Design and control of entrainer-assisted reactive distillation for N-propyl propionate production

Accepted Manuscript Title: Design and Control of Entrainer-Assisted Reactive Distillation for N-Propyl Propionate production Authors: Hui Xia, Xin Dai, Qing Ye, Shenyao Feng, Rui Li, Xiaomeng Suo PII: DOI: Reference:

S0098-1354(17)30295-8 http://dx.doi.org/doi:10.1016/j.compchemeng.2017.08.003 CACE 5870

To appear in:

Computers and Chemical Engineering

Received date: Revised date: Accepted date:

14-3-2017 29-7-2017 3-8-2017

Please cite this article as: Xia, Hui., Dai, Xin., Ye, Qing., Feng, Shenyao., Li, Rui., & Suo, Xiaomeng., Design and Control of Entrainer-Assisted Reactive Distillation for N-Propyl Propionate production.Computers and Chemical Engineering http://dx.doi.org/10.1016/j.compchemeng.2017.08.003 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Design and Control of Entrainer-Assisted Reactive Distillation for N-Propyl Propionate production Hui Xia, Xin Dai, Qing Ye*, Shenyao Feng, Rui Li , and Xiaomeng Suo

Jiangsu Key Laboratory of Advanced Catalytic Materials and Technology, School of Petrochemical Engineering, Changzhou University, Changzhou, Jiangsu 213164, China

AUTHOR INFORMATION *Tel.: +86 519 86330355. Fax: +86 519 86330355. E-mail: [email protected].

Highlights: 

We study an entrainer assisted reactive distillation process for the N-Propyl Propionate production.



This Entrainer assisted reactive distillation process can save 46.11% of TAC compared with the two-column process.



The entrainer assisted reactive distillation process can reduce the reboiler duty by 41.40% compared with the two-column process.



Two control structures are proposed to control the entrainer assisted reactive distillation process for the N-Propyl Propionate production.

Abstract: An entrainer-assisted reactive distillation process is proposed to produce high-purity N-propyl propionate from propionic acid and N-propanol. The E-RD process can take advantages of both the heterogeneous azeotropic distillation (HAD) and reactive distillation (RD). Cyclohexane is selected as the proper entrainer in the E-RD process. And the E-RD process is optimized by calculating the minimum total annual cost (TAC). The optimal results reveal that the E-RD process can save 46.11% of TAC and 41.40% of reboiler duty compared with the two-column process. Furthermore, two control structures for the E-RD process are considered. The dynamic performances demonstrate that the improved control structure (CS2) can solve the problem of disturbances and maintain the product purities close to the set points with small deviations and short settling times. Key words: esterification, propyl propionate, entrainer-assisted reactive distillation, temperature control

NOMENCLATURE E-RD

Entrainer-assisted reactive distillation

Propro

N-Propyl propionate

POH

N-propanol

ProAc

Propionic acid

HAD

Heterogeneous azeotropic distillation

RD

Reactive distillation

RDC

Reactive distillation column

RC

Recovery column

ri

Reaction rate, mol·eq-1·s-1

aProAc

Liquid phase activity of the ProAc

aPropro

Liquid phase activity of the Propro

awater

Liquid phase activity of the water

aPOH

Liquid phase activity of the POH

mcat,dry

Mass of the dry catalyst, g

vi

The stoichiometric coefficient of the ith component

ccat

Concentration of active sites on the dry catalyst, eq·kg-1

Keq(T)

Activity based equilibrium constant, mol·eq-1·s-1

k1

Rate constant of the forward reaction, mol·eq-1·s-1

R

Universal gas constant , J·mol-1·K-1

TAC

Total annual cost

BP

Boiling points

VLE

Vapor-liquid equilibrium

VLLE

Vapor-liquid-liquid equilibrium

RCM

Residue curve map

TOC

Total operating cost

TCC

Total capital cost

TAC

Total annual cost

PI

Proportional and integral

QB

Reboiler duty

Nr

Numbers of the rectifying stages

Nrz

Numbers of the reaction stages

Ns

Number of the stripping stages

Kc

Gain

Ti

Integral time

TC

temperature controller

KU

ultimate gains

PU

ultimate periods

1. Introduction N-propyl propionate (Propro) is a widely used chemical solvent. It has many applications in paints, coatings, and other industrial products. Propro can be produced by the liquid-phase esterification of N-propanol (POH) and propionic acid (ProAc) (Altman et al., 2012; Buchaly et al., 2012; Gooch, 2007). The esterification reaction is a kind of reversible reaction and it is limited by chemical equilibrium. Propro is produced in a batch or continuous reactor in a homogeneous or heterogeneous system catalyzed by acids followed by several distillation columns (Cruz-Díaz et al., 2012; Duarte et al., 2006). It has many problems: it has low conversion rate; it consumes large capital and energy cost and the corrosion of the equipment is serious. Hence, reactive distillation (RD) is an effective method to solve the problems. RD is an integration of reaction and separation into a single column. It allows the simultaneous production and removal of the products, it can also improve the productivity and reduce capital costs and energy consumption (Kiss, 2013). It has been applied to the industrial production of MTBE, ETBE and TAME (Al-Arfaj and Luyben, 2004; Domingues et al., 2014; Huang et al., 2008). Several studies reported the production of Propro in a reactive distillation column (RDC). Kotora et al. (2008) studied the experiment for the Propro production in a pilot-scale RDC, the purity of Propro in the process is 0.698; Keller et al. (2011) studied the experiment of the production of Propro in a pilot-scale RDC assisted with

a liquid−liquid phase separator. The purity of Propro in the process is 0.521. Based on the processes, the experimental purities of Propro are below 0.7, this is because POH and water form a minimum-boiling azeotrope so that most of POH is removed from the reaction zone of the RDC. Since POH is not totally reacted in the reactive section, and the chemical equilibrium of the esterification is limited, an extra separation unit is necessary to recover and recycle the unreacted alcohol. Xu et al. (2014) proposed a two-column process to synthesize Propro, they concluded that it is hard to produce high-purity Propro in a single RDC because it could not operate ‘neat’. The two-column process feathers a RDC coupled with a decanter and a recovery column (RC). Consequently, the purity of Propro in the process is 0.9975. Though high-purity Propro can be obtained in the two-column process, the two-column process needs larger capital cost and energy consumption. To reduce the capital cost and energy consumption, Dimian et al. (2002) introduced a novel entrainer-assisted RD process by adding an entrainer to the reactive distillation system. The entrainer is able to form a minimum ternary heterogeneous azeotrope with alcohol and H2O. The formation of azeotrope can enhance water removal and improve the concentration of reactants. A second column is unnecessary for alcohol recovery. Recently, some researchers studied the entrainer-assisted reactive distillation process. Wang et al. (2011) concluded that the entrainer assisted reactive distillation process combines both the advantages of heterogeneous azeotropic distillation (HAD) and RD. Jong et al. (2008) studied the entrainer assisted RD process and the conventional RD process for the fatty acid esters production. They concluded that the E-RD process is able to reach the required conversion of 99%, and the E-RD process needs fewer reactive stages and energy consumption than that of the conventional RD process. Hu et al. (2011) investigated an entrainer assisted RD process for the ethyl acetate production, in the process, N-butyl acetate is selected as the entrainer to carry out H2O from the RD column. The entrainer assisted reactive distillation process can save 32% of energy consumption compared with the conventional RD process. As is concluded in the literature (de Jong et al., 2008; Dimian et al., 2002; Hu et al., 2011; Wang and Huang, 2011), the entrainer-assisted reactive distillation process can not only produce

high purity products, but also reduce capital investment and operation costs. To reduce capital cost and energy consumption of the two-column process, the entrainer-assisted reactive distillation (E-RD) process can be used to produce Propro. Dynamic control is another important aspect in the E-RD process. The control structure of E-RD process is more complex than the control structures of the RD systems and the azeotropic distillation systems. The control of RDC and the azeotropic distillation column have been deeply investigated by many researchers. But the control structure of the entrainer-assisted reactive distillation column has attracted less attention. Chen et al. (2016) proposed a temperature control structure to control a RD process for the Methyl Valerate production. Based on the control structure, the reboiler duty is utilized to control the tray temperature of the column. In terms of dynamic performance, the control structure was able to deal with disturbances, maintain the methyl valerate purity, and get to steady state very fast. Huang et al. (2004) proposed a temperature control structure to control the heterogeneous reactive distillation process. The ratio scheme between the two reactants is used to keep the balance of reactants and the reboiler heat duty is used to control the tray temperature of the column. The dynamic results show that the control structure can reject throughput disturbances very fast and maintain the product purity. Wang et al. (2006) investigated a temperature control structure to control the entrainer-added RD process for the fatty ester production. In the process, the tray temperature of the entrainer-added RD is controlled by manipulating the reboiler duty. The dynamic results show that the temperature control scheme has good dynamic performance. Hung et al. (2014) investigated the tray-temperature control strategy to control the triacetin reactive distillation process for the utilization of glycerol. The dynamic results show that the proposed tray-temperature control strategy is able to maintain product purity despite disturbances from throughput and feed composition changes. Since temperature control structure has good dynamic performance of reactive distillation, heterogeneous azeotropic distillation and entrainer-assisted reactive distillation. Thus the temperature control structure can be used to the control the E-RD process for the production of Propro.

Though many researches illustrate the advantages of the E-RD process, the E-RD process hasn’t been studied for Propro production, so far. The aim of the research is to investigate the synthesis of Propro by the esterification of POH and ProAc in the E-RD process. A proper entrainer is selected for the E-RD process. And the E-RD process is optimized through calculating the minimum total annual cost (TAC). Moreover, two control structures are proposed to evaluate the stability and controllability of the E-RD process.

2. Kinetics and thermodynamics Propro is synthetized by the reversible liquid-phase esterification reaction of ProAc and POH, the reaction equation is shown as: O

O O

Cat.

HO

+

OH

+

O

H

H

C3H8O

C3H6O2

C6H12O2

H2O

propanol

propionic acid

Propyl propionate

water

97.2 ℃

141.17℃

122.5℃

100℃

(Eqa.1) The esterification reaction is a reversible reaction. It requires to be catalyzed by acidic cation exchange resin (Amberlyst 46TM). Amberlyst 46TM has the maximum operating temperature of 120 ℃ (Ilgen, 2014). We assume the catalyst occupies 50% of the tray holdup volume and the density of the catalyst is 770 kg/m3 (Huang et al., 2004). The kinetic equation provided by Duarte et al. (2006) is used: 𝑟𝑖 = 𝑚𝑐𝑎𝑡,𝑑𝑟𝑦 𝑣𝑖 𝑐𝑎𝑐𝑡 (𝑘1 (𝑇)𝑎𝑃𝑟𝑜𝐴𝑐 𝑎𝑃𝑂𝐻 −

𝑘1 (𝑇) 𝐾𝑒𝑞 (𝑇)

−5.9630 × 104 𝑘1 (𝑇) = 7.381 × 107 𝑒𝑥𝑝 ( ) 𝑅𝑇

(Eqa.3) 4.519 × 103 𝐾𝑒𝑞 (𝑇) = 6.263𝑒𝑥𝑝 ( ) 𝑅𝑇

(Eqa.4)

𝑎𝑃𝑟𝑜𝑝𝑟𝑜 𝑎𝑤𝑎𝑡𝑒𝑟 ) (Eqa.2)

The reaction rate ri has the unit of mol﹒eq-1﹒s-1 and

aProAc, aPOH, aPropro and awater

is the liquid phase activity of the ProAc, POH, Propro and water, respectively. mcat, dry is the dry catalyst weight in grams, vi is the stoichiometric coefficient of the ith component; ccat is the concentration of active sites on the dry catalyst with unit of eq ﹒kg-1, Keq(T) is the activity based equilibrium constat with unit of mol﹒eq-1﹒s-1 and k1 is rate constant of the forward reaction with unit of mol﹒eq-1﹒s-1 Where, R is the universal gas constant with units of J·mol-1·K-1, T is the temperature with unit of K. The boiling points (BP) of the components and azeotropes in this system are shown in Table 1. The data in Table 1 are extracted from Aspen Plus. The lightest component is the ternary azeotrope formed by POH, water and Propro while the heaviest component is the ProAc. The Propro esterification reactive system is a strong non-ideal system due to the existence of 4 azeotropes. Altman (2011) measured the vapor-liquid equilibrium (VLE) and liquid-liquid equilibrium (LLE) data for reactive distillation of propyl propionate. And the VLE and LLE data were fitted using the UNIQUAC model to account for non-idealities in the liquid phase while the Hayden O’Connell equation of state was used to account for non-idealities in the vapor phase. As a result, the models are in good agreement with the experimental data. Thus, the UNIQUAC model is employed for the liquid activity coefficient (Abrams and Prausnitz, 1975) in order to describe the non-ideal behavior of vapor-liquid equilibrium or vapor-liquid-liquid equilibrium. The Hayden-O’Connell model is applied to depict the non-ideal behavior of the vapor phase (Hayden and O’Connell, 1975).

3. Two-column process Xu et al. (2014) investigated a two-column process for the Propro production. Fig 1 shows the two-column flowsheet. The process consisted of a RDC coupled with a decanter and a recovery column (RC). The RDC was divided into rectifying section,

reaction zone and stripping section. ProAc and POH were fed to the top tray and bottom tray of the reaction zone, respectively. The vapor of RDC was mainly the ternary and binary azeotrope formed by POH, water and Propro. Obviously, the azeotrope carried most of POH at the top of RDC, thus, the reaction equilibrium was limited. In order to enhance the conversion of ProAc, POH was fed to the RDC in excess. The flow rate of POH and ProAc was 60 kmol/h and 50 kmol/h, respectively. Vapor stream from the top of RDC condensed in condenser and separated into organic phase and aqueous phase in a decanter. The organic phase which contains mass of unreacted POH was recycled back to the RDC and the aqueous phase which contains almost pure water was carried out as a byproduct. The bottom product of the RDC was a mixture of Propro and POH, and it was fed to the RC, Propro was obtained at the bottom of RC, POH was obtained at the top of RC and was sent to the RDC. The purity specification of water was set above 99.6 mol% and the purity specification of Propro was set above 99.7 mol%. The design parameters of the two-column process were optimized by minimum of total annual cost (TAC), Fig 1 shows the optimal design and operating conditions of the two-column process. Table 4 shows the optimal results of this process.

4. Design of entrainer-assisted reactive distillation (E-RD) process In this section, an entrainer-assisted reactive distillation (E-RD) process was designed to reduce TAC compared to the two-column process. E-RD process is a RD process by adding an external entrainer stream to the system. The entrainer can break the water-alcohol azeotrope and enhance the removal of water, thus the separation efficiency of the system can be increased, and the RC for alcohol recovery is unnecessary. 4.1 Entrainer selecting The entrainer selecting is a key step in the E-RD process. In order to select a proper entrainer, we make a comparison among toluene, cyclohexane, 1,2-dichloroethane and benzene. The physical properties of four entrainers are listed in Table 2. The data in Table 2 are extracted from Aspen Plus. As is shown in Table 2, all four candidate entrainers form a heterogeneous azeotrope with POH and H2O. The

order of the azeotropic temperature is as follows: Cyclohexane-POH-water(66.82℃) ＜Benzene-POH-water(68.24℃) ＜ 1,2-dichloroethane-POH-water(73.2℃) ＜Toluene-POH-water(79.46℃) (Eqa.5) The azeotropic temperature is critical in choosing a proper entrainer because it decides the temperature difference (△T) among the azeotrope and other components in the system. Large △T makes the separation easier while small △T makes the separation harder. Thus, toluene is inappropriate for using as the entrainer in this system due to the small temperature difference between the azeotrope and other components in this system. Fig. 2a, b, c, d show the residue curve maps (RCM) for the Benzene-POH-water, 1,2-dichloroethane-POH-water, cyclohexane-POH-water and toluene-POH-water, respectively. All RCMs display a ternary heterogeneous azeotrope and a large immiscibility region. All distillation curves start from the heterogeneous ternary azeotrope, thus the distillation vapor stream is close to the heterogeneous azeotrope formed by water, entrainer and POH. In Fig 2a, c and d, the entrainer is placed near the entrainer vertex, indicating that the entrainer phase may include little water, the direction of the tie-lines in Fig 2a, c,d demonstrates that the aqueous phase would include little alcohol and entrainer. However, the tie-lines in Fig 2b point to the water vertex, but the immiscibility region in Fig 2b is not placed near the entrainer vertex, the entrainer phase may include some water, as a result, high water concentration in the entrainer phase results in a decline ability of removing water from the reaction zone when 1,2-n is used as the entrainer. It is concluded that the 1,2 –dichloroethane is not the proper entrainer in the system. Compared with benzene, cyclohexane is more appropriate because benzene is toxic. Based on the above analysis, cyclohexane is selected as the entrainer in the system. 4.2 Process description In the two-column process, the azeotrope at the top of RDC carries most of POH out of the reaction zone, and the reaction equilibrium is limited due to the formation

of azeotrope. An excess POH is fed to RDC to convert ProAc to products. Thus a RC is needed to recover unreacted POH and purify propro. However, in the E-RD process, the ternary heterogeneous azeotrope carries mass of H2O from the reaction zone in the entrainer assisted reactive distillation column (E-RD), and the reaction equilibrium is promoted. Thus, ProAc and POH are fed to the E-RD with the stoichiometric balance. And the RC is not needed for recovering unreacted POH. Fig. 9 presents the flowsheet of the E-RD process for the synthesis of propro with stoichiometric feed ratio of POH and ProAc. The flowsheet features of a E-RD coupled with a decanter. The E-RD is divided into rectifying zone, reaction zone and stripping zone. ProAc and POH are fed to the top tray and the bottom tray of the reaction zone in E-RD, respectively. The feed flow rate for POH and ProAc are both 50 kmol/h, respectively. Cyclohexane is fed to the decanter. The top vapor stream of E-RD is close to the ternary heterogeneous azeotrope formed by cyclohexane, water and POH, the top vapor stream will condense in condenser and split into aqueous phase and organic phase in a decanter. The aqueous phase including almost pure water is removed as the distillate and the organic phase which contains rich cyclohexane is refluxed to the E-RD. The use of cyclohexane increases water removal and promotes the reaction equilibrium. Reaction rate is enhanced because the concentration reactants are kept high in the reaction zone. Therefore, high-purity Propro can be produced at the bottom of the E-RD and high-purity water can be produced of the aqueous phase. In order to make compensations for the cyclohexane loss in the aqueous phase, the flow rate of the cyclohexane makeup is designed as 0.005 kmol/h. The top pressure of the first tray in E-RD is 1 atm. The pressure drop of each tray is 0.0068atm. The purity of propro was set above 99.7 mol% and the purity of water was set above 99.6 mol%. 4.3 Optimization of design variables There are some process variables that need to be optimized, such as the reactive stage (Nrz), the rectifying stage (Nr), the stripping stage (Ns). TAC is applied as the objective function in order to calculate a minimum TAC by adjusting Nrz, Nr and Ns. The TAC is defined as:

TAC =

Total capital cost + Total operating cost Payback period

Here, the payback period is designed as 3 years. The annual operating cost (AOC) consists of the costs of cooling water, catalyst and steam. The total capital cost (TCC) including the costs of the shell, trays, reboiler, and condenser. The cost of the catalyst is 40.74$/kg and the catalyst occupies 50% of the tray holdup volume, the density of the catalyst is 770 kg/m3 and a catalyst life of 3 months is assumed (Huang et al., 2004). Luyben (2013) introduce the method for calculating TAC. Table 3 shows the detailed method. The method for calculating the total capital cost in Table 3 is suitable for carbon steel, however, carbon steel is improper for construction due to the existence of the propionic acid in the system, thus, we choose stainless steel because it is resistant to chemical corrosion (Perry and Green, 2008). Thus we choose stainless steel for construction. Silla (2003) provides the materials cost factors relative to carbon steel. A detailed iterative sequential procedure for optimization was applied to find the optimal process variables, the design variables are studied to get the minimization of TAC, the optimization diagram is shown in Fig 3. Fig 4 shows how TAC as well as energy consumption changes with the number of rectifying section (Nr). With fewer than 8, TAC shows a decreasing trend, this is because increasing Nr would reduce the reflux, and the TOC can be greatly reduced. However, with larger than 8, TAC shows an increasing trend, this is because more stages means more capital cost. Therefore, there is a minimum TAC. when Nr equals to 8. It is observed that the energy consumption shows an increasing trend. Fig 5 shows how TAC as well as energy consumption change as number of stripping section (Ns) is increased. Increasing Ns from 8 to 13 causes a decrease in energy consumption. TAC shows a decreasing trend with fewer than 11, this is because increasing Ns would reduce the vapor boilup, and the TOC can be greatly reduced. However, with larger than 11, TAC shows an increasing trend, this is because more stages means more capital cost. Therefore, there is a minimum TAC when Ns equals to 11.

Fig 6 demonstrates the effect of changing the number of reactive trays (Nrz) with TAC as well as energy consumption. It can be observed in Fig.6 that the energy consumption have a decreasing trend when the Nrz increases. TAC shows a decreasing trend when Nrz changes from 21 to 24, this is because increasing Nrz would increase reactive hold up, which would make the reboiler duty lower, And TOC can be greatly reduced. However, TAC exhibits an increasing tread when Nrz changes from 24 to 25, this is because more stages will increase the total capital cost. Therefore, there is a minimum TAC when Nrz equal to 24. By TAC analysis, the optimal parameters are obtained. The E-RDC has a total of eight rectifying stages, twenty-four reactive stages and eleven stripping stages. Fig 7 (a) shows the temperature profile of the E-RD. we mentioned above that the maximum operation temperature of Amberlyst 46TM is 120℃，it can be observed that the temperatures in the reactive section are over the limit of the catalyst temperature. The limitation can be overcome by changing the position of the reactive section (Tung and Yu, 2007). Fig 7(b) shows the temperature profile when Nrz equals to 22, it is observed that the temperatures in the reactive section are lower the limit of the catalyst temperature. As a result, the optimal parameters with smaller TAC under the maximum operating temperature of the catalyst are obtained. The E-RDC has a total of eight rectifying stages, twenty-two reactive stages and eleven stripping stages. Fig 8 (a,b) show the vapor composition profile and liquid composition profile of the E-RD process with smaller TAC under the maximum operating temperature of the catalyst. The temperature and composition profiles exhibit a sudden change around the feed location, as is shown from the vapor composition profile, water is entrained by cyclohexane towards the column top; the water concentration in the column top is larger than 0.5. Most of ProAc reacts with POH in the reaction zone. It is clear that high purity Propro can be obtained at the bottom of the E-RDC. Fig 9 shows the optimal operating condition of the E-RD process.

5. Comparisons of the two-column process and E-RDC process Table 4 summarizes the optimal design and operating parameters of the

two-column process and the E-RD process. The two-column process has two columns with a total of 79 trays while the E-RD process has only one column with a total of 41 trays. E-RD process has fewer reactive stages than that of the two-column process. It can be observed that the E-RD process can reduce the total reboiler duty greatly than that of the two-column process. The E-RD process can reduce the total heat transfer area of condensers and reboilers compared with the two-column process. The E-RD process can reduce the catalyst cost compared with the two-column process. Thus, the total capital cost (TCC) of the E-RD process can be greatly reduced. The E-RD process can save 51.45% of TCC and 41.86% of TOC than that of the two-column process. Additionally, the E-RD process can save 46.11% of TAC compared with the two-column process. Consequently, the E-RD process can not only reduce the TCC but also reduce the TAC. Moreover, it also reduces the reboiler input, thus, the E-RD process is a more economic process for the production of Propro.

6. Control structure In this part, the control structure of the E-RD process is developed. Aspen Dynamics can be utilized for the purpose of control study. The tray sizing tool can be applied to calculate the diameter of the E-RD. For the reboiler of the E-RD, it is sized to have the holdup time of 10 min with 50% liquid level. For the decanter, it is sized to have 20 min hold time. The main control issue is to maintain the propro purity at 99.7% and maintain the water purity at 99.6%. The disturbances considered include the feed flow rate changes and the feed temperature changes in the feed streams. The tray temperature control structure will be considered. 6.1 Selecting of temperature control trays In the temperature control structure, the most important issue is to select the best tray which is the most sensitive. The methods have been illustrates in many literatures so far. Luyben et al. ( 2010) introduces the sensitivity criterion to find the tray where there is the largest changes in tray temperature for a very small variety in the manipulated variable. Based on the sensitivity criterion, we have tested for the

sensitivity criterion not only by changing of the reboiler duty but also ratio between refluxed flow rate (R) and the distillated flow rate (D). As the reboiler duty and R/D change by ±0.1%, two substantial changes are observed. As is shown in Fig 10, it is demonstrated that the 34th tray is the most “sensitive” because it has the largest temperature deviation. Therefore, the 34th tray is selected as the controlled tray in the E-RD process. 6.2 Control Structure for the E-RD process 6.2.1 Basic control structure 1 (CS1) Fig 11 predicts the control structure (CS1) as well as the control panel of the E-RD process. The detailed control structure can be summarized as follows: (1) FProAc is flow controlled (reverse acting). (2) FPOH is flow controlled (reverse acting). The controller of FPOH is in ratio with the signal of FProAc (FPOH= FProAc×1), so that the molar flow ratio of the FProAc and FPOH can be kept constant. (3) FC6H12 is flow controlled (reverse acting). A ratio scheme is inserted so that the controller of FC6H12 is in cascade with the signal of FProAc. (4) The bottom level of E-RD is controlled by adjusting the bottom product flow rate (direct acting). (5) The level of the aqueous phase from decanter is controlled by adjusting the aqueous products flow rate, the level of the organic phase from decanter is controlled by adjusting the organic phase flow rate.(direct acting) (6) The top pressure of the E-RD is controlled by adjusting the top vapor stream flow rate. (reverse acting) (7) The temperature of the decanter is controlled by adjusting the heat duty of the heat exchanger (reverse acting). (8) The 34th tray temperature of the E-RD is controlled by adjusting the reboiler duty (reverse acting). (9) For the temperature control loops, a 1 min dead time is inserted to fit for the practical operation.

Luyben et al. (2008) introduced the method for developing control structures using Aspen Dynamics. In this control structure, P-only controllers with gain (Kc) =2 are used for all the level controllers in the E-RD process. PI controllers with Kc=20 and integral time (𝜏) =12min is used for the pressure controller of the E-RD process. PI controllers with Kc= 0.5 and 𝜏 =0.3 min are applied for all the flow controllers in the E-RD process. For the temperature control loop in the E-RD process, the Tyreus-Luben tuning method can be used to obtain the ultimate gain (Ku) and ultimate periods (Pu) .Table 5 shows the detailed parameters of the temperature controllers. On the basis of the control structure, in order to assess the dynamic capability of the temperature control structure, disturbances in the feed flow rate of ProAc (FProAc) and feed temperature of ProAc (TProAc) were introduced. Fig 12 shows the dynamic responses (CS1) to ±20% disturbances in the feed flow rate of ProAc after 0.5 h. As can be seen from the Fig12, the flow rate of ProAc (FProAc) as well as the flow rate of C6H12 (FC6H12) is able to change to a new level very fast in terms of ±20% disturbances in the feed flow rate of ProAc. The flow rate of Propro (FPropro) and the flow rate of water (Fwater) changes correspondingly as well within 3.5 h. XPropro exhibited large transient deviation. For a +20% disturbance in the FProAc, the transient deviation of XPropro approaches 0.989, and XPropro can return to a new steady level within 5 h. For a -20% disturbance in the FProAc, XPropro exhibited large transient deviation, and XPropro approaches 0.9975 at a new steady level within 5 h. The overall transient deviation of XPropro is about 0.011. For a +20% disturbance in the FProAc, the transient deviation of Xwater approaches 0.9925, and Xwater approaches

0.9960 at a new steady level within 5 h. For a -20% disturbance in the FProAc, Xwater exhibited very small transient deviation, and Xwater approaches a new steady level within 5 h. the overall transient deviation of Xwater is about 0.0037. The feed temperature of ProAc (TProAc) is kept constant while the temperature of the 34th stage (T34) of the E-RD process returned to steady state at about 4 h. For a +20% disturbance in the FProAc, the transient deviation of QReb approaches 1388.89 kW and QReb is able to reach to a new level within 4.5 h. For a -20% disturbance in the FProAc, the transient deviation of QReb approaches 916.61kW and QReb is able to reach to a new level within 4.5 h. The dynamic responses of CS1 to ±10℃ disturbances in the ProAc feed temperature at 0.5 h can be seen in Figure 13. FProAc and FC6H12 can return to their designed value very fast. Both of FPropro and Fwater return to their designed value within 8 h. For a +10℃ disturbances in the feed temperature of ProAc, the transient deviation of XPropro approaches 0.9967, and XPropro returned to 0.9975 within 4h. For a -10℃ disturbances in the feed temperature of ProAc, XPropro changed from 0.9975 to 0.9978, and returned to 0.9975 within 4h. The overall transient deviation of XPropro is about 0.0011. For a +10℃ disturbances in the feed temperature of ProAc, Xwater has a slight drop, the transient deviation of Xwater is less than 0.0002, and Xwater is able to reach a new steady state within 3 h. For a -10℃ disturbances in the feed temperature of ProAc, Xwater changed from 0.996 to 0.99612, and Xwater reached a steady state within 3 h. The overall transient deviation of Xwater is about 0.0003. T34 is able to return to the stable state which is close to its set points. For a +10℃ disturbances in the feed temperature of ProAc, the transient deviation of QReb approaches 1159.75 kW and QReb is able to reach to a new level within 5 h. For a -10℃ disturbances in the feed temperature of ProAc, the transient deviation of QReb approaches 1141.66kW and QReb is able to reach to a new level within 4 h. Compared with the dynamic performance of the two-column process proposed by Xu et al. (2014) in terms of ±10℃ disturbances in the ProAc feed temperature, CS1 can control Xwater and XPropro close to their designed values, but CS1 performs large transient deviation and needs longer settling time.

6.2.1 Dual temperature control structure (CS2) To reduce the large transient deviations and the settling time, CS1 must be improved. We developed the dual temperature control structure (CS2) of the E-RD process. Fig 14 shows the control structure (CS2) and the control panel. As is shown in Fig 14, the temperature on tray 6 is controlled by manipulating the ratio of FPOH to FProAc. Fig 15 and Fig 16 show the dynamic responses to disturbances in the feed flow rate of ProAc (FProAc) and the feed temperature of ProAc (TProAc). Fig 15 shows the dynamic responses (CS2) to ±20% disturbances in the feed flow rate of ProAc, FProAc and FC6H12 is able to reach a new level very fast in terms of ±20% disturbances in the FProAc. FPropro and the Fwater change correspondingly as well within 2 h. For a +20% disturbance in the FProAc, the transient deviation of XPropro approaches 0.9918, and XPropro returned to a new steady level within 5 h. For a -20% disturbance in the FProAc, XPropro exhibited small transient deviation, and XPropro approaches 0.998 at a new steady level within 5 h. The overall transient deviation of XPropro is about 0.0062. For a +20% disturbance in the FProAc, the transient deviation of Xwater approaches 0.99526, and Xwater returned to a new steady level within 5 h. For a -20% disturbance in the FProAc, the transient deviation of Xwater approaches 0.99624, and Xwater approaches a new steady level within 5 h. the overall transient deviation of Xwater is less than 0.001. T34 of the E-RD process returns to a steady state at about 3 h. For a +20% disturbance in the FProAc, the transient deviation of QReb approaches 1388.73 kW and QReb is able to reach to a new level within 4 h. For a -20% disturbance in the FProAc, the transient deviation of QReb approaches 917.23kW and QReb is able to reach to a new level at about 3.5 h. Compared with the dynamic performance of CS1 in terms of the ±20% disturbance in the FProAc, CS2 is able to reduce the overall transient deviation of Xwater and XPropro compared with CS1. Thus, CS2 is more effective than CS1 in terms of the ±20% disturbance in the FProAc. The dynamic responses of CS2 to ±10℃ disturbances in the ProAc feed temperature at 0.5 h can be seen in Figure 16. FProAc is able to return to its designed value very fast while FC6H12 is able to return to its designed value within 3 h. Both of

FPropro and Fwater return to their set value within 5 h. For a +10℃ disturbances in the feed temperature of ProAc, the transient deviation of XPropro approaches 0.99718, and XPropro returned to 0.9975 within 3h. For a -10℃ disturbances in the feed temperature of ProAc, XPropro changed from 0.9975 to 0.9977, and returned to 0.9975 within 4h. The overall transient deviation of XPropro is about 0.00052. For a +10℃ disturbances in the feed temperature of ProAc, Xwater has a slight drop, the transient deviation of Xwater is about 0.99593, and Xwater returned to a new state within 3 h. For a -10℃ disturbances in the feed temperature of ProAc, Xwater changed from 0.996 to 0.99606, and Xwater reached a steady state within 3 h. The overall transient deviation of Xwater is about 0.00013. T34 is able to return to the stable state which is close to its set points. For a ＋10℃ disturbances in the feed temperature of ProAc, the transient deviation of QReb approaches 1159.75 kW and QReb is able to reach to a new level within 5 h. For a -10℃ disturbances in the feed temperature of ProAc, the transient deviation of QReb approaches 1141.64kW and QReb is able to reach to a new level within 4 h. It is concluded that the control structure can deal with the disturbances in the ProAc feed temperature. Compared with the dynamic performance of CS1 in terms of the ±10℃ disturbances in the feed temperature of ProAc, but CS2 is able to reduce the overall transient deviation of Xwater and XPropro compared with CS1. Thus, CS2 is more effective than CS1 in terms of the ±10℃ disturbances in the feed temperature of ProAc.

7. Conclusion In this paper, an entrainer-assisted reactive distillation (E-RD) process is investigated in order to produce high-purity Propro from the esterification of N-Propanol and propionic acid. Cyclohexane is selected as the entrainer in the E-RD process. The optimal design and operating parameters of the E-RD process is conducted by calculating the minimum of total annual cost (TAC). Compared with the two-column process proposed by Xu et al, the E-RD process can save 46.11% of TAC.

Furthermore, two control structures of the E-RD process are proposed for comparison. The dynamic results revealed that CS2 is useful for reducing transient deviations and the settling time, returning the purity to its designed value. Thus, CS2 has the better control performance. Acknowledgment Comments and suggestions from two anonymous reviewers are gratefully acknowledged. We are thankful for the assistance from the staff at the Jiangsu Key Laboratory of Advanced Catalytic Materials and Technology from the School of Petrochemical Engineering (Changzhou University).

References Abrams, D.S., Prausnitz, J.M., 1975. Statistical thermodynamics of liquid mixtures: A new expression for the excess Gibbs energy of partly or completely miscible systems. AIChE J. 21, 116–128. doi:10.1002/aic.690210115 Al-Arfaj, M.A., Luyben, W.L., 2004. Plantwide control for TAME production using reactive distillation. AIChE J. 50, 1462–1473. doi:10.1002/aic.10138 Altman, E., Stefanidis, G.D., van Gerven, T., Stankiewicz, A., 2012. Microwave-Promoted Synthesis of n -Propyl Propionate using Homogeneous Zinc Triflate Catalyst. Ind. Eng. Chem. Res. 51, 1612–1619. doi:10.1021/ie200687m Altman, E., Stefanidis, G.D., van Gerven, T., Stankiewicz, A.I., 2011. Phase Equilibria for Reactive Distillation of Propyl Propanoate. Pure Component Property Data, Vapor−Liquid Equilibria, and Liquid−Liquid Equilibria. J. Chem. Eng. Data 56, 2322–2328. doi:10.1021/je101302p Buchaly, C., Kreis, P., Górak, A., 2012. n -Propyl Propionate Synthesis via Catalytic Distillation - Experimental Investigation in Pilot-Scale. Ind. Eng. Chem. Res. 51, 891–899. doi:10.1021/ie102225t Chen, C.-L., Chung, Y.-H., Lee, H.-Y., 2016. Design and Control of Reactive Distillation Process for the Production of Methyl Valerate. Ind. Eng. Chem. Res. 55, 1347–1360. doi:10.1021/acs.iecr.5b03495 Cruz-Díaz, M., Buchaly, C., Kreis, P., Pérez-Cisneros, E.S., Lobo-Oehmichen, R., Górak, A., 2012. Synthesis of n-propyl propionate in a pilot-plant reactive distillation column: Experimental study and simulation. Comput. Chem. Eng. 39, 118–128. doi:10.1016/j.compchemeng.2012.01.004 De Jong, M.C., Dimian, A.C., de Haan, A.B., 2008. Entrainer-based reactive distillation versus conventional reactive distillation for the synthesis of fatty acid esters. Comput. Aided Chem. Eng. 25, 199–204. doi:10.1016/S1570-7946(08)80038-7 Dimian, A.C., Omota, F., Bliek, A., 2002. Entrainer-Assisted Reactive Distillation.

Comput. Aided Chem. Eng. 10, 175–180. doi:10.1016/S1570-7946(02)80057-8 Domingues, L., Pinheiro, C.I.C., Oliveira, N.M.C., 2014. Optimal design of reactive distillation systems: Application to the production of ethyl tert-butyl ether (ETBE). Comput. Chem. Eng. 64, 81–94. doi:10.1016/j.compchemeng.2014.01.014 Duarte, C., Buchaly, C., Kreis, P., 2006. Esterification of propionic acid with n-propanol catalytic and non-catalytic kinetic study. Inzynieria Chem. Proces. 27, 273–286. Gooch, J.W. (Ed.), 2007. Propyl propionate, in: Encyclopedic Dictionary of Polymers. Springer New York, New York, NY, pp. 797–797. doi:10.1007/978-0-387-30160-0_9366 Hayden, J.G., O’Connell, J.P., 1975. A Generalized Method for Predicting Second Virial Coefficients. Ind. Eng. Chem. Process Des. Dev. 14, 209–216. doi:10.1021/i260055a003 Hu, S., Zhang, B., Hou, X., Li, D., Chen, Q., 2011. Design and simulation of an entrainer-enhanced ethyl acetate reactive distillation process. Chem. Eng. Process. Process Intensif. 50, 1252–1265. doi:10.1016/j.cep.2011.07.012 Huang, K., Wang, S.-J., Ding, W., 2008. Towards further internal heat integration in design of reactive distillation columns—Part III: Application to a MTBE reactive distillation column. Chem. Eng. Sci. 63, 2119–2134. doi:10.1016/j.ces.2008.01.006 Huang, S.-G., Kuo, C.-L., Hung, S.-B., Chen, Y.-W., Yu, C.-C., 2004. Temperature control of heterogeneous reactive distillation. AIChE J. 50, 2203–2216. doi:10.1002/aic.10247 Hung, S.-K., Lee, C.-C., Lee, H.-Y., Lee, C.-L., Chien, I.-L., 2014. Improved Design and Control of Triacetin Reactive Distillation Process for the Utilization of Glycerol. Ind. Eng. Chem. Res. 53, 11989–12002. doi:10.1021/ie500346w Ilgen, O., 2014. Investigation of reaction parameters, kinetics and mechanism of oleic acid esterification with methanol by using Amberlyst 46 as a catalyst. Fuel

Process. Technol. 124, 134–139. doi:10.1016/j.fuproc.2014.02.023 Keller, T., Muendges, J., Jantharasuk, A., Gónzalez-Rugerio, C.A., Moritz, H., Kreis, P., Górak, A., 2011. Experimental model validation for n-propyl propionate synthesis in a reactive distillation column coupled with a liquid–liquid phase separator. Chem. Eng. Sci. 66, 4889–4900. doi:10.1016/j.ces.2011.06.056 Kiss, A.A., 2013. Advanced distillation technologies: design, control, and applications. Wiley, a John Wiley & Sons, Ltd., Publication, Chichester, West Sussex. Kotora, M., Buchaly, C., Kreis, P., Górak, A., Markoš, J., 2008. Reactive distillation — experimental data for propyl propionate synthesis. Chem. Pap. 62. doi:10.2478/s11696-007-0080-x Luyben, W.L., 2013. Distillation design and control using Aspen simulation, 2nd ed. ed. John Wiley & Sons, Hoboken, N.J. Luyben, W.L., Chien, I.-L., 2010. Design and control of distillation systems for separating azeotropes. John Wiley, Hoboken, N.J. Luyben, W.L., Yu, C.-C., 2008. Reactive distillation design and control. Wiley : AIChE, Hoboken, N.J. Perry, R.H., Green, D.W. (Eds.), 2008. Perry’s chemical engineers’ handbook, 8th ed. ed. McGraw-Hill, New York. Silla, H., 2003. Chemical process engineering: design and economics. M. Dekker, New York. Tung, S.-T., Yu, C.-C., 2007. Effects of relative volatility ranking to the design of reactive distillation. AIChE J. 53, 1278–1297. doi:10.1002/aic.11168 Wang, S.-J., Huang, H.-P., 2011. Design of entrainer-enhanced reactive distillation for the synthesis of butyl cellosolve acetate. Chem. Eng. Process. Process Intensif. 50, 709–717. doi:10.1016/j.cep.2011.04.007 Wang, S.-J., Wong, D.S.H., 2006. Design and Control of Entrainer-Added Reactive Distillation for Fatty Ester Production. Ind. Eng. Chem. Res. 45, 9042–9049. doi:10.1021/ie060626j Xu, H., Ye, Q., Zhang, H., Qin, J., Li, N., 2014. Design and control of reactive distillation–recovery distillation flowsheet with a decanter for synthesis of

N-propyl propionate. Chem. Eng. Process. Process Intensif. 85, 38–47. doi:10.1016/j.cep.2014.08.001

POH=0.9999 mf ProAc=1.871e-21 mf H2O=9.898e-06 mf Propro=9.010e-05 mf RF= 10.13kmol/hr

25℃

Qc=-0.1502 GJ/hr

RR2=7.134 FProAc:50kmol/h 25℃ 1.2atm D=1.036m QR=5.226 GJ/hr Qc=-2.490 GJ/hr Nrz=33 FPOH:60kmol/h 25℃ 1.2atm

Nr=6 7

RDC

67℃ 1atm POH=0.024mf ProAc=0.009mf H2O=0.017mf Propro=0.95mf RR=59.096 kmol/hr

66.7℃ 1atm POH=0.0015 mf ProAc=0.0017 mf H2O=0.9960 mf Propro=0.0008 mf D1=49.99 kmol/hr

97.2℃ 1atm

20 39 Ns=5

122.5℃ 1.14atm POH=0.1688 mf ProAc=0.002 mf Steam H2O=1.671e-06 mf Propro=0.8292mf B1=60.00 kmol/hr

D=0.686m QR=1.840 GJ/hr Qc=1.819 GJ/hr NT=35

Steam

130.2℃ 1atm POH=2.031e-5 mf ProAc=0.0025 mf H2O=1.096e-22 mf Propro=0.9975 mf B2=49.87 kmo/h

Fig 1 Two-column process for synthesis of Propro

a

b

c

d

Fig.2 Residue curve maps of (a) Benzene-POH-water; (b) 1,2-dichloroethane-POH-water; (c) cyclohexane-POH-water; (d) Toluene-POH-water

Set FPOH/FProAc=1

Fix the tray number in the rectifying section (Nr)

No TAC is minimized with Nr Select the tray number in the stripping section (Ns)

Yes Select the tray number in the reactive section (Nrz) Calculate the TAC of the E-RD

Place NFProAc and NFPOH on the top and the lowest tray of the reaction zone Get optimal parameters: Nr, Ns and Nrz

Change the reboiler duty (Qb) until the product specification is met. End

No TAC is minimized with Nrz

Yes TAC is minimized with Ns

No

Yes

Fig.3 Sequential iterative optimization procedure in ERDC

5.01

4.68 TAC Energy consumption

4.66 4.64

4.99

4.62 4.98 4.60 4.97 4.58 4.96

4.56

4.95

4.54

4.94

4.52 5

6

7

8

9

Energy consumption/(GJ/h)

TAC/(105$)

5.00

10

Stage

Fig 4. Effects of the numbers of stages in the rectifying section on the TAC and energy consumption TAC Energy consumption

5.15

4.9

5.10

TAC/(105$)

5.0

4.8

5.05

4.7

5.00

4.6

4.95

4.5

4.90

Energy consumption/(GJ/h)

5.20

4.4

4.85 8

9

10

11

12

13

Stage

Fig 5. Effects of the numbers of stages in the stripping section on the TAC and energy consumption 4.838

4.25

4.834

TAC/(105$)

4.30

4.832 4.20 4.830 4.828

4.15

Energy consumption/(GJ/h)

TAC Energy consumption

4.836

4.826 4.10

4.824 21

22

23

24

25

Stage

Fig 6. Effects of the numbers of stages in the reaction section on the TAC and energy consumption

(b)

(a)

FProAc

140

FPOH

FPOH Temperature

Temperature

135

130

130 125

125

120

120

Temperature (℃)

Temperature profile

FProAc

135

115 110 105 100 95

115 110 105

90

100

85

95

80

90

75

85

70 0

5

10

15

20

25

30

35

40

45

0

5

10

15

20

Stage

25

30

35

40

Stage

Fig.7 Temperature profile（(a) Temperature profile when Nrz=24; (b) Temperature profile when Nrz=22） (a)

(b) FPOH

FProAc

1.0

Vapor Composition Profiles

0.8 0.7

0.9

Liquid Composition Profiles

0.9

FPOH

FProAc

1.0

POH PROAC H2O PROPRO C6H12-01

0.6 0.5 0.4 0.3 0.2

0.8 0.7

POH PROAC H2O PROPRO C6H12-01

0.6 0.5 0.4 0.3 0.2 0.1

0.1

0.0

0.0 0

5

10

15

20

Stage

25

30

35

40

0

5

10

15

20

25

30

35

40

45

Stage

Fig.8 Composition profile（(a) Vapor composition profile; (b) Liquid composition profile）

89.677kmol/h POH:0.003mf ProAc:0.002 mf Water:0.547 mf Propro: 0.259 mf C6H12:0.19 mf FProAc=50 kmol/h 25℃ 1.2atm

D=1.03m QR=4.14GJ/hr Qc=-0.81GJ/hr Nrz=22

Nr=8 Aqueous phase 50kmol/h 56℃ 1 atm POH=0.0015 mf ProAc=0.0017 mf Water=0.9960 mf Propro=0.0008 mf C6H12=10ppm

Organic phase 39.682kmol/h 56.5℃ 1atm POH:0.006 mf ProAc:0.003 mf Water: 0.004 mf Propro:0.569 mf C6H12:0.418 mf

9

E-RD

FPOH=50 kmol/h 25℃ 1.2atm

C6H12:0.005kmol/h 25℃ 1atm

87.1℃ 1atm

30

Ns=11 Steam

132.4℃ 1 atm 50kmol/h POH=2.031e-5 mf ProAc=0.0025 mf Water=1.096e-22 mf Propro=0.9975 mf C6H12=0

Fig.9 Optimal operating condition of the E-RD process 0.08

+0.1% reboiler duty -0.1% reboiler duty

0.08

Temperature difference/℃

Temperature difference/℃

+0.1% R/D -0.1% R/D

0.06

0.06 0.04 0.02 0.00 -0.02 -0.04

0.04 0.02 0.00 -0.02 -0.04 -0.06

-0.06

-0.08

-0.08 0

5

10

15

20

25

30

35

40

45

0

5

10

15

Stage

20

25

30

35

40

45

Stage

Fig.10 Open-loop dynamic responses for a 0.1% step change of the reboiler duty and R/D (black line,+0.1% change; red line,-0.1% change)

Ratio FC

TC

PC

C6H12 Makeup

Condenser FC

Ns=8

ProAc Feed

LC

Decanter LC

Nrz=24

Ratio

FC

POH Feed TC

Nr=11

Reboiler LC

Fig.11 Basic control structure 64

+20% feed flow -20% feed flow

57 54 51 48 45 42

+20% feed flow -20% feed flow

0.0056

0.0052

0.0048

0.0044

1

2

3

4

5

6

7

8

9

10

0

Time/hours

64

1

2

3

4

5

6

7

8

9

56

52

48

44

5

0.9960 0.9945 0.9930 0.9915

6

7

8

9

10

2

3

4

2

3

4

Temperature of the stage 34/℃

26 25 24 23

7

8

9

10

+20% feed flow -20% feed flow

0.9955 0.9950 0.9945 0.9940 0.9935 0.9930

5

6

7

8

9

0

10

1

2

3

+20% feed flow -20% feed flow

121.5

4

5

6

7

8

9

10

Time/hours

Time/hours

122.0

27

6

0.9920 1

122.5

+20% feed flow -20% feed flow

28

5

0.9925

0

Time/hours

Feed temperature of ProAc/℃

1

Time/hours

+20% feed flow -20% feed flow

1400

121.0

1300

120.5

Reboiler duty/kW

4

0

0.9960

0.9885 3

44

0.9965

40 2

48

0.9975

0.9900

1

52

10

+20% feed flow -20% feed flow

0.9990

Mole fraction of propro

60

0

56

Time/hours

+20% feed flow -20% feed flow

Mole fraction of water

0

+20% feed flow -20% feed flow

60

40

0.0040

39

Mole flow of water/kmol h-1

Mole flow rate of propro/kmol h-1

0.0060

Mole flow rate of C6H12/kmol h-1

Mole flow rate of ProAc/kmol h-1

60

120.0 119.5 119.0 118.5 118.0 117.5

1200

1100

1000

117.0 22

116.5 0

1

2

3

4

5

Time/hours

6

7

8

9

10

0

1

2

3

4

5

Time/hours

6

7

8

9

10

900 0

1

2

3

4

5

6

7

8

9

Time/hours

Fig.12 Dynamic responses (CS1) to ±20% disturbances in the feed flow rate of ProAc

10

Mole flow rate of C6H12/kmol h-1

Mole flow rate of ProAc/kmol h-1

50.07 50.04 50.01 49.98 49.95 49.92 49.89

50.20

+10℃feed temperature -10℃feed temperature

0.005010

Mole flow rate of propro/kmol h-1

+10℃feed temperature -10℃feed temperature

50.10

0.005007 0.005004 0.005001 0.004998 0.004995 0.004992 0.004989

0

1

2

3

4

5

6

7

8

9

10

50.10 50.05 50.00 49.95 49.90 49.85

0

1

2

3

4

Time/hours

5

6

7

8

9

10

0

Time/hours

1

2

3

4

5

6

7

8

9

10

Time/hours

50.6

0.99615

-10℃ feed temperature +10℃ feed temperature

+10℃ feed temperature -10℃ feed temperature

0.9978

50.4

50.0

49.8

49.6

Mole fraction of water

50.2

+10℃ feed temperature -10℃ feed temperature

0.99610

0.9976

Mole fraction of propro

Mole flow rate of water/kmol h-1

-10℃ feed temperature +10℃ feed temperature

50.15

0.9974

0.9972

0.9970

0.99605 0.99600 0.99595 0.99590 0.99585

0.9968 0.99580

49.4

0.9966 0

1

2

3

4

5

6

7

8

9

0

10

1

2

3

4

5

6

7

8

9

10

0

1

2

3

4

5

6

7

8

9

10

Time/hours

Time/hours

Time/hours 119.85

25

20

119.75

1156

119.70

1154

119.65 119.60 119.55

1152 1150 1148 1146 1144

119.50 15

+10℃feed temperature -10℃feed temperature

1158

Reboiler duty/kW

+10℃ feed temperature -10℃ feed temperature

30

1160

-10℃ feed temperature +10℃ feed temperature

119.80

Temperature of the 34th stage/℃

Feed temperature of ProAc/℃

35

1142

119.45

1140 0

1

2

3

4

5

6

7

8

9

0

10

1

2

3

4

5

6

7

8

9

10

0

1

2

3

Time/hours

Time/hours

4

5

6

7

Time/hours

Fig 13 Dynamic responses (CS1) to ±10℃ disturbances in the feed temperature of ProAc Ratio

TC

FC

PC

Condenser

C6H12 Makeup

FC

Ns=8

ProAc Feed

LC

Decanter LC FC

Nrz=22

Ratio

POH Feed TC6

Nr=11

TC34

Reboiler LC

Fig.14 Dual temperature control structure (CS2)

8

9

10

65

+20% feed flow - 20% feed flow

0.0060

+20% feed flow -20% feed flow

56

52

48

44

40

60

+20% feed flow -20% feed flow

0.0056

0.0052

0.0048

0.0044

0

50

45

40

1

2

3

4

5

6

7

8

9

10

0

1

2

3

4

Time/hours

5

6

7

8

9

35

10

0

1

2

3

4

Time/hours 0.999

+20% feed flow -20% feed flow

5

6

7

8

9

10

Time/ hour +20% feed flow rate -20% feed flow rate

0.998

60

0.99624

+20% feed flow rate -20% feed flow rate

0.99610

-1

Mole fraction of propro

0.997 55

50

45

40

Mole fraction of water

Flow rate of water/kmol h

55

0.0040

65

0.996 0.995 0.994 0.993

1

2

3

4

5

6

7

8

9

1

2

3

4

+20% feed flow -20% feed flow

5

6

25 24 23 22

8

9

10

0

1

2

3

4

5

6

7

8

9

10

+20% feed flow -20% feed flow

1400

120.2

+20% feed flow -20% feed flow

1300 120.0

Reboiler duty/kW

Temperature of the 34th tray/℃

26

7

Time/hours

120.4

27

0.99554

Time/hours

Time/ hour 28

0.99568

0.99512 0

10

0.99582

0.99526

0.991 0

0.99596

0.99540

0.992

35

Feed temperature of ProAc/℃

Feed flow of propro/kmol h -1

Mole flow rate of C6H12/kmol h-1

Mole flow rate of ProAc/kmol h-1

60

119.8 119.6 119.4

1200

1100

1000

119.2 119.0

900 0

1

2

3

4

5

6

7

8

9

0

10

1

2

3

4

5

6

7

8

9

0

10

1

2

3

4

5

6

7

8

9

10

Time/hours

Time/hours

Time/hours

Fig.15 Dynamic responses (CS2) to ±20% disturbances in the feed flow rate of ProAc 0.005012

50.07 50.04 50.01 49.98 49.95 49.92

+10℃ feed temperature -10℃ feed temperature

0.005008

Mole flow rate of Propro/kmol h-1

Mole flow rate of C6H12/kmol h-1

Mole flow rate of ProAc/kmol h-1

50.10

+10℃ feed temperature -10℃ feed temperature

50.10

0.005004

0.005000

0.004996

0.004992

+10℃ feed temperature -10℃ feed temperature

50.07 50.04 50.01 49.98 49.95 49.92

49.89 0.004988 0

1

2

3

4

5

6

7

8

9

10

0

1

2

3

4

Time/hours

6

7

8

9

49.89

10

0

Time/hours

50.25

+10℃ feed temperature -10℃ feed temperature

50.20

-10℃ feed temperature +10℃ feed temperature

0.99768

1

2

3

4

5

6

7

8

9

10

Times/hours +10℃feed temperature -10℃ feed temperature

0.99606

50.15

0.99604

50.10 50.05 50.00 49.95 49.90 49.85

0.99760

Mole fraction of water

Mole fraction of propro

Mole flow rate of water/kmol h-1

5

0.99752 0.99744 0.99736

0.99602 0.99600 0.99598 0.99596

0.99728

49.80

0.99594 0.99720

49.75

0.99592 0

49.70 0

1

2

3

4

5

6

7

8

9

1

2

3

4

10

5

6

7

8

9

10

0

1

2

3

4

5

+10℃ feed temperature -10℃ feed temperature

30

25

20

1158

8

9

10

1156

119.8

Reboile duty/kW

Temperature of the 34th tray

Feed temperature of ProAc/℃

-10℃ feed temperature +10℃ feed temperature

119.9

7

+10℃ feed temperature -10℃ feed temperature

1160 35

6

Time/hours

Time/hours

Time/hours

119.7

119.6

1154 1152 1150 1148 1146

119.5 1144 1142

119.4

15

0 0

1

2

3

4

5

Time/hours

6

7

8

9

10

1

2

3

4

5

Time/hours

6

7

8

9

10

1140 0

1

2

3

4

5

6

7

Time/hours

Fig 16 Dynamic responses (CS2) to ±10℃ disturbances in the feed temperature of ProAc

8

9

10

Table 1. Boiling Pointing Ranking for Pure Components and azeotropes for this system at 1 atm Component

Azeotrope type

Boiling point (℃）

Mole basis

POH/H2O/Propro

Heterogeneous

86.84

0.2783/0.5959/0.1277

POH/H2O

Homogeneous

87.7

0.4069/0.5931

H2O/Propro

Heterogeneous

89.26

0.6713/0.3287

POH

97.2

H2O

100

ProAc/H2O

Homogeneous

100.01

Propro

122.4

ProAc

141.14

0.0078/0.9922

Table 2. boiling point and composition of azeotrope Entrainer Component ID

Ternary heterogeneous

Azeotrope

azeotrope

Binary heterogeneous azeotrope

Azeotrope

BP(℃)

Water

POH

Entrainer

BP(℃)

Water

Entrainer

BP (℃)

Toluene

110.68

0.4395

0.2496

0.3109

79.46

0.5578

0.4422

84.43

Cyclohexane

80.78

0.2569

0.1471

0.596

66.82

0.3

0.7

69.41

1,2-dichloroethane

83.67

0.3463

0.0919

0.5618

73.2

0.3632

0.6368

73.94

Benzene

80.13

0.2782

0.1086

0.6132

68.24

0.2976

0.7024

69.24

Table 3. Basis of economics Column vessel costs Capital cost=17640 D1.066 L0.802 where D is using aspen tray sizing while L=(NT-1)×0.61+6 Condensers costs heat-transfer coefficient=0.852kW/K·m2 Capital cost=7296Ac0.65, where area is in squared meters Reboilers costs heat-transfer coefficient=0.568kW/K·m2 Capital cost=7296AR0.65, where area is in squared meters Annual steam cost Steam cost=Cs×Q×8000×3600 Where Cs=$7.72 per GJ /LP steam (433K) Annual cooling water cost Cooling water cost=0.03×Qc/(ΔTW×4.183×1000)×8000×3600 Stainless steel cost factors relative to carbon steel=1.35 (Silla, 2003)

Table 4. Comparisons of the two-column process and the E-RD process Two-column process

E-RD Process

RDC

RC

E-RD

Total number of total trays (NT)

44

35

41

Reactive stages (Nrz)

33

/

22

POH/60

/

POH/50

ProAc/50

/

ProAc/50

Column diameter (m)

1.036

0.686

1.026

Total reboiler duty (GJ/h)

5.226

1.84

4.141

Heat transfer area of condensers (m2)

58.4

42.67

19.05

Capital costs of condensers (105 $)

1.385

1.13

0.668

Heat transfer area of reboilers (m2)

73.44

25.86

58.2

Capital costs of reboilers (105 $)

1.608

0.8154

1.382

Capital cost of columns (105 $)

3.946

3.009

3.722

Catalyst cost (105 $)

0.411

/

0.288

Total capital cost (105 $)

11.894

5.774(51.45%)

Total operating cost (105 $)

4.986

2.899(41.86%)

3 years of TAC (105 $)

8.951

4.824(46.11%)

Feed flow rate (kmol/h)

Table 5. Parameters of All Temperature Controllers Parameters

TC-C

TC-HX

Controlled variable

T34

TDecanter

Manipulated variable

QR

QHX

KU (%/%)

12.2144

16.4576

PU (min)/min

4.2

5.4

KC

3.817

5.143

9.24

11.88

(%/%)

τI/min