Novel Control Structure Design of Differential Pressure Thermally Coupled Reactive Distillation for Methyl Acetate Hydrolysis

Li Jun; Zhou Hao; Huang Xiaoqiao,2; Zhao Tianlong; Ma Zhanhua; Sun Lanyi

(1. State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Qingdao 266580;2. PetroChina Fuel Oil Company Limited Research Institute, Beijing 100195)

Abstract: In this paper, the novel control structures of differential pressure thermally coupled reactive distillation process for methyl acetate hydrolysis were proposed. The RadFrac module of Aspen Plus was adopted in the steady-state simulation. Sensitive analysis was applied to find the stable initial value and provide a basis for the improved control structure design. The Aspen Dynamics software was adopted to study the process dynamic behaviors, and two novel control structures provided with feed ratio controllers and sensitive tray temperature controllers were proposed. The reflux ratio controllers were applied in the improved novel control structures. Both control structures abandoned the composition controllers that were replaced by simpler controllers with which the product purity could meet the specification requiring under a ± 20%disturbance to the total feed flowrate / MeAc composition.

Key words: reactive distillation; process control; hydrolysis; thermally coupled; methyl acetate

1 Introduction

Distillation column is one of the most important operating units for separating liquid mixtures in the chemical industry. By using distillation columns, the high-purity components can be obtained. With the aim of saving energy and separating close-boiling mixtures, the differential pressure distillation (DPD) process has been developed[1-2], which can operate two or more columns at different pressure levels to separate azeotropic or close-boiling mixtures. Due to the pressure difference between the columns in the DPD process, the condenser temperature of the high-pressure column may be higher than the reboiler temperature of the low-pressure column.By coupling the two heat exchangers mentioned above,the differential pressure thermally coupled distillation(DPTCD) process was proposed[3]. The DPTCD process can obtain high-purity components with higher efficiency and less energy consumption[4]. Li, et al.[5]investigated the DPTCD processes for the separation of benzenefluorobenzene, benzene-heptane, benzene-toluene, and benzene-chlorobenzene mixtures to achieve by 90.64%,73.92%, 46.43%, and 35.69% of energy reduction,respectively, and concluded that the DPTCD processes could save more energy in a system with lower relative volatility.

Methyl acetate (MeAc) can be obtained as a byproduct from the purified terephthalic acid (PTA) plants and polyvinyl alcohol (PVA) plants. The MeAc hydrolysis process with reactive distillation (RD) columns is commonly used to produce its higher-value hydrolysis products[6-7]. The RD column can bring about some advantages on separation efficiency and reduction of facility cost[8-11]by combining the phase separation and chemical reaction in a single column, which is widely used in chemical equilibrium-limited reactions. Since the MeAc hydrolysis reaction is an equilibrium-limited reaction[12], different types of RD columns are used to hydrolyze MeAc, and one method is to use the reactive dividing wall column (RDWC)[13]. RDWC combines the reaction and separation in a dividing wall column(DWC) to achieve some advantages on total annual cost(TAC) and energy consumption cost. Several processes have been proposed to apply the RDWC to MeAc hydrolysis[14-15], but the RDWC works better in the high relative volatility system. Since the boiling points of MeAc and MeOH are very close, it is thus difficult to obtain both high-purity MeOH and HAc products via the RDWC process in the MeAc hydrolysis system[15-16]. To overcome the problems mentioned above, the differential pressure thermally coupled reactive distillation(DPTCRD) process was developed[17]to combine the DPTCD process and the RD process in a column with a reaction zone. Li, et al.[17]gave us an example of using the DPTCRD process to synthesize n-butyl acetate, which achieved a reduction of 26.02% on energy consumption.Compared with the DPTCD process or the RD process,the DPTCRD process is more suitable for the equilibriumlimited reaction in the low relative volatility system,such as the hydrolysis of MeAc. In one of our previous papers, Zhai, et al.[18]used the DPTCRD process on MeAc hydrolysis, which saved the energy consumption by 36.1% and achieved a reduction of 41.3% on TAC,as compared with the conventional reactive distillation(CRD) process. However, the product of this process was a mixture of MeOH, acetic acid (HAc) and water, which should need further treatments. In our further research,Li, et al.[19]provided a process with several distillation columns to obtain the high-purity products, which could achieve reductions of 7.49% of TAC and 40.07% of energy consumption.

Since the DPTCRD process is a combination of RD and DPTCD processes, the control structure research on RD and DPTCD processes is beneficial to the study of DPTCRD process control behaviors[8,20-21].Luyben[8,21]proposed control structures for a RD process and a DPTCD process separately, where the feed ratio controllers were applied in the DPTCD process to compensate for the feed disturbance, and the composition controllers were installed on the RD process when the column temperature controllers could not ensure the expected control behaviors. Since the DPTCRD process is a thermally coupled process, it is difficult to control the reboiler/condenser duty by temperature controllers.Li, et al.[19]proposed a control structure for a MeAc hydrolysis DPTCRD process, which contained a feed ratio controller and a composition controller on cascade.They also indicated that the control behaviors could be improved with a temperature controller installed in the low-pressure column. However, the control structures with composition controllers are not good choices in industrial production practice, since the composition controller is much more expensive and shows slower response than other controllers, like a ratio controller or a temperature controller[22], and, especially, such longer integral time of composition controller may lead to the system instability. To overcome the disadvantages of the composition controller in the DPTCRD process, novel control structures of the MeAc hydrolysis DPTCRD process without composition controllers were proposed in this paper. Based on the sensitivity analysis results, an alternative variable was controlled in place of the product composition. A ±20% disturbance to the total feed flowrate/MeAc composition was added to the system, and control behaviors of the novel structures were studied in this paper.

2 Process Description

Different types of catalyzers can be applied to different processes of MeAc hydrolysis[13-15]. In this study, the NKC-9 cation-exchange resin was used as the catalyzer.The hydrolysis reaction can be described as follows:

The kinetic model of this reaction has been proposed by Quan, et al.[23]:

where r is the reaction rate, kfand krare the rate constants of the forward and the reserve reactions, respectively,with T being the absolute temperature and R being the gas constant, and xiis the molarity of component i. The liquid hold-up is set as 0.063 m3in the reactive section.

Figure 1 DPTCRD structure for MeAc hydrolysis

A DPTCRD process structure has been proposed for MeAc hydrolysis in our previous paper[19], as shown in Figure 1, in which four columns are used in the DPTCRD process. The HP column is a high-pressure reactive distillation column with the catalyzer on its trays and the LP column is a conventional low-pressure distillation column. A mixture of MeAc and MeOH is introduced to the bottom of the HP column, while water is fed to the top of the HP column. The hydrolysis reaction of MeAc and the separation between the reactant and product take place in the HP column at the same time. Unreacted MeAc with some water and MeOH from the top of the HP column is recycled and mixed with the MeAc/MeOH feed stream. The liquid outflow from the last tray of the HP column, which contains more product than the top stream of the HP column, is transferred to the top of the LP column for further separation, while the vapor from the first tray of the LP column is fed to the bottom of the HP column via a compressor. The bottom liquid outflow of the LP column is fed to the HAc column. HAc product is obtained from the bottom of the HAc column with 99±0.5% purity in mole fraction. A mixture of water and MeOH from the top of the HAc column is fed to the MeOH column, where MeOH product with a 99±0.5%purity in mole fraction is obtained from the top, and a dilute HAc solution is obtained from the bottom as a byproduct. Since the pressure of the HP column is much higher than that of the LP column, the temperature at the top of the HP column is thus higher than the bottom temperature of the LP column. Hence, in order to reuse the energy of the top vapor stream of the HP column,this stream is routed to the reboiler of the LP column as a heat source. Besides, a trim condenser is added to the hot stream, which comes from the reboiler of the LP column to ensure total condensation.

3 Simulation and Sensitivity Analysis

In this paper, we set up the operating parameters for the DPTCRD MeAc hydrolysis process based on the previous studies[18-19,24]. To analyze the process behaviors, we used the RadFrac module in Aspen Plus, which is a rigorous distillation module and can work well for reactive distillation simulation. We have chosen the UNIQ-HOC property method built in the Aspen Plus, which uses the Hayden-O’Connell equation of state and can well describe the behavior of HAc associated compound. The results of the steady-state simulation are presented in Table 1, Figure 2 and Table 2.

Table 1 Operating parameters for all the columns

Table 1 shows the basic operating parameters for the steady-state simulation. The temperature profiles of the four columns from the steady-state simulation are shown in Figure 2, which provides the basis of sensitive tray selection.The composition profiles of the three product streams and quality targets are shown in Table 2. It can be seen that the product purity could meet the specification with the operating parameters given in Table 1. Finally, the process was imported to Aspen Plus Dynamics to continue the design of control structures.

Table 2 Product composition of the DPTCRD process

4 Design of Control Structure

Control of the DPTCRD process is more difficult than that of the conventional process[8]. In the DPTCRD process of this study, the reboiler/condenser duty is uncontrollable because the reboiler energy of the LP column should be fully provided by the top vapor stream of the HP column. The overhead vapor stream of the HP column is mixed with the feed stream, which would increase the complexity of the system. In our previous research[19], the control structures for the MeAc hydrolysis DPTCRD process were proposed, with the control behaviors studied. Although the previous control structures using composition controllers (CCs) to control the product purity can obtain good control behaviors, they can hardly be implemented in industrial production. In this study, two novel control structures were investigated.Both the novel structures had abandoned CCs, and their control behaviors were studied and compared.

The design of the novel control structures was based on the simulation results of Aspen Plus Dynamics. Since it is complex and difficult to converge by using the thermally coupled process structure in Aspen Plus directly, two equations were provided to achieve the fully thermally coupled process[8]. As the process in this paper is a partial thermally coupled process, the equations need to be modified. The modified equations can be summarized as follows:

Figure 2 Temperature profiles of the HP columns

in which QR,LPmeans the reboiler duty of the LP column,QC,HPand QA,HPare the condenser duty and the trim condenser duty of the HP column, respectively. UA[8]is the overall heat transfer coefficient which is calculated by means of the initial temperature difference and the reboiler duty of the LP column, and TR,HPand TB,LPare the top and the bottom temperatures of the HP and the LP column, respectively. Equation (4) calculates the energy demand and sets the reboiler duty of the LP column. The trim condenser duty is calculated by using the condenser and the reboiler duties through equation (5). Since the trim condenser controls the top pressure of the HP column, the trim condenser duty is set as the control output directly.Both the equations were inputted into the Aspen Plus Dynamics using the program language as shown in Figure 3,and the related fixed variables were set to “free”.

4.1 Basic control structure

In the conventional distillation column, using the tray temperature to control the reboiler/condenser duty is a viable control strategy. However, the thermally coupled structure makes reboiler or condenser duty uncontrollable.

Furthermore, the temperature change in the HP column is too small to provide reliable control behaviors in this system. Based on the reasons mentioned above, the basic control structure of MeAc hydrolysis DPTCRD process is finally proposed as shown in Figure 4.

The basic control structure shown in Figure 4 contains several simple controllers, and the flowrate controllers(FCs) are added to the water feed stream and the MeAc/MeOH feed stream to control the feed mole flowrate,while the liquid level controllers (LCs) are added to the top/bottom of the columns to control the reflux drum/sump liquid levels, and the pressure controllers (PCs)are added to the top of the HAc column and the MeOH column to control the overhead pressure.

Figure 3 Equations using program language in Aspen Plus Dynamics

Figure 4 Basic control structure for MeAc hydrolysis DPTCRD process

To improve the stability of the process control behaviors, more controllers are required for the said control structure. Since the LP column does not have a condenser, the pressure controller added to the top of the column can control the compressor brake power to act a substitute for the said condenser. Due to the particularity of thermally coupled structure, the pressure controller output signals of the HP column equations are presented in Figure 3. The set point of the water feed stream flowrate controller is controlled by the mole flowrate of MeAc/MeOH feed stream with FC1 set on cascade.The tray temperature of the HAc column is studied, and the sensitive tray temperature controllers are added to the system. The sensitive tray means the tray with the largest temperature difference among all trays, when the column operating parameter changes. A ±1% disturbance to the feed flowrate is added to the system, and the tray temperature difference obtained after 0.5 h of operation are studied to find the sensitive trays. The temperature profiles of the HAc and the MeOH columns with a ±1%disturbance to the feed flowrate added to the system are shown in Figure 5. The stage 22 with the largest temperature difference is set as the sensitive tray for the HAc column, and then a temperature controller is thereby set on the stage 22 to control the reboiler duty. Based on the same principle, the stage 19 is set as the sensitive tray of the MeOH column, and another temperature controller is provided to control the reboiler duty.

Figure 5 Temperature difference with a ±1% disturbance to feed flowrate in the HAc column and the MeOH column after 0.5 hour of operation

The controller parameters were then set as follows.

All the flowrate controllers were set with the gain KC= 0.5, and the integral time τIwas equal to 0.3 min. All the liquid level controllers were set with a gain KCof 2,and the integral time τIwas equal to 9 999 min. All the pressure controllers were set with a gain KCof 20, and the integral time τIwas equal to 12 min. As for the temperature controllers, the Tyreus-Luyben turning settings[8]were used.TC1 was set with a gain KCof 0.635, and the integral time τIwas equal to 5.28 min. TC2 was set with a gain KCof 0.2488, and the integral time τIwas equal to 7.92 min. The process was run in a dynamic environment until it reached a steady state. After that, a ±20% disturbance to the total feed flowrate/MeAc composition was added to the system,with the dynamic behaviors of some key parameters of the system shown in Figure 6.

It can be seen from Figure 6 that both the HAc product and the MeOH product can reach a new stable quality value compared with the target purity via the basic control structure. However, the top temperature of the LP column and the MeAc composition of the HP column bottom stream show bad control behaviors after the disturbance.Although the two parameters can reach the new stable values after several hours, such differences between the initial and the final values may lead to the system instability.

4.2 Improved control structure

The basic control structure does not show good control behaviors, so it is critical to propose an improved control structure. As mentioned above, using the composition controller (CC) has many disadvantages, which may make the system uncontrollable under more frequent disturbances[25]. Therefore, investigating a control structure without CC is of great importance.

To provide the information on control design, we optimized free parameters in the steady-state simulation using the sensitivity analysis tool of Aspen Plus. In the sensitivity analysis, the manipulated variables vary within specified ranges and the responses of the measured variables were analyzed to find a reasonable and high-efficiency operating condition. In this paper, we only present the sensitivity analysis results of the reflux ratio of the LP column, which provides the useful information for control purpose. For the sensitivity analysis of other important variables, please see papers of our previous work[18-19].

Figure 6 Control effect of basic control structure with: a ±20% disturbance to total feed flowrate and a ±20% disturbance to MeAc feed composition

Since the LP column does not have a condenser, the reflux ratio of the LP column denotes the ratio of the liquid stream mole flowrate from the HP column versus the vapor stream mole flowrate from the LP column. The reflux ratio of the LP column has an influence on the MeAc mole concentration in the LP bottom product, the trim condenser duty and the compressor duty, which were all selected as measured variables. The reflux ratio of the LP column was specified as the manipulated variable. The reflux ratio varied from 1.1 to 1.6, with its effect shown in Figure 7.

Figure 7 Effect of the LP column reflux ratio on MeAc mole fraction of the LP column bottom product and compressor and trim condenser duty

Figure 7 shows that with the increase of the reflux ratio,the MeAc mole concentration in the LP bottom stream will increase (with drop of final MeAc product purity),and both the trim condenser duty and the compressor duty will also decrease. To reduce the energy consumption and improve the product purity, it is important to find a stable operating point of the reflux ratio. In Figure 7, the MeAc mole concentration has a sharp increase when the reflux ratio reaches 1.40, while the trim condenser/compressor duty decreases slowly when the reflux ratio is above 1.20.Upon considering the stability of the system, the rates of variation in MeAc mole fraction and total duty were studied. The second-order finite difference of two variables was calculated to characterize the change rates. Ranges of the two variables in the domain of 1.20 to 1.40 were used to calculate the weight of two change rates. Finally, the weighted average change rates were calculated, and 1.24 with the lowest change rate was set as the operating value of the reflux ratio of the LP column.

Based on the sensitivity analysis, the reflux ratio of the LP column has a direct influence on the product purity, while the flowrate controller has a much quicker response than the composition controller. We then proposed an improved control structure without CCs, as shown in Figure 8. In the improved control structure, flowrate signals of the LP column top vapor stream and the HP column bottom liquid stream are transported to a ratio controller model,which can control the control point of the water feed controller on cascade. Such control structure shows better control behaviors as shown in Figure 9.

It can be seen from Figure 9 that the improved control structure shows better control behaviors than the basic control structure. The top temperature of the LP column and the MeAc composition of the HP column bottom stream show much smaller differences between the initial and the final values with the change in disturbances to the total feed flowrate. With the change in disturbances to the MeAc composition, the parameters shown above do not display obvious improvement as compared to that of the basic control structure. A possible reason is that CC is not contained in the novel control structures, while the reflux ratio of the LP column does not show a direct response with the change in feed composition.

Figure 8 Improved control structure for MeAc hydrolysis DPTCRD process

However, either with the feed disturbance or with the composition disturbance, the product purity could reach the target level. This means that the novel control structures without CC are feasible, and the improved control structure is especially suitable for systems with small disturbance to the feed composition.

5 Conclusions

In this paper, the novel control structures for methyl acetate hydrolysis differential pressure thermally coupled reactive distillation process were proposed. The novel control structures had abandoned CCs and used simpler controllers with which the product purity could meet the specification. The operating parameters of the steady-state simulation were based on our previous research work[18-19].The temperature profiles of four columns were provided in this paper. The steady-state simulation results showed that the parameters could reach the expected product purity, and the simulation could also provide a basis for control structure design.

Based on the steady-state results, two novel control structures were proposed. The basic control structure used the sensitive tray temperature controllers to ensure the product purity, and a feed/feed ratio controller was applied to keep the ratio of reactant. The results showed that the product purity could reach the goal under a±20% disturbance to the total feed flowrate or the MeAc composition, and the changes in the LP column temperature and the bottom composition showed the system’s instability. In the improved control structure,the sensitivity analysis was applied to analyze the effect of the LP column reflux ratio, and then a reflux ratio controller which could control the water feed flowrate on cascade was added to the top of the LP column. Such structure with a simple controller showed a more stable performance to cope with the changes in disturbance to the total feed flowrate. The composition and temperature changes were obviously reduced as compared with that of the basic control structure, and the two novel control structures were more suitable for industry process with less composition change in comparison with the control structure provided with composition controllers.

Acknowledgement:This work is supported financially by the Fundamental Research Funds for the Central Universities (Grant No. 18CX02120A), the Promotive Research Fund for Excellent Young and Middle-aged Scientists of Shandong Province(Grant No. BS2014NJ010) and the National Natural Science Foundation of China (Grant No. 21506255).