H2 RECOVERY PROCESSES COMPARED

T.R. Tomlinson, A. J. Finn
Costain Engineering Ltd.
Manchester, England

Choosing between cryogenic and membrane technologies for the recovery of hydrogen from industrial offgases is influenced by such factors as feed pressure, product pressure, feed composition, plant capacity, and operating costs.

A look at three examples in which both technologies have been used illustrates the need for a thorough understanding of the engineering principles involved.

The first examines purge gas recovery on ammonia plants where both technologies have extensive track records. The second discusses hydrogen recovery from a refinery offgas, and the third looks at hydrogen recovery utilizing "free" pressure energy.

HIGH PERMEATION RATE

The separation of hydrogen is particularly suited to membrane technology as this gas has a very high permeation rate relative to other gases. A comparison of separation processes for the recovery of hydrogen should highlight the factors that favor membranes for any gas separation.

The available technology for gas separation and purification includes a number of competing processes: cryogenics (with or without distillation), absorption, pressure and thermal swing adsorption, and metal hydrides.

During the last 10 years, the use of membranes has become established and competes favorably in many areas. Membrane processes are well placed in gas sweetening, dewpoint control, air separation, and helium purification as well as hydrogen recovery.

A comparison of conventional gas-separation technologies is given in Table 1.

PRINCIPLES-MEMBRANES

Synthetic membranes are made from a variety of polymers, including polyethylene, polyamides, polyimides, cellulose acetate, polysulfone, and polydimethylsiloxan. Generally the membrane is manufactured as either a flat film or a hollow fiber.

There are two basic types of modules in which membranes are contained: hollow fiber bundles in a metal module and spirally wound bundles in a metal module.

For gas separation, the majority of membranes are still made with hollow fiber but with increasing use of spirally wound types.

Hollow-fiber membranes, though less robust, have a larger area/volume ratio, Typically, spirally wound modules have a packing density of 1,500 sq m/cu m surface compared with about 9,000 sq m/cu m for hollow-fiber membranes.

Usually, membranes are made up of composite materials to give them good mechanical and selectivity properties. In such cases, the mechanism of permeation becomes more complex, relying on solution diffusion and diffusion through pores due to a pressure gradient.

To calculate performance of the membrane, an equation can be drawn up assuming Fickian diffusion of a dissolved gas in a dense membrane:

For each component:

[SEE FORMULA]

Transport across the membrance can be assumed to consist of solution of molecules at the high partial pressure surface, diffusion across the membrane due to the pressure gradient, and desorption at the low partial pressure surface.

Thus the permeability of each component is a function of solubility and diffusion coefficients. Clearly it is important to pack as much area as possible into each bundle of hollow fibers. The fibers themselves are, therefore, very narrow in bore, typically, about 0.2-0.4 mm diameter.

The thinner the membrane the higher the gas flux but the weaker the membrane itself.

Designers of membranes seek a happy medium for a given application.

Different gases have widely differing permeabilities, and different membrane construction or formulation can yield very different flux rates.

Typical permeation rates of various gases through different types of membranes are given in Table 2.

The relative basis in this table is unity for 02 on a typical polysulfone membrane.

From the table it can be seen that low-molecular-weight gases and strongly polar gases have high relative permeabilities. These are often called "fast gases." In contrast, "slow gases" have higher molecular weight and symmetrical molecules.

Separation of gases is achieved by the difference in the rates at which different gases permeate through the membrane.

The membrane allows fast gases, such as hydrogen, to be separated from slow gases, such as methane. The driving force for the permeation of the fast gas (and hence the separation of the fast gas from the other slower components) is the difference in partial pressure from one side of the membrane to the other.

Hence, for recovery of hydrogen the product stream must be at a substantially lower pressure than the feed stream.

PRINCIPLES-CRYOGENICS

The separation of the gas molecules relies on the relative volatilities of the components at low temperatures. The volatilities depend on the operating pressure but are usually very high for hydrogen separation.

Typical relative volatilities of various gases at 90 K. and 70 bar are given in Table 3.

Separation is achieved when the feed gas is cooled and partially condensed. The vapor forms the hydrogen product, and the separated liquid is expanded and evaporated to become waste or fuel gas.

The driving force for the process is the temperature drop due to the Joule-Thomson effect when the condensed liquid is reduced in pressure. This results in the hydrogen product being available at essentially feed-gas pressure.

AMMONIA-PLANT PURGE GAS

In the conventional ammonia synthesis loop, it is necessary to provide a means of preventing buildup of impurities.

This is achieved by a continuous purge of a portion of hydrogen-rich recycle gas which keeps the concentration of argon and methane down to an acceptable level. The purge gas is usually available at a pressure of more than 140 bar.

The hydrogen product would be expected to return at the maximum pressure possible to conserve energy of recompression while the reject gases would leave at a lower pressure for use in the plant's fuel system.

The overall mass balances for a cryogenic unit and membrane system are shown in Table 4. Detailed comparisons of the benefits of the competing processes have been carried out.

CRYOGENIC UNIT

A simplified process flow-sheet with operating pressures is shown in Fig. 1a.

Because energy of a gas is a valuable form of energy, it is surprising that the feed gas is let down from 140 to 70 bar across a control valve. In practice, however, mechanical limitations prevent the use of aluminum plate-fin exchangers at much greater than 70 bar, and thus the purge gas must be let down to an intermediate pressure.

This expansion is the first and only major energy loss in the process.

The designer does have alternatives that can reduce this loss, but they all involve considerable expenditure. Development of new, higher pressure exchangers represents the best solution to further improvements.2

The cryogenic section of the process consists of a single exchanger-separator system. Refrigeration for the process is achieved by reheating of the hydrogen and fuel streams.

A positive temperature difference is achieved by expansion of the liquid fuel from 70 bar to 4 bar with the Joule-Thomson effect producing a fall in temperature from expansion. This gives the necessary driving force between cooling and warming streams.

Expansion of a liquid across a control valVE! in this way is more efficient than expansion of a gas stream. A single-stage separation gives good hydrogen product purity and recovery due to the high volatility of hydrogen with respect to the other components.

The cryogenic process when applied to hydrogen recovery from ammonia purge gas offers the following advantages:

Product gas from the cold box is available at essentially the same pressure as feed gas to the cold box.
With high rejection of inerts, purge rate is kept low.
No external energy is required to achieve the desired separation.
The process is more competitive at higher capacities because of economies of scale.

The main disadvantages of the cryogenic process are the following:

Because of the mechanical limitations of cryogenic exchangers, pressure energy is lost across a control valve and is not usefully employed in the process.
This mechanical limitation results in significant recompression power being needed to return the hydrogen product at synthesis loop pressure.

MEMBRANE UNIT

Since its introduction, the membrane process has also been extensively used for hydrogen recovery from ammonia-plant purge gas particularly for purge gas from small to medium sized ammonia plants, (i.e., less than 1,000 tons/day of ammonia).

A major reason for its popularity is that the feed gas is available at high pressure which is clearly desirable for any membrane process.

A process flowsheet of a typical membrane unit is shown in Fig. 1 b.

The membrane process relies on a large pressure difference between the hydrogen product and the feed gas to achieve separation. Basically, hydrogen permeates the membrane preferentially to any of the other components which leads to a hydrogen-rich product.

A two-stage separation is utilized to achieve a high recovery of hydrogen.

This results, however, in a significant portion of hydrogen product only being available at low pressure at 91 bar (0.645 x 140) and the product hydrogen partial pressures at 63 bar (0.9 x 70) and 22.5 bar (0.9 x 25).

The separation is achieved by a reduction of the pressure of the product stream while the fuel-gas stream is produced at essentially feed-gas pressure before being let down across a control valve into the fuel system.

As has already been mentioned, gas energy is a very valuable form of energy, and using gas energy to achieve separation results in a relatively energy-intensive process.

The membrane process when applied to hydrogen recovery from ammonia purge gas offers the following adVantages:

It is simple.
The membranes are supplied as several modules. This results in low initial investment for smaller throughput plants due to the linear relationship between throughput and investment.
It has a short delivery time.

The main disadvantages of the membrane process are the following:

Because of the inefficient nature of the process, i.e., the major loss of high-pressure gas energy, there is a considerable energy requirement for hydrogen-product compression.
Because there are reduced economies of scale, higher capacities lead to a less competitive process.

REFINERY OFFGAS

Both membrane and cryogenic technologies can be considered where the hydrogen product is required at medium-to-high pressures.

This is often a requirement in refineries, an example being hydrogen recovery from catalytic reformer offgas. The application of cryogenic technology for these types of separations has been reviewed .3

A comparison of the operating conditions of both membrane and cryogenic processes for a residfiner project is shown in Fig. 2. The main difference is the lower product pressure for the membrane process which results in a higher overall utilities consumption. The overall mass balance and a comparison of the utilities consumption is given in Table 5.

"FREE-PRESSURE" ENERGY

It is apparent that hydrogen recovery utilizing hydrogen partial pressure as the driving force is energy intensive. Therefore, in cases where free-pressure energy is available between feed and hydrogen product streams, it could be considered that a membrane unit would be more competitive than a cryogenic unit.

Fig. 3 shows a typical process flowsheet for a cryogenic unit where free-pressure energy is available. This energy is utilized in a process-gas turboexpander which work-expands hydrogen rich gas, thereby producing "free" refrigeration.

This type of unit can achieve up to 98% hydrogen purity. A mass balance is shown in Table 6. Both the turboexpander and the letdown of the fuel liquid (to produce Joule-Thomson effect) contribute to the refrigeration requirement and the work of separation making a high-purity product possible.

The hydrogen partial pressure of the feed gas is 30 bar (0.748 x 41), while the product partial pressure is 26 bar (0.978 x 27). This pressure difference would not provide adequate driving force for a membrane unit. Hence, achieving the same separation would require the hydrogen product to be obtained from the membrane unit at much lower pressure than 27 bar. Consequently considerable energy input would be required for product recompression.

As the selectivity of membranes increases and the development of thinner binders improves permeation rate, membranes should become more dominant in certain areas of gas separation.

Applications that are most likely to suit a membrane have one or more of the following characteristics:

When a combination of high purity and high recovery is not required.
When the feed gas contains a high concentration of the product gas.
When the flowrates are relatively small.
When the feed gas pressure is high, and either the "fast" gas product is required at low pressure or the "slow" gas product is required at high pressure.
When the selectivity of the membrane can be matched with the desired duty.
When the gas does not attack or alter the performance of the membrane.

REFERENCES

Isalski, W.H., "25 Years of Purge Gas Recovery," Nitrogen No. 152, November-December 1984.
Finn, A.J., "Cryogenic Purge Gas Recovery Boosts Ammonia Plant Productivity and Efficiency," Nitrogen, No. 175, September-October 1988.
Banks, R., and Isalski, W.H., "Excess Fuel Gas? Recover H2/LPG," Hydrocarbon Processing, October 1987.