shown in Table 3 are used in the pairwise comparison matrices. The input data can be achieved by pairwise comparison. For example, for the experience rule ? ?favor easy separation first?, we can judge this rule by the relative 2.5 1.4 15.0 1.02 1.5 1.17 2.40 4.00 2.50 3.00 2.40 ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. The optimal flowsheet Rule2 Rule1 S1 Figure 3. process. Rule3 S2 Rule4 S3 Rule5 S4 The model of the analytic hierarchy Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj 909 910 C. LI, G. WOZNY AND K. SUZUKI Asia-Pacific Journal of Chemical Engineering Table 3. Scales of relative importance. Intensity of importance Definition Explanation 1 3 Equal importance Weak importance of one over the other 5 Essential or strong importance 7 Demonstrated importance 9 Absolute importance 2,4,6,8 Intermediate values between the two adjacent judgments Distillation Two activities contribute equally to the objective Experience and judgment strongly favor one activity over another Experience and judgment strongly favor one activity over another An activity is strongly favored and its dominance demonstrated in practice The evidence favoring one activity over another is of the highest possible order of affirmation When compromise is needed Table 4. Problem specifications for example 1. Membrane Fraction (mass) Vapor pressure (kPa) Relative volatility (relative to Benzene) Relative volatility 0.091 0.169 0.176 0.019 0.030 0.177 0.112 0.226 269.70 151.31 115.71 73.63 55.41 50.62 37.44 24.30 11.10 6.23 4.76 3.03 2.28 2.08 1.61 1.00 1.78 1.31 1.57 1.33 1.09 1.35 1.54 A A B C D E F G H B C D E F G H B B C B C D E F G C D D E F G E F E F G E F G H Figure 4. The final promising design for example 1. Distillation A B C D E A B C D E Crystallization 22-DMP i -Pentane n-Pentane 22-DMB 23-DMB 2-MP n-Hexane Benzene Table 5. The dealing specifications for example 1. A Relative volatility Vapor (relative Relative Fraction pressure/kPa to Benzene) volatility B C D C D E Figure 5. The final promising design for example 2. volatilities. For the mixture ABC, if the relative volatility of A/B is larger than B/C, then the first separation point is A/(BC), as regards the specific scale value is 3, 5, or 6, we should trade off between all the separation point, then decide the final value according to Table. 3 Of course, if we use only this rule to decide the separation point, the AHP method will not necessary, but if we also consider another experience rule, for example the experience rule ? ?favor 50/50 split?, in the same manner, the separation mixture ABC, if the mass content of A, B, C is 10, 20, and 70 percent respectively, then according to the rule the first separation point should ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. 22-DMP i -Pentane n-Pentane 22-DMB 23-DMB 2-MP n-Hexane Benzene 0.091 0.169 0.176 0.019 0.207 0.112 0.226 269.70 151.31 115.71 73.63 55.41 50.62 37.44 24.30 11.10 6.23 4.76 3.03 2.28 1.78 1.31 1.57 1.33 1.61 1.00 1.35 1.54 be (AB)/C. Then we will find results from the two different rules to be different. To integrate the value, the pairwise comparison matrices based on AHP will be the best choice. Several sets of pairwise comparison matrices of elements in the same level, which attribute to accomplishing the goals of the parent element in the next higher level are finally obtained as in Eqn (1). The derived pairwise comparisons of relative importance, Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj Asia-Pacific Journal of Chemical Engineering SEPARATION PROCESS BY HYBRID METHOD Table 6. The pairwise comparison matrices and the ordering vectors for example 1. Pairwise comparison matrices ? 1 3 5 9? 0.33 1 2 5? AO?R = ? 0.2 0.5 1 2 0.11 0.2 0.5 1 ? AR1 ?B 1 ? 0.25 ? 0.5 =? ? 0.25 ? 0.25 0.5 4 1 2 1 1 2 ? AR2 ?B 1 0.33 1 ?3 ?3 1 ? =? 2 ?5 5 2 7 3 ? AR3 ?B 1 0.33 1 ?3 ?5 2 =? ?9 4 ? 5 2 3 1 ? AR4 ?B 1 ?1 ?1 =? ?1 ? 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 0.5 1 0.5 0.5 1 4 1 2 1 1 2 4 1 2 1 1 2 Consistency check ?max = 4.033 ?max = 6 [0.36,0.09,0.18,0.09,0.09,0.18] CI = 0 RI = 0 CR = 0 ? 0.14 0.33 ? 0.33 ? ? 0.5 ? ? 0.5 1 ?max = 6 [0.04,0.11,0.11,0.20,0.20,0.34] CI = 0 RI = 0 CR = 0 ? 0.20 0.11 0.20 0.33 0.5 0.25 0.5 1 ? 1 0.5 1 2 ? ? 2 1 2 4 ? ? 1 0.5 1 2 0.5 0.25 0.5 1 1 1 1 1 1 1 [0.59,0.23,0.12,0.06] CI = 0.011 RI = 0.89 CR = 0.012 ? 2 0.5 ? 1 ? ? 0.5 ? ? 0.5 1 0.33 0.2 0.2 1 0.5 0.5 1 0.5 0.5 2 1 1 2 1 1 3 2 2 1 1 1 1 1 1 The total ordering vectors ?max = 6.006 [0.04,0.10,0.19,0.38,0.19,0.10] CI = 0.001 RI = 1.86 CR = 0.001 ? 1 1? 1? ? 1? ? 1 1 ?max = 6 [0.17, 0.17, 0.17, 0.17, 0.17, 0.17] CI = 0 RI = 0 CR = 0 Table 7. The total ordering vectors of the pairwise comparison matrices for example 1. The total ordering matrices ? 0.36 0.09 0.18 0.09 ? 0.04 0.11 0.11 0.20 0.04 0.10 0.19 0.38 0.17 0.17 0.17 0.17 0.09 0.20 0.19 0.17 Consistency check The total ordering vectors CI = 0.012 [0.24, 0.10, 0.16, 0.15, 0.13, 0.21] 0.18 ? 0.34 ? 0.10 0.17 RI = 0.90 CR = 0.013 aij = wi /wj , for all decision elements and their reciprocals, aji = 1/aij , are inserted into a reciprocal square matrix A = {aij } as shown in Eqn (1). The analytical ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. solution of Eqn (2) then provides the relative weights for each decision element. According to the eigenvalue method, the normalized right eigenvector (W = Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj 911 912 C. LI, G. WOZNY AND K. SUZUKI Asia-Pacific Journal of Chemical Engineering {w1 , w2 , . . . , wn }T ) associated with the largest eigenvalue (?max ) of the square matrix A provides the weighting values for all decision elements. ? 1 ? w2 /w1 A=? ? ... wn /w1 AW = ?max W w1 /w2 1 .. . wn /w2 ? и и и w1 /wn и и и w2 /wn ? ? ? иии иии (1) 1 (2) In this step, the final pairwise comparison matrices will be under the influence of subjective factors on some degree and, as far as we know, has seldom been addressed in the literature before. The AHP uses a principal eigenvalue method to derive priority vectors, Following Saaty,[15] the priority vector has two meanings: ?The first is a numerical ranking of the alternatives, which indicates an order of preference among them. The other is that the ordering should also reflect intensity or cardinal preference as indicated by the ratios of the numerical values.? This second meaning requires, in our view, that these ratios preserve, whenever possible, the order of the respective preference intensities, which is not always the case for AHP priority vectors; this can be considered as a basic drawback of the AHP. But for our application in this study, the construction of pairwise comparison matrices will bases on the thermodynamic and experience, and the drawback will be weaken to some extent. The distillation separation sequence. In the step, the ordering vectors of the pairwise comparison matrices (A shown in Eqn (1)) and the final ordering vectors (W shown in Eqn (2)) are calculated. Normalize these vectors and gain the initial separation sequence. Step 5. Separation pseudo-component For the pseudo-component separation, it belongs to the research field of azeotrope separation synthesis, and lots of synthesis methods are published,[15,16] which need special information from the forgoing experience, and also based on the factual industrialization application. For this topic, making concrete analysis of concrete problems is the best method. Step 6. Optimize the final separation sequence In this step, the final separation sequence is optimized by the simulation tool, and the final operation parameters can be available. An energy saving methodology also can be applied in this step. But the detailed simulation results from the integrated approach are not provided in this study. They can be obtained from the authors. ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. APPLICATION EXAMPLES Example 1: separation of a light-end refinery mixture The first example deals with a light-end refinery mixture which is separated into eight products.[9,14] The vapor pressure and relative volatility of the components are calculated The data is presented in Table 4. From the calculating results, the relative volatility of 23-DMB and 2-MP is close to 1. According to the proposed method, if two components form azeotrope, the two components are considered as one pseudo-component which is separated in the last step. After this, the new calculation results are listed as in Table 5. From Table 5, it can be seen that the relative volatility of all the mixture are larger than 1.2. So based on pure component and mixture properties analysis, the crystallization and membrane separation and other separation techniques are eliminated. The distillation technique is preferred. Thus we can start from Step 4 of our proposed procedure. The rules selected for the separation point are according to the examples; different examples can use different rules. The rules[3] used in the example are listed in the following. Rule 1. Favor easy separations first. Specifically, arrange the components separately according to their relative volatilities as separation factors in the ordered components in the feed vary widely. Sequence the splits in the order of decreasing adjacent relative volatility. Rule 2. Remove the most plentiful component first. A product composing a large fraction of the feed should Table 8. Problem specifications for example 2. phenol o-cresol m-cresol p-cresol 2,4 -xylenol Fraction Vapor pressure (kPa) Relative volatility (relative to 2,4 -xylenol) Relative volatility 0.35 0.20 0.15 0.15 0.10 0.1690 0.1317 0.0834 0.0833 0.0545 3.10 2.42 1.53 1.53 1 1.28 1.58 1.00 1.53 Table 9. the dealing specifications for example 2. phenol o-cresol m-cresol p-cresol 2,4 -xylenol Fraction Vapor pressure (kPa) Relative volatility (relative to 2,4 -xylenol) Relative volatility 0.35 0.20 0.30 0.1690 0.1317 0.0833 3.10 2.42 1.53 1.28 1.58 0.15 0.0545 1 1.53 Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj Asia-Pacific Journal of Chemical Engineering SEPARATION PROCESS BY HYBRID METHOD Table 10. The pairwise comparison matrices and the ordering vectors for example 2. Pairwise comparison matrices ? 1 3 5 9? 0.33 1 2 5? AO?R = ? 0.2 0.5 1 3 0.11 0.2 0.33 1 AR1 ?B = AR2 ?B = AR3 ?B = AR4 ?B = 1 5 4 0.20 0.25 1 2 0.5 1 1 2 0.5 1 0.2 0.33 5 3 1 1 0.5 2 1 0.2 0.14 5 7 1 1 5 0.2 1 0.11 0.33 Consistency check The ordering vectors ?max = 4.026 [0.59, 0.23, 0.13, 0.05] CI = 0.003 RI = 0.89 CR = 0.003 ?max = 3.04 [0.10, 0.57, 0.33] CI = 0.02 RI = 0.58 CR = 0.03 ?max = 3.02 [0.58, 0.31, 0.11] CI = 0.01 RI = 0.58 CR = 0.02 ?max = 3.05 9 3 1 [0.33, 0.59, 0.08] CI 0.03 RI 0.58 CR = 0.05 ?max = 3.032 [0.75, 0.18, 0.07] CI = 0.016 RI = 0.58 CR = 0.03 Table 11. The total ordering vectors of the pairwise comparison matrices for example 2. The total ordering matrices ? 0.10 0.57 0.33 ? ? 0.58 0.31 0.11 ? 0.33 0.59 0.08 0.75 0.18 0.07 Consistency check CI = 0.043 The total ordering vectors W = [0.28, 0.49, 0.23] RI = 1.47 CR = 0.03 be separated first, provided that the separation factor or relative volatility is reasonable for the separation. Rule 3. Favor 50/50 split. If component compositions do not vary widely, sequences which give a more nearly 50/50 or equimolar split of the feed between distillate (D) and bottom (B) products should be favored, provided that the separation factor or relative volatility is reasonable for the split. Rule 4. Direct sequence rule. During distillation neither the relative volatility nor the molar percentage in the feed varies widely. Remove the components one by one as distillate products. The resulting sequence is commonly known as the direct sequence, in which the ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. optimum operation pressure tends to be highest in the first separator and reduces in each subsequent separator. Based on Rules 1?4, construct the AHP model as Fig. 3; in this model, the top level is the optimal flowsheet (final results), the middle level is rules; for this study, there are only four rules. And the bottom level is separation points, totally six. The constructed pairwise comparison matrices and the calculated ordering vectors of the pairwise comparison matrices are listed in Table 6. We take Rule 1 as a sample to explain how to construct the matrices AR1?B . First, according to the calculation results of relative volatilities, we evaluate the relative importance for the Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj 913 914 C. LI, G. WOZNY AND K. SUZUKI Asia-Pacific Journal of Chemical Engineering Table 12. Energy consumption of different separation sequence for example 2. Separation process Distillation Crystallization A I A B C D E B B C D E 1 C D E 2 II 1 A B C D E 1 Top/bottom Column pressure (atm) Trays Top/bottom 1 ??1 60 182/199 ?0.0577/0.0654 2 ??1 40 190/204 ?0.0433/0.0434 3 ??1 40 202/211 ?0.0511/0.0510 E ?0.1521/0.1598 Crystallization A B A B 2 C D E C C D 1 ??1 40 184/204 ?0.0312/0.0388 2 ??1 60 182/190 ?0.0659/0.0660 3 ??1 40 201/211 ?0.0415/0.0415 D E 3 Distillation III D Heat duty (MMkcal/h) 3 Distillation A B C D E C C D Temperature (? C) ?0.1386/1463 Crystallization 1 ??1 40 190/210 ?0.1052/0.1127 2 ??1 60 182/197 ?0.0659/0.0661 3 ??1 40 190/202 ?0.0330/0.0330 A A B C D B B C D E 2 C D 3 ?0.2039/0.2118 Distillation IV A B C D E 1 A B C D E 2 A B 1 ??1 40 190/210 ?0.1052/0.1127 B 2 ??1 40 184/202 ?0.0998/0.0998 C 3 ??1 60 182/190 ?0.0659/0.0660 A 3 C D D Crystallization six separation points by pairwise comparison and give the value respectively, but in the same matrices all the value should have consistency, for example, if B and C are two and four times for A, then C should be two times for B during pairwise comparison process. In this procedure, it will include some subject factors whose influence in the whole process is small in relation to ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. ?0.2709/0.2785 other methods. After all, the pairwise comparison matrices are decided and the total ordering vectors of the pairwise comparison matrices are calculated by Eqn (2) and listed in Table 7. According to the total ordering vectors of the pairwise comparison matrices, from the result of consistency checking (C .R.total < 0.1) we can conclude that the Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj Asia-Pacific Journal of Chemical Engineering results have a good consistency. From the final vectors, we can construct the initial separation sequence from the larger values to smaller values. Lastly, for the separation of the pseudo-component (23-DMB, 2-MP), thermodynamic analysis, crystallization, or membrane[20] are all possible separation techniques; by trading-off, it is the second choice. The final promising design for Example 1 is shown in Fig 4, which is a little different from the result of the BekPedersen[9] and Shah,[14] It is because they did not consider azeotrope. Example 2: separation of phenolic mixture The second example deals with the mixture which contains phenol(A), o-cresol(B), p-cresol(C), and mcresol(D), which come in oil or fine chemical process. The vapor pressure and relative volatility of the components are calculated. The original data is presented in Table 8. From the calculated results, the relative volatility of m-cresol and p-cresol is close to 1, so the two components are considered as one pseudocomponent which is separated in the last step. The data are listed in Table 9. Based on Rules 1?4, construct the AHP model as Fig. 3; the constructed pairwise comparison matrices and the calculated ordering vectors of the pairwise comparison matrices are listed in Table 10. The total ordering vectors of the pairwise comparison matrices are listed in Table 11. From the result of consistency checking (C .R.total < 0.1) we can conclude that the results have a good consistency. For the pseudo-component (m-cresol and p-cresol), based on thermodynamic analysis according to the proposed method of Li,[21] crystallization is selected as the separation technique, so the final promising design for the Example 2 is shown as Fig 5. We also carried out a simple simulation (without optimization) for the different promising designs of the sample; the results were shown in Table. 12 The purities of all the products are above 98%. The energy consumption for the separation of C/D by crystallization method was not considered, which will not influence the final results. It can be found that the final promising design based on our proposed method has low energy consumption. It should be mentioned that the results of examples 1 and 2 are only relevant for the analyses. For another mixture including the same kinds of components, the results may be different even if the components are the same. ? 2009 Curtin University of Technology and John Wiley & Sons, Ltd. SEPARATION PROCESS BY HYBRID METHOD CONCLUSION A hybrid method integrating the AHP with heuristic approaches and thermodynamic insights for separation synthesis is proposed. The AHP method is used to decide the initial separation sequence and help capture both subjective and objective evaluation measures for this process. Useful rules and thermodynamic calculation are used to reduce the size of the search space of the algorithmic methods and construct the AHP model. The hybrid method is applied to two illustrative examples, a light-end refinery mixture and phenol system containing phenol, o-cresol, p-cresol, m-cresol. The results show that the methodology can prove as a guide for the initial separation process design and it can be easily applied. REFERENCES [1] J.L. Humphrey, G.E. Keller. Separation Process Technology, McGraw-Hill: New York, 1997; pp.50?101. [2] R.W. Thompson, C.J. King. AIChE J., 1972; 18, 491?496. [3] V.M. Nadgir, Y.A. Liu. AIChE J., 1983; 29(6), 926?934. [4] S.D. Barnicki, J.R. Fair. Ind. Eng. Chem. Res., 1990; 29, 421?428. [5] J.M. Douglas. AIChE J., 1995; 41, 252?257. [6] B.C. Shi, J.H. Wang. J. Chem. Ind. Eng., 1997; 48(2), 175?179. [7] X.N. Li, A. Kraslawski. Chem. Eng. Process., 2004; 43(5), 583?594. [8] S.D. Barnicki, J.R. Fair. Ind. Eng. Chem. Res., 1992; 31(7), 1679?1694. [9] E. Bek-Pedersen, R. Gani. Comput. Chem. Eng., 2000; 24, 253?259. [10] E. Bek-Pedersen, R. Gani. Chem. Eng. Process., 2004; 43, 251?262. [11] C.A. Jaksland, R. Gani, K.M. Lien. Comput. Eng. Sci., 1995; 50(3), 511?530. [12] I.E. Grossmann, M.M. Daichendt. Comput. Chem. Eng., 1996; 20(6?7), 665?683. [13] L. Sangbum, I.E. Grossmann. Comput. Chem. Eng., 2003; 27(11), 1557?1575. [14] P.B. Shah, D.K. Antonis. AIChE J., 2002; 48(3), 527?550. [15] T.L. Saaty. The Analytic Hierarchy Process, McGrawHill: New York, Extended in RWS Publication, 1980. [16] C.H. Zhao. Sci. Technol. Chem. Ind., 2002; 10(2), 68?71. [17] H. William. Eur. J. Oper. Res., 2008; 186(1), 211?228. [18] G.E. Keller, R.A. Anderson, C.M. Yon. In Adsorption in Handbook of Separation Process Technology (Ed.: R.W.. Rosseau), Wiley: New York, 1987; pp.210?260. [19] Oak Ridge National Laboratory, DOE-EERE-Industrial Technologies Program (ITP)-Materials for Separation Technologies: Energy and Emission Reduction Opportunities. 2005.5.4. [20] T. Matsufuji, K. Watanabe, N. Nishiyama. Ind. Eng. Chem. Res., 2000; 39(7), 2434?2438. [21] C.S. Li, X.P. Zhang, X.Z. He, S.J. Zhang. J. Clean. Prod., 2007; 15, 690?698. Asia-Pac. J. Chem. Eng. 2009; 4: 905?915 DOI: 10.1002/apj 915

1/--страниц