Membrane Technology for Hydrogen Separation in Ethylene Plants

A. A. Al-Rabiah, K. D. Timmerhaus, and R. D. Noble

Department of Chemical Engineering University of Colorado Boulder, Colorado 80309, USA

The use of membranes for hydrogen separation has been applied commercially in recent years. Several new membrane materials, which are now commercially available, are seriously being considered for the separation of hydrogen both effectively and efficiently within the ethylene process. This study investigates the feasibility of hydrogen separation from the cracked gas in an optimal manner before entering the lowtemperature section of the ethylene plant. This separation would consequently reduce the refrigeration load as well as the equipment size of the cold-box section. Polyimide membrane materials are very selective for hydrogen transport as compared to other hydrocarbons, such as methane and ethylene. Polysulfone has also proven to be selective for the separation of hydrogen from hydrocarbons. This study evaluates two new commercial polyimide membranes as well as a new polysulfone membrane and determines the feasibility of hydrogen separation before entering the low-temperature section of the ethylene plant. The performance of the membranes and their effects on the overall ethylene process are also presented.

INTRODUCTION A quick review of the chemical literature indicates that ethylene is one of the most important as well as one of the largest volume petrochemicals in the world today and serves as a key building block in the petrochemical industry. Conventional ethylene production involves the cracking of a hydrocarbon feed to form a mixture of hydrogen, methane, ethylene, ethane and heavier components that are separated by expensive cooling and distillation processes. The realization that the separation and purification process steps in the ethylene production consume more than 70% of the total energy required, provides a strong motivation for evaluating the impact of new technology on this part of the production process. Membrane technology has many advantages over other conventional technologies. These include lower capital and operating costs, low maintenance cost and the ease of installation and operation [1]. Membrane technology for hydrogen separation from other gases was successfully applied in the last few years to recover hydrogen from tail gases in oil refineries. The first-large commercial application for membranebased hydrogen systems was the separation of hydrogen from nitrogen in ammonia plants. Membrane technology is also used commercially for hydrogen/carbon monoxide (synthesis gas) ratio adjustment [2]. The membrane-based hydrogen separation in the past was based on low selective materials such as cellulose acetate polymers. However, new polymer membranes with improved selectivity and flux rate are now available in the market. For example, polyimides (Ube, Praxair), brominated polysulfone (Permea) are new selective membranes [3]. The selective membranes provide a major opportunity to improve the economics for different gas separation applications. This study investigates the use of membrane-based technology for hydrogen separation in a commercial ethylene process. The goal has been to study the feasibility of separating the hydrogen from the cracked gas in an optimal manner before it enters the cryogenic section, to further decrease the refrigeration load in the latter unit. Different membrane materials available from recent studies are evaluated for this separation to obtain the maximum selectivity and flux performance. PROCESS DESCRIPTION The ethylene plant used in this study is a state-of-the-art design of the propane cracking process with an annual production of 500,000 metric tons of 99.9% polymer grade ethylene. The conventional ethylene process employs a front-end deethanization sequence with a front-end acetylene reactor. The refrigerants

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001
-141 oC 65 psia

Hydrogen Rich Gas Tail Gas Cold-Box

Expander

Tail Gas
-121 oC 15 oC 514 psia C3 Ref -46 oC C2 Ref -73 oC C2 Ref -101 oC -96 oC -71 oC -43 oC

C2 Ref -101 oC

C2 Splitter Feed
-4 oC 506 psia

Cracked Gas
C3 Ref 10 oC

Demethanizer Figure 1: Chilling and demethanization section of the front-end deethanizer ethylene process. used in this process are ethylene and propylene. Hydrogen-rich gas is separated from other hydrocarbon components in the chilling and demethanization section as shown in Figure 1. After leaving the front-end acetylene reactor, the cracked gas is compressed in the last compressor stage, cooled with the overhead of the deethanizer and then sent to the cold-box. The latter consists of four plate-fin exchangers to cool the cracked gas with the expanded off-gas streams. Propylene and ethylene refrigerants are also used to chill the cracked gas in three shell-and-tube heat exchangers before sending the gas to the demethanizer column. Condensates, which are removed from the cracked gas stream after each stage of refrigeration, are separately fed to the high-pressure demethanizer at appropriate stage locations. Most of the hydrogen component is separated in the last flash drum. The hydrogen-rich stream from the last drum contains about 62.8 mol% H2, 36.8 mol% CH4, and 0.4 mol% (C2H4 + CO). This stream leaves the reflux drum at -121 oC and passes through the platefin exchangers of the cold-box to chill the incoming cracked gas. The hydrogen-rich gas is usually credited as a high H2 content by-product. The composition and operating conditions of the cracked gas stream are shown in Table 1. The cracked gas feed temperature and pressure are 15 oC and 35.44 bar (514 psia), respectively. The cracked gas feed pressure is optimal for the membrane-based gas separation process.

Table 1: Composition and operating conditions of the cracked gas stream. Temperature (oC) Pressure (bar) Vapor mole fraction Total flow rate (mol/sec) Volumetric flow rate (scfm) Composition (vol%) Hydrogen, H2 Methane, CH4 Ethylene, C2H4 Ethane, C2H6 Carbon monoxide, CO 15 35.44 1 1768 83965 19.25 39.65 34.86 6.24 Trace

MEMBRANE MATERIALS Hydrogen has a relatively small molecular size compared to other gases and exhibits high selectivity ratios in glassy polymers [4]. The ideal membrane possesses a high flux as well as a high selectivity. However, there

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001 is a trade-off between selectivity and flux rate. The degree of separation is highly dependent on both membrane selectivity and permeability. For our application, the separation goals require selection of membranes that provide high selectivity for hydrogen transport over ethylene to decrease the ethylene loss in the permeate stream. On the other hand, the membrane should also have a relatively high flux rate for hydrogen transport to lower both the capital and operating costs of the process. Additionally, the membrane must be stable at the operating conditions of the process. The fluxes and selectivities of hydrogen are provided for three new commercial membranes in Table 2 [2,5]. Table 2: Hydrogen separation membranes. Selectivity H2/CH4 H2/C2H4 250 125 33 200 250 35 Hydrogen Permeance (GPU)* 100 500 100

Membrane Type Polyimide A Polyimide B-H Polysulfone

H2/CO 100 56 1.7

H2/C2H6 1000 590 50

cm3 (STP) m3 (STP) *1 GPU = 10-6 = 7.501 10-12 (cm2 s cm Hg) (m2 s Pa)

The first polyimide membrane (polyimide A) has higher selectivities for hydrogen over CO, CH4 and C2H6 compared to the second polyimide membrane (polyimide B-H). The latter, however, has better selectivity for hydrogen over ethylene ( H2 / C2 H4 = 250 ). On the other hand, the hydrogen permeance of the

polyimide B-H membrane is five times higher than that for the polyimide A membrane. This has the advantage of a high flux membrane but at the same time the ethylene flux is higher compared to the first polyimide membrane. The polysulfone provides no advantages over the two polyimide membranes. The three membranes can operate over wide pressure and temperature ranges. The polyimide membrane of Ube can operate for a feed gas pressure of up to 40 barg with a maximum temperature of 150 oC and from 40 to 150 barg feed pressure with a maximum temperature of 100 oC [5]. The lifetime of the polyimide membrane is in the range of seven to ten years.
PROCESS DESIGN CONSIDERATIONS The feed to the membrane unit must be maintained at a temperature similar to that of the experimental data. In general, a low feed temperature increases the selectivity, but the flux rate is also decreased. The cracked gas which leaves the front-end acetylene reactor is cooled from 94 oC to 60 oC using cooling water before being sent to the membrane unit. On the other hand, the retentate stream (cracked gas with low hydrogen content) that leaves the membrane unit is cooled against the overhead of the deethanizer before sending this stream to the cold-box. In this way, the modified design does not require a steam heat exchanger and the cracked gas can be sent to the cold-box at a temperature of 15 oC. The recovery and purity of the ethylene product are the most important specifications by which we can make an equivalent comparison between the membrane-based hydrogen technology and the conventional cooling and distillation system. Usually in the ethylene industry, the overall ethylene recovery loss is minimized to less than 1% in order to achieve an economical process. The goal of this study is to design a membrane system such that the hydrogen recovery using the membrane-based technology is equivalent to the hydrogen recovery obtained from the conventional process. At the same time, the ethylene loss in the hydrogen rich gas (permeate stream) must be minimized to that level of ethylene loss obtained in the conventional process. To obtain the complete recovery of hydrogen with a minimum loss of ethylene, a cascade membrane system is required. The operating temperature is maintained at the same temperature as that used in obtaining the experimental data. Furthermore, the hydrogen permeate stream is maintained at a pressure of 3.45 to 6.89 barg (50-100 psig) for export and further processing. Credit for the hydrogen rich gas product is based on the concentration. MEMBRANE MODEL The three membranes used in this study are nonporous polymer membranes. The transport of gases through nonporous polymer membranes can be described by the solution-diffusion mechanism. This transport is a concentration gradient driven process, which is generally well described by Fick's first law

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001

dci (1) dx where J i is the flux of species i through the membrane, Di is the diffusion coefficient of species i in the membrane and dci / dx is the local concentration gradient of species i at a given position in the membrane. Equation 1 can be expressed in terms of the partial pressure of species i by using Henrys law as given by ci = Si Pi (2) where Si is the Henry's law solubility coefficient of species i and Pi is the partial pressure of species i outside the membrane. A more common way of measuring the flux of different species through the membrane is by applying the permeability coefficient P i = Di Si (3) The permeability coefficient, P i , expresses the overall degree of permeation regardless of the actual transport mechanism. Integrating Equation 1 with the assumption of constant permeability coefficient yields J i = Di ( Ph xi Pl yi ) (4) where is the membrane thickness, Ph and Pl are the pressures on the feed side and the permeate side, respectively, xi and yi are the mole fractions of species i on the feed and permeate side, respectively. The ability of a membrane to separate two gases i and j is obtained from the ratio of their permeabilities. This ratio is defined as the membrane selectivity ij

J i = Pi

ij =

Pi DS = i i P j DjS j

(5)

The performance of the membrane module can be described by a number of flow patterns: complete mixing, counter-current flow, co-current flow, and cross flow. For gas separation, counter-current and cross flow are normally used [6]. In addition, different module types are available. However, tubular hollow-fiber and spiral-wound modules are commercially used for gas separation applications. The counter-current flow pattern shown in Figure 2 best describes the flow in a hollow-fiber module. The latter is normally used in applications involving hydrogen separations [2]. The multicomponent transport through a membrane involves material balances along with a pressure drop calculation. The resulting model in differential forms can be solved by various computational algorithms such as approximate solution [7], finite elements [8] and collocation method [9]. The latter technique used in this study converts the boundary-value differential equations into a set of algebraic equations that can be solved by Newton's procedure.

Feed

Retentate

Permeate

Figure 2: Schematic diagram of counter-current membrane module.

MEMBRANE PERFORMANCE The performance of polyimide and polysulfone membranes has been investigated for the separation of hydrogen from the cracked gas in an ethylene plant. The hydrogen product leaves the membrane process as the lower-pressure permeate stream. The feed pressure and compositions are fixed in all calculations. The simulations were accomplished to evaluate the design variables such as permeate pressure on the performance of the membrane systems.

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001
There is a trade-off between the purity and the recovery of hydrogen. At the same time, there is a tradeoff between the purity and recovery of hydrogen and the permeate pressure. Figure 3 shows the effect of the permeate pressure on the recovery and purity of hydrogen on the membrane permeate side for the polyimide A membrane. A decrease in permeate pressure increases both the recovery and purity of hydrogen. However, the compression cost increases when the permeate pressure is decreased. Figure 3 also shows that the mole fraction of hydrogen increases on the permeate side is associated with a decrease in hydrogen recovery. Concentration profiles of gases on the permeate side are presented in Figure 4 as a function of the hydrogen recovery. This figure shows that the ethylene mole fraction increases with an increase in the hydrogen recovery. In fact, the ethylene mole fraction could increase to 0.2 in the permeate stream when complete hydrogen recovery is under consideration. Figure 5 shows the hydrogen recovery as a function of the relative membrane area and ethylene loss for a permeate pressure of 3.5 bar. Ethylene loss is defined as the ratio of ethylene in the permeate side to the ethylene in the feed side. Figure 5 reemphasizes the point that the ethylene loss increases as the hydrogen recovery is increased. The performance of the polyimide B-H membrane is illustrated in Figure 6. Equivalent permeate pressures were used for this figure to provide a comparison with the performance of the polyimide A membrane shown in Figure 3. Polyimide B-H membrane provides a high hydrogen recovery but with a low hydrogen concentration in the permeate when compared to polyimide A. Note that the feed operating conditions are identical for both membranes. Figure 7 shows a comparison between polyimide A and polyimide B-H with respect to relative membrane area and hydrogen purity on the permeate side as a function of the hydrogen recovery for a permeate pressure of 3.5 bar. Polyimide B-H membrane can provide a high hydrogen recovery using a relatively small membrane area compared to the polyimide A membrane. The relatively small area of polyimide B-H is a result of the high flux rate of the polyimide B-H membrane (see Table 2). On the other hand, polyimide A can provide a higher purity hydrogen compared to polyimide B-H but requires a larger membrane area. This suggests that the polyimide B-H membrane is the better choice for high hydrogen recovery since it requires a smaller membrane area. However, polyimide A is the choice for separations requiring high hydrogen purity. The membrane area is a function of several design variables including the permeate pressure and the hydrogen feed concentration. The required membrane area decreases with an increase in the hydrogen feed concentration as well as a decrease in the permeate pressure. Note that the comparison shown in Figure 7 was performed for a hydrogen feed concentration of 19.25 mol% and a permeate pressure of 3.5 bar. The performance of the polysulfone membrane does not compare to the performance of the polyimide membranes. The low flux as well as the low selectivities indicate that the polysulfone membrane is a poor choice for this particular hydrogen separation.
HYDROGEN SEPARATION SCHEME The goal of this hydrogen separation process has been to construct a unique scheme that provided an equivalent hydrogen recovery compared to the conventional scheme with a minimum ethylene loss. The new separation scheme, however, should provide a reduction in the overall cost involving both capital and operating expenses. Figure 8 represents an innovative scheme for the separation of hydrogen in an ethylene plant using a twostage membrane system with recycling. This scheme uses the polyimide B-H membrane in the first separation stage and polyimide A membrane in the second stage. The polyimide B-H is used for bulk separation in the first stage since it requires only a small membrane area for high hydrogen recovery. Polyimide A is used in the second stage for a sharp separation to produce a high purity hydrogen with minor ethylene losses. The second stage does not require a large membrane area since the hydrogen feed is already at a higher concentration. Hydrogen is recovered from the cracked gas in the first stage and then cooled using cooling water before being sent to the compressor. The hydrogen composition in the permeate stream from the first stage is 61.5 mol%. This permeate stream is compressed to 35.44 bar, cooled in a second heat exchanger and sent to the

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001

120

100

Hydrogen recovery ( % )

80

60

P F = 3 5 .4 4 b a r P p = 2 .0 b a r P p = 2 .5 b a r P p = 3 .0 b a r P p = 3 .5 b a r P p = 4 .0 b a r P p = 5 .0 b a r P p = 6 .0 b a r

40

20

0 0 .5 5 0 .6 0 0 .6 5 0 .7 0 0 .7 5 0 .8 0 0 .8 5 0 .9 0 0 .9 5 1 .0 0

H y d ro g e n m o le fra c tio n

Figure 3: Impact of the permeate pressure on the recovery and purity of hydrogen using polyimide A membrane.
1 .0 0 .9 0 .8 0 .7 0 .6 0 .5 0 .4 0 .3 0 .2 0 .1

H2

Permeate mole fractions

P F = 3 5 .4 4 b a r P p = 3 .5 b a r

C 2H 4 CH4 C 2H 6

0 .0 -0 .1 30 40 50 60 70

CO
80 90 100

H y d ro g e n re c o v e ry ( % )

Figure 4 Concentration profiles of gases as a function of the hydrogen recovery in the permeate side of the membrane using polyimide A membrane. second stage provided with polyimide A membrane. Hydrogen rich gas is produced in the second stage with a composition of 98 mol% and a recovery which is the same as that for the conventional ethylene process. The ethylene composition in the permeate stream is about 0.7 mol%, providing a minimum ethylene loss. The operating conditions and compositions of streams shown in Figure 8 are presented in Table 3.

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001

1.0 0.9 0.8 0.7

1.0

Ethylene loss Relative membrane area

PF = 35.44 bar, Pp = 3.5 bar

0.9

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 30 40 50 60 70 80 90 100

0.6 0.5 0.4 0.3 0.2 0.1 0.0

Hydrogen recovery (%)

Figure 5: Ethylene loss and relative membrane area as a function of the hydrogen recovery using polyimide A membrane.

110

100

Hydrogen recovery (%)

90
Pf = 35.5 bar

80

Pp = 2.0 bar Pp = 2.5 bar Pp = 3.0 bar Pp = 3.5 bar

70

Pp = 4.0 bar Pp = 5.0 bar Pp = 6.0 bar

60 0.55 0.60 0.65 0.70 0.75 0.80

Hydrogen mole fraction

Figure 6: Impact of the permeate pressure on the recovery and purity of hydrogen using polyimide B-H membrane. For this configuration, the cracked gas volume in the retentate stream is decreased by about 18% compared to the conventional process. The refrigeration loads in the chilling and demethanization sections of ethylene plant have been reduced. For example, the ethylene-cooled exchanger which required ethylene refrigerant at -101 oC was eliminated with the new membrane scheme. In addition, a reduction of 57% in the condenser duty for the demethanizer column was obtained when the number of stages remained the same as

Relative membrane area

0.8

Ethylene loss

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001

1.00 0.95 0.90 PF = 35.44 bar, Pp = 3.5 bar

1.0 0.9

H2 mole fraction

0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 30 40 50 60 70 80 90 100 H2 mole fraction [polyimide A] H2 mole fraction [polyimide B-H] Relative area [polyimide A] Relative area [polyimide B-H]

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

H2 recovery (%)

Figure 7: Comparison between polyimide A and polyimide B-H in relative membrane area and hydrogen purity in the permeate side as a function of hydrogen recovery for a permeate pressure of 3.5 bar. those used in the conventional process. The compression duty also dropped significantly to about 6,800 hp in the ethylene refrigeration system and to nearly 10,700 hp in the propylene refrigeration system. On the other hand, the new membrane cascade compressor requires an additional 4,970 hp. Nevertheless, the net power saving in the entire plant is about 12,500 hp.
PROCESS ECONOMICS This study has evaluated the new membrane separation scheme for a new grass-roots ethylene plant as well as for an existing plant. The assumptions used for the economic evaluations are listed in Table 4. The membrane capital cost is based on the use of the hollow-fiber module for both stages and assuming the same unit cost per square meter for both of the polyimide membranes. Although the lifetime of the membrane is more than seven years, a membrane replacement was assumed after five years. The new grass-roots ethylene plant involves a significant reduction in the capital costs as well as in the operating costs when the membrane scheme is implemented. The demethanizer column with the new membrane scheme was fixed number to the conventional process since reduction in condenser refrigeration has a greater effect on the entire process. Consequently, the size of the equipment for the refrigeration systems was reduced along with the steam requirement. The ethylene process with the two-stage membrane system shows an annual processing savings of about 9,242,000 US dollars and an incremental return on investment of about 26 %. The average annual pre-tax return on investment, ROI, of the new modified ethylene process is 43.9%. The average annual after-tax return on investment of this new process is 28.9%.

Table 3: Operating conditions of the two-stage membrane system with recycling shown in Figure 8.
Stream Pressure-bar Flow rate-mol/sec Component mole fractions Hydrogen Methane Ethylene Ethane 1 2 3 4 35.44 6.00 35.44 35.44 1767.53 557.96 1445.80 557.96 5 5.45 321.74 6 35.44 236.23

0.192 0.397 0.349 0.062

0.615 0.257 0.119 0.009

0.017 0.482 0.425 0.076

0.615 0.257 0.119 0.009

0.980 0.013 0.007 0.000

Relative membrane area

0.8

0.118 0.589 0.271 0.022

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001
H2 Gas Polyimide-B-H Membrane Feed
60 oC 35.4 bar CW

Polyimide-A Membrane
o CW 60 C 35.4 bar

33 oC 5.45 bar CW

33 oC 6 bar

Compressor 6 to Cold-Box 3 Non-Permeate Gas

Figure 8: Two-stage membrane system with recycling for hydrogen production from cracked gas in an ethylene plant. The two-stage membrane cascade scheme would save about 0.855 / lb of ethylene produced. In revamping an existing ethylene plant, the equipment of the chilling and demethanization section and the equipment associated with the refrigeration and steam systems were not changed. The revamped ethylene process with the two-stage membrane system gives a net annual savings of about 2,899,000 US dollars per year and requires a capital investment of about 13,521,000 US dollars. For the highest membrane unit cost, as shown in Table 4, the payout period is about 3.1 years. The payout period is decreased to 1.84 years when the average membrane unit cost is reduced to 21 US $/ m2.
CONCLUSIONS Economic evaluation of the cascade membrane systems shows that the two-stage cascade with recycling is a more profitable scheme when compared to the conventional designs. Implementation of this membrane technology in ethylene plants for improving the hydrogen separations shows considerable potential in either conventional grass-root ethylene plants or retrofits of existing ethylene plants. The savings in operating and capital costs for ethylene plants utilizing the proposed membrane systems are annually more than 9 million US dollars when the higher cost membranes are included in the economic analysis.

Table 4: Economic parameters and assumptions used in the evaluation of the membrane-based hydrogen separation system.
FIXED CAPITAL INVESTMENT (FCI) (US $) Membrane module cost $ 50 / m2 Membrane replacement cost $ 25 / m2 Life of membrane 5 years Life of other associated equipment 10 years PRODUCTION COST (PC) (US $) Fuel gas 3.04 $/M ft3 (2.94 $/MMBTU) Electricity 4.3 /kwh Cooling water 8.50 /Mgal

Compressor efficiency

0.85 (polytropic)

Maintenance Taxes and insurance

4% of FCI / year 2% of FCI / year

References

6th World Congress of Chemical Engineering Melbourne, Australia 23-27 September 2001
1- Sengupta, A., & Sirkar, K.K. (1995), Analysis and Design of Membrane Permeators for Gas Separation, Membrane Separation Technology- Principles and Applications, Noble, R.D., & Stern, S.A., eds., Elsevier Science, Amsterdam, 499 2- Zolandz, R.R., & Fleming, G.K. (1992), Gas Permeation, Membrane Handbook, Ho, W.S.W., & Sirkar, K.K., eds., Van Nostrand Reinhold, New York, 78 3- Baker, R.W., (2000), Membrane Technology and Applications, McGraw-Hill, New York, 311 4- Mulder, M. (1996), Basic Principles of Membrane Technology, Kluwer Academic Publishers, Dordrecht, the Netherlands, 324 5- Ube Industries Ltd., (1996), Ube Gas Separation System by Polyimide Membrane, Product Brochure 6- Gottschlich, D.L., (1989), Economic Comparison of Several Membrane Configurations for H2/N2 Separation, Gas Separation & Purification, 3(12), 170-179 7- Peterson, T., & Lien, K.M., (1994), A New Robust Design Model for Gas Separating Membrane Modules, Based on an Analogy with Countercurrent Heat Exchangers, Computers chem. Engng., 18(5), 427-439 8- Coker, D.T., et al., (1998), Modeling Multicomponent Gas Separation Using Hollow-Fiber Membrane Contactors, AIChE J., 44(6), 1289-1302 9- Tessendorf, S., et al., (1999), Modeling, Simulation and Optimization of Membrane-Based Gas Separation Systems, Chemical Engineering Science, 54, 943-955