Ethane/ethylene and propane/propylene separation in hybrid...

15
Separation and Purification Technology 73 (2010) 377–390 Contents lists available at ScienceDirect Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur Ethane/ethylene and propane/propylene separation in hybrid membrane distillation systems: Optimization and economic analysis Marzouk Benali , Bora Aydin 1 Natural Resources Canada, CanmetENERGY, 1615 Lionel-Boulet Blvd., Varennes, Quebec J3X 1S6, Canada article info Article history: Received 16 September 2009 Received in revised form 27 April 2010 Accepted 28 April 2010 Keywords: Multicomponent distillation Ethylene Propylene Hybrid system Membrane separation Economic analysis abstract The combination of membrane and distillation processes to form a hybrid separation system is proposed as an alternative design to replace the current distillation technology. The main objective of this work is to scrutinize the feasibility of numerous hybrid membrane distillation (HMD) schemes through simula- tion. Silver nitrate is used for modeling/simulation as membrane carriers at a concentration of 6 mol L 1 . Different configurations are suggested to obtain highest product (ethylene or propylene) purity. These technologies are compared in terms of capital, operating and utility costs with the conventional C2- and C3-splitters. For the ethane/ethylene separation, the membrane cascade system resulted in the highest ethylene purity. However the series configuration is more economical than the membrane cascade sys- tem. For propane/propylene separation, the top configuration outperformed the conventional C3-splitter and other HMD configurations in terms of propylene purity. Crown Copyright © 2010 Published by Elsevier B.V. All rights reserved. 1. Introduction Ethylene (C 2 H 4 ) is one of the most important and one of the largest volume petrochemicals produced in the world today. The importance of ethylene is derived from the double bond in its molecular structure making it reactive and industrially convert- ible to a variety of intermediate and end products. It serves as the principal building block of the petrochemical industry. Worldwide demand for ethylene has grown steadily in the past and is expected to reach 140 million tons per year by 2010. The ethylene produc- tion system is very capital intensive; the capital cost is strongly dependent on the nature of the feedstock. In an ethylene plant, the feedstock is subjected to thermal cracking to produce a mixture of hydrocarbons ranging from hydrogen and methane to gasoline and heavier components. In general, the production of ethylene generates liquid, gaseous and solid wastes that have to be man- aged in a safe manner. The amounts and types of waste streams depend on the type of hydrocarbon feedstock. The major waste Abbreviations: CEPCI, Chemical Engineering Plant Cost Index; FT, facilitated transport; HIDiC, heat integrated distillation column; ICIS, International Chemical Information Service; M1, membrane 1; M2, membrane 2; PR, Peng–Robinson; SCDS, simultaneous correction distillation column; SRK, Soave–Redlich–Kwong; VR, vapor recompression. Corresponding author. Tel.: +1 450 652 5533; fax: +1 450 652 5198. E-mail addresses: [email protected] (M. Benali), [email protected] (B. Aydin). 1 Present address: Delft University of Technology, Process and Energy Laboratory, Leeghwaterstraat 44, 2628 CA Delft, The Netherlands. streams include caustic scrubber effluent, dilution steam conden- sate, coke and tare at the condensate separators, and system vent gases during start-ups and shutdowns. The waste gas streams are generally burned either in a furnace or in a flare. The flare is one of the most important and expensive parts of the ethylene plant facility, which has to be optimized. The optimization opportuni- ties include design changes, product purity enhancement, energy reduction, production capacity increase, by-products minimizing, revamp economics, and reduction of greenhouse gases. Thermal cracking of propane and/or ethane as feedstocks in the presence of steam remains one of the most important and widely employed processes for ethylene production, which consists basically of four distinct processes, namely: thermal cracking and quenching, com- pression and acid gas removal (i.e. H 2 S, CO 2 ), subcooling and product separation, and refrigeration. Fig. 1 illustrates a block flow diagram for a typical ethylene plant. The feedstocks are fed to a bank of parallel pyrolysis furnaces. In the convection zone of the fur- nace, the feed is preheated to about 600 C and is then diluted with steam that reduces coking and improves product selectivity. In the radiation zone of the furnace, the feedstock–steam mixture passes through vertical coils where pyrolysis takes place at temperatures above 600 C. At the exit of the cracking furnace, the outputs are immediately quenched to about 350 C in a transfer-line exchanger to stop reactions and recover the waste heat for steam generation. The cracked gases are cooled to about 40 C in a water quench tower to condense the heavy products (e.g. fuel oil) and most of the dilu- tion steam. The cooled gases are then compressed to about 3.5 MPa in four compression stages. Between the 1st and 3rd compression stages, the gases are scrubbed with caustic gas to remove H 2 S and 1383-5866/$ – see front matter. Crown Copyright © 2010 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.seppur.2010.04.027

Transcript of Ethane/ethylene and propane/propylene separation in hybrid...

Page 1: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

Ed

MN

a

ARRA

KMEPHME

1

limipdttdfoagad

tIsr

(

L

1d

Separation and Purification Technology 73 (2010) 377–390

Contents lists available at ScienceDirect

Separation and Purification Technology

journa l homepage: www.e lsev ier .com/ locate /seppur

thane/ethylene and propane/propylene separation in hybrid membraneistillation systems: Optimization and economic analysis

arzouk Benali ∗, Bora Aydin1

atural Resources Canada, CanmetENERGY, 1615 Lionel-Boulet Blvd., Varennes, Quebec J3X 1S6, Canada

r t i c l e i n f o

rticle history:eceived 16 September 2009eceived in revised form 27 April 2010ccepted 28 April 2010

a b s t r a c t

The combination of membrane and distillation processes to form a hybrid separation system is proposedas an alternative design to replace the current distillation technology. The main objective of this work isto scrutinize the feasibility of numerous hybrid membrane distillation (HMD) schemes through simula-tion. Silver nitrate is used for modeling/simulation as membrane carriers at a concentration of 6 mol L−1.

eywords:ulticomponent distillation

thyleneropyleneybrid system

Different configurations are suggested to obtain highest product (ethylene or propylene) purity. Thesetechnologies are compared in terms of capital, operating and utility costs with the conventional C2- andC3-splitters. For the ethane/ethylene separation, the membrane cascade system resulted in the highestethylene purity. However the series configuration is more economical than the membrane cascade sys-tem. For propane/propylene separation, the top configuration outperformed the conventional C3-splitter

tions

embrane separationconomic analysis

and other HMD configura

. Introduction

Ethylene (C2H4) is one of the most important and one of theargest volume petrochemicals produced in the world today. Themportance of ethylene is derived from the double bond in its

olecular structure making it reactive and industrially convert-ble to a variety of intermediate and end products. It serves as therincipal building block of the petrochemical industry. Worldwideemand for ethylene has grown steadily in the past and is expectedo reach 140 million tons per year by 2010. The ethylene produc-ion system is very capital intensive; the capital cost is stronglyependent on the nature of the feedstock. In an ethylene plant, theeedstock is subjected to thermal cracking to produce a mixturef hydrocarbons ranging from hydrogen and methane to gasoline

nd heavier components. In general, the production of ethyleneenerates liquid, gaseous and solid wastes that have to be man-ged in a safe manner. The amounts and types of waste streamsepend on the type of hydrocarbon feedstock. The major waste

Abbreviations: CEPCI, Chemical Engineering Plant Cost Index; FT, facilitatedransport; HIDiC, heat integrated distillation column; ICIS, International Chemicalnformation Service; M1, membrane 1; M2, membrane 2; PR, Peng–Robinson; SCDS,imultaneous correction distillation column; SRK, Soave–Redlich–Kwong; VR, vaporecompression.∗ Corresponding author. Tel.: +1 450 652 5533; fax: +1 450 652 5198.

E-mail addresses: [email protected] (M. Benali), [email protected]. Aydin).

1 Present address: Delft University of Technology, Process and Energy Laboratory,eeghwaterstraat 44, 2628 CA Delft, The Netherlands.

383-5866/$ – see front matter. Crown Copyright © 2010 Published by Elsevier B.V. All rioi:10.1016/j.seppur.2010.04.027

in terms of propylene purity.Crown Copyright © 2010 Published by Elsevier B.V. All rights reserved.

streams include caustic scrubber effluent, dilution steam conden-sate, coke and tare at the condensate separators, and system ventgases during start-ups and shutdowns. The waste gas streams aregenerally burned either in a furnace or in a flare. The flare is oneof the most important and expensive parts of the ethylene plantfacility, which has to be optimized. The optimization opportuni-ties include design changes, product purity enhancement, energyreduction, production capacity increase, by-products minimizing,revamp economics, and reduction of greenhouse gases. Thermalcracking of propane and/or ethane as feedstocks in the presenceof steam remains one of the most important and widely employedprocesses for ethylene production, which consists basically of fourdistinct processes, namely: thermal cracking and quenching, com-pression and acid gas removal (i.e. H2S, CO2), subcooling andproduct separation, and refrigeration. Fig. 1 illustrates a block flowdiagram for a typical ethylene plant. The feedstocks are fed to a bankof parallel pyrolysis furnaces. In the convection zone of the fur-nace, the feed is preheated to about 600 ◦C and is then diluted withsteam that reduces coking and improves product selectivity. In theradiation zone of the furnace, the feedstock–steam mixture passesthrough vertical coils where pyrolysis takes place at temperaturesabove 600 ◦C. At the exit of the cracking furnace, the outputs areimmediately quenched to about 350 ◦C in a transfer-line exchangerto stop reactions and recover the waste heat for steam generation.

The cracked gases are cooled to about 40 ◦C in a water quench towerto condense the heavy products (e.g. fuel oil) and most of the dilu-tion steam. The cooled gases are then compressed to about 3.5 MPain four compression stages. Between the 1st and 3rd compressionstages, the gases are scrubbed with caustic gas to remove H2S and

ghts reserved.

Page 2: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

378 M. Benali, B. Aydin / Separation and Purifica

Nomenclature

Ci concentration of component i (mol m−3)D diffusion coefficient (m2 s−1)F feed flow rate (mol s−1)J flux (mol m2 s−1)Keq equilibrium constant (m3 mol−1)kH Henry’s constant (mol m−3 Pa−1)� thickness of membrane (m)Mcut membrane module cut raten molar flow rate (mol s−1)P permeate flow rate (mol s−1)p pressure (Pa)Pc critical pressure (Pa)RR retentate in Eq. (7) (mol s−1)R gas constant in Eqs. (16)–(18)S membrane surface area (m2)T temperature (K)Tc critical temperature (K)Tr reduced temperatureV volume (m3)xi

F mole fraction of component i in the feedxi

P mole fraction of component i in the permeatexi

R mole fraction of component i in the retentatez membrane axis

Greek letters˛i/j separation factorε porosity� tortuosityω acentric factor (−log(P/Pc)Tr=0.7) − 1

Subscriptsc critical valueeq equilibriums–e silver–ethylidenetot total

Fig. 1. Block flow diagram of a

tion Technology 73 (2010) 377–390

CO2. After the 4th compression stage, the gases are cooled to about15 ◦C with propylene refrigerant and dehydrated with molecularsieves. The dried gases are cooled to low temperatures in a seriesof heat exchangers before they enter the separation section. Thebottoms of the primary and secondary deethanizers are carried to adepropanizer where C3 components are separated from the heaviercomponents (C4+). The secondary deethanizer overhead is hydro-genated in a catalytic reactor (diameter and volume are 1.818 mand 5.094 m3, respectively) to convert acetylene into ethylene. Asdescribed above, in the purification/separation subsystem of theplant, ethylene is separated from its by-products via a sequence offractionation steps. Among these steps, several require a cascadeof propylene/ethylene refrigeration loops. Propylene refrigerant isused in the chilling–demethanizer subsystem. When the relativevolatility between the key components is less than 1.2, the sepa-ration process becomes difficult. Subsequently, the distillation inthis case requires high-energy consumption, increased refrigera-tion capacity as well as an increased number of stages, leadingobviously to enhanced capital and operating costs. About 70% of therequired energy is consumed in the purification/separation subsys-tem. Therefore, any reduction in the refrigeration load will result ina significant decrease of the operating cost of the propylene refrig-eration closed-cycle system as well as the entire plant. As stated inthe literature [1], ethane/ethylene, and propane/propylene separa-tions are potential steps for an energy saving of 33% through hybridtechnologies involving both membranes and distillation processes.The selection of the best technologies for separation of ethylenefrom ethane, and propylene from propane is therefore conse-quential to minimize plant energy consumption and capital cost.Different technologies were investigated to intensify the traditionaldistillation processes. Ghosh et al. [2] investigated the potentialof a hybrid adsorption–distillation system for propane/propyleneseparation. According to their findings, although there is a reduc-tion in energy consumption there is a need for an innovativeadsorption–desorption process or for adsorbents having high selec-tivity. Schmal et al. [3] compared a heat integrated distillation

column (HIDiC) with a vapor recompression (VR) distillation col-umn. They found HIDiC to be 14% more economical than VRin operating cost. Finally, membranes combined with distillationcolumns were demonstrated in the literature [4–8] as a technolog-

typical ethylene plant.

Page 3: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

M. Benali, B. Aydin / Separation and Purifica

iagbeibm

2

olbmcp

fdthtamtasbtr[fdctwdsws

2

(smfdo

Fig. 2. Schematic diagram of the facilitated transport membrane system.

cal option to optimize the performance of hydrocarbon separationnd purification. In this work, membrane separations, which areenerally less energy-intensive than distillation processes, haveeen considered as promising alternatives to enhance the purity ofthylene and propylene as well as to reduce the capital and operat-ng costs of separation processes. In general, such an approach cane also applied for separation of various close-boiling hydrocarbonsixtures as styrene/ethylbenzene, and styrene/xylene isomers.

. Methods and design of hybrid separation flow diagrams

Membrane separation technologies have several advantagesver existing mass transfer processes including high selectivity,ow energy consumption and simple design [8]. The hybrid mem-rane processes can be designed in two different configurations: aembrane process combined with a conventional separation pro-

ess and a membrane process combined with another membranerocess.

The combination of membrane and distillation technologies toorm a hybrid separation system is proposed as an alternativeesign to replace the current distillation technology. Facilitatedransport (FT) membrane technology is used to design severalybrid membrane distillation (HMD) configurations studied inhis work. The appropriateness of FT membranes for ethylenend propylene recovery has been proven in [1], and [9–13]. FTembranes involve carrier-mediated transportation in addition

o permeate physical dissolution and diffusion. The presence ofcarrier that can react reversibly with permeate results in high

electivity and high permeability. There are two types of FT mem-ranes: one is a mobile carrier membrane (liquid membrane) wherehe carrier can diffuse into the membrane; the other is a fixed car-ier membrane where the carrier is immobilized in the membrane14]. The main objective of the present work is to scrutinize theeasibility of numerous HMD schemes. The hybrid configurationsiffer in the location of the membrane relative to the distillationolumn. Top, parallel, bottom, top-bottom and series configura-ions, along with a membrane cascade system with recycling andithout a distillation step are investigated. Agrawal [15] alreadyiscussed different possible membrane cascade configurations toeparate multicomponent gaseous mixtures. These configurationsere drawn based on the analogy with multicomponent distillation

chemes.

.1. Properties and equations of FT membrane separation process

Fig. 2 shows the general principle of an ethane (C2H6)/ethyleneC2H4) FT membrane system consisting of an aqueous solution of

ilver nitrate (AgNO3) transported by capillarity in a microporousembrane. At high pressure, C2H4 reacts with silver ion (Ag+) to

orm a complex component named silver–ethylidene ion, whichiffuses freely across the membrane. Once this complex reaches thepposite low pressure side of the membrane, the complex decom-

tion Technology 73 (2010) 377–390 379

poses and, subsequently, releases C2H4 in the permeate and thecarrier Ag+, which is then returned to the membrane feed side. Thechemical equilibrium can be written as follows:

Ag+ + C2H4 � (Ag · C2H4)+complex (aq) (1)

The equilibrium constant is

Keq = [Ag · C2H+4 ]

[Ag+][C2H4]= Cs–e

CAg+ CC2H4

(2)

The equilibrium constant of the silver–ethylidene ion formationhas been evaluated by several authors [16–18]. Obviously, sucha constant varies with the concentration of AgNO3 solution. Fora given temperature and concentration of AgNO3, the higher theC2H4 pressure, the greater the amount of C2H4 absorbed in the formof the silver–ethylidene complex. On the other hand, for a givenconcentration of AgNO3, the lower the temperature, the greaterthe amount of C2H4 absorbed in the form of the silver–ethylidenecomplex.

The total carrier concentration within the membrane module isobtained from the following balance:

Ctot = CAg+ + Cs–e (3)

Combining Eqs. (2) and (3), the concentration can be calculated asfollows:

Cs–e = KeqCC2H4 Ctot

1 + (KeqCC2H4 )(4)

By assuming that the equilibrium occurs at the feed side and thepermeate side, the boundary concentrations of silver–ethylideneion are determined as a function of known boundary concentrationsof ethylene:⎧⎪⎪⎨⎪⎪⎩

Cs–e(z=0) = KeqCC2H4(z=0)Ctot

1 + (KeqCC2H4(z=0))

Cs–e(z=�) = KeqCC2H4(z=�)Ctot

1 + (Keq × CC2H4(z=�))

(5)

where CC2H4(z=0)and CC2H4(z=�) correspond to the solubility of C2H4at the feed side and the permeate side, respectively. These boundaryconcentrations are determined using Henry’s law, expressing thepressure effects on the solubility of a gas in a saturated solution:{

CC2H4(z=0) = (kH)C2H4pC2H4(z=0)

CC2H4(z=�) = (kH)C2H4pC2H4(z=�)

(6)

2.2. Membrane model

The membrane model is based on the following assumptions:

- Cross-flow along the permeate side of the membrane- Plug flow along the feed side of the membrane- Steady state process- Isothermal operation- Instantaneous ethylene/silver reaction.- The local permeate-side composition is a function only of the local

feed-side composition.

The cross-flow pattern is used along all HMD configurations.The permeate composition changes along the membrane lengthand it is not affected by the composition of any other point along

the membrane side. It is only affected by the local flux at that point.The overall mass balance for a membrane module is:{

F = P + RR

FxiF = Pxi

P + RRxiRR

(7)

Page 4: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

3 urifica

Tm

T

T

IdiiFc

J

Tfip[

Id

J

CimE

J[Keq(

Fb

J

Fcsa

uMmbmFiEmeC

80 M. Benali, B. Aydin / Separation and P

he overall mass balance across a differential area of a membraneodule is given by

dn

dS= −J = −

n∑i=1

Ji (8)

he component mass balance is therefore given by

d(nxiF)

dS= −Ji (9)

ndxiF

dS+ xi

Fdn

dS= −Jxi

P (10)

he combination of Eqs. (8) and (10) gives:

dxiF

dS= J

n(xi

F − xiP) (11)

n contrary to distillation, in which separation is based on thermo-ynamics, the separations are inherently based on rate of transport,

.e. they depend on diffusion. The basic flux across the membranes equal to the flux across a thin film as proposed by Whitman [19].or ethylene, the flux ji becomes JC2H4 , which is proportional to theoncentration gradient and obeys to Fick’s law:

C2H4 = DC2H4

[CC2H4(z=0) − CC2H4(z=�)

](12)

he quantity C2H4/� obviously corresponds to a mass transfer coef-cient. Due to the microscopic properties of the membrane (i.e.orosity and tortuosity of the pores) and considering the film model19], this coefficient can be defined as follows:

DC2H4

�≡ DC2H4

ε

�(13)

n the case of a FT membrane, the flux should include the transportiffusion related to the silver–ethylidene ion. Eq. (12) becomes:

C2H4 = DC2H4

ε

[CC2H4(z=0) − CC2H4(z=�)

]

+Ds–e

ε

[Cs–e(z=0) − Cs–e(z=�)

](14)

onsidering that the membrane separation feature is stronglynfluenced by the partition of the solute (i.e. C2H4) between the

embrane and the adjacent solution (i.e. AgNO3), substitution ofqs. (5) and (6) in Eq. (14) gives:

C2H4 = DC2H4

ε

{(kH)C2H4

[pC2H4(z=0) − pC2H4(z=�)

]}+ Ds–e

ε

{KeqCtot(kH)s–e

[1 +

or ethane, which does not react with silver ion, the flux is giveny:

C2H6 = DC2H6

ε

{(kH)C2H6

[pC2H6(z=0) − pC2H6(z=�)

]}(16)

or C3H6/C3H8 separation, similar equations can be written to cal-ulate the local composition of each component in the permeateide. The physicochemical properties of ethylene and propylene inn aqueous silver solution are summarized in Table 1.

The membrane porosity, also termed the membrane void vol-me, is an important parameter affecting the membrane flux.embranes with higher porosity exhibit greater surface area. Theembrane porosity usually lies between 30 and 85%. The mem-

rane tortuosity is the average length of the pores compared toembrane thickness. It is in the order of 2–4. The thickness of the

T membrane is 170 �m and the ratio of porosity to the tortuos-

ty used for both C2H4/C2H6 and C3H6/C3H8 separation is 0.258.qs. (8) and (11) are solved numerically. The composition of per-eate as well as the flux of permeate are calculated from transport

quations (14) and (15). The numerical data are incorporated intoHEMCAD [22] by using “Excel Integration” feature. The integration

tion Technology 73 (2010) 377–390

pC2H4(z=0)

kH)C2H4pC2H4(z=0)]

− pC2H4(z=�)

1 + [Keq(kH)C2H4pC2H4(z=�)]

]}(15)

into CHEMCAD is based on the following sequence: load CHEMCAD,put data into CHEMCAD and run, load data. The Visual Basic serverinterface allows starting CHEMCAD, loading a simulation case, andtransferring information back and forth from Excel in an automatedfashion.

2.3. Ethane/ethylene hybrid separation

The HMD system may be arranged in several configurations.Operating conditions may be varied for each configuration in orderto achieve optimum performance. However, the concentration ofsilver nitrate is maintained constant at 6 mol L−1. A series con-figuration is investigated in which two membranes in series arefollowed by a distillation column. In addition, a membrane cascadesystem is explored as an alternative to a conventional C2-splitter torecover ethylene from ethane. Fig. 3 illustrates a two-stage mem-brane cascade based on two single-stage compressors for recyclingand a multistage compressor required to compress the ethyleneproduct to the desired pressure, as in the case of a conventionalC2-splitter. The fresh stream (208, coming from the demethanizer),composed of 75.66% C2H4, 12.61% C2H6, 8.58% C3H6, 1.34% C3H8,1.80% CH4, and 0.01% C3H4 enters at a pressure of 2.61 MPa and atemperature of 28 ◦C, and then is mixed with the recycled stream(207 and thereafter is fed to the first membrane (M1) at a pressureof 2.61 MPa and temperature of 31 ◦C. In M1, the recycled ethaneis separated as the retentate and directed to the cracking reactor.The recovered ethylene is concentrated as permeate at a pressureof 0.83 MPa and directed, after a compression step, to the secondmembrane M2 as a feed at a pressure of 1.34 MPa and a temperatureof 33 ◦C. The entire amount of C3H8, C3H6, CH4, and C3H4 is recov-ered in the retentate (stream 201). Thus, the stream 212 containsonly 99.99% C2H4 and 0.01% C2H6. High pure C2H4 is therefore pro-duced in M2 at a permeate pressure of 0.10 MPa and a temperatureof 20 ◦C. The retentate of M2 is compressed from 1.34 to 2.62 MPabefore being mixed with the fresh feed. The surface area of M1is obviously greater than the surface area of M2 since the highestrecovery of C2H4 occurs in the first separation stage. The ratio ofsurface area of M2 to surface area of M1 has been varied from 0.10to 0.80, and the optimal value is 0.40 corresponding to surface areasof 3243 and 8150 m2, respectively.

Fig. 4 illustrates the series configuration of the hybrid system. Inthis case, the fresh feed (3.50 MPa; −4 ◦C) is first flashed at a pres-

sure of 2.61 MPa to reach a temperature of −16 ◦C required to besent to two membranes in series. For optimal membrane cut ratesof 0.99 and 0.90, respectively for M1 and M2, the heat duties ofexchangers 102 and 103 are—0.67 and 8.46 MW, respectively. Incase of series configuration, the permeate (stream 107) of mem-brane M1 contains 78.08% C2H4, 7.86% C2H6, 10.29% C3H6, 1.60%C3H8, 2.16% CH4, and 0.01% C3H4 while the permeate (stream 108)of membrane M2 contains only 99.99% C2H4 and 0.01% C2H6. Theresulting permeate is recovered as ethylene product and the reten-tate leaves the second membrane at a pressure of 0.66 MPa and atemperature of 50 ◦C, and sent as the feed to the distillation column.A high pressure difference is required to provide an enhanced driv-ing force as well as a high feed concentration to produce high purity

ethylene product. The feed to the first membrane stage comes fromthe bottom of the demethanizer and flashes to a moderate pressurebefore entering the membrane. The refrigeration load associatedwith the feed stream is exchanged with other process streams tominimize the overall quantity of propylene refrigerant used in the
Page 5: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

M. Benali, B. Aydin / Separation and Purification Technology 73 (2010) 377–390 381

Table 1Physicochemical properties of components involved in a membrane module.

Component Diffusivity (m2 s−1) Chemical equilibrium constant (m3 mol−1) Henry’s constant (mol m−3 Pa−1)

C2H4 1.87 × 10−9 [20] – 5.61 × 10−5 [present work]C2H6 – – 2.39 × 10−5 [present work](Ag·C2H4)+ 1.13 × 10−9 [present work] 0.129 [present work] –Ag+ 1.66 × 10−9 [20] – –C3H6 1.63 × 10−9 [21] – 6.17 × 10−5 [present work]C3H8 1.61 × 10−9 [21] – 2.15 × 10−5 [present work](Ag·C3H6)+ 1.06 × 10−9 [21] 0.100 [present work] –

e syst

ptnrd

Fig. 3. Two-stage membrane cascad

lant. The feed temperature to the membrane system is set to main-ain the membrane operation above the freezing point of the silveritrate solution. The feed from the first stage is concentrated byemoving part of the recycled ethane as a retentate. The feed is thenirected to the second membrane where the separation yields a

Fig. 4. Hybrid membrane distillation series confi

em for ethane/ethylene separation.

high purity ethylene product. The retentate from the second mem-brane is directed to a low pressure column while high pressuredistillation column is used in the conventional C2-splitter. The keybenefits of such a configuration are downsizing the column, reduc-ing the condenser duty, and minimizing the amount of refrigerant.

guration for ethane/ethylene separation.

Page 6: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

382 M. Benali, B. Aydin / Separation and Purifica

Table 2Design details of the conventional C3-splitter (two columns in parallel).

Column 1 Column 2

Number of trays 115 90Tray efficiency 0.85 0.85

Tac

2

pbm(T1a1iitsfmstTccprdtciiioptibmrotwiro

st0Cdhmt

hand, lower selectivity (due to lower compressor load) leads to

Column diameter (m) 5.30 5.30Column pressure (MPa) 1.64 1.72Column reflux ratio 20.33 507.51

he diameter of the distillation column and the number of stagesre reduced by 67.0% and 78.7%, respectively, in comparison to aonventional C2-splitter system.

.4. Propane/propylene hybrid separation

Four different configurations were tested forropane/propylene separation: top, parallel, bottom and top-ottom. The concentration of silver nitrate is 6 mol L−1. Theembrane can be located at the top of the distillation column

Fig. 5). In this case, the membrane performs the final purification.he overhead of distillation column is fed to the membrane at.64 MPa and a temperature of 40 ◦C. The retentate is recyclednd fed at the middle of the distillation column at a pressure of.66 MPa and a temperature of 48 ◦C while the permeate pressure

s 0.10 MPa. The pressure drop along the feed side of the membranes assumed negligible. The membrane may also be combined withhe distillation column in a parallel configuration (Fig. 6). Inuch a configuration, the membrane feed stream is withdrawnrom one of the intermediate stages of the distillation column,

ore specifically at the 40th stage. The permeate and retentatetreams leaving the membrane are compressed and fed backo the distillation column at 20th and 60th stages, respectively.hese two streams enter the distillation column where the streamomposition is the same as the tray composition. In the bottomonfiguration, the membrane performs the final purification of theropane sent from the distillation column (Fig. 7). Refrigerant isequired to cool the permeate to its dew point before entering theistillation column. This configuration differs from the top one inhat a larger amount of refrigerant is required for the permeateondenser. The feed to the membrane is a saturated liquid. Fig. 8llustrates the top-bottom configuration in which one membranes located at the top of the distillation column while another ones located at the bottom of the column. The optimal conditionsf top-bottom HMD configuration corresponds to the permeateressure set at 0.34 MPa for the top membrane and 1.66 MPa forhe bottom membrane. The separation within the top membrane isn gas permeation as the membrane feed is a saturated vapor. Theottom membrane is operated in a pervaporation mode since theembrane feed is a saturated liquid. After a compression step, the

etentate of top membrane is fed at the 10th stage (1.67 MPa; 44 ◦C)f the distillation column while the permeate of bottom is fed athe 2nd stage (1.09 MPa; 159 ◦C). In all HMD systems associatedith C3H6/C3H8 separation, the diameter of the distillation column

s kept the same (i.e. 5.30 m) as in the conventional C3-splitter foretrofit purpose whereas it is reduced by 25.7% for design purposef a new plant.

For all configurations, the fresh feed stream entering the hybrideparation system is obviously the same as the one of the conven-ional C3-splitter and is composed of 73.37% C3H6, 14.83% C3H8,.03% C2H6, 0.20% C2H2, 1.80% CH4, and 0.01% C3H4, 7.89% 1,3-H , 1.64% 1,-C H , 0.11% n-C H , and 1.92% i-C H . The design

4 6 4 8 4 10 4 10

etails of the conventional C3-splitter are given in Table 2. Eachybrid system has its distinctive minimum reflux ratio. Two mainembrane features have a significant effect on the selectivity and

he flux rate: silver nitrate concentration and permeate pressure.

tion Technology 73 (2010) 377–390

For the top configuration, the surface area of membrane was variedfrom 100 to 1000 m2 whereas the for bottom and parallel configu-rations it was ranged from 400 to 4000 m2. The optimal values aregiven in Table 5 for each HMD system.

2.5. Separation factor of membrane module

The parameter to describe the separation efficiency forethane/ethylene, and propane/propylene mixtures fed to the mem-brane is the separation factor, ˛i/j, which is also called the selectivityfactor of the membrane. It can be formulated as below:

˛i/j =xP

i/xP

j

xFi/xF

j

(17)

The membrane module cut rate is one of the major independentvariables for HMD systems. In practical terms, this variable repre-sents the ratio of how much permeate is recovered per unit feedto the membrane module. There is always a trade-off between thepurity and the flow rate of permeate and retentate streams. For agiven membrane module cut rate, the flow patterns have a signifi-cant effect on the degree of separation as well as on the membranearea required to do so. When a perfect mixing is assumed on bothsides of the membrane module, the degree of separation decreaseswith an increase of the membrane module cut rate. For the cross-flow, as considered in the present work, the feed flows through theupstream of membrane surface in plug flow with no longitudinalmixing. There is no flow of permeate along the membrane surface.Thus, the degree of separation increases very slightly (in the orderof magnitude of 2–10%) at low membrane module cut rate (typi-cally 0 ≤ ˛i/j ≤ 0.5). As ˛i/j � 0.5, the degree of separation increasessignificantly.

3. Results and discussion

The distillation column is simulated with CHEMCAD usingthe simultaneous correction distillation (SCDS) column typeand Soave–Redlich–Kwong (SRK) state equation. SCDS is mainlydesigned to simulate non-ideal K-value chemical systems. It cal-culates the derivatives of each equation rigorously, including thederivative of K-value with respect to composition (ıK/ıX) ( ) term,which is significant in chemical system simulation. The equationsrelated to SRK model are given as follows:

p = RT

V − b− a

V2 + bV(18)

where

a = 0.42748R2T2

c [1 + f�(1 − T0.5r )]

2

pc(19)

b = 0.08664RTc

pc(20)

fω = 0.48 + 1.574ω − 0.176ω2 (21)

The simulation and optimization of the HMD system use two lim-iting permeate pressures of 0.10 and 0.34 MPa. The lower value of0.10 MPa provides high selectivity which minimizes the membranearea required by the HMD system. However, the compression dutywas observed to increase when 0.10 MPa was selected as a per-meate pressure. Contrarily, the compressor duty load decreasedwith the upper permeate pressure limit (0.34 MPa). On the other

increased membrane area. The economic performance of variousmembrane hybrid configurations and membrane cascade systemhas been assessed and compared to an ethylene production plantwith a yearly throughput of 500,000 metric-tons.

Page 7: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

M. Benali, B. Aydin / Separation and Purification Technology 73 (2010) 377–390 383

n for propane/propylene separation.

3

3

itnthal

Table 3Comparison of the design specifications of the conventional C2-splitter and seriesconfiguration hybrid systems.

Parameters ConventionalC2-splitter

Series configurationhybrid system

Number of columns 2 1Number of trays per column 94 40

Fig. 5. Top hybrid configuratio

.1. HMD system analysis

.1.1. Ethane/ethylene separationHMD systems are characterized by smaller equipment siz-

ng and relatively low operating cost. As seen in Table 3, forhe series configuration hybrid system, the number of columns,

umber of trays and column diameter are lower than in conven-ional C2-splitters. This results in a lower operating pressure andigher reflux ratio. The optimal configuration could be defineds producing the desired component at the highest purity andowest operation cost. As explained in detail in the next section,

Tray efficiency 0.85 0.85Column diameter (m) 4.545 1.500Column pressure (MPa) 2.00 0.65Column reflux ratio 0.09 1.38Product purity 0.9965 0.9979

Fig. 6. Parallel hybrid configuration for propane/propylene separation.

Page 8: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

384 M. Benali, B. Aydin / Separation and Purification Technology 73 (2010) 377–390

ion fo

hlu

flftppfiboofifiu

Fig. 7. Bottom hybrid configurat

igh product purity could be obtained with HMD systems at aower cost in comparison to the conventional C2-splitter config-ration.

Fig. 9 shows the effect of membrane module cut rate on theow rate of C2H4 in the product and recovery lines (stream 113)

or the series configuration. In all HMD and membrane cascade sys-ems, the membrane module cut rate is defined as the ratio of theermeate to the membrane feed flow rate. The flow rate in theroduct line increases with the increasing module cut rate of therst membrane at constant module cut rate of the second mem-rane. At constant module cut rate of the first membrane the effect

f the module cut rate of the second membrane on the flow ratef C2H4 in product line is minor. Higher module cut rate signi-es higher amount of C2H4 in the permeate (stream 107) of therst membrane. Thus, higher amount of C2H4 is sent to the prod-ct line via the permeate of the second membrane. In the recovery

Fig. 8. Top-bottom hybrid configuration

r propane/propylene separation.

line, the flow rate of C2H4 decreases with the module cut rate ofthe first and second membrane. A minor effect of the module cutrate on the recovery line is observed for the second membrane atthe cut rate values higher than 0.7. However the second membraneshould be operated at a membrane module cut rate of 0.9 to achievemaximum throughput of C2H4 in the product line. Fig. 10 showsthe effect of the membrane module cut rate on the separation fac-tor for the series configuration. The membrane module cut rate isexpressed as the ratio of the permeate flow rate to the feed flowrate. The mole fractions of C2H4 and C2H6 in the permeate and feedwere calculated for each membrane module cut rate to determine

the separation factor, which is directly proportional to the molefraction of C2H4 in the permeate, i.e. enrichment of C2H4. The sepa-ration factor increases linearly with the membrane module cut ratefor both membranes. High separation factor values for the secondmembrane can be essentially explained with higher mole fraction

for propane/propylene separation.

Page 9: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

M. Benali, B. Aydin / Separation and Purification Technology 73 (2010) 377–390 385

Fig. 9. Effect of first membrane module cut rate (on the flow rate of C2H4 in theproduct and recovery lines for the series configuration.

o1

Ccictpcu(ihflfomiadsocmoic

Fig. 11. Influence of membrane module cut rate of (a) first membrane and (b) secondmembrane on the flow rate of C2H4 in the product and recovery lines for the cascadesystem.

Fig. 14 shows the influence of the membrane module cut rate on

Fig. 10. Membrane module cut rate versus separation factor for the series.

f C2H4 and lower mole fraction of C2H6 in the permeate (stream08) of the second membrane.

The effect of membrane module cut rate on the flow rate of2H4 in the product and recovery lines (stream 212) for the cas-ade system is shown in Fig. 11(a) and (b). The flow rate of C2H4n the permeate of the first membrane increases with its moduleut rate. The permeate of the first membrane (stream 202) is fed tohe second membrane and then it is sent to the product line via theermeate of the second membrane (stream 210). At a given moduleut rate of the second membrane the flow rate of C2H4 in the prod-ct line increases with the module cut rate of the first membraneFig. 11a). Additionally the flow rate of C2H4 in the product linencreases with the module cut rate of the second membrane due toigher flow rate of C2H4 in the permeate. In the recovery line theow rate decreases abruptly with the membrane module cut rate

or both membranes. For the optimum module cut rate of the sec-nd membrane (0.90), an increase in the module cut rate of the firstembrane from 0.45 to 0.76 leads to a decrease of 99.7% of C2H4

n the recovery line (Fig. 11b). Similarly for the second membrane;n increase of the module cut rate from 0.50 to 0.90 results in aecrease of 85.3% of C2H4 in the recovery line which confirms theeparation efficiency of both membranes. Fig. 12 shows the effectf the membrane module cut rate on the separation factor for theascade system. The separation factor increases linearly with the

embrane module cut rate for both membranes. The mole fraction

f C2H4 in the permeate increases while the mole fraction of C2H6n the permeate decreases with the increasing membrane moduleut rate. Higher separation factor values for the first membrane are

Fig. 12. Membrane module cut rate versus separation factor for the cascade system.

due to lower mole fraction of C2H4 and higher mole fraction of C2H6in the feed (stream 209) of the first membrane.

The effect of membrane module cut rate on the total actualcompressor power and heat exchanger duty for the series config-uration is illustrated in Fig. 13. The heat exchanger duty increaseswith increasingly membrane module cut rate for both membraneswhere the actual compressor power decreases with the membranemodule cut rate. The effect of membrane module cut rate on theactual compressor power is more pronounced for the second mem-brane.

the total actual compressor power and heat exchanger duty for thecascade system. The actual compressor power and heat exchangerduty increase when the membrane module cut rate increases forboth membranes where the effect of the membrane module cut

Page 10: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

386 M. Benali, B. Aydin / Separation and Purification Technology 73 (2010) 377–390

Fig. 13. Effect of first membrane module cut rate (a) and second membrane modulecut rate (b) on the actual compressor power and heat exchanger duty for the seriesconfiguration.

Fa

ridcomtfl

3

prwb

fl1wdCgfl

((fib

Fig. 15. Influence of membrane module cut rate on the flow rate of C3H6 in theproduct and recovery lines for top configuration.

Fig. 16. The flow rate of C3H6 in the product and recovery lines for different mem-brane module cut rate values of bottom configuration.

ig. 14. Effect of first membrane module cut rate on the actual compressor powernd heat exchanger duty for the cascade system.

ate is more pronounced for the first membrane. The significantncrease of the total actual compressor power and heat exchangeruty is essentially due to the increase of 42.1% in the power of theompressor 208 and the duty of the heat exchanger 209 locatedn the product line (stream 212). Similar analysis for the secondembrane shows that the increase in the compressor power and

he heat exchanger duty is 24.3%. The corresponding increase of theow rate of C2H4 in the permeate is 24.3%.

.1.2. Propane/propylene separationA systematic analysis as above was performed for the

ropane/propylene separation. The composition of the product andecovery lines as well as compressor and heat exchanger dutiesere investigated for all HMD configurations at different mem-

rane module cut rate values.Fig. 15 shows the influence of membrane module cut rate on the

ow rate of C3H6 in the product and recovery lines (streams 109 and02, respectively) for the top configuration. The flow rate increasesith increasing membrane module cut rate. The overhead of theistillation column is the feed of the membrane. The flow rate of3H6 depends only on the membrane selectivity which could beiven as an explanation for the linear relation between the C3H6ow rate and membrane module cut rate.

As shown in Fig. 16, the flow rate of C3H6 in the product linestream 206) increases whereas C3H6 flow rate in the recovery line

stream 203) decreases sharply with membrane module cut rateor bottom HMD configuration. The bottoms of the distillation units fed to the membrane of which separation efficiency is measuredy the amount of C3H6 recovered and recycled to the distillation

Fig. 17. The flow rate of C3H6 in the product and recovery lines for different mem-brane module cut rate values of parallel configuration.

column through the permeate of the membrane (stream 204). Asthe membrane module cut rate increases, higher flow rate of C3H6in the permeate is obtained, which results in the decrease of C3H6flow rate in the recovery line by 98.5%. The permeate stream 204 isobviously recycled to a stage in C3-splitter where the compositionis the same as that of permeate.

Fig. 17 exhibits the effect of membrane module cut rate on theflow rate of C3H6 in the product and recovery lines (streams 308and 309 respectively) for the parallel configuration. It is clearlyindicated that the membrane module cut rate does not affect sig-

Page 11: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

M. Benali, B. Aydin / Separation and Purification Technology 73 (2010) 377–390 387

Fcc

ntrrwo

tafliHtvCoimwpmtflup

fFrItf

Fig. 19. Membrane module cut rate versus separation factor for top, parallel andtop-bottom configurations.

Fig. 20. Effect of membrane module cut rate on the compressor power and heatexchanger duty for top configuration.

pressor power and heat exchanger duty for different HMD

ig. 18. Effect of first membrane module cut rate (a) and second membrane moduleut rate (b) on the flow rate of C3H6 in the product and recovery lines for top-bottomonfiguration.

ificantly the amount of C3H6 in the product and recovery lines:he increase of the membrane module cut rate from 0.49 to 0.78esults in only an increase of 0.34% and 0.87% for the product andecovery lines respectively. The minimum number of stages occurshen the membrane feed composition is close to the composition

f C3-splitter feed stage.Fig. 18a and b depicts the effect of membrane module cut rate on

he flow rate of C3H6 in the product and recovery lines (streams 410nd 403 respectively) for the top-bottom HMD configuration. Theow rate in the product and recovery lines increases with increas-

ng membrane module cut rate for the first membrane (Fig. 18a).owever the flow rate of C3H6 in the recovery line is very low and

he increase is inconsequential. C3H6 is sent to the product lineia the permeate stream of the first membrane (stream 407). Thus,3H6 flow rate in this permeate increases with the module cut ratef the first membrane. The flow rate of C3H6 in the product linencreases and it decreases in the recovery line with increasingly

embrane module cut rate for the second membrane (Fig. 18b),hich favours the recycle of C3H6 to the distillation column. Theermeate stream 404; whose composition is proportional to theembrane module cut rate, is fed to the 2nd stage of the distilla-

ion column. As most C3H6 is sent to the column via permeate, theow rate of C3H6 in the recovery line decreases as membrane mod-le cut rate increases to maximize the yield of the polymer graderopylene.

The effect of the membrane module cut rate on the separationactor for top, parallel and top-bottom configurations is shown inig. 19. The separation factor values for the top-bottom configu-

ation belongs to the membrane placed at the top of the column.t should be indicated that the similar separation factor values forop and parallel configurations does not result from similar moleraction values of C2H4 and C2H6 in the permeate and feed. The sep-

Fig. 21. Compressor power vs. membrane module cut rate for bottom configuration.

aration factor values for the top-bottom configuration are higherthan that for the top configuration. This confirms the separationefficiency of the membrane placed at the top of the column for thetop-bottom configuration.

The effect of membrane module cut rate on the actual com-

configurations is illustrated in Figs. 20–23. For the top config-uration, the actual compressor power and heat exchanger dutyaugment with membrane module cut rate (Fig. 20). Higher mem-brane module cut rate results in a higher C3H6 concentration in the

Page 12: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

388 M. Benali, B. Aydin / Separation and Purifica

Fig. 22. Effect of membrane module cut rate on the total actual power and heatexchanger duty for parallel configuration.

Fa

ptpmid

trttieidof

ccarmOeCbt

ig. 23. Effect of first membrane module cut rate on the actual compressor powernd heat exchanger duty for top-bottom configuration.

ermeate. The increase in the total actual compressor power is dueo increase in the actual power of the compressor located in theroduct line (equipment 105). For an increase of the membraneodule cut rate from 0.54 to 0.85 the compressor actual power

ncreases by 35.21% whereas the heat exchanger duty increases,ue to higher C3H6 flow rate in the permeate (stream 106), by 15.6%.

As seen in Fig. 21, the actual compressor power increases withhe increase in membrane module cut rate for the bottom configu-ation. Fig. 22 shows the effect of membrane module cut rate on theotal actual power of the compressor and heat exchanger duty forhe parallel configuration. The actual power decreases by 79.58%ncreasing membrane module cut rate in contrast with the heatxchanger duty which increases by 36.0%. The substantial decreasen the total actual power is mainly explained by the tremendousecrease of 88.7% in the actual power of the compressor 305 locatedn the retentate line as the membrane module cut rate increasesrom 0.49 to 0.78.

The effect of the membrane module cut rate on the total actualompressor power and heat exchanger duty for the top-bottomonfiguration is shown in Fig. 23. The actual compressor powernd heat exchanger duty increase with the membrane module cutate for the first membrane. The flow rate of C3H6 in the per-eate increases with increasing first membrane module cut rate.

bviously, higher compression power (equipment 405) and heatxchanger duty (equipment 406) is required for higher flow rate of3H6. The power and duty values are increasing function of mem-rane module cut rate for the second membrane. The permeate ofhe second membrane is composed of C3H6. As all C3H6 coming

tion Technology 73 (2010) 377–390

from the bottoms of the column should be recycled to the col-umn via the permeate of the second membrane; higher membranemodule cut rate relatively higher flow rate of C3H6 requires highercompressor power and heat exchanger duty.

3.1.3. Economic analysisFor ethane/ethylene separation with the membrane cascade

system, ethylene purity of 0.9999 is obtained and the cost saving isfound to be 54.41%. For the series configuration a purity of 0.9979and a cost saving of 61.45% are found (Table 4). The saving is for-mulated as the ratio of total investment cost difference between ahybrid configuration and a conventional column to the total invest-ment cost of a conventional column. The expression for the totalinvestment cost is given below:Total investment cost = Capital cost + Operating cost + Utility cost (22)

where

Capital cost = 1.82 × Installed cost (23)

Operating cost = 0.20 × Installed cost (24)

The installed cost was calculated with CHEMCAD by introducingJanuary 2009 Chemical Engineering Plant Cost Index (CEPCI). Thecost estimation calculation is based on the area of the equipment.A rigorous costing routine was defined by the Parser programminglanguage for each equipment separately. Capital and operatingcosts were calculated based on the definitions given by Douglas[23]. The utility cost consists of electricity and steam prices for eachseparation system. The cost of ethylene and propylene consideredin this work are 1.65 and 1.85 $ kg−1 (2008-basis), respectively. Theprices of these two products were taken from the ICIS (Interna-tional Chemical Information Service) database [24]. The membranemodule cost was taken as 215 $ m−2 including all related elements.The membrane replacement cost was assumed to be 95 $ m−2 for alifetime of 3 years.

The comparison of the HMD systems with the conventional C2-splitter is given in detail in Table 4. As mentioned above, the mainobjective of the HMD system is to obtain the desired product athigh purity and low cost. For the series configuration, the purity ofethylene is 0.9979 and the cost savings is 61.45%. The membranemodule cut rates are of 0.93 and 0.01 for M1 and M2, respectively.For the membrane cascade, the purity of ethylene is enhanced to0.9999 with membrane module cut rates of 0.75 and 0.90 for M1and M2, respectively. Even though the total cost savings favoursthe series configuration, the higher purity of ethylene is obtainedwith the membrane cascade, showing a cost saving of 54.41%. Thus,for a grass root design, the membrane cascade configuration can beconsidered as the best alternative configuration to the conventionalC2-splitter. On the other hand, it cannot be recommended for aretrofit of an existing plant since its integration will be costly. One ofthe disadvantages of series and membrane cascade configurationsis, however, that more compression duty is required for ethyleneproduct. In comparison to a conventional C2-splitter, the utility costis 2 times and 4 times higher for membrane cascade and the seriesconfigurations respectively.

For propane/propylene separation, different HMD configura-tions were compared with two columns in series. For a given HMDconfiguration, the effect of the number of stages on the productpurity and the capital cost was analyzed. High propylene purity wasobtained for the simulation of a column with 90 and 100 stages. Thedifference in the propylene purity between 90 and 100 is negligi-ble, and 90 was selected as the optimum number of stages for all

HMD systems. The top configuration results in a propylene purityof 0.9996 and a saving of 29.84% (Table 5). For the top-bottom con-figuration, propylene purity and saving were found to be 0.9985and 36.57%, respectively. For bottom and parallel configurations,the propylene purity is only 0.924. The top-bottom HMD system
Page 13: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

M.Benali,B.A

ydin/Separation

andPurification

Technology73 (2010) 377–390

389

Table 4Ethane/ethylene separation: cost comparison.

Configuration Ethylene purity Membrane area (m2) Separation factor Capital cost (US$) ×106 Operating cost (US$) ×106 Utility cost (US$) ×106 Total savings (%)

Series 0.9979 M1: 8150 M2: 890 M1: 1.92 M2: 99.92 7.88 0.87 1.86 61.45%Membrane cascade 0.9999 M1: 8150 M2: 3243 M1: 998.98M2: 1.80 10.48 1.15 0.92 54.41%Conventional C2-splitter 0.9950 – – 24.4 2.7 0.5 –

Table 5Propane/propylene separation: cost comparison for retrofit purpose.

Configuration Distillation column Propylene purity Membrane area (m2) Separation Factor Capital cost (US$) ×106 Operating cost (US$) ×106 Utility cost (US$) ×106 Total investmentcost savings (%)

Diameter (m) Number of trays

Top 5.30 90 0.9996 182 4.92 26.01 2.86 3.41 29.84Bottom 5.30 90 0.9245 790 – 25.98 2.86 3.35 30.03Top-bottom 5.30 90 0.9985 Top: 346Bottom: 527 19.70 23.11 2.54 3.53 36.57

Parallel 5.30 90 0.9239 2086 4.93 26.32 2.89 3.52 28.84152 0.9948 3652 – 32.58 3.58 4.79 10.98

Conventional C3-splitter 5.30 205 0.9815 – – 37.8 4.2 4.00 –

Page 14: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

3 urifica

itamgsbtsC

isctbwtbohi[c

sopps1csuuc

4

ntiareTcnmbtoCsocscp

[

[

[

[

[

[

[

[

[

[

[

[

[[23] J.M. Douglas, Conceptual Design of Chemical Processes, McGraw-Hill, USA,

90 M. Benali, B. Aydin / Separation and P

s the optimum hybrid system for propane/propylene separationo obtain propylene at high purity with a remarkable cost savingt optimal surface areas of 346 and 527 m2 for top and bottomembranes, respectively. In case of designing a new plant inte-

rating top-bottom HMD system, the internal diameter of the 90tages distillation column can be reduced from 5.30 to 3.94 com-ined with two membranes having the same surface areas. Based onhese specifications, the estimated savings are 53.64%, which makesuch HMD very competitive than two-stage distillation system for3H6/C3H8 separation/purification.

As seen in Table 5 parallel configuration is the optimum designn terms of the total investment cost saving for propane/propyleneeparation. However, the product purity is low for this suggestedonfiguration. For higher propylene purity (≥0.99), a hybrid sys-em with a distillation column of more than 90 trays, shoulde used. Similar configuration was presented by Kookos [25] inhich a distillation column having a diameter of 3.30 m and 250

rays including the condenser and reboiler combined with a mem-rane of 2245 m2 was suggested to obtain a propylene purityf 0.99. A 17.1% of total cost savings were obtained with theybrid system in comparison to a single column of 248 trays

ncluding the condenser and reboiler and 3.86 m in diameter25]. The major savings were resulted from the cut in steamost.

In the present work a comparison was done between a hybridystem with a distillation column of 90 trays and a FT membranef 2086 m2, and a hybrid system with a column of 152 trays forarallel configuration with a FT membrane of 3652 m2 (Table 5). Aropylene purity of 0.9948 was obtained by using a parallel hybridystem with a column of 152 trays yielding total cost savings of0.98% in comparison to conventional C3-splitter composed of twoolumns having 115 and 90 stages. The product purity and total costaving for top-bottom configuration is close to the parallel config-ration involving a distillation column of 152 trays. However, thetility cost is increased by 26.3% in comparison to the top-bottomonfiguration.

. Concluding remarks

For high product purity and incremented both energy and eco-omic savings, the HMD system is proposed as an alternative designo replace current distillation technology. FT membrane technologys clearly a favourable option for the intensification of C2H4/C2H6nd C3H6/C3H8 separation/purification. Different hybrid configu-ations were compared with conventional C2- and C3-splitters forthane/ethylene and propane/propylene separation, respectively.o obtain high product purity at relatively low cost, the membraneascade system is proposed as the optimum configuration for aew ethane/ethylene separation plant. This is achieved at increasedembrane requirements corresponding to higher the utility cost

y 45.6% in comparison with conventional C2-splitter. However,he separation of C2H4 from a multicomponent mixture composedf CH4 (1.80%), C2H4 (75.66%), C2H6 (12.61%), C3H8 (1.34%), and3H6 (8.58%), can be efficiently achieved in the membrane cascadecheme providing ultra-high purity (99.99%) as well as high yield

f C2H4. Such two-stage membrane cascade integrated with threeompressors can provide a competitive solution than the multi-tage distillation of such a multicomponent mixture. HMD seriesonfiguration is found to be suitable for the retrofit of an ethylenelant. The top-bottom HMD system is proposed as the preferred

[

[

tion Technology 73 (2010) 377–390

design for propane/propylene separation from propylene purityand yield point of view even though the total savings are slightlyfavourable for parallel HMD system

Acknowledgements

The authors acknowledge the financial support provided by theProgram of Energy Research and Development of Natural ResourcesCanada. The authors note their gratitude to Chemstations for thediscussions with their technical support.

References

[1] G.E. Keller, R.R. Beebe, R.J. Fruehan, N.N. Li, E.L. Menger, G.P. Pez, P.H. Pfromm,R.W. Rousseau, M.P. Thomas, Separation Technologies for the Industries of theFuture, National Academy Press, USA, 1999, p. 128.

[2] T.K. Ghosh, H.D. Lin, A.L. Hines, Hybrid adsorption–distillation process for sep-arating propane and propylene, Ind. Eng. Chem. Res. 32 (1993) 2390–2399.

[3] J.P. Schmal, H.J. van der Kooi, A. de Rijke, Z. Olujic, P.J. Jansens, Internal versusexternal heat integration: operational and economic analysis, Chem. Eng. Res.Des. 84 (2006) 374–380.

[4] S. Moganti, R.D. Noble, C.A. Koval, Analysis of a membrane/distillation columnhybrid process, J. Membr. Sci. 93 (1994) 31–34.

[5] T. Pettersen, A. Argo, R.D. Noble, C.A. Koval, Design of combined membrane anddistillation processes, Sep. Technol. 6 (1996) 175–187.

[6] I.K. Kookos, Optimal design of membrane/distillation column hybrid processes,Ind. Eng. Chem. Res. 42 (2003) 1731–1733.

[7] T.G. Pressly, K.M. Ng, A break-even analysis of distillation–membrane hybrids,AIChE J. 44 (2004) 93–105.

[8] D.E. Suk, T. Matsuura, Membrane-based hybrid processes: a review, Sep. Sci.Technol. 41 (2006) 595–626.

[9] R.J. Valus, R. Eshraghi, A.E. Velikoff, J.C. Davis, High pressure facilitated mem-branes for selective separation and process for the use thereof, US Patent5,057,641 (1991).

10] I. Pinnau, L.G. Toy, C. Casillas, Olefin separation membrane and process, USPatent 5,670,051 (1997).

11] J.H. Kim, Y.S. Kang, J. Wong, Silver polymer electrolyte membranes for facili-tated olefin transport: carrier properties, transport mechanism and separationperformance, Macromol. Res. 12 (2004) 145–155.

12] S.W. Kang, J.H. Kim, K.S. Oh, J. Won, K. Char, H.S. Kim, Y.S. Kang, Highly stabilizedsilver polymer electrolytes and their application to facilitated olefin transportmembranes, J. Membr. Sci. 236 (2004) 163–169.

13] D.E. Gottschlich, D.L. Roberts, Energy minimization of separation processesusing conventional/membrane hybrid systems, in: DOE Final Report, 1990, p.157.

14] L.G. Wua, J.N. Shenb, H.L. Chenb, C.J. Gaoa, CO2 facilitated transport throughan acrylamide and maleic anhydride copolymer membrane, Desalination 193(2006) 313–320.

15] R. Agrawal, Membrane cascade schemes for multicomponent gas separation,Ind. Eng. Chem. Res. 35 (1996) 3607–3617.

16] K.N. Trueblood, H.J. Lucas, Coordination of silver ion with unsaturated com-pounds. V. Ethylene and propene, J. Am. Chem. Soc. 74 (1952) 1338–1339.

17] M. Teramoto, N. Takeuchi, T. Maki, H. Matsuyama, Ethylene/ethane separa-tion by facilitated transport membrane accompanied by permeation of aqueoussilver nitrate solution, Sep. Purif. Technol. 28 (2002) 117–124.

18] M. Teramoto, S. Shimizu, H. Matsuyama, N. Matsumiya, Ethylene/ethane sep-aration and concentration by hollow fiber facilitated transport membranemodule with permeation of silver nitrate solution, Sep. Purif. Technol. 44 (2005)19–29.

19] W.G. Whitman, The two-film theory of gas-absorption, Chem. Met. Eng. 29(1923) 146–148.

20] M. Terramoto, H. Matsuyama, T. Yamashiro, Y. Katayama, Separation of ethy-lene from ethane by supported liquid membranes containing silver nitrate asa carrier, J. Chem. Eng. Jpn. 19 (1986) 419–424.

21] R.D. Hughes, J.A. Mahoney, E.F. Steigelmann, Olefin separation by facilitatedtransport membranes, in: N.N. Li, J.M. Calo (Eds.), Recent Developments inSeparation Science, vol. IX, CRC Press, Cleveland, USA, 1982.

22] CHEMCAD V6, Chemstations Inc., Houston, USA, 2008.

1988, p. 601.24] ICIS, Chemical Prices and Chemical Industry Trends, 2008, www.icispricing.

com.25] I.K. Kookos, Optimal design of membrane/distillation column hybrid processes,

Ind. Eng. Chem. Res. 42 (2003) 1731–1738.

Page 15: Ethane/ethylene and propane/propylene separation in hybrid ...download.xuebalib.com/a6jBNov4l0c.pdf · The hybrid mem-brane processes can be designed in two different configurations:

本文献由“学霸图书馆-文献云下载”收集自网络,仅供学习交流使用。

学霸图书馆(www.xuebalib.com)是一个“整合众多图书馆数据库资源,

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研,提供最强文献下载服务。

图书馆导航:

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具