188alvarado morales

Upload: ctomey

Post on 14-Apr-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/27/2019 188Alvarado Morales

    1/6

    20th European Symposium on Computer Aided Process Engineering ESCAPE20S. Pierucci and G. Buzzi Ferraris (Editors) 2010 Elsevier B.V. All rights reserved.

    Synthesis, Design and Analysis of DownstreamSeparation in Bio-refinery Processes through a

    Group-Contribution Approach

    Merlin Alvarado-Morales, Krist V. Gernaey, John M. Woodley, Rafiqul Gani

    Department of Chemical and Biochemical Engineering, Technical University of

    Denmark, DK-2800 Kgs. Lyngby, Denmark, [email protected]

    Abstract

    In this paper, a novel systematic approach to simultaneously model, design, and

    synthesize chemical and biochemical processes is presented. The core idea behind thisapproach is to apply the principles of the group-contribution approach for purecomponent property prediction to the synthesis and design of chemical processflowsheets. The method is highlighted through a bio-refinery case study involving theproduction of bioethanol (bioEtOH), succinic acid (SA) and diethyl succinate (DES),for which, energy efficient processing options have been identified.

    Keywords: process synthesis, flowsheet modeling, systematic method, process-group

    1.IntroductionIn the group-contribution method for pure component/mixture property prediction,molecular identity is described by means of a set of groups bonded together to form a

    specified molecular structure. By analogy, for flowsheet property prediction, a processflowsheet can be described by means of a set of process-groups bonded together to

    represent a specified flowsheet structure. The process-groups represent either a singleunit operation (such as a reactor, distillation, flash, etc.) or a set of unit operations (suchas extractive distillation, pressure swing distillation, etc). The bonds among the process-groups represent the streams and/or recycles, in an analogous way to the bonds that link

    molecular groups. Consequently, each process-group provides a contribution to theproperties of the flowsheet. The properties can be the performance in terms of energyconsumption or operating cost or profit, etc. In this way, once the flowsheet is describedby the process-groups, the property of interest can be calculated. Therefore, based onthis premise we have applied the group-contribution approach to systematically model,synthesize/design and analyze downstream separation from chemical and biochemical

    processes. The core idea of the approach in this paper, is based on the process-groupapproach developed by d Anterroches and Gani (2005) to solve synthesis/designproblems related to chemical processes. In order to represent an extended set of unit

    operations, an extended set of corresponding process-groups has been developed. Theapplication range of the new set of process-groups is highlighted by means of a bio-refinery case study involving the production of bio-ethanol (bioEtOH), succinic acid

    (SA) and diethyl succinate (DES).

    2.Overview of the methodThe method presented in this paper, the process-group contribution based approach,consists of the following seven steps: (1) synthesis problem definition, (2) synthesis

  • 7/27/2019 188Alvarado Morales

    2/6

    AlvaradoMorales et al.

    problem analysis, (3) process-group selection, (4) generation of candidates, (5)ranking/selection of candidates, (6) reverse simulation, and (7) final verification. Step 1

    involves the structural definition of the process inputs (raw materials) and outputs(desired products) of the process flowsheet as well as the definition of the flowsheet

    property targets. In step 2, in order to gain usable knowledge for the subsequent steps,reaction and pure component/mixture property analysis are performed. By identification

    of the component identities in the desired product that are not available as reactants (rawmaterials), a database search is performed to find the chemical reactions yielding thosecomponents as products. The pure component/mixture property analysis is performedby means of the thermodynamic insights based method developed by Jaksland (1996).

    This method is based on the principle that for each process operation task, the propertiesof the species to be separated can be associated in order to provide information related

    to the feasibility of a process separation task for a given separation technique. In step 3,the process-groups are matched with the feasible operation tasks and the separationtechniques identified in the previous step. The objective in step 4 is to combine the

    process-groups selected in step 3 according to the set of connectivity rules andspecifications proposed by d Anterroches and Gani (2005) to generate flowsheetstructures. Each process-group has output specifications, which are guaranteed to be

    met if the connectivity rules of the process-groups are satisfied. In step 5, the generatedflowsheet candidates are tested with respect to their target property values defined instep 1, using the corresponding flowsheet property model. Step 6 involves two tasks, theresolution of the mass balance for each process-group and the calculation of theflowsheet design parameters of the unit operations in the process flowsheet. Here,

    reverse simulation is used. In reverse simulation, knowledge of the state variablescorresponding to the inputs and outputs of a unit operation, i.e., individual flowrates,pressures and temperatures, are used to back calculate the design parameters of thecorresponding unit operation (e.g. number of stages, feed location, reflux ratio,residence time, volume). The reverse simulation for separation process-groups (such as

    distillation, extractive distillation, flash) is based on the driving force concept proposedby BekPedersen and Gani (2004). For the reactor process-group it is based on the workof Horn (1964) on the attainable region concept. In step 7, all the necessary information

    to perform the final verification through rigorous simulation is available. The use of acommercial simulator allows further fine-tuning of the alternatives and the possibility to

    perform optimization of the design parameters for the most promising candidates.

    3.Application of the method3.1.Case studyThe starting point of the case study is a slurry generated by a saccharification processwhere the conversion of cellulose to glucose is catalyzed by a cocktail of enzymes. Theavailable glucose in the slurry to be converted into bioEtOH and SA is equal to 32110

    kg/h. In order to calculate the amount of glucose that should be converted into SA, amass balance is done based on a SA production capacity of 190.34 kgSA/h and an annualload equal to 8406 h/yr. This SA production capacity corresponds to 10% of the currentworldwide SA production according to the BREW Project (Patel, 2006). Assuming aSA yield on glucose equal to 0.775 (Song et al., 2007), the amount of glucose needed toproduce this amount of SA is equal to 246 kg/h. From mass balance calculations, the

    amount of saccharified slurry containing this amount of glucose corresponds to 2811.6kg/h. The remaining slurry is then sent to the bioEtOH production plant where bioEtOH

    is produced by fermentation.

  • 7/27/2019 188Alvarado Morales

    3/6

    Synthesis, Design and Analysis of Downstream Separation in Chemical and

    BioProcesses through a GroupContribution Approach

    3.1.1.Synthesis problem definitionBased on the above analysis, the structural definition of the synthesis problem related to

    the downstream separation of bioEtOH from the effluent of the fermentor has beenformulated as follows: one input process-group initialized with the mixture produced by

    the fermentor and one output process-group initialized with the desired product,bioEtOH. The flowsheet property target is the minimization of the energy consumption.

    The structural definition of the synthesis problem related to SA production isformulated as follows: one input process-group initialized with the mixture produced bythe saccharification reactor, and one output process-group initialized with the desiredproduct, SA. For the DES production process the structural definition is formulated as

    follows: one input process-group initialized with bioEtOH and SA streams resultingfrom their respective production processes, and one output process-group initialized

    with the DES. Figure 1 shows a schematic representation of the synthesis problem.

    Figure 1. A schematic representation of the synthesis problem.

    3.1.2.Synthesis problem analysisThe reactions in the SA production process taking place in the fermentor were reportedby Song et al. (2007), and the reactions for the DES production process were taken fromKolah et al. (2008). From the pure component/mixture property analysis for thebioEtOH process, flash separation has been identified to separate CO2 and O2 from theother components. Flash/evaporation and distillation were identified as alternatives toperform the separation of water from the other components, except for the ethanol-waterbinary pair which forms an azeotrope. Further analysis of this binary azeotrope

    indicated that ethanol can be recovered by using liquid membrane, pervaporation, gasadsorption or extractive distillation. For the SA process, formic acid, pyruvic acid, CO2,H2, O2, ammonia and microbial cells were found to be either products and/or reactantsin the reactions. A pure component/mixture analysis was performed taking into accountthese components. Flash, distillation, and pervaporation have been identified to performthe separation of CO2, H2, O2, and ammonia from the other components. Crystallization

    has been identified to separate SA from the rest of the components. However, due to thehighly dilute nature of the mixture, liquid-liquid extraction has also been identified assuitable for the separation of water from the rest of the components in the mixture. AsDES is produced from SA via the intermediate formation of monoethyl succinate(MES), this component was also taken into account for the pure component/mixtureanalysis. Flash/evaporation, liquid membrane, and pervaporation, have been identified

    to perform the separation of water and ethanol from the other components in the DESsynthesis problem. Crystallization has been identified as a feasible separation techniqueto perform the separation of SA from DES and MES, while liquid adsorption, liquid

  • 7/27/2019 188Alvarado Morales

    4/6

    AlvaradoMorales et al.

    membrane, and pervaporation have been identified to perform the separation of DESand MES.

    3.1.3.Process-group selectionThe corresponding process-groups are selected from the database according to thefeasible separation techniques identified in the previous step, and are matched with thecomponents in the mixture.

    3.1.4.Generation of candidatesFor the downstream separation design of a bioEtOH production process 288 candidates

    have been generated through the combination of the process-groups selected in theprevious step. However, out of the 288 design alternatives, only 4 candidates were ofinterest, since only they satisfied the connectivity rules as well as the structuraldefinition of the synthesis problem. For the SA production process, 65 designcandidates have been generated through the combination of the process-groups. Out ofthese 65 candidates, 16 were found to be feasible based on the structural constraints

    defined in the problem. For the DES production process, 54 feasible candidates havebeen generated.

    3.1.5.Ranking/selection of candidatesThe design candidate using the solvent-based azeotropic separation process-group in thedownstream separation design for the bioEtOH process is considered to highlight the

    workflow in this step of the method. Note that the solvent identity is not known so far.Therefore, a CAMD (Computer Aided Molecular Design) problem formulation can be

    set up to find a matching solvent. The ProCAMD tool (Gani et al., 1997) has been usedto find potential candidates. With respect to ionic liquids (IL), the potential candidateswere found through a search in the open literature (Seileret al., 2004; Jorket al., 2004;Wang et al., 2007). Table 1 shows the performance of the design candidates in terms of

    energy consumption.

    Table 1. Design candidate results.

    Candidate Solventfraction

    Drivingforce

    Predictedenergy

    (MkJ/h/kmole)

    Energy demand(MkJ/h/kmole)

    Seileret al.

    (2004)

    Glycerol 0.63 0.48 0.0322

    Ethylene glycol (EG) 0.52 0.48 0.0317 0.0335

    Triethylene glycol 0.63 0.25 0.0618

    [EMIM]+[BF4] 0.375 0.37 0.0352 0.0333

    [BMIM]+[Cl] 0.45 0.42 0.0260

    [EMIM]+[EtSO4] 0.40 0.31 0.0386

    [EMIM]+[DMP] 0.40 0.38 0.0318

    In the case of the SA production process, only one candidate using a liquid-liquid basedseparation process-group (representing liquid-liquid extraction) is selected for furtheranalysis. For the bioEtOH process, the solvent identity is not known so far. Performing

    a database search, the following candidates have been identified as potential extractiveagents: n-decyl acetate and n-butyl acetate. The performance of the potential candidates

  • 7/27/2019 188Alvarado Morales

    5/6

    Synthesis, Design and Analysis of Downstream Separation in Chemical and

    BioProcesses through a GroupContribution Approach

    has been tested throughPT-flash (multi-phase) rigorous calculation using ICAS (Gani etal., 1997). Based on this calculation, n-decyl acetate was selected as the best extractive

    agent since it is totally immiscible with water and it can promote a higher driving forcethan n-butyl acetate. Note that additional criteria such as cost of the solvent or toxicity

    of the solvent can be taken into account as well before making the final selection. Sincethe solvent needs to be recovered (for recycle), crystallization has been identified to

    separate the solvent from SA. In the case of the DES production process, the onecandidate using the pervaporation process-group is selected for further analysis.

    3.1.6.Reverse simulationThe feasible flowsheet shown in Figure 2 is considered for the reverse simulation step.In Figure 2 each process unit operation is represented by its corresponding process-

    group. Once the solvent identity is known, the solvent flowrate is calculated by massbalance.

    Figure 2. Process flowsheet for the selected design candidate.

    A mass balance was made for each process-group for the process flowsheet in Figure 2.The downstream separation in the bioEtOH process using EG as an entrainer and the

    downstream separation in the SA process using n-decyl acetate as an extractive agent,were considered for reverse simulation. The reverse simulation of the flash anddistillation process-groups was performed using the driving force based method (Bek-Pedersen and Gani, 2004) and the results are given in Table 2.

    Table 2. Design parameters of the distillation columns.

    Distillation column Extractive column Recovery column

    Number of stages 32 30 15Feed stage 17 22 5

    Reflux ratio 3.2 0.52 0.54

    DFmax 0.35 0.48 0.59

    In the case of the liquid-liquid based separation process-group, the number of stages canbe determined by using the phase diagram, plottingX(kgsolute/kgcarrier) versus Y(kgsolute/kgsolvent) as illustrated Figure 3.

  • 7/27/2019 188Alvarado Morales

    6/6

    AlvaradoMorales et al.

    0.00 0.04 0.08 0.12 0.16 0. 20 0.24 0.28

    0.000

    0.005

    0.010

    0.015

    0.020

    0.025

    0.030

    0.035

    0.040

    0.045

    0.050

    Operation line

    Equilibrium curve

    R

    AXF

    AX4

    3

    2

    Y(kg

    SuccinicAcid/kg

    Solvent)

    X(kgSuccinic Acid/kgWater)

    1

    fs (target solubility)

    Figure 3. Graphical determination of the number of equilibrium stages for the liquid-liquidextraction process-group.

    3.1.7.Final verificationThe candidates with the best performance in terms of energy consumption were verifiedthrough rigorous simulation using an appropriate simulator. The optimal integratedsequence uses extractive distillation with IL for BioEtOH recovery, liquid-liquidextraction and crystallization for SA recovery, and pervaporation for DES recovery. The

    total predicted energy consumption is equal to 0.329MkJ/h/kmole.

    4.ConclusionsA novel systematic approach based on the group-contribution concept has beenpresented. One important feature of the method is its versatility, since it can be extended

    by adding new process-groups representing all types of process unit operations. Thus, itis possible to simultaneously model, design, and synthesize novel products and

    processes as is demonstrated in this paper. On the other hand, the ability to predict aflowsheet property energy consumption without the need for rigorous simulationoffers a lot of advantages as it opens the possibility to screen a lot of process optionsvery quickly and with high accuracy, a feature which has been demonstrated fordownstream separation from the bioEtOH and SA processes. Finally, the results also

    show that the method provides a fast, efficient, and systematic process design approachby first solving the mass balance based on the process-group specifications, followed by

    calculation of the design parameters of the unit operations through reverse simulation.

    References

    E. Bek-Pedersen, R. Gani, Chem. Eng. Process., 43 (2004) 251.

    L. dAnterroches, R Gani, Fluid Phase Equilib., 228-229 (2005) 141.R. Gani, G. Hytoft, C. Jaksland, A.K. Jensen, Comput. Chem. Eng., 21 (1997) 1135.F. Horn, (eds.), In Proc. 3rd Eur. Symp. on Chemical Reaction Engineering, 1964.C. Jaksland, PhD Thesis, Department of Chemical Engineering, DTU, Denmark, 1996.C. Jork, M. Seiler, Y.A. Beste, W. Arlt, J. Chem. Eng. Data, 49 (2004) 852.A.K. Kolah, N.S. Asthana, D.T. Vu, C.T. Lira, D.J. Miller, Ind. Eng. Chem. Res., 47

    (2008) 5313.M. Seiler, C. Jork, A. Kavarnou, W. Arlt, R. Hirsch, AIChE Journal, 50 (2004) 2439.H. Song, Y.S. Huh, S.Y. Lee, H.W. Hong, Y.K. Hong, J. Biotechnol., 132 (2007) 445.M. Patel, The BREW Project, Final report, Utrecht University, 2006.JF. Wang, CX. Li, ZH. Wang, ZJ. Li, YB. Jiang, Fluid Phase Equilib., 255 (2007) 186.