Jie Li,Zhijian Peng, *,Chunshan Li *,Ping Li,Rafiqul Gani

1 School of Engineering and Technology,China University of Geosciences,Beijing 100083,China

2 Beijing Key Laboratory of Ionic Liquids Clean Process,Key Laboratory of Green Process and Engineering,Key Laboratory of Multiphase Complex Systems,Zhongke Langfang Institute of Process Engineering,Institute of Process Engineering,Chinese Academy of Sciences,Beijing 100190,China

3 State Key Laboratory of High-efficiency Utilization of Coal and Green Chemical Engineering,School of Chemistry and Chemical Engineering,Ningxia University,Yinchuan 750021,China

4 PSE for SPEED,Skyttemosen 6,DK-3450 Allerod,Denmark

Keywords:Extraction Methacrylic acid Optimization Techno-economic analysis

ABSTRACT In this work,a techno-economic study for the solvent based extraction of methacrylic acid from an aqueous solution is presented.The involved phase equilibrium calculations in process design are verified by measured experimental data.First,experiments are conducted with different solvent candidates to measure LLE(liquidliquid equilibrium)data and to establish the effects of extraction temperature and dosage of solvent.Next,the binary interaction parameters for the UNIQUAC model to be used for equilibrium calculations are fine-tuned with measured data.Then,a process for the solvent based extraction of methacrylic acid recovery is designed and verified through simulation with the regressed UNIQUAC model parameters.The optimal configuration of the process flowsheet is determined by minimizing the total annualized cost.Among the three solvent candidates considered-cyclohexane,hexane and toluene-the highest efficiency and the lowest total annualized cost is found with toluene as the solvent.

1.Introduction

Methacrylic acid(MAA)is a typical carboxylic acid and an important monomer that is widely used in the production of fibers,adhesives,paints,and plexiglass[1-4].The green synthesis route of MAA from coal-based route has attracted extensive attention in recent years[5-7].In this route for synthesis of MAA,a large quantity of water(～50 wt%)exits as the reactor outflow.Thus,the separation of MAA from an aqueous solution is an important separation step that needs to be further investigated.

The separation techniques are determined by concentration of carboxylic acids.Distillation is a widely used method to recover carboxylic acids from aqueous solutions when the concentration of carboxylic acids is above 50 wt%.While,if it is below 50 wt%,separation by distillation should evaporate a large amount of water which is a high energy consumption process.Furthermore,because of the tendency of carboxylic acids to polymerize at temperatures higher than the boiling point of water,distillation is not the best option for recovery of low concentration of carboxylic acids.A variety of alternatives to distillation for recovering carboxylic acid from aqueous solutions have been proposed,such as absorption,liquid-liquid extraction,reactive extraction[8],ion exchange[9],electrodialysis[10],and reverse osmosis[11].Compared to distillation,absorption and ion exchange are the low-energy extraction methods.However,the application of these methods is prevented by difficulties with respect to product recovery,material life,disposal of used material,and regeneration capacity of additives such as resin[12].Also,even though membrane-based separation has been developed in recent years[13,14],their vulnerability to contamination of membranes leads to the possibility of lower separation efficiencies with time of operation.

The technology of liquid-liquid extraction is a separation scheme that is easy to industrialize and has relatively low energy consumption[15,16],where selection of the appropriate solvent is a primary task.The solvents for recovery of carboxylic acid from aqueous solutions can be classified into two categories[17,18]:physical solvents and reaction solvents.Physical solvents are mainly alkanes,esters,aromatic hydrocarbons and so on[19]，which have no complexation reaction with carboxylic acids.However,reaction solvents have strong interaction with carboxylic acids and from complexes.Because of this strong interaction,the recovery of reaction solvents has become more difficult than physical solvents.Recent developments of ILs as novel solvents have attracted much attention in recovery of carboxylic acids such as acetic acid[20],propionic acid[21],lactic acid[22],succinic acids[23],and hydroxycinnamic acids[24]because of their environment-friendly properties,ease of recovery,re-usability and many more.A series of ILs have been investigated with different cation and anion combinations as solvent candidates for MAA recovery from a dilute aqueous solution[25].The extraction performance of ILs is found to be higher than hexane for low concentration of MAA(1 wt%).However,since the concentration of MAA is around 20 wt%～35 wt%from reactor in industrial productions,the use of ILs becomes economically infeasible.In addition,the price of ILs is much higher than that of organic solvents.Therefore,ILs as solvents for MAA extraction in industrial productions is not yet the best choice.Thus,organic solvents are the preferred extractants for industrialization.However,the extraction-step alone cannot give the desired product.In order to satisfy the technoeconomic requirements,the extraction-step integrated with the solvent recovery-step,where distillation may be employed,needs to be investigated.

The objective of industrial production is to maximize profits[26].However,the economic analysis needs to be integrated with process evaluation in order to determine the optimal process that gives the best profit subject to the process specifications.Economic analysis includes capital costs,operating costs,as well as analysis of payback period[27,28].Many cost models exist for economic analysis,such as Happle and Jordan[29],Guthrie[30],Page[31],Garrett[32],and Humphreys[33].In this work,the Guthrie method is used for its simplicity,consistency and wide applications.

Although some organic solvents have been reported in patents[19,34],in this paper,the objective has been to find better solvents together with the associated extraction that correspond to the optimal economic objective.In this work,the effect of MAA extraction by different solvents is studied;the optimal process design together with the best extractant is determined and verified through experiments.The simulation results used for the techno-economic evaluation are based on regressed thermodynamic model parameters with the required data obtained through experiments.The simulationbased process design and economic analysis has been done through Aspen Plus with regressed UNIQUAC binary interaction parameters.Three organic extractants have been studied and the optimal economic MAA separation process has been found with toluene as the solvent.

2.Experimental Method and Data

2.1.Chemicals

Cyclohexane(≥99.5%,)，n-hexane(≥95%),acetic acid(≥99.5%,HAC),and toluene(≥99.5%)are purchased from Beijing Chemical Works(China).Ethanol(≥99.7%)is bought from Sinopharm Chemical Reagent Co.,Ltd.(China).Methacrylic acid(≥99.0%)is purchased from Aladdin Industrial Corporation(China).All of the chemicals are of analytical-reagent grade.

2.2.Extraction experiment

A fixed amount of MAA-HAC-H2O-solvent mixture is added to the liquid-liquid equilibrium(LLE)still(supplied by Zhejiang University),which is equipped with magnetic stirrer and water bath.The mixture is stirred sufficiently for 2 h in the LLE still and settled in a water bath,which is kept at a constant temperature until equilibrium is reached.After at least a further 3 h,samples are collected from the two liquid phases for analysis.

All the samples are analyzed using gas chromatography(2010-GC Plus)with a BID detector.Ethanol is selected for addition to the samples as an internal standard and for calibration.Each experiment is repeated three times to guarantee the repeatability of the measured data.The final reported result is the average of the three measured data.

2.3.Extraction efficiency of MAA

The extraction efficiency of MAA is defined as:

where E(%)is the percentage extraction efficiency of MAA;[MAA]org and morgare the mass fraction of MAA and the total mass of organic phase,respectively.[MAA]0and m0are the mass fraction of MAA and the mass that is fed into the separation system.

2.4.Regression of UNIQUAC model parameters

The UNIQUAC model,as an equation derived from the two-liquid theory,is widely used in the simulation of extraction and separation processes[35].The good predictive effect of the application in carboxylic acid system has been proved by many researches[36,37].Therefore,liquid-liquid equilibrium(LLE)experimental data for quaternary MAA+H2O+HAC+extractant mixture are selected to regress the LLE binary interaction parameters for UNIQUAC.

To ensure reliable performance of the regressed UINQUAC model within Aspen Plus the absolute deviation(AD)values of extraction efficiency of MAA are calculated between simulation obtained by regressed binary interaction parameters and experiments.

where i is the point at a constant condition;refer to the simulated extraction efficiency of MAA and experimental extraction efficiency of MAA,respectively.

2.5.Separation process design and cost data

The feed composition,operation conditions,and MAA purity requirement based on industrial process requirements are listed in Table 1.A typical process flowsheet extractant based MAA separation is shown in Fig.1.The dosage of solvent is determined with the requirement of 99 wt%recovery of MAA from the feed stream.

Table 1 Feed conditions and product purity

In the work,the total annualized cost(TAC)of the process is used as the objective function,and the calculation of this involves the operation cost including the cost of utilities(streams and cold water)and required solvent and the equipment investment consisting of the purchased cost and the installed cost.The TAC is determined through correlations which are given below.

2.5.1.Equipment size

In this part,the main equipment including extraction &recovery column and heat exchangers are chosen to calculate the investment.

2.5.1.1.Columns.The column is designed to have a sieve tray with 2 ft tray spacing.The column diameter is determined by the column flooding condition,which determines the upper limit of the vapor velocity.The operating velocity is usually between 70% and 90%of the flooding velocity[38].In this study,80% of the flooding velocity is used as the default.The diameters of the columns are calculated by sizing-option Aspen Plus,where the internal geometry is adjusted to the calculated hydraulic flow inside the column.The column height is obtained by the following equation:

Fig.1.Separation process of MAA.

where H is the column height;NTdenotes the number of theoretical plates;ETis the plate efficiency,and the value of ETis normally 60%.

2.5.1.2.Heat exchangers.The area of the heat exchanger is obtained for an exchanger minimum approach temperature of 10°C.

2.5.2.Cost

The cost is obtained by Guthrie correlation formula[30].

Column:

The prices of solvents refer to the average price of each solvent in the industrial scale trading market.

TAC=(equipment cost/depreciation year)+operation cost,depreciation year=10 years

2.6.Process simulation and optimization

Typically,the process design and optimization are based on economic objectives,for example,the TAC defined above.The optimization of process is conducted based on the sequential quadratic programming(SQP)algorithm[39,40]in Aspen Plus.The SQP method is effectively used to solve nonlinearly constrained optimization problems;it has been successfully proved efficient in many studies[28,41-45].The process simulation needed for optimization is also performed in Aspen Plus using the regressed UNIQUAC model for the prediction of LLE.The process design parameters and cost model parameters given above are used in the simulation and optimization of the separation of MAA with different solvents.

3.Results and Discussion

3.1.Extraction experiment analysis

In this work,cyclohexane,toluene,and hexane are selected to extract MAA from the aqueous solution whose composition is listed in Table 1.The extraction efficiencies are calculated to evaluate the extraction performance of the three solvents.

The extraction temperature and dosage of solvents are investigated through the LLE based extraction experiments.The extraction efficiencies of MAA,calculated based on Eq.(1),are shown in Fig.2,which are the average of the three measured data The extraction temperature is varied from 10 to 50°C.The dosage of solvents is expressed by the mass ratio of the extractant to MAA in raw material.

As shown in Fig.2(a),the extraction efficiency of MAA decreases with the increase of extraction temperature.This phenomenon is observed for toluene and hexane as extractants(as shown in Fig.2(b,c)).As the mass ratio of the extractant to MAA becomes larger,the extractant shows a better extraction efficiency.For cyclohexane and toluene,the extraction efficiency increases slowly when the mass ratio is greater than 0.8(see Fig.2(a,b)),while for hexane(Fig.2(c)),this value is close to 1.0.For a deeply understanding of the extraction efficiencies for the three solvents,the measured data at temperatures of 20°C and 30°C,which are typically used in industry,are compared in Fig.3.It can be noted from Fig.3 that toluene as the solvent shows the best extraction effect,followed by hexane and cyclohexane.

Fig.2.The extraction efficiency of MAA for(a)cyclohexane,(b)toluene,(c)hexane.

3.2.Experiment data regression and evaluation

Fig.3.Comparison of the performance of the three solvents.

To enable simulations,the binary interaction parameters for a liquid activity coefficient model are necessary.In the work,the UNIQUAC model is applied for simulation-based techno-economic evaluation.The binary interaction parameters in Aspen Plus are selected for simulation of the extraction MAA by the decanter model.As the available binary interaction parameters within Aspen Plus for the UNIQUAC model did not give a reasonable match of the measured data.The source of the binary interaction parameters is changed to verify the binary interaction parameters,the good match between simulation and experimental is determined by the binary parameters of MAA-H2O.The details are presented in supporting information.Thus,the binary parameters of MAA-H2O should be obtained by regression from experiment data.The experiment data corresponding to extraction by toluene(given in Table 2)is regressed to obtain the binary interaction parameters for the MAA-H2O interaction(given in Table 3).Other binary interaction parameters are from Aspen shown in supporting information.

Based on the above analysis,the comparison between experiment and simulation at 20°C and 30°C are chosen to evaluate the regression results.As can be seen from Fig.4,the AD values are within 5%,except for one point at 0.5 used by hexane(30°C),the value is close to 5%.The low AD values indicate that the regression data has an acceptable agreement with experiment.

3.3.Process optimization and comparison

The separation process of MAA is shown in Fig.1.The extraction column model is selected to simulate the extraction process.In this column,the ratio of solvent to the feed flow rate(E/F)is an important variable in solvent-based extraction design.High E/F values give high extraction efficiencies,but it leads to the requirements of larger energies in the distillation step for recovery and recycle of the solvent.Since the energy requirements is obtained totally from the distillation step in the separation process(shown in Fig.1),the E/F should be optimized first.Fig.5 shows the results of extractor optimization under the constraint that the recovery of MAA is greater than 99%.The dosage of solvent decreases as the number of stages(extraction column)increases for all three solvents.The amount of solvent must be minimized to reduce the energy required in the distillation step.However,the increased number of stages results in the higher capital cost.Thus,the balance of two factors should be considered in the optimization process.As shown in Fig.5,there is an obvious reduction in solvents requirement with the increased number of stages in a certain range,which are 3 to 7,3 to 6,and 3 to 7 for cyclohexane,toluene and hexane as solvents,respectively.As the values of number of stages are beyond the certain range,the variation of E/F is relatively small.It results in relatively insignificant reduction of solvents required.Based on the above analysis,the optimum numbers of stages for extractor are respectively 7,6,and 7 using cyclohexane,toluene and hexane as solvents for subsequent studies.

Table 2 Experiment LLE data of MAA(1)-H2O(2)-HAC(3)-Toluene(4)

Table 3 LLE binary interaction parameters of UNIQUAC model for MAA-H2O

Fig.4.Comparison of experiment and simulation.

The objective function with minimization of TAC mainly involves the equipment(column and heat exchangers)and operation cost.Two variables are chosen to optimize the extractor:(1)E/F,(2)the number of extractor stages.The optimal results of extractor have been discussed above.There are also two variables in the distillation step:(1)the number of stages,(2)reflux ratio.To satisfy the purity requirement of MAA(≥99 wt%),optimization is conducted using the SQP method and sensitivity in Aspen Plus.Before optimization,the maximum recovery of MAA on different number of stages was determined on the basis of MAA purity above 99.9 wt%,the details are provided in supporting information.The optimized values of number of stages and reflux ratio are achieved as follows:

Fig.5.E/F optimization with number of extractor stages.

First,the MAA concentration in the bottom of the distillation column is investigated with the variables of number of stages and reflux ratio.The better values of number of stages and reflux ratio are selected under the conditions where MAA concentration is greater than 99 wt%.Then TAC is achieved based on the estimated correlations in Section 2 from the better values.The values corresponding to the minimization TAC are the optimal number of stages and reflux ratio for the distillation column.In addition,the optimal feed stages were obtained by sensitivity,which are presented in supporting information.

Fig.6 shows the effects of number of stages and reflux ratio on the purity of MAA in the bottom product of the distillation column.The MAA concentration in the bottom is determined by reflux ratio.The ranges of the reflux ratio are from 0.1 to 0.2,0.2 to 0.5,and 0.03 to 0.1 for cyclohexane,toluene and hexane,respectively.Based on the optimization procedure discussed above,the TAC is shown in Fig.7.The optimal results are summarized in Table 4.Comparing the TAC in Table 4,the process using toluene as an extractant has the lowest TAC.The optimal values for toluene extraction are 18 for number of stages,0.24 for reflux ratio,and 1.29815×105USD for TAC.

4.Conclusions

A techno-economic analysis of the process for recovery of MAA from an aqueous solution has been proposed in this work.The experiments with extractants have been conducted to obtain LLE data,which is used for modeling,solvent based extraction process for recovery of MAA.For the selected phase equilibrium UNIQUAC model,the measured quaternary LLE data has been used to fit the 6 involved binary pairs.Only one pair,MAA-water pair needed to be fine-tuned,while changes in all other pairs gave insignificant improvements.Note that these fine-tuned parameters can only be used for the specific quaternary systems studied in this work Process simulation with the regressed model parameters have been performed to obtain the initial design,which has been optimized by the minimizing TAC.Compared with cyclohexane and hexane,toluene exhibits the best extraction effect and lowest TAC cost.The minimum value of TAC cost is 1.29815×105USD.The optimal numbers of stages are 6 for extractor and 18 for distillation.The values of E/F ratio and reflux ratio are 0.35 and 0.24,respectively.The results reveal that it provides a significant guidance for the separation methacrylic acid from aqueous solution in industrial production.

Fig.6.The effects of number of stages and reflux ratio on the purity of MAA in the bottom using(a)cyclohexane,(b)toluene,(c)hexane.

Fig.7.TAC cost of(a)cyclohexane extraction(b)toluene extraction(c)hexane extraction.

Table 4 Summary of optimal results

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2019.02.014.