Stefano Ciannella,Arthur Siqueira Damasceno,ícaro Cazé Nunes,Gilvan Wanderley de Farias Neto,Wagner Brand?o Ramos*,Romildo Pereira Brito

Department of Chemical Engineering,Federal University of Campina Grande,Campina Grande 58109-970,Paraíba,Brazil

1.Introduction and Problem Definition

The study and development of efficient control structures for specific variables of a distillation system,meaning mainly a column's top pressure and temperature,are of extreme importance to reduce the effect of potential disturbances which may occur during the operation of any distillation column,whether coming from changes in nominal conditions or failures in upstream or downstream processes that are directly or indirectly connected to the main separation unit.

Fluctuations in a column's top pressure and feed conditions around their operational values have a strong relation to undesired changes in temperature and composition pro files throughout the column,as well as with separation efficiency,since these factors are related to relative volatility between components in each stage.Thus,variations in these variables will induce vaporization of heavy components and condensation of light components,thereby obtaining a non-conforming final product and ultimately affecting process economy.

Pressure is the most important variable to control in distillation columns,because it affects condensation,vaporization temperatures,volatilities and almost all the processes that occur in the column[1].However,the assumption of constant pressure is often justified because pressure is tightly controlled,but overall pressure dynamics and their effect on the column behavior are not well understood[2].

In fact,there are few studies about how pressure affects the dynamic behavior of distillation columns and their control system.An efficient control of a column's top pressure in a distillation unit can minimize potential compensations in temperature throughout its operations due to disturbances or relevant noise,as well as preventing flooding within the column[3].A fair number of configurations for pressure control in vacuum and low/average-pressure distillation columns were studied,pointing out their main characteristics and requirements for a successful application;however,only a qualitative discussion is presented.The importance of pressure control on dividing wall columns(DWCs)to keep the vapor flow rate on either side of the wall as designed was also studied,but no numerical study is presented[4].

There are several ways to control column pressure,depending on how the column is configured,but the most common pressure control strategy in a distillation column involves the manipulation of cooling fluid flow rate.However,since water is the main cooling fluid used,this strategy is usually avoided,as it can accelerate fouling and corrosion in condensers[1].

Although manipulation of cooling fluid flow rate strategy works well in a simulation environment for pressure control[5–14],it is a difficult solution for the pressure control problem in most real distillation systems,because the flow rate of cooling fluid is an input variable which must be manipulated at high flow rates,which can be difficult or even unfeasible in real plants,as it is hard to manipulate a necessarily large valve for such a huge demand of cooling fluid.

A commonly used alternative to avoid manipulation of cooling fluid flow rate is to partially flood the condenser with process liquid by placing a control valve after the condenser;changes in the liquid level change the heat-transfer area,which changes the condenser duty;the cooling fluid flow rate is fixed[15].On the other hand,another study does not recommend the use of a flooded condenser operation with a valve after the condenser,because the heat-transfer area is a nonlinear function of the amount of process liquid in the condenser[16].

The use of hot-vapor bypass system for pressure control of distillation columns does not require the direct manipulation of the cooling fluid flow rate;it assumes the split of a hot stream such that a portion of this stream is redirected to promote fast dynamics in a specific part of the process,thus offering a fair condition for the control of a desired variable.

An adequate usage of the hot-vapor bypass strategy for air-cooled condensers is a satisfactory alternative for the column pressure problem among other explored approaches,such as direct modulation of the overhead stream valve and cooling fluid manipulation in the condenser,since the latter have considerable implementation and maintenance issues[17].

The implementation of control schemes which include hot-vapor bypass for heating and cooling of liquid streams through heat exchangers was investigated[18];according to the results,the addition of a hot-vapor bypass stream around the heat exchanger can promote robust control of the outlet temperature by manipulating the bypass flow rate,providing a product stream in a suitable temperature for a specific downstream process.In a distillation unit using hot-vapor bypass,the hot stream is the overhead vapor coming from the top of the column,with a portion of it being redirected to the re flux drum,as demonstrated in Fig.1.

Fig.1.Flow sheet(PFD)for a conventional distillation system with hot-vapor bypass.

According to Fig.1,the dynamics of top column pressure are affected by the manipulation of hot-vapor bypass stream valve(V-HVBP)or vapor stream valve(V3),thus making it possible to control pressure in such a manner that when the column pressure rises,the pressure controller manipulates the valve(s)to reduce vapor pressure within the column.

The present investigation discusses alternative control systems for pressure control in a distillation column through the adequate inclusion of a hot-vapor bypass stream connecting the column and re flux drum,keeping both the heat-transfer area of the condenser and the cooling fluid flow rate constant.Through dynamic simulations,it was observed that column and re flux drum pressure control were successfully achieved,while also evaluating the Integral Absolute Error(IAE)norm as the control performance index when considering various disturbances in the column's feed stream.This is a decent contribution to the field since there is a lack of scientific investigation exploring the simulation of hot-vapor bypass as a practical and consistent option for performing pressure control in reality.

2.Process Studied and Control Con figurations

The PFD presented in Fig.1 was implemented with Aspen Plus?(steady-state)using the RadFrac routine.The simulated column has a total of 29 stages(counted from top to bottom),including the reboiler,while the condenser and re flux drum were decoupled from the columns to obtain a more rigorous and realistic model.The condenser and re flux drum were simulated with Heater and Flash2 routines,respectively;moreover,total condensation was assumed,leading to no vapor production in the re flux drum.Phase equilibrium(VLE)was represented through a γ ? φ approach with NRTL model for activity coefficient calculations(γ).The vapor phase was considered ideal(φ =1)due to low system pressure.

The diameters of the column were calculated using the Column Internal/Interactive Sizing mode from Aspen Plus?,while the length and diameter of the re flux drum and sump height were calculated for a5-minutehold-up[19];the column diameter was1.84 m,sump height was 1.78 m and the re flux drum volume was 6.1 m3.

In order to reach the top and bottom specifications,re flux and boil up ratios were set as equal to 4.01 and 1.47,respectively,which corresponds to a reboiler heat duty equal to 8059.7 k W and a condenser heat duty of 7149.9 kW.The data for the global distillation column streams are shown in Table 1.

Table 1 Data for global streams

Once the steady-state model converges numerically and no physical inconsistencies are found,it is then exported to the dynamic platform Aspen Plus Dynamics?,where the control loops are implemented and properly tested.Fig.2 presents the control structure with manipulation of cooling fluid flow rate,called the base-case,established as the following:

(1)Sump liquid level is controlled by manipulating bottom flow rate;

(2)Re flux drum liquid level is controlled by distillate flow rate manipulation;

(3)The ratio between feed and re flux flow rates was kept constant;

(4)Temperature control is performed through manipulation of steam flow rate in the reboiler;

(5)Top pressure is controlled by manipulating cooling fluid flow rate;

(6)A flow rate controller was added in the feed stream to introduce disturbances.

The slope criteria was applied to determine the best stage to have its temperature controlled,which consists in selecting the tray where considerable temperature variations exist from tray to tray;the column temperature of stage 28 is recommended to be controlled.

Fig.2.Control structure manipulating the cooling fluid flow rate to pressure control(base case).

Only level controllers are proportional(P),while pressure, flow rate and temperature controllers are proportional-integral(PI).A dead time block of 2 min was added before the temperature controller.The controllers were tuned[20],and the obtained results(the same values for all cases)are summarized in Table 2.

Table 2 Controller parameters

For an approach(difference bet ween fluid process inlet temperature and cooling fluid outlet temperature)of 15 K,the UA product(overall heat transfer coefficient x heat-transfer area)is 168.9 kW·K?1.Condenser heat duty is removed by 408566 kg·h?1of incoming cooling water at 301 K.The investigation of the hot-vapor bypass strategy for the column's pressure control was performed considering three distinct configurations of manipulated variables:

?Con figuration I:manipulation of overhead vapor flow rate to the condenser;

?Con figuration II:manipulation of overhead vapor flow rate to the hotvapor bypass stream;

?Con figuration III:simultaneous manipulation of overhead vapor flow rates to the condenser and to the hot-vapor bypass stream.

Control configuration I aims to keep the pressure constant at the top of the column through manipulation of overhead vapor coming from the column and entering the condenser unit,as shown in Fig.3.Such manipulation is performed by modulating the condenser stream valve's degree of opening,while the hot-vapor bypass valve is unaltered.

Fig.3.Control structure manipulating the vapor flow rate to condenser(configuration I).

Furthermore,the following procedures are recommended to obtain a stable column top pressure upon simulating disturbances in the feed stream:

?Deploy the regulatory control;

?Deploy the ratio control between feed and re flux flow rates;

?Fix re flux drum and column sump liquid level at 50%;

?Run the simulation until the level reaches the set point;

?Deploy temperature controller and perform tuning actions;

?Deploy pressure controller and perform tuning actions.

In control configuration II(as depicted in Fig.4),the pressure on the top stage of the column is kept constant through manipulation of overhead vapor coming from the column and is redirected into the re flux drum through the bypass stream;manipulation of bypass vapor flow rate is executed by modulating the bypass stream valve's degree of opening,while condenser stream valve remains unchanged.The same procedures for configuration I are recommended for a stable simulation with configuration II.

Fig.4.Control structure manipulating the vapor flow rate to hot-vapor bypass(configuration II).

Control configuration III was designed to perform pres sure control in both the top of the column and the re flux drum using the condenser stream and bypass stream valves simultaneously,thus consisting of a multi variable 2×2 control system.Fig.5 shows the control scheme of configuration III.The same procedures for configuration I are recommended for a stable simulation of the hot-vapor bypass strategy following this configuration.

Fig.5.Control structure with dual pressure control(configuration III).

The use of hot-vapor bypass for pressure control requires subcooled liquid from the condenser to avoid vent from the re flux drum to atmosphere or increasing the drum pressure;this results in condensers with larger heat-transfer areas.Fig.6 shows the heat-transfer area and split fraction redirected to hot-vapor bypass as a function of sub cooling degree.

Fig.6.Split fraction redirected to hot-vapor bypass and heat-transfer area as a function of the subcooling degree.

3.Dynamic Results

Fig.7.Open loop top pressure responses for feed disturbance in the base case.

Before presenting the results for the three proposed control configurations,Figs.7–9 show the dynamic behavior of top pressure and composition(ethanol)at the top and bottom of the column,respectively,for±10%disturbances in the feed stream,with open loop top pressure for the base case:

?The flow rate was increased to 660 kmol·h?1at time equal to 2 h,and it was reduced to 540 kmol·h?1at 8 h;

?The composition(water)was increased to 88 mol%at time equal to 2 h and it was reduced to 72 mol%at 8 h.

Due to the small change(from 87500Pa to 1.15×105Pa)as depicted in Fig.7,the effect of pressure on the K-valuescan beneglected.However,even this small change in pressure(not only in the top,but throughout the column)is sufficient to cause large deviations in the distillate(Fig.8a),and mainly in the bottom(Fig.8b)compositions.These deviations are due to modifications in mass and energy balances after changes in feed conditions( flow rateor composition),in addition to the action of temperature,level and ratio controllers.

For positive disturbance,the mass flow rate inside the column increases,and the reboiler heat duty increases to reach the temperature set-point;as a result,the vapor flow rate at the top of the column also increases.As the cooling water flow rate is kept constant(controller in manual mode),the temperature(dewpoint)in the condenser outlet stream increases by 3°C;then the stream has two phases into the re flux drum for some minutes,however,the vapor phase disappears when pressure increases.In summary,pressure behavior throughout the column strongly depends on the withdrawal of energy by the condenser.

The UA product value chosen for hot-vapor bypass configurations is 264.2 kW·K?1(30 K of subcooling);this value is 56.5%greater than the base case,for a split fraction close to 9%.Condenser heat duty is the same as the base case.The UA value was chosen by comparing all hot-vapor bypass configurations using the same condenser heat transfer area.Table 3 depicts the results for stable top pressure control as a function of the subcooling degree for±10%disturbances in feed flow rate and composition.

According to Table 3,it is possible to use a hot-vapor bypass configuration(II and III)with 15 K degrees of subcooling,which would result in a lower heat transfer area(UA=202.7 kW·K?1)and is 20%greater than the base case.

Fig.9 presents the top pressure dynamic responses(closed-loop)for the three proposed configurations.According to the results,all hot-vapor bypass configurations could keep the top pressure in its set-point.Compared to the base case,hot-vapor bypass configurations present superior control performance,i.e.lower overshoots.

Fig.8.Dynamic responses of ethanol mole fraction at distillate(a)and bottom(b).Open-loop top pressure for base case.

Fig.9.Dynamic responses of top pressure for feed flow rate(a,b and c)and feed composition(d,e and f)disturbances.

Table 3 Stable pressure control as a function of subcooling degree in the condenser

Due to the elevated number of results for top and bottom ethanol compositions,the comparison between hot-vapor bypass and base case will only be presented for configuration I,which presented lower performance.As depicted in Figs.10 and 11,the performance for configuration I and the base case are identical;the lower overshoot of configuration I does not affect top or bottom compositions.

For hot-vapor bypass configurations,the cooling fluid flow rate is kept constant;therefore,only two curves will be presented in comparing the base case's dynamic behavior of this variable.

Fig.12 presents the dynamic behavior of the cooling fluid flow rate,and as can be observed after the positive disturbance in the feed flow rate(Fig.12a),the flow rate increases from 408 to 926 t·h?1to keep the top pressure at its set-point.This would require a fairly large valve,which in practical terms might be infeasible or would demand high maintenance values.For negative disturbance(Fig.12b),the cooling fluid flow rate changes from 408 to 249 t·h?1,which is another large modification.An identical behavior for the cooling flow rate is observed for disturbance in the feed composition.In some plants,throttling the cooling water flow rate is undesirable because fouling problems arising from low velocity can allow solids to deposit in the cooling water[17].

From a process point of view,the dynamic behavior of the cooling flow rate is proportional to the vapor flow rate from the top of the column.Since a large change is observed,it is very common to operate the condensers in an industrial plant with the cooling fluid flow rate valve completely open(100%);but this can result in high energy consumption.

Fig.10.Dynamic responses of ethanol mole fraction in the distillate(a)and bottom(b)for feed flow rate disturbance.

Fig.11.Dynamic responses of ethanol mole fraction in the distillate(a)and bottom(b)for feed composition disturbance.

Fig.12.Dynamic responses of the cooling fluid flow rate for feed flow rate(a)and(b)feed composition(a)disturbances.

Fig.13.Dynamic responses of cooling fluid outlet temperature for feed flow rate(a)and feed composition(b)disturbances.

Another important variable is the outlet temperature of cooling fluid.Fig.13 presents the results for the base case and all configurations using hot-vapor bypass.As can be observed,the outlet temperature for the base case reaches values superior to 323 K;this does not occur with the hot-vapor bypass configurations.It is important emphasize that high temperatures increase the corrosion in the condenser[17].In addition,the cooling fluid returning to the cooling tower at a high temperature may result in some problems such as a high evaporation rate.

Fig.14.Dynamic responses of cooling fluid flow rate and manipulated variable for feed flow rate(a)and feed composition(b)disturbances:configuration II.

It is important to emphasize that it is possible to use a smaller cooling fluid flow rate for the base case than what is presented in Fig.12,but one of two options should be chosen for the condenser:increase the temperature approach or the heat-transfer area.However,the bigger impact is caused by the temperature approach and as previously mentioned,the cooling fluid outlet temperature value is already very high.

Another important aspect of a control strategy is the behavior of the manipulated variable(MV),which is normally a valve.Fig.14 presents the dynamic behavior of the MV for disturbances in the feed flow rate(14a)and feed composition(14b)for configuration II,where it can be noted that the change in the valve opening is less than 10%for both disturbances.Furthermore,the change in the hot-vapor bypass flow rate is also close to 10%for both disturbances.From an industrial point of view this means using small valves.

4.Control Performance Analysis

Considering all the results obtained by hot-vapor bypass configurations and due to its operational and implementation simplicity,the best choice would be configuration I or II;i.e.only manipulating one valve,because configuration III assumes a higher complexity compared to the others as it undergoes a multi variable 2×2 system in which(inherent to this sort of control problem)there are interact ions between each control loop.However,also considering the economical point of view,the best configuration is II because of the lower heat transfer area.

Control performance indexes are commonly encountered as integral functions and basically evaluate the numerical discrepancy between a measured variable and its set-point within a time length;several indexes may be considered upon analyzing the performance of a control system,such as the integral of absolute error(IAE),integral of time multiplied by absolute error(ITAE),integral of squared error(ISE),and integral of time multiplied by squared error(ITSE)[21].In order to evaluate disturbance rejection for the three control schemes proposed in this work,the IAE index was chosen as the quantifier for control performance in each performed simulation,because it relates to process variance and how close the set-point can be to controlled variable specification,thus being mathematically defined by Eq.(1):

In Eq.(1)y is the controlled variable value and yspis the set-point value;t0and t are the initial and final time.Table 4 provides numerical values for the IAE index obtained in each simulation for all configurations,relating each configuration to their disturbance variables through the IAE.

Table 4IAE(×102Pa)index for performance test with each hot-vapor bypass pressure control configuration

According to Table 4,the smallest value in magnitude for the IAE average is related to configuration II,or the“pure hot-vapor bypass control configuration”.Con figuration III(dual pressure control)demonstrated a better performance when compared to configuration I,indicating that the adoption of hot-vapor bypass valve modulation as the manipulated variable affected the closed-loop response form of top pressure,since configuration I demonstrated dynamic response with considerable overshoots and undershoots.In summary,configurations that assume the hot-vapor bypass valve as a manipulated variable made the top column pressure response a smoother form.This outcome is rather interesting and supports the use of the hot-vapor bypass to change the dynamics of the re flux section in a distillation column,showing the potential of its use to simplify the form of a closed-loop response,and thus turn the system into an easier one to control or manipulate set points.

5.Concluding Remarks

The present work contributes to the development of alternative options to accomplish pressure control in a realistic and practical fashion considering the problems of a distillation process,and provides explanations for the implications of inserting a hot-vapor bypass stream for pressure control.

In general,all three hot-vapor bypass control configurations investigated in this work showed decent and relevant results as alternatives for distillation pressure control against the usual strategy(base case),which involves the dynamic manipulation of the cooling fluid flow rate by t·h?1.However,configurations II and III assume the hot-vapor bypass valve as a manipulated variable,and have made the dynamic response of pressure softer.Furthermore,configurations II and III work with a smaller heat-transfer area.

Between II and III,modulating the hot-vapor bypass valve only in configuration II provided the most satisfactory control performance considering deviation of the top column pressure to its set point,besides configuration III assumes a higher complexity as it undergoes a multi-variable system in which there are interactions between each control loop.

Acknowledgements

The authors thank the Conselho Nacional de Desenvolvimento Cientí fico e Tecnológico(CNPq)for financial support for this work.

[1]H.Z.Kister,D.W.Hanson,Control column pressure via hot-vapor bypass,Chem.Eng.Prog.111(2005)35–45.

[2]S.Skogestad,Dynamics and control of distillation column:A critical survey,IFACSymposium,197,1997.

[3]A.W.Sloley,Effectively control column pressure,Chem.Eng.Prog.97(2001)38–48.

[4]D.Staak,T.Grutzner,Process integration by application of an extractive dividingwall column:an industrial case study,Chem.Eng.Res.Des.123(2017)120–129.

[5]Q.Wang,B.Yu,C.Xu,Design and control of distillation system for methylal/methanol separation part 1:Extactive distillation using DMF as an entrainer,Ind.Eng.Chem.Res.51(2011)1281–1292.

[6]S.Wang,C.Yu,H.Huang,Plant-wide design and control of DMC synthesis process via reactive distillation and thermally coupled extractive distillation,Comput.Chem.Eng.34(2010)361–373.

[7]I.D.Gil,J.M.Gómez,G.Rodríguez,Control of an extractive distillation process to dehydrate ethanol using glycerol as entrainer,Comput.Chem.Eng.39(2012)129–142.

[8]Z.Fan,X.Zhang,W.Cai,F.Wang,Design and control of extraction distillation for dehydration of tetrahydrofuran,Chem.Eng.Technol.36(2013)829–839.

[9]E.Hosgor,T.Kucuk,I.N.Oksal,D.Kaymak,Design and control of distillation processes for methanol-chloroform separation,Comput.Chem.Eng.67(2014)166–167.

[10]H.Luo,K.Liang,W.Li,Y.Li,M.Xia,C.Xu,Comparison of pressure-swing distillation and extractive distillation methods for isopropyl alcohol/diisopropyl ether,Ind.Eng.Chem.Res.53(2014)15167–15182.

[11]L.Sun,Q.Wang,L.Li,J.Zhai,Y.Liu,Design and control of extractive dividing wall column for separating benzene/cyclohexane mixture,Ind.Eng.Chem.Res.53(2014)8120–8131.

[12]S.Tututi-Avila,A.Jiménez-Gutiérrez,J.Hahn,Control analysis of an extractive dividing-wall column used for ethanol dehydration,Chem.Eng.Process.82(2014)88–100.

[13]Y.C.Wu,H.Lee,H.Huang,I.Chien,Energy-saving dividing-wall column design and control for heterogeneous azeotropic distillation systems,Ind.Eng.Chem.53(2014)1537–1552.

[14]W.B.Ramos,M.F.Figueirêdo,K.D.Brito,S.Cianella,L.G.S.Vasconcelos,R.P.Brito,Effect of solvent content and heat integration on the controllability of extractive distillation process for anhydrous ethanol production,Ind.Eng.Chem.Res.55(43)(2016)11315–11328.

[15]W.L.Luyben,Dynamic simulation of flooded condensers,Chem.Eng.Res.Des.118(2017)12–20.

[16]F.G.Shinskey,Distillation Control,McGraw-Hill,N.Y,1977.

[17]W.L.Luyben,Quantitative comparison of alternative control schemes for air-cooled condensers,AIChE J.52(2006)611–622.

[18]W.L.Luyben,Heat-exchanger bypass control,Ind.Eng.Chem.Res.50(2011)965–973.

[19]W.L.Luyben,Distillation Design and Control Using Aspen Simulation,John Wiley&Sons,Inc.,NewYork,2006.

[20]B.D.Tyreus,W.L.Luyben,Tuning of PI controllers for integrator dead time processes,Ind.Eng.Chem.Res.31(1992)2625–2628.

[21]R.A.Krohling,J.P.Rey,Design of optimal disturbance rejection PID controllers using genetic algorithms,IEEE Trans.Evol.Comput.5(2001)78–82.