A review of internally heat integrated distillation column☆

Jing Fang*,Xiaomin ChengZhongyang LiHao LiChunli Li

1 School of Chemical Engineering and Technology,Hebei University of Technology,Tianjin 300130,China

2 National-Local Joint Engineering Laboratory for Energy Conservation of Chemical Process Integration and Resources Utilization,Tianjin 300130,China

Keywords:HIDIC Distillation Heat integrated Energy consumption Energy conservation

A B S T R A C T The energy consumption of distillation operation determines the amount of energy consumption throughout the chemical separation process.A heat integrated distillation column(HIDiC)could greatly reduce the irreversibility of the distillation process,so it gradually becomes a research hotspot.There are two major techniques for heat integration in HIDiC:internally and externally.This review paper describes the major research aspects of an internally heat integrated distillation column (IHIDiC), including the heat transfer models, various design structures(including the two-column HIDiC,Concentric HIDiC,Shell and tube HIDiC,Plate-fin HIDiC and the Super HIDiC, etc.),experimental research,simulation and optimization,process control research,as well as industrial plants and potential industrial applications.Among them,the heat transfer performance of various structures was analyzed of the various design structures based on experimental research,the effects of different factors(including relative volatility,reflux ratio,compression ratio,etc.)on HIDiC energy consumption or TAC is summarized in the simulation part.The calculation methods of the overall heat transfer coefficient and heat transfer models are summarized.The various optimization algorithms and optimization results of simplified HIDiC are summarized in the optimization part.The research status and the key technical issues in various aspects of HIDiC are summarized in this paper.In order to meet the requirements of industrial energy efficiency,the selection of multi-component separation distillation configurations needs to be considered more diversified,and internal complex coupling relationship of HIDiC needs to be further studied.

1.Introduction

Distillation units dominate the separation process in the chemical industry and their energy consumption reaches over 40%of the energy consumption of the entire chemical industry [1]. The irreversibility of the separation process resulted by the concentration difference,temperature difference and pressure difference caused a loss of energy in the distillation process.The larger mass transfer driving force and larger heat transfer driving force of the conventional distillation column at the top and the bottom of the column caused the large loss of effective energy,so the thermodynamic efficiency of a conventional distillation column is very low[2].An intermediate reboiler is provided at a certain position in a conventional rectification column,according to the thermodynamic analysis, the irreversibility of the separation process is reduced.If this idea is extended,an intermediate condenser is added to each stage of the rectifying section,and an intermediate reboiler is added to each stage of the reboiler to form a diathermic distillation,the entire distillation process can maximize the reduction of the irreversibility of the separation process.The heat integrated distillation column(HIDiC)can be regarded as the combination of heat pumps and diathermy distillation. With the gradual application of the vapor recompression column(VRC)and heat pumps in industrial separation processes, HIDiC was considered for energy saving. Bruinsma and Spoelstra[3]compared the energy efficiency of heat pump distillation with HIDiC, and proposed that the combination of VRC, HIDiC and novel heat pumps would make distillation process more energy efficient.Mah[2]proposed secondary reflux and vaporization(SRV)for the first time to increase the thermodynamic efficiency of distillation columns in 1977; SRV can be regarded as the prototype of HIDiC.Since then HIDiC (Fig. 1) was widely concerned by researchers at home and abroad.The conventional distillation column is divided into two columns including the stripping section and the rectifying section.The rising steam from the stripping section is supplied to the rectifying section through the compressor.The liquid from the stripping section flows through the throttling valve to the stripping section.The pressure difference and temperature difference between the rectifying section and the stripping section in the vertical position are generated by the compressor,which is similar to the diathermic distillation,so the irreversibility of the separation process is reduced for HIDiC.The rising steam of the rectifying section continuously condenses to provide the vaporization latent heat for the descending liquid of the stripping section. The degree of heat transfer determines the reduction in the heat loaded on the condenser and reboiler.Studies have shown that HIDiC has great potential for energy conservation[4,5].

Fig.1.The configuration of HIDiC.

Domestic and foreign researchers have done a lot of research on HIDiC[6-11]in modeling,configuration design,experimental research,simulation optimization,and controller design.The teams in Japan and the Netherlands have done numerous work on configuration design and experimental research.The teams in our country (such as Kejin Huang,Xinggao Liu,and Xigang Yuan)conducted model establishment,simulation optimization,and controller design.The design calculation and simulation methods of HIDiC including heat transfer models,parameter optimization,and heat transfer optimization,are summarized in this paper,it is convenient for researchers to make choices in the future research process.At the same time, this paper introduces the experimental research and industrial application of various structures of HIDiC, which provides theoretical guidance for industrial design,so the more comprehensive energy-saving guidance of HIDiC from design to industrial applications is available.Possible industrial applications of HIDiC are analyzed, and key scientific and technical problems in the existence of HIDiC are proposed in this paper.

2.The Heat Transfer Models

2.1.Mathematical models

There are steady state mathematical model and dynamic mathematical model for the mathematic model of the distillation columns.It is divided into an equilibrium model and a non-equilibrium model according to whether the tray efficiency is 100%.Mah et al.[2]established a steady-state mathematical model based on the equilibrium model firstly:

Component material balance for component i on stage j:

Vapor-liquid equilibrium equation for component i:

Normalized equation:

Thermal calculation on stage j:

The overall material balance for the rectifying section including stage j:

Or the overall material balance for the stripping section including stage f:

Among them,the definition equation of heat transfer is:

where ΔT is the temperature difference between the paired stages.A modified Wang-Henke method[12]was used to solve these equations.

All heat transfer models are based on mathematical models.However, the column performs high nonlinearity and asymmetry due to the high thermal coupling between the rectifying section and the stripping section of HIDiC,which makes it difficult to design the modeling and controllers[6].In order to make the model better conform to the dynamic behavior in HIDiC, Hwang [13,14] established a nonlinear wave theory for conventional binary distillation columns.Xinggao Liu[15,16] proposed that a nonlinear wave model, the profile functions and the wave velocity equations are the core part of the model,and it is better to meet the complicated dynamic behavior in the column.Composition in each stage should be measured for the estimation of wave velocity and there were many differential terms in the wave model,so some errors in the variables to be differentiated might lead to amplification of the error,the model is improved in the latter,highpurity control of the product can be achieved using an improved model.

2.2.The heat transfer models

The key to the establishment of the heat transfer model is how to obtain the heat transfer amount of each stage and the overall heat transfer coefficient between the rectifying section and the stripping section.The Mah team [17] proposed a plate-fin heat-coupling column heat transfer model in 1986,and verified the feasibility of the model through experiments.The researchers[2,18-21]regard the overall heat transfer coefficient as a constant value usually and calculate the heat transfer of each stage according to Eq.(7).However,the experiments of literatures[22-24]show that the vapor flow rate of the rectifying section and heat transfer temperature difference affect the value of the overall heat transfer coefficient,in which the heat transfer temperature difference is the main influencing factor,and the heat transfer temperature difference between the rectifying section and the stripping section is determined by the compression ratio. In order to make the overall heat transfer coefficient more in line with the actual heat transfer conditions of HIDiC, Dawei Chen et al. [25] established an annular structured packed HIDiC with an ethanol-water system,and based on the experimental data,they obtained corrected overall heat transfer coefficient as a function of compression ratio based on experimental data.Some researchers obtain the value of the overall heat transfer coefficients by applying different relevant formulas. For example, Ponce et al. [26]calculated the overall coefficient using empirical heat transfer coefficient formulas that accord with the vapor-liquid flow conditions in the column.Our research team[27]estimated the overall heat transfer coefficient by calculating the phase change heat transfer coefficients of the rectifying section and the stripping section respectively,the phase change heat transfer coefficients are calculated by material thermodynamic property parameters, and established a pilot-scale concentric HIDiC with an ethanol-water system,the accuracy of the model was verified.However,the effect of the liquid film thickness along the wall on the heat transfer coefficients of the rectifying section and the stripping section was not considered.The phase change heat transfer coefficients are not only related to the thermodynamic environment of the column,but also affected by the dynamic behavior of the column.

There are two methods for calculating heat distribution along the vertical height of the rectification section and stripping section currently: the uniform heat transfer area method and the uniform heat distribution method.In the early times,the method of the uniform heat transfer area was generally adopted[2,7,28,29].The heat transfer amount between the two stages along the vertical height of the column is proportional to the heat transfer temperature difference between the stage of the rectifying section and the stage of the stripping section,and the heat exchange area was regarded as constant.?.Oluji? et al.[18]used the uniform heat distribution method to calculate the required heat transfer area for different heights of the rectifying section, and they calculated the size of the heat exchange plate installed between the rectifying section and the stripping section according to the heat transfer area.The Gadalla team[8,30]obtained the temperature distribution of all stages in the stripping section and the rectifying section through Aspen Plus simulation.The pinch analysis was used to analyze the existence of pinch points in the middle of the column of the i-HIDiC.The uniform heat distribution method was used to calculate the pinch points. The required heat transfer area is very large, and there is difficulty in convergence during simulation,so a variable heat transfer rate design leads to better designs with minimum total cost compared with constant heat transfer rate schemes. Suphanit [9] proposed a new Aspen Plus simulation method,the mixture system was simulated with a conventional distillation column firstly,the lower heat load value of the reboiler and the condenser is used as the initial value of the total heat transfer between the rectifying section and the stripping section,and the average heat transfer amount required for corresponding stages was calculated using the uniform heat distribution method, the obtained new total heat transfer amount is used as the initial value of the next simulation,repeat the same calculation simulation steps until the heat transfer amount of each stage converged. Meanwhile, the advantages of the uniform heat transfer area method and the uniform heat distribution method were compared using the case of a benzenetoluene system,which obtained the same conclusion with Gadalla that the cost performance of HIDiC with the uniform heat transfer area method was better.

3.Design Structures

As mentioned before,HIDiC can reduce the irreversibility of distillation process and improve energy efficiency,solve its key technical problem is how to reduce the heat loads of the reboiler and condenser to 0 as far as possible.The key factors that affect the mass transfer and heat transfer performance of HIDiC are column structure and hardware inside the column[31].Various HIDiC configurations have been proposed by researchers from different countries.

Fig.2.The two-column HIDiC.

In 1977, Haselden [32] proposed the two-column configuration shown in Fig.2.The rectifying section and the stripping section were coupled through a series of tubular heat exchange hardware.The rising vapors from the high-pressure distillation column at different heights passed through the tube and reached the corresponding position of the low-pressure distillation column,in this way,the rising vapor in the rectifying section and the falling liquid in the stripping section achieve heat transfer through the tube wall,and excess vapor returns to the rectifying section through the pipe.Since the vapor-transporting pipe is exposed outside the column, the heat loss is severe. Seader[33,34] proposed the divided column, as shown in Fig. 3. The heat pipes installed in the middle serve as the heat transfer apparatus for transferring the thermal energy from the rectifying section to the stripping section. The column is divided vertically into the rectifying section and the stripping section. The case has been shown to save about 50%energy compared to conventional distillation columns.Seader and Baer[35]constructed a bench-scale distillation apparatus in order to investigate the potential method of performing intracolumn heat transfer. The heat transfer coefficient is 4717 W·m-2·K-1over a range in the temperature-driving-force of 12.4 to 22.7 K,the separation between the rectifying section and the stripping section is pure water.The transient experiments revealed that the heat pipes with water as a hydraulic resistor rapidly reacts to abrupt changes in the intracolumn temperature-driving-force.The whole process can obtain a good heat transfer effect under a small temperature difference. However, the temperature in the rectifying section cannot be too high,and the experiments are based on the distillation of pure system.

Fig.3.The divided column HIDiC.

Inspired by the literature[36]in 1986,Hugill[37]proposed a platefin HIDiC for the first time. The rectifying section and the stripping section within the column are alternately arranged,both the rectifying section and the stripping section can be regarded as heat exchange plates,so the heat transfer area is greatly increased,the fin of the plates spacing is not adjustable,as shown in Fig.4.The fin material used in the column is arranged in a shape similar to that of a structured packing.Not only does the regular fin material greatly increase the heat transfer area between the two column sections,but also its rough surface also strengthens the mass transfer process in the column to some extent.Other researchers[38,39]have controlled the heat transfer process between the rectification section and stripping section by changing the length and position of the applied fin flexibly,but the redistribution of vapors and liquids between the parallel column sections is difficult.

Fig.4.The plate-fin HIDiC.

Due to the complicated structure of the HIDiC heat exchanger mentioned above,Govind[40]proposed a concentric HIDiC shown in Fig.5 in 1986.The heat is transmitted from the internal high pressure column through a common wall to the external low pressure column.Our research team [27]carried out experiments in pilot scale for the concentric HIDiC,the calculated value of the overall heat transfer coefficient for the ethanol-water system was 300 to 800 W·m-2·K-1.Noda et al. [22] established a concentric HIDiC with the benzene-toluene system.Experiments have shown that the overall heat transfer coefficient is related to the vapor flow rate of the inner column.The overall heat transfer coefficient of the three inner diameters is all around 500 W·m-2·K-1.Lianghua Xu et al.[25]designed an annular structured HIDiC by considering the uneven vapor-liquid distribution along the column.The experimental results show that the overall heat transfer coefficient ranges from approximately 750 to 1000 W·m-2·K-1in the range of the investigated compression ratio,but the phenomenon of wall flow is more serious.Since the heat exchange area of the concentric HIDiC is only the area of the wall surface,in order to increase the heat exchange area, Graauw et al. [41] added special structure heat exchanger plates shown in Fig.6 based on the concentric HIDiC.The heat exchanger plates can be installed in the rectifying section or in the stripping section, and the heat exchange area can be changed flexibly according to the heat transfer temperature difference.Scholars in Netherlands[42]analyzed the concentric tray-HIDiC installed special heat exchanger plates'heat transfer performance with the cyclohexane-nheptane system, the experiment shows that the pressure drop of the column does not increase after adding the heat exchanger plates, the overall heat transfer coefficient is 350-1500 W·m-2·K-1.

Bruinsma[43]in Netherlands proposed the structured HIDiC similar to the plate-fin HIDiC structure.Plate packing and plate-fin can be used for heat exchange inside the structured HIDiC.Heating or cooling of the structured HIDiC is achieved by controlling the temperature of the external heat transfer oil. Experiments have shown that the thermal coefficient is about 300 W·m-2·K-1.

The shell and tube HIDiC proposed by Aso et al. [44] is shown in Fig.7.Similar to the concentric HIDiC,the heat transfer area is increased greatly.Due to the small diameter of the inner tube, it is suitable for packing inner column, less heat loss is lost at the same time. On the other hand, structural design is difficult due to the need to fix each small diameter concentric rectifying section. Tianjin University [45]established a simplified simulation model for the shell and tube HIDiC,the overall heat transfer coefficient is considered as a constant 500 W·m-2·K-1during the simulation.The simulation results show that the compression ratio is greatly affected by the boiling point difference of the mixture.

In 2011,Toyo Engineering Corp.of Japan invented a new energysaving distillation configuration: Super HIDiC [46]. The rectifying section and the stripping section of the conventional distillation column are inverted,the rectifying section is in the lower part,and the stripping section is in the upper part, and heat exchange is performed in the middle part of each section.Thermosiphon systems are used to circulate the mixture without pumping operations,and a compressor is used to increase the pressure and temperature in the column.Up to 50%energy savings compared to conventional distillation columns,the Super HIDiC is easier to maintain compared to other HIDiC.The new system can be used in a wide range of applications.This technology was improved and expanded to industrial applications by Toyo Engineering Corp.of Japan in collaboration with the National Institute of Advanced Industrial Science and Technology(AIST).

Fig.5.The concentric HIDiC.

Fig.6.Heat transfer panels in concentric HIDiC.

Fig.7.Shell and tube HIDiC.

In terms of fine chemicals and food production,the product demand may be less and the product purity requirements are higher. Batch distillation is more advantageous than continuous distillation in these respects. Currently there is less research on batch distillation [47].Nakaiwa [48] pointed out that the batch internal energy integrated distillation is more energy-efficient than the continuous internal energy integrated distillation energy utilization.However,because the HIDiC used in batch distillation is similar to the divided column in Fig. 3,the heat exchange structure is complex and it is difficult to extend it to industrial applications.Recently,Amiya and Jana proposed[49]a simpler structure,a new divided-wall HIDiC,which is the same as divided wall HIDiC,but there is no heat exchanger in the middle wall.A typical tray arrangement was designed to realize the heat transfer between the two sides.The analysis of the methanol-water case shows that the energy consumption and cost are more advantageous than the conventional HIDiC.

4.Simulation and Optimization

4.1.Effects of operating variables

The influence of design and operating variables on the performance of HIDiC is usually evaluated by economic evaluation of TAC or energy efficiency[50].Olujic et al.[4]compared the energy efficiency of the conventional binary distillation column,the vapor recompression,and HIDiC through the exergy loss analysis.It can be seen that the relative volatility of the mixture system has a greater impact on the energy saving efficiency of HIDiC,it has obvious advantages in the separation of close boiling point systems,i.e.,systems with relative volatility close to 1 are more favorable in terms of the energy saving.Some azeotropic mixtures,such as tetrahydrofuran-water and acetonitrile-water,are separated with conventional distillation columns must be added with an extractant or applied pressure swing distillation column, more energy is consumed. The idea of pressure swing distillation can be applied to HIDiC.The azeotropic point reached at one end of the lower pressure column and it will change under operating conditions of the higher pressure column,thus azeotropy of the mixture can be avoided.Nakaiwa and Huang[51,52]applied the above idea to design the separation of acetonitrile-water in HIDiC.It breaks binary azeotropes without the addition of any solvents; the desired component product is obtained at the end of the high pressure section and the low pressure section respectively.Some researchers[29] also studied the energysaving effect of HIDiC for multi-component separation.The energy consumption for the separation of benzene-toluene-p-xylene systems is only 70% of the energy consumption with a conventional distillation column.

The design and operating parameters of HIDiC include total stages,pressure difference,feed thermal condition,feed flow rates,feed composition, and feed location. Nakaiwa et al. [50] proposed operating cost functions with feed flow rate as a varying parameter; HIDiC is more economical than its conventional counterparts within a limited region of the feed flow rate through mathematical calculations.So the feed flow rate cannot be too large or too small for the separation of two-component mixtures.In order to meet the specifications of both end products, the pressure difference needs to increase as the feed flow rate increases without changing other operating conditions.Equipment costs increase with the increasing of total stages,but the operating cost and pressure difference increase with the decrease of total stages,therefore,the feed flow rates and total stages should be determined according to the actual separation process. Tianjin University [53] obtained the best total number of stages by calculating the TAC under the optimal compression ratio,and a balance between the two costs can be achieved by optimizing the numbers of the added stages, or equivalently the heat load of the HIDiC. While the feed composition has a relatively small effect on the differential pressure,it has almost no effect on the process energy efficiency. The team represented by Nakaiwa and Huang[54,55]also has more work on studying the impact of designing and operating parameters in the ideal HIDiC.For different thermal conditions, the mixture should be divided into two streams for feeding,and the vapor is best fed in the middle of the column section,the liquid is best fed in the middle of the column section,the vapor-liquid mass transfer in the column is enhanced in this way.In general,the researchers[18,30]considered HIDiC as ideal HIDiC when simulated with Aspen Plus,i.e.,the heat load of both the reboiler and the condenser was set to zero. But Rijke[42]reported that the secondary reflux in the rectifying section,caused by internal condensation against the heat transfer surface,was not sufficient to attain the top product specification,requiring the primary reflux produced by the condenser to be increased, and the installation of a reboiler or condenser can increase the operational flexibility of HIDiC [50]. Simulations by Bandaru et al.[56]showed that stage pressure drop had less effect on product purity and energy consumption. Product purity increased with the increase of heat transfer(UA),and the UA has great effect on the heat load of the reboiler,but less impact on the heat load of the condenser.In addition,the choice of reflux ratio for conventional distillation columns is relatively simple,but given the complex dynamics and thermodynamic coupling of HIDiC, the reflux ratio is also important for the energy-saving effect of HIDiC.The Mexican researchers[57]proposed a novel stochastic optimization algorithm called Boltzmann univariate marginal distribution algorithm(BUMDA)with the constraints handling to optimize the HIDiC sequences.With the reflux ratio as a continuous optimization variable, an exhaustive exploration is promoted by the Reset mechanism,the optimum reflux ratio can be obtained more accurately.

Whether a two-component or multi-component mixture is separated,pressure difference is the most important design and operating variable affecting the performance of HIDiC[29,50,53].Iwakabe et al.[29] pointed out that the sensitivity of the compression ratio to the energy saving efficiency of HIDiC varies with the boiling point difference of the mixture.Compression ratio mainly affects the flow rate and temperature distribution in the column.Gustavo et al.[26]studied the effect of compression ratio on the performance of HIDiC with heat panels in detail.The results show that the temperature and heat transfer coefficient on each stage increase with the increase of compression ratio.Because the bottom of the rectifying section receives the compressor steam,the increase of the compression ratio has a great influence on the central flow of the HIDiC.Considering the power demand of the compressor and the heat load costs of the reboiler and the condenser,the compression ratio is not the greater the more energy-efficient.The simulation results of Gustavo also meet this conclusion.

As can be seen clearly, the initial value of design and operation parameters lays an important foundation for subsequent optimization of HIDiC.

4.2.Optimization of design variables

In order to realize the industrial application of HIDiC,reducing its design difficulty and improving its operational flexibility,the researchers used different methods to simplify the design of HIDiC. After Huang[52]proposed a pressure swing external thermal coupling double distillation column,the simplified design of the number of heat exchangers on the side of column and the optimization of the heat transfer position between the rectifying section and the stripping section have become a hot topic at home and abroad gradually.The methods are divided into two types roughly: one is the mathematical programming method[19,58,59],and the other is the thermodynamic graph method such as the CGCC method[5,60].

The mathematical programming method treats the compression ratio and the heat load of the condenser, the reboiler, and the sideexchange heat exchanger as continuous variables and treats the number of theoretical stages and the stages to install the side heat exchangers as discrete variables. An optimum solution is sought or obtained by the SQP or MINLP method.Because these equations are nonlinear and non-convex equations,the search for an optimal solution is more complex.Suphanit[19]restricts the thermal coupling position between the rectifying section and the stripping section to the same height,which reduces the discrete variables and simplifies the SQP problem,but this may miss the optimal thermal coupling position.In order to solve the MINLP problem as a MILP problem,Alcántara[61]proposed compensatory terms to compensate the reduced energy at a condenser or reboiler less than the amount of the removed or supplied energy in the rectifying or stripping section,but reduced the energy efficiency of the reboiler.It can be seen that a long time to be consumed for solving optimization problems through the MINLP method.

Dhole and Linnhoff [60] proposed the column grand composite curve(CGCC),which can be used to analyze the possibility of thermal coupling between the two columns.The mass transfer and heat transfer in the column are assumed to be reversible in this method, so the number of theoretical plates needs to be infinite.Since the number of optimized HIDiC plates is limited in the actual distillation column,Wakabayashi[5]shows that the energy demand error is large by this method.Gadalla[8]proposed a design method in which the thermodynamic design is based on pinch technology and tray hydraulic design,the minimum heat exchange heat temperature difference and the heat transfer limit of the tray are obtained through the pinch analysis,but the difference in the minimum heat transfer temperature determines the number of heat exchangers,the minimum heat transfer temperature difference is a fixed value in this paper,so the optimal number of heat exchangers cannot be guaranteed.The effect of the total theoretical stage number and the position of the side heat exchanger on the heat load of the reboiler or the condenser is not considered for the methods based on CGCC,so the optimum heat transfer position of the side heat exchanger could not be determined.Wakabayashi and Hasebe[10]proposed a new graphical design approach based on the Ponchon-Savarit diagram in the latter.The number of heat exchangers was determined by the T-xy diagram, and the optimal thermal coupling position was optimized using the H-xy diagram.The calculation results are in good agreement with the conditions obtained through numerical optimization, but this method is only used for a binary system. In 2015,Wakabayashi and Hasebe[62]also proposed a new simplified optimization design method based on the Ponchon-Savarit diagram,which can be applied to multi-component systems successfully.The calculation of the number of heat exchangers in these two methods is based on the uniform heat transfer area and uniform heat distribution of the heat exchanger; no-uniform heat transfer area or no-uniform heat distribution method for HIDiC energy-saving efficiency remains to be further studied.

5.Process Control

Due to the strong internal thermal coupling of HIDiC,its nonlinear and complex dynamic behaviors cause difficulties for designing of the HIDiC control scheme.A good control system can compensate greatly for the morbidity and dynamic characteristics of the process; it can also maximize the energy-saving potential of the process. On the contrary, unsuitable control systems can degrade process dynamics and cause instability or system divergence[63,64].So the study of the HIDiC control scheme is the key to its transformation into practical applications.

The Japanese team first began the control research of HIDiC.For a complete internal energy-integrated distillation column(ideal HIDiC),unlike the conventional rectification column,no reboiler and condenser,the manipulated variable needs to be reset. They used the pressure difference between the two columns and the feed thermal condition as manipulated variables to control the composition of the product at the top and bottom of the column [65-67]. Nakaiwa [68] proposed and compared five kinds of control strategies:single loop composition control system,multi-loop composition control system,multivariable internal model control (IMC) system, modified multivariable IMC control (MIMC) system and nonlinear process model based control(NPMC)system,where NPMC has the best control performance.However,these control systems use direct component control and require on-line analyzers,which are costly.Since the temperature of the sensitive stage is influenced by the continuous variations in the column and cannot be used as a control variable,Huang[69]proposed a temperature difference control program for the first time in 2007,the heating steam of the feed preheater was set as a manipulated variable,however,the purity of the top product can only be guaranteed without guaranteeing the bottom of the column.In 2010,Huang[70]studied the dual temperature control system.When there are non-stationary disturbances in the system (such as feed flow), due to continuous pressure changes in the column caused by non-equilibrium disturbances, the balance between the temperature of the stages and the product composition is destroyed. They proposed inference compensators to achieve the required purity of the top and bottom products.However,the steady state residual error at the top of the column and the bottom of the column needs to be further reduced. For partial internal energyintegrated distillation columns(general HIDiC),the manipulated variable increase is caused by the increase of the reboiler and condenser.The double ratio control scheme(L/D,V/B)of the common rectification column can be used for general HIDiC, but the sensitivity to nonstationary disturbances is not very good. When using the pressure difference between the two columns and the feed heat condition as manipulated variables,there is a complex coupling relationship between them,and Huang[71]proposed that multivariable control methods can be considered.In 2001 Huang[72]proposed to mitigate the process nonlinearity through variable scaling,multi-model representation of process dynamics and feedforward compensation. The results show that the method has high robustness with respect to variation of operating conditions.

The Xinggao Liu team also conducted extensive research on the dynamic characteristics analysis and control of HIDiC.In 2005,Liu[73]first discovered a distinct inverse response of a high-purity internallinkable distillation column (ITCDIC) in dynamic behaviors; they proposed Modified IMC and multivariable PID control with singular value decomposition to improve control performance.For the pathological characteristics including high sensitivity to disturbances, strong asymmetric nonlinearity and a distinct inverse response of ITCDIC,they proposed generic model control(GMC)[74],generalized predictive control(GPC)[75],a model predictive control scheme(MPC)[76],and various expansion model control schemes [11,77-81], and they carried on the analysis to each control scheme performance. Among them, GMC is only applicable to handle the model with a relative order of 1,but the expanded model of the GMC is applicable to processes with any number of relative order.The product concentration of the top of the column up to 99.99%used RMPC control scheme,but the process of the calculation meeting closed-loop stability condition is more complicated. Liu [82] proposed a dynamic matrix control(DMC)method in 2014,compared with the traditional PID and an adaptive multivariable generalized predictive control(AM-GPC),the control method shows better control performance in terms of response speed,servo control and regulatory control.Due to the difficulty of on-line concentration measurement,many of the control models proposed by the team can be used for on-line concentration assessment,where EGMC performs well and its service and regulatory performances are better than the other control schemes.

Since the purity of the top of the column and the bottom of the column is affected by the coupling effect of multivariate factors,Meyer[83]first proposed multivariable control strategies,he established and compared the control performance of the fully decentralized control scheme based on PI-control and the centralized control system based on supervisory.Multi-variable control strategy based on MPC shows better control performance,and the integral absolute error(IAE)of the top and bottom product components is reduced to varying degrees,while the control strategy based on PI-control are not very effective for handling inter-actions among loops.

6.Industrial Plants and Potential Industrial Applications

6.1.Industrial plants

At present,there are few HIDiC devices for industrialization.Japan has carried out large-scale national research projects.HIDiC has been used in tests for cryogenic air separation.According to Matsuda et al.[84],two sets of pilot scale HIDiC have been established,the first set of devices (Double-tube type packed HIDiC) was established for the separation of multi-component systems,the height of the column is 27 m.In order to adapt to changes in vapor-liquid flow along the column,the inner column tube was modified in three levels of reduction and operated continuously for 1000 h.The results show that its energy consumption is 60%lower than that of traditional distillation columns;the second set of devices[85]established a shell and tube packed column.The addition of the reboiler and heat exchanger is more in line with the factory operation,and multicomponent industrial separation was achieved,the energy saving advantage is obvious.In 2016,Super HIDiC has been successfully applied to the distillation system of the methyl ethyl ketone production facility of Maruzen Petrochemical Company in Chiba Prefecture, Japan, with the aim of saving energy substantially. This is also the first industrial application of “Super HIDiC”technology.Different from Japanese scholars,Dutch researchers mainly studied tray columns of HIDiC [42], a concentric HIDiC was established to separate the cyclohexane-n-heptane mixture,and the rectification section and stripping section were in the full reflux state without compressor connection.After the addition of heat exchanger plates,the separation efficiency was increased,and the pressure drop was almost unchanged. This provides a theoretical basis for HIDiC to achieve industrial high pressure operation.Based on experimental studies[24,42,43,87-90],the interior of the structured packing plate is far superior to the plate-fin structure with porous fins in terms of heat transfer and mass transfer,and the efficiency is significantly improved with the increase in the process throughput,but it also brings about an increase in pressure drop.Compared with other HIDiC columns,the compact HIDiC has a smaller overall heat transfer coefficient,but the heat exchange area is larger than that of the concentric and shell and tube HIDiC. Researchers have studied more about the ideal HIDiC in model building,simplified optimization design,and dynamic control,and some researchers [91,92] proposed a simplified design of ideal HIDiC, so the author of this paper thinks that the general HIDiC proposed by Huang[52]and Super HIDiC have better industrial application prospects.

6.2.Potential industrial applications

As a high-energy-saving column, HIDiC can be extended to other distillation separation processes in addition to the separation of close boiling point systems and air separation. Reactive distillation can be used for esterification generally, trans-esterification, and hydrolysis[93]. Usually, these reactions are reversible reactions. The reaction products can be removed in a reactive distillation column in order to increase the reaction conversion and yield. On the other hand,the heat released by the reaction can be used to reduce the heat load of the reboiler. However, some of the reactants and products may azeotrope during the reaction.For example,methyl acetate and methanol mixture react with n-butanol to form n-butyl acetate. There is azeotropic boiling of methyl acetate and methanol among the mixture;the reaction and separation cannot be achieved at the same time using conventional reactive distillation columns. Some researchers [94-96]used a thermally coupled reactive distillation column to solve this problem and achieved the simultaneous completion of the reaction and separation process. HIDiC is more energy efficient than thermal coupling columns due to the fact that there is no reboiler, condenser,or small heat load. And Huang [52] pointed out that HIDiC can be used for the separation of azeotropes, and HIDiC or the simplified HIDiC can be considered in future studies to be applied to the azeotropic reaction distillation process and has great potential for energy saving.In addition,as mentioned in Section 3,batch distillation can be considered for smaller production volumes, and researchers[97] designed the nonlinear controller of internally heat integrated batch distillation in the latter, it shows better control performance.So batch operation can be considered in the industrial application of HIDiC.

7.Conclusions

In order to make this paper more clear,the main aspects investigated of HIDiC are summarized in Appendix A.

As the focus of energy conservation research, the energy efficiency of HIDiC is mainly affected by parameters such as compression ratio, feed position, and reflux ratio. It can be said that the compression ratio plays a decisive role. At present, the scientific research achievement of HIDiC can successfully balance the cost of electricity consumed by the compressor and the heat loads of the reboiler and condenser, in addition, the improved HIDiC-VRC or HIDiC-VRCIR[98,99]can achieve lower cost.However,it should be noted that the compressor is relatively expensive,some researchers[86]are studying the use of a compressor-free HIDiC system to save energy,it is necessary to consider comprehensively the implementation of HIDiC in industrial applications compared to other distillation schemes.

The operability of the two-column HIDiC, the divided column HIDiC,the plate-in HIDiC,the structured HIDiC,and other heat integration columns mentioned in this paper has been verified by experiments in Japan, Netherlands, and the United Kingdom. The heat transfer coefficient almost increases with the increase of the F factor,but the specific heat-mass coupling mechanism of the process has not yet been solved.The theoretical research and process design of HIDiC are based on the uniform heat transfer area or uniform heat distribution,which causes a certain deviation between the simulation of the internal energy integrated distillation process and the actual process, it will directly affect the accuracy of the process steady-state model and the dynamic model,also brings great difficulties to the separation efficiency,process optimization,and accurate calculation of energy consumption. Therefore, establishing an accurate heat transfer model and revealing the thermal-mechanical coupling mechanism of the process are important foundations for the calculation of the thermodynamic efficiency and energy-saving effect of the internal energy integrated distillation system and the design of the process and equipment,in the meanwhile theoretical basis and technical guidance for the industrial application of distillation energy integration process could be provided. Our research group took into consideration that the heat transfer model of the internal energy-integrated distillation column not only includes the longitudinal heat balances of the plate-by-plate in the conventional distillation column,but also involves the horizontal heat transfer between the rectifying section and the stripping section.We will study the effect of heat transfer on mass transfer and conduct a rigorous thermodynamic analysis of the process, aiming to establish the HIDiC trans-wall heat transfer model.

Due to the complexity of the internal mass transfer and heat transfer coupling mechanism, the technical difficulties for the HIDiC process control are posed. Change in feed flow rate or feed composition will inevitably lead to fluctuations in the temperature and product concentration of stages of the HIDiC.The feasibility of the control model determines the dynamic response performance of both ends of the column,and the control model is based on the dynamic model.It can be seen that the establishment of an accurate dynamic model has important guiding significance for realizing industrial applications of HIDiC.The influence of heat transfer on mass transfer and dynamic behavior has not been considered in the establishment of control models,and the control variables,measurable disturbances,and unmeasurable disturbances have a complex coupling relationship to the impact on the entire system. For multivariable interference to the system, there is a lack of studies,so researchers need to propose a more accurate predictive control model for multivariate interference.

In short,due to the complexity of design,optimization,and manipulation of HIDiC,it's large-scale industrialization requires more in-depth research,and it is necessary to further solve the scientific bottlenecks of HIDiC.

Nomenclature

Ajheat exchange area between the paired stages,m2

B bottom flow rate,mol·s-1

D distillate flow rate,mol·s-1

Fjflow rate of feed to stage j,mol·s-1

Fijflow rate of component i in the feed to stage j,mol·s-1

Hjenthalpy of the vapor on stage j,J·mol-1

hjenthalpy of the liquid on stage j,J·mol-1

Kijvaporization equilibrium ratio for component i on stage j

Ljliquid flow rate out of stage j,mol·s-1

n number of stages

Qjheat removal rate from stage j,J·s-1

Ujoverall heat transfer coefficient,J·m-2·s-1·K-1

Vjvapor flow rate out of stage j,mol·s-1

Wjvapor side stream withdraw rate from stage j,mol·s-1

xijliquid mole fraction of component i on stage j

yijvapor mole fraction of component i on stage j

Appendix A

The main aspects investigated of HIDiC are summarized in Table A1.

Table A1The main aspects investigated of HIDiC