DETERMINATION OF OPTIMAL ENERGY EFFICIENT SEPARATION SCHEMES BASED ON DRIVING FORCES Abstract Erik Bek-Pedersen, Rafiqul Gani CAPEC, Department of Chemical Engineering, Technical University of Denmark, DK-00 Lyngby, Denmark Olivier Levaux Belsim, S. A., Allée des Noisetiers, 03 Liège, Belgium A new integrated approach for synthesis, design and operation of separation schemes is presented. This integrated approach is based on driving forces that promote the desired separation for different separation techniques. A set of algorithms needed by the integrated approach for sequencing and design of distillation columns and for generating hybrid separation schemes are presented. The main feature of these algorithms is that they provide a visual solution which also appears to be near optimal in terms of energy consumption. Several illustrative examples highlighting the application of the integrated approach are also presented. Introduction Synthesis and design of process flowsheets involve generation and identification of feasible alternatives, process/solvent design, determination of optimal condition of operation together with the optimal process flowsheet and many more. In an integrated approach, the objective is to consider various aspects related to, for example, synthesis, design and operation simultaneously. Also, preferably, some of the design decisions that may also affect other problems (such as synthesis and operation) are considered in the early steps of the integrated approach. In this paper, the concept of driving forces is employed to develop an integrated approach for the determination of energy efficient separation schemes. The driving force is defined here as a function of the difference in composition of a component between two coexisting phases and can be determined from differences in thermodynamic properties of the mixture compounds. By using the insights obtained through an analysis of the driving forces, the integrated approach proposes to make early decisions regarding flowsheet configuration, design and condition of operation. Since the composition differences of a component in two co-existing phases can be created in different ways, driving forces are different on the basis of properties. For example, when the relative volatility of two compounds in a mixture differ significantly from unity, there exist driving forces between the liquid and the vapour phase making distillation feasible as a separation technique. In the case of pervaporation, a driving force exists because of the partial pressure difference across the membrane for the component being removed. In this case, the feasibility of the separation depends on the temperature of the feed mixture and the pressure on the permeate side of the membrane. Most separation techniques commonly used in the process industry employ some form of driving forces to achieve the required separation. Therefore, in the generation
of a process flowsheet, combining or sequencing separation techniques that lead to the largest total driving force may have consequences related to use of energy (if the driving force is influenced or caused by an external energy source), feasibility (separation is not possible if the driving force is zero) and condition of operation (driving forces are usually functions of intensive variables that define conditions of operation). Note also that the use of driving forces provide a visualisation feature as it allows a visual determination of near optimal (if not optimal) solutions of the integrated synthesis, design and operation problem. In this paper, the main features of the integrated approach are presented. The application of the integrated approach is highlighted through several illustrative examples dealing with determination of separation schemes. Driving Force Based Integrated Approach The driving force as defined by Gani and Bek-Pedersen (999) is given by, x () x i ij F ij y i x i i ( ij ) i In the above equation, x i and y i are the compositions of component i in two coexisting phases, F ij is the driving force for component i, ij is the relative separability factor for component i with respect to property (or separation technique) j. Note that ij = f(t, P, composition, ), where indicates external factors such as resistance to mass transfer and heat transfer. From Eq., it can also be noted that at fixed P (or T), two-dimensional plots of F ij versus x i (or y i ) can be made where each data point may also indicate a different T (or P). Therefore, these diagrams can be used to design and configure separation schemes, including conditions of operation. Gani and Bek-Pedersen (999) have already shown how the driving force diagrams with respect to relative volatility can be used for near optimal (with respect to energy consumption) single distillation column design. Bek-Pedersen et al. (999) have shown how the driving force diagrams can be used to obtain near optimal (with respect to energy consumption) distillation column configuration. In this paper, these algorithms have modified and extended to allow for presence of azeotropes in the multi-component mixtures, to generate hybrid separation schemes and to allow for scaling in the distillation column design when "extreme" conditions for the feed mixture or product compositions are specified. Three different algorithms are briefly presented. The first one deals with the scaled single column distillation, the second deals with the sequencing of simple distillation columns for multi-component separation and the operating conditions of columns through distillation trains. The third one deals with generation of hybrid separation schemes. Algorithms and are illustrated in Figure.
Algorithm : Single distillation column design.. Generate or retrieve from a database, the vapor-liquid data for the binary system in the column. For a multicomponent system, select the two key components to define the "split" and use them as the binary (key) mixture.. Compute F ij using Eq. and plot F ij as a function of x i. 3. Identify the point D x. (See Figure ).. Determine N F from N F = ( - D x ) N P. (Stages counted from the top). 5. Determine whether a scaling factor needs to be applied. If condition or is satisfied, scaling is needed and go to, otherwise stop.. If condition is satisfied, then go to.. Else go to.... If condition a is satisfied, then relocate N F between 5 and 0 % up in the column. Else condition b is satisfied, then relocate N F between 5 and 0 % down in the column... If condition a is satisfied, then relocate N F 0 % up. Else, if condition b is satisfied, then relocate N F 5 % up. Else, if condition c is satisfied, then relocate N F 5 % down. Else, if condition d is satisfied, then relocate N F 0 % down. Condition : a) x HK, Z 0. and D x < 0.7 b) x LK, Z 0. and D x > 0.3 Condition : LK,D a) 0. 0 HK,B LK,D b) 0. HK,B LK,D c) 0. HK,B LK,D d) 0. 0 HK,B and D x < 0.7 and D x < 0.7 and D x > 0.3 and D x > 0.3 Algorithm a: Sequencing of distillation columns.. Retrieve the vapour-liquid data available for each pair of adjacent key components.. List all the components in the mixture, NC, according to their relative separability, ij. 3. Calculate the driving force diagrams for the adjacent components, preferably all at the same operating condition. In total, driving force diagrams for NC- binary pairs are calculated. Set k =. 3.. If a binary pair forms an azeotrope, multiply the maximum driving force by a Dy, min penalty weight D to make the value the smallest among the binary pairs. Dy,max y,azeotrope 3
. For split k, select the adjacent pair for which the maximum driving force has the largest value. 5. Remove the split between the selected adjacent pair from the list. Set k = k+ and repeat the algorithm from step until only one split remains to be allocated.. Draw the flowsheet based on the selected order of the splits. 7. If allocation of column operating pressures is desired go to algorithm b. Otherwise go to step.. For each column, perform the distillation column design through algorithm. Note that this sequencing algorithm provides the largest total driving force for the generated flowsheet. Algorithm b: Allocation of column operating pressure.. Calculate data for the pxy diagram for the two key components in the first distillation column at the bubble point temperature of the feed.. Draw the driving force curve from the data calculated in step together with the bubble point curve. (The driving force diagram, in this case, is the horizontal distance from the dew point line to the bubble point line in the pxy diagram, as a function of x i (or y i ))... Identify the point D x as the composition x i, where the driving force reaches its maximum value (D y )... Identify the bubble point pressure at the point D x. Allocate this pressure as the operating pressure for the condenser in the distillation column..3. Based on a specified pressure drop per plate, determine the reboiler pressure 3. Calculate pxy diagrams for the two key components in the next distillation column as a function of temperature. Identify the temperature at which the pxy diagram gives the maximum bubble point pressure to within 5 % of the reboiler pressure of the previous column is selected. 3.. Repeat step + to determine the column pressures in the condenser and reboiler.. Repeat step 3 until all the condenser and reboiler pressures have been allocated in the distillation column sequence. 5. Note: If the pressure in one or more of the columns in the sequence results in a significant vacuum in the condenser, consider starting at a higher pressure in the first column. If the final condenser pressure in the last column is higher than ambient pressure and the separation can be performed at ambient pressure, consider starting at a lower pressure in the first column. Algorithm 3: Generation of hybrid separation sequence.. For a binary pair that forms an azeotrope (or eutectic points or exhibits mutual solubility), retrieve all sets of two-phase composition data, where each set correspond to a different separation technique. Note: Two sets of data at different operating condition for the same separation technique is considered here as different separation techniques.. Calculate and plot all the corresponding driving force diagrams. 3. For the specified product purities, identify all feasible paths, allowing switch from one separation technique to another if necessary, by moving along the driving force curves (see Figure a). Note: if one separation technique is unable to achieve the desired separation, switch to another is considered.
. For each feasible path, identify the corresponding separation techniques and operating conditions from the driving force diagram. 5. Select as the initial flowsheet, the one with the largest total driving force.. Use the information from steps and 5 to formulate and solve a structural optimisation problem to determine the optimal flowsheet. Application Examples Four illustrative examples highlighting the application of the integrated approach is presented. Example involves an analysis of the reverse extractive separation scheme originally proposed by Hunek et al (99). Example involves the determination of a hybrid separation scheme for the separation of a binary azeotropic mixture. Example 3 involves the sequencing of simple distillation columns for the separation of a multicomponent mixture (Shah and Kokossis, 997). Example deals with both the sequencing of a distillation train and the determination of operating conditions for the distillation columns in processing of hydrocarbons. (PRO/II User s Guide, 99). The detailed simulation results validating the results from the integrated approach are not provided here. They can be obtained from the authors. Example : The flowsheet for the reverse extractive distillation proposed by Hunek et al (99) (see Figure 3) has been analyzed with the driving force diagrams in terms of feasibility of the proposed separation and the design of the distillation columns. Hunek et al proposed that a sufficient amount of methanol in an aqueous mixture of alcohols (ethanol, propanol, and higher) is able to break the azeotropes between ethanol-water and -propanol-water. Consequently, simple distillation columns can be used to remove water and obtain high purity alcohols without involving external solvents or complex distillation column operations. The driving force diagrams from Figures a and b confirm that indeed this separation scheme is feasible. It also shows that in such a scheme, first the water should be removed from the lighter alcohols and then the lighter alcohols should be separated from the remaining water and the heavier alcohols. Note that since the azeotropes between water and the heavier alcohols are not broken, they will go with water in the second column. These same figures are also used to design the individual distillation columns as proposed by algorithm. The feed mixture of water and alcohols is given in Table. Simulations with PRO/II (PRO/II User s Guide, 99) confirm that these simulations are indeed feasible. Example : In this example, the binary azeotropic mixture of methanol and MTBE is to be separated into pure component products. First, applying the algorithm of Jaksland et al. (995), different separation techniques that can be considered are identified. These are azeotropic distillation, extractive distillation, pressure swing distillation, pervaporation, and crystallization. Next, for solvent based separation (extractive or azeotropic distillation), solvents are identified. In the same step, for external agent based separation techniques, the external agents (membranes) are identified. Figure a shows the driving force diagrams for extractive distillation (solvent-free basis), pervaporation, pressure-swing distillation and crystallization. It is clear from figure a that the desired separation cannot be achieved through one (unit) operation, with the exception of crystallization. This is because for single distillation the driving force becomes zero before the desired product can be reached. Also, for solvent based separation, a solvent recovery operation is needed while membrane based separation is limited by a maximum permeate flux. Thus a hybrid separation 5
scheme is needed. By applying the integrated approach (algorithm 3) it has been possible to visually configure a feasible flowsheet together with the condition of operation (see Figures a and b). For each case, the design corresponding to the largest driving force has been obtained. In this case, it can be seen that separation scheme in Figure b, which employs distillation followed by pervaporation, is potentially more energy efficient than any two-column separation scheme because pervaporation requires less energy than distillation. As shown in Figure b, the first separation is performed with distillation at pressure = atm followed by pervaporation at pressure = 0.0 atm The top and bottom products in the distillation column is given as a mixture near the azeotropic composition and 9 % pure methanol respectively, and the permeate composition from pervaporation is 9 % pure MTBE. The distillation column configuration (N F, N, R, Product specification) is determined through algorithm. Thus, the energy efficient, feasible and near optimal solution is obtained visually through Figure a, without any detailed calculations. Detailed simulations have shown that energy consumption for separation scheme in Figure b is at least 35% less than any other feasible scheme that can be generated from Figure a. Example 3: In this example, a multi-component mixture is separated into the constituent pure component products in a sequence of simple distillation columns (Shah and Kokossis, 997). Application of the integrated approach means generation of 7 driving force diagrams for the 7 pairs of components (note that only 7 pairs, arranged in terms of boiling points need to be considered). These pairs indicate the separation tasks or splits for each column. Note that n-hexane and benzene form a binary azeotrope. The driving force diagrams for the 7 binary pairs are shown in Figure 5a. Since pair 7 forms a binary azeotrope but has a driving force which is third largest, it is removed from the list and placed at the end, by multiplying with a penalty factor of 0.. Applying algorithm a and then algorithm, the flowsheet shown in Figure 5b is obtained together with the design specifications (reflux ratio, feed location, product purity, etc.). Note that this flowsheet is different from the optimal proposed by Shah and Kokossis (997). Simulations of the flowsheet in Figure 5b and that proposed by Shah and Kokossis (997) however shows that the one obtained with the integrated approach requires less energy (about %). It is important to note, however, this flowsheet has been obtained visually without any detailed simulation and optimization. Example : This example deals with the separation of light-end alkanes through a sequence of distillation columns. The alkanes are C through C+ with a small amount of non-condensable gases. Details of the feed mixture is given in table. Four products are desired from the separation sequence (PRO/II User s Guide, 99). This leads to 3 driving force diagrams between the adjacent key components. Algorithm a is applied to predict the sequence of the 3 distillation columns. In the next step algorithm b is applied to determine the column pressures and algorithm is applied to determine the single column design. The driving force diagrams between the key components in each column confirm the sequence given in Figure, which is also the sequence given in the reference (PRO/II User s Guide, 99). Note that the operating pressures determined through algorithm b are different from the operating pressures given in the reference. It should be noted that the total operating cost given in table 3 is lower for the design proposed by algorithms, a and b. Further it can be noticed that the third column is operating at atmospheric pressure and not at vacuum
conditions, so the operating cost of this column will also be reduced, thereby making the flowsheet predicted by this new integrated approach even more advantageous. Conclusions A new integrated approach for separation synthesis/design and operation has been proposed. The approach is based on three algorithms that in an integrated manner enables the visual determination of the near-optimal (if not optimal) separation sequence together with the corresponding condition of operation and design of distillation columns. The integrated approach is also able to generate hybrid separation schemes where other separation techniques are also allowed. The new methodology is solely based on driving forces between two co-existing phases. Therefore only phase composition data are needed and the integrated approach requires no rigorous simulation or optimisation. The methodology not only identifies feasibility of different separation techniques for a given separation task, but also indicates the optimum methods of separation. Taking this into consideration, it is possible to make early decisions on separation sequences as well as distillation configuration that are near-optimum solutions. Results obtained from this new methodology are directly usable to formulate and solve a structural optimisation problem. Finally, the results appear to confirm the theory that separation at the highest driving force is the easiest separation and requires a near minimum of energy. Since energy is needed to create the driving force, this conclusion is not surprising. References Bek-Pedersen, E., Eden, Mario R., Jørgensen, Sten Bay, Gani, R., 999, A Driving Force Based Visual Technique for Separation Process Synthesis AIChE Annual Meeting, d, Dallas, Oct. 3-Nov. 5, 999. Gani, R. and Bek-Pedersen, E., 999, A Simple New Algorithm for Distillation Column Design Submitted to AIChE J, 999. Hunek, J., Gal, S., Posel, F., Glavic, P., 99, Separation of an azeotropic mixture by reverse extractive distillation AIChE J. Vol. 35, No. 7, pp. 07-09. Jaksland, C. A., R. Gani, K. Lien, 995, Separation process design and synthesis based on thermodynamic insights, Chem Eng Sci, 50, 5. PRO/II User s Guide, 99, Simulation Science Inc. Shah, P. B. and Kokossis, A. C., 997, Conceptual Programming: Towards the development of new targeting and screening technology using engineering and optimisation methods, AIChE topical conference on separation science andd technologies, Los Angeles, Nov. -, 997. 7
S M S7 E S7 C 3 5 7 9 0 3 5 7 9 0 3 S9 S E S C 0 C5 3 5 7 9 0 3 S5 S S C3 3 5 7 9 0 3 5 7 9 0 3 5 7 9 S3 30 S S S0 SC S3 S C 0 0 30 3 3 3 3 0 50 5 5 5 0 M S S5 0.3. Dy Top pressure in Column K ABS(yi-xi) Top pressure in Column K+ Bubble Point Pressure 0.0 0.0.0 Dx x (Light key component) Figure : Driving force diagram for constant ij = 3. Driving force based separation efficiency diagram on solvent (MeOH) free basis for Column T, P = atm. Driving force based separation efficiency on solvent (MeOH) free basis for Column T, P = atm. 0.35 0.0 0.30 0.5 Water Ethanol Water Isopropanol 0.35 0.30 -b- Water Ethanol Water Isopropanol Water -Propanol Water -Butanol ABS(yi-xi) 0.0 0.5 -a- ABS(yi-xi) 0.5 0.0 Water -Methyl--Butanol Water -Butanol Water -Methyl--Propanol 0.0 0.05 0.00 Dx = 0.7 0.00 0.0 0.0 0.30 0.0 0.50 0.0 0.70 0.0 0.90.00 xi 0.5 0.0 0.05 0.00 Dx = 0.3 Water -Pentanol 0.00 0.0 0.0 0.30 0.0 0.50 0.0 0.70 0.0 0.90.00 xi Figure : Driving force diagrams on solvent (MeOH) free basis (a) and driving force diagrams for all the alcohols with water (b) 3 5 7 9 0 3 5 7 9 0 3 5 7 9 30 3 3 33 3 35 3 37 3 39 5 59 Figure 3: Flowsheet for reverse extractive distillation proposed by Hunek et al (99).
0.3 Distillation Column -a- Pervaporation Unit VLE, atm. VLE, 5 atm. S 0.5 Permeate Feed Solvent free VLE Pervaporation Crystallisation 0 0. ABS(yi-xi) 0.5 Pervaporation Distillation 0 30 3 S3 -b- 3 0. S 3 3 0.05 Bottom Product 0 9 S7 S DISTILLATION 50 EXPANDER COMPRESSOR S PERVAPORATOR S5 0 0 0. 0. 0.3 0. 0.5 0. 0.7 0. 0.9 x (Methanol) Figure : Driving force diagrams for various separation techniques in example (a), and the flowsheet corresponding to the optimum flowsheet (b). Driving force based separation effciency curves at P = atm. A B 0.,-Dimethypropane i-pentane i-pentane n-pentane T T3 0. 0. -a- n-pentane,-dimethylbutane,-dimethylbutane,3-dimethylbutane,3-dimethylbutane -Methylpentane C 0. -Methylpentane n-hexane n-hexane Benzene T D E ABS(yi-xi) 0.0 0.0 T5 T7 0.0 T F 0.0 0.0 -b- T G 0.00 0.00 0.0 0.0 0.30 0.0 0.50 0.0 0.70 0.0 0.90.00 xi H Figure 5: Driving force diagrams for adjacent pairs (a) and the corresponding flowsheet (b), example 3. 9
C C3 T- Deethanizer T-3 Depropanizer T- Debutanizer C's C5+ V-3 V- Figure : Optimum sequence of columns in example. 0
Column N N F Specification I T 3 all alcohols = 0.9 T 0 7 Methanol through Isopropanol = 0.999 T3 30 Methanol = 0.9997 T 0 Ethanol = 0.99 Specification II water = 0.9 -Propanol through -Pentanol = 0.9995 Ethanol, Isopropanol = 0.99 Isopropanol = 0.97 Table : Column configurations for the flowsheet of Hunek et al (99). R Q R (GJ/hr).37 75.00.00 9.9 3.37 0.30 95.9.3 Column T T T3 T T5 T T7 Separation task,- Dimethylpropane / n-pentane n-pentane /,-Dimethylbutane i-pentane / n-pentane -Methylpentane / n-hexane,-dimethylbutane /,3- Dimethylbutane n-hexane / Benzene,3-Dimethylbutane / -Methylpentane N P N F R Specification I Specification II 7.7 3 5.5 30. 3 3 0. 0.5 39.0 3 73 59.0 Table : Column specifications for the optimum flowsheet of example 3.,- Dimethylpropane = 0.997 i-pentane through n-pentane = 0.999 i-pentane = 0.955,-Dimethylbutane through -Methylpentane = 0.995,-Dimethylbutane = 0.99 n-hexane = 0.95,3-Dimethylbutane = 0.9 i-pentane through Benzene = 0.99,- Dimethylbutane through Benzene = 099 i-pentane = 0.95 n-hexane through Benzene = 0.995,3-Dimethylbutane through -Methylpentane = 0.995 Benzene = 0.9 -Methylpentane = 0.99
Deethanizer, T- Depropanizer, T-3 Debutanizer, T- Data from Provision user s guide (99) Condenser pressure (kpa) 930 75 79 Feed location, N F 3 Reboiler Duty (GJ/hr) 0.059 9.57.5739 Data obtained from algorithms -3 Condenser pressure (kpa) 500 550 000 Feed location, N F 3 Reboiler Duty (GJ/hr).5. 7.059 Saving (%) 3..9-5.70 Total Saving (%).3 Table 3: Comparison of results from the new integrated approach with the results from the PRO/II User s Guise (99). Note that the extra cost for the vacuum operation of column T in the reference has not been considered. Components Example Example 3 Example Feed Feed composition Components composition Components (kmole/hr) (kmole/hr) Methanol 0.,- Dimethylpropane (A) Feed composition (kmole/hr) 3.5 Nitrogen 0.050 Ethanol.0 i-pentane (B) 59.97 Carbondioxide 0.99 i-propanol 5.9 n-pentane (C).7 Methane.70 Water 75,- Dimethylbutane (D). Ethane 5.903 -Propanol.0 -Butanol 0.,3- Dimethylbutane (E) -Methylpentane (F) 0.0 Propane.935.7 i-butane 5.0 -Methyl-- Propanol 3.3 n-hexane (G) 39. n-butane 0. -Butanol 0.5 Benzene (H) 0.33 i-pentane.5 -Pentanol. n-pentane.953 -Methyl-- Butanol 0. Hexane+ 7.73 Table : Feed compositions for examples,3 and.