INTRODUCTION

As hydrogen use has become more widespread in refineries, hydrogen production has moved
from the status of a high-technology specialty operation to an integral feature of most
refineries. This has been made necessary by the increase in hydrotreating and hydro- cracking,
including the treatment of progressively heavier feedstocks.

Steam reforming is the dominant method for hydrogen production. This is usually com- bined
with pressure-swing adsorption (PSA) to purify the hydrogen to greater than 99 vol %. As
hydrogen production grows, a better understanding of the capabilities and require- ments of
the modern hydrogen plant becomes ever more useful to the refiner. This will help the refiner
get the most from existing or planned units and make the best use of hydro-

gen supplies in the refinery.

USES OF HYDROGEN

Overview

There has been a continual increase in refinery hydrogen demand over the last several
decades. This is a result of two outside forces acting on the refining industry: environ- mental
regulations and feedstock shortages. These are driving the refining industry to con- vert from
distillation to conversion of petroleum. Changes in product slate, particularly outside the
United States, are also important. Refiners are left with an oversupply of heavy, high-sulfur oil,
and in order to make lighter, cleaner, and more salable products, they need to add hydrogen
or reject carbon.

Within this trend there are many individual factors depending on location, complexity of the
refinery, etc.

Hydrogen Demand

The early use of hydrogen was in naphtha hydrotreating, as feed pretreatment for catalyt- ic
reforming (which in turn was producing hydrogen as a by-product). As environmental
regulations tightened, the technology matured and heavier streams were hydrotreated. These
included light and heavy distillates and even vacuum residue.
Hydrotreating has also been used to saturate olefins and make more stable products. For
example, the liquids from a coker generally require hydrotreating, to prevent the for- mation
of polymers.

At the same time that demand for cleaner distillates has increased, the demand for heavy fuel
oil has dropped. This has led to wider use of hydrocracking, which causes a fur- ther large
increase in the demand for hydrogen.

HYDROGEN PRODUCTION

Hydrogen has historically been produced in catalytic reforming, as a by-product of the pro-
duction of the high-octane aromatic compounds used in gasoline. Changes in this process have
had a large impact on the refinery hydrogen balance.

As reforming has changed from fixed-bed to cyclic to continuous regeneration, pres- sures
have dropped and hydrogen production per barrel of reformate has increased. Recent changes
in gasoline composition, due to environmental concerns, have tended to reduce hydrogen
production, however. Besides limits on aromatics, requirements for oxygenates in gasoline
have resulted in reduced reforming severity, as the high-octane oxygenates have displaced
reformate from the gasoline pool. The only safe statement is that the situ- ation will continue
to change.

Where by-product hydrogen production has not been adequate, hydrogen has been
manufactured by steam reforming. In some cases partial oxidation has been used, particularly
where heavy oil is available at low cost. However, oxygen is then required, and the capital cost
for the oxygen plant makes partial oxidation high in capital cost.

Figure 6.1.1 shows a typical modern hydrogen plant. This unit produces 82 million SCFD (at
60°F and 14.7 lb/in2 absolute) [92,000 (N) m3/h (N represents normal tempera- ture and
pressure at 0°C and 1.0332 kg/cm2 absolute)] of hydrogen from natural gas for a Far Eastern
refinery, at a purity of 99.9 vol %. The Foster Wheeler Terrace Wall* steam reforming furnace
is visible in the background, with the 12 absorbers and two surge drums of the pressure-swing
adsorption unit in the foreground.

Chemistry

Steam Reforming. In steam reforming, light hydrocarbons such as methane are reacted with
steam to form hydrogen
The reaction is typically carried out at approximately 1600°F (870°C) over a nickel cata- lyst
packed into the tubes of a reforming furnace. Because of the high temperature, hydro-
carbons also undergo a complex series of cracking reactions, plus the reaction of carbon with
steam. These can be summarized as

Carbon is produced on the catalyst at the same time that hydrocarbon is reformed to hydro-
gen and CO. With natural gas or similar feedstock, reforming predominates and the car- bon
can be removed by reaction with steam as fast as it is formed. When heavier feedstocks are
used, the carbon is not removed fast enough and builds up. Carbon can also be formed where
the reforming reaction does not keep pace with heat input, and a hot spot is formed. To avoid
carbon buildup, alkali materials, usually some form of potash, are added to the catalyst when
heavy feeds are to be used. These promote the carbon-steam reaction and help keep the
catalyst clean. The reforming furnace is also designed to produce uniform

heat input to the catalyst tubes, to avoid coking from local hot spots.

Even with promoted catalyst, cracking of the feedstock limits the process to hydrocar- bons
with a boiling point of 350°F (180°C) or less: natural gas, propane, butane, and light naphtha.
Heavier hydrocarbons result in coke building up on the catalyst. Prereforming, which uses an
adiabatic catalyst bed operating at a lower temperature, can be used as a pre- treatment to
allow slightly heavier feeds to be used without coking. A prereformer will also make the fired
reformer more tolerant of variations in heat input.

After reforming, the CO in the gas is reacted with steam to form additional hydrogen, in the
water-gas shift reaction

This leaves a mixture consisting primarily of hydrogen and CO2. After CO2 removal, which we
shall discuss later, many plants use methanation—the reverse of reforming—to remove the
remaining traces of carbon oxides:

Partial Oxidation. Hydrogen can also be produced by partial oxidation of hydrocarbons:

The shift reaction also participates so the result is a mixture of CO and CO2 in addition to H2.
Temperature in partial oxidation is not limited by catalyst tube materials, so higher
temperature may be used, which results in reduced methane slippage.

Steam Reforming/Wet Scrubbing

Figure 6.1.2 shows the flow sheet for a wet scrubbing plant, based on steam reforming of
natural gas. Plants with similar configurations came into widespread use about 1960, when
high-pressure steam reforming became economical. They were built until the mid-1980s, when
they were generally supplanted by plants using PSA.

Feedstock at 450 lb/in2 (31 bar) gage is preheated and purified to remove traces of sul- fur and
halogens in order to protect the reformer catalyst. The most common impurity is H2S; this is
removed by reaction with ZnO. Organic sulfur may also be present; in this case recycled
product hydrogen is mixed with the feed and reacted over a hydrogenation cata- lyst
(generally cobalt/molybdenum) to convert the organic sulfur to H2S. If chlorides are present,
they are also hydrogenated and then reacted with a chloride adsorbent.

The feed is then mixed with steam, preheated further, and reacted over nickel catalyst in the
tubes of the reformer to produce synthesis gas—an equilibrium mixture of H2, CO, and CO2.
The steam/carbon ratio is a key parameter, since high steam levels aid in methane conversion.
Residual methane in the synthesis gas will pass through the plant unchanged (along with any
N2 in the feed). This will reduce the hydrogen purity so it is important to ensure a low
methane leakage. A high steam/carbon ratio and high reforming temperatures are used for
this reason. Excess steam is also used to prevent coke formation on the cata- lyst. Typical
reformer outlet conditions for hydrogen production are 1600°F and 300 lb/in2 (870°C and 21
bar) gage.

Much of the design and operation of hydrogen plants involves protecting the reforming
catalyst and the catalyst tubes. The extreme temperatures and the sensitivity of the catalyst
tend to magnify small upsets. Minor variations in feed composition or operating conditions can
have significant effects on the life of the catalyst or the reformer itself. This is partic- ularly true
of changes in molecular weight of the feed gas, or poor distribution of heat to the catalyst
tubes.

The synthesis gas passes through the reformer waste heat exchanger, which cools the gas and
generates steam for use in the reformer; the surplus is exported. The cooled gas [still at about
650°F (345°C)] is reacted over a fixed bed of iron oxide catalyst in the high- temperature shift
converter, where the bulk of the CO is reacted, then cooled again and reacted over a bed of
copper zinc low-temperature shift catalyst to convert additional CO. The raw hydrogen stream
is next scrubbed with a solution of a weak base to remove CO2. The flow scheme in Fig. 6.1.2 is
based on use of a potassium carbonate solution in water to react with the CO2; a similar
process uses an ethanolamine solution. CO2 in the gas reacts reversibly with potassium
carbonate to form potassium bicarbonate. The solu- tion is depressured and steam-stripped to
release CO2, with the heat for the regenerator reboiler coming from the hot synthesis gas. The
regenerator overhead stream is then cooled to condense water. The CO2 is available for
recovery or can be vented.

The raw hydrogen leaving the CO2 removal section still contains approximately 0.5 per- cent
CO and 0.1 percent CO2 by volume. These will act as catalyst poisons to most hydrogen
consumers, so they must be removed, down to very low levels. This is done by methanation,
the reverse of reforming. As in reforming, a nickel catalyst is used, but as a fixed bed.

Typical final hydrogen purity is 97 vol %, with the remaining impurities consisting mainly of
methane and nitrogen. Carbon oxide content is less than 50 vol ppm.
Product hydrogen is delivered from the methanator at approximately 250 lb/in2 (17 bar) gage
and must generally be compressed before final use. This is done in a reciprocating compressor.
Centrifugal compressors are not feasible because of the low molecular weight; the pressure
rise per foot of head is too low, and too many stages would be required.

Steam Reforming/PSA

Plants built since the mid-1980s are generally based on steam reforming followed by pres-
sure-swing adsorption. PSA is a cyclic process which uses beds of solid adsorbent to remove
impurities from the gas. The hydrogen itself passes through the adsorbent beds with only a
tiny fraction absorbed. The beds are regenerated by depressurization, followed by purging at
low pressure.

When the beds are depressured, a waste gas (or “tail gas”) stream is produced, consist- ing of
the impurities from the feed (CO2, CO, CH4, N2) plus some hydrogen. This stream is burned in
the reformer as fuel. Reformer operating conditions in a PSA-based plant are set so that the
tail gas provides no more than about 85 percent of the reformer fuel. This limit is important for
good burner control because the tail gas is more difficult to burn than regular fuel gas. The
high CO2 content can make it difficult to produce a stable flame.

As the reformer operating temperature is increased, the reforming equilibrium shifts, resulting
in more hydrogen and less methane in the reformer outlet and hence less methane in the tail
gas. Actual operating conditions can be further optimized according to the rela- tive cost of
feed, fuel, and export steam.

The flow sheet for a typical PSA-based hydrogen plant is shown in Fig. 6.1.3. As in the wet
scrubbing process, the feed is purified and reformed, followed by shift conversion.

Only a single stage of shift conversion is used, since a very low CO residual is not required. Any
CO remaining in the raw hydrogen will be removed and recovered as reformer fuel. After
cooling, the gas is purified in the PSA unit.

The PSA unit is simpler to operate than a wet scrubbing system, since it has no rotat- ing
equipment or circulating solutions. In addition, the adsorbent will remove methane and
nitrogen, which could not be removed by the wet scrubbing process. Typical hydrogen
recoveries in a PSA unit are in the 80 to 90 percent range, with product purity generally

99.9 vol %.

Because of the loss of hydrogen to the PSA tail gas, the reformer and front end of a PSA plant
are larger than in a wet scrubbing plant. A PSA plant uses less process steam and does not
require heat for the reboiler; this leaves additional steam available for export. Capital cost is
generally lower for the design with PSA. The additional export steam can provide a strong
utility cost advantage for the PSA plant in addition to its purity and oper- ability advantages.
Product Properties

Hydrogen purity depends primarily on the purification method. This is illustrated in Table

6.1.3. In wet scrubbing, the major impurities are methane and nitrogen. Methane in the
product is the residual left after reforming, or is formed in the methanator from residual CO or
CO2. Nitrogen in the feed is carried through the plant unchanged, although there is a dilution
effect because of the larger volume of hydrogen compared to the feedstock.

In a PSA plant, most impurities can be removed to any desired level. Table 6.1.4 shows the
difficulty of removal of impurities. Removal of a more difficult impurity will generally ensure
virtually complete removal of easier impurities. Nitrogen is the most difficult to remove of the
common impurities, and removing it completely requires additional adsorbent. Since it acts
mainly as a diluent, it is usually left in the product. The exception occurs where the hydrogen is
to be used in a very high-pressure system such as a hydrocracker. In that case the extra cost
for nitrogen removal is justified by the savings in hydrogen purge losses.

In the case of a nitrogen-free feedstock such as liquefied petroleum gas (LPG) or naph- tha, a
purity of 99.99 percent can be readily achieved. In this case, carbon monoxide is the usual
limiting component. Since CO must be removed to ppm levels, the other impuri- ties—CO2 and
H2O—are removed to virtually undetectable levels. A typical residual of about 100 ppm CH4
remains because of inefficiencies in the purge system.

Operating Variables

Operating Conditions. The critical variables for steam reforming are temperature, pressure,
and the steam/hydrocarbon ratio. Picking the operating conditions for a particular plant
involves an economic trade-off among these three factors.

Steam reforming is an equilibrium reaction, and conversion of the hydrocarbon feed- stock is
favored by high temperature, which in turn carries a fuel penalty. Because of the volume
increase in the reaction, conversion is also favored by low pressure, which con- flicts with the
need to supply the hydrogen at high pressure. In practice, temperature and pressure are
limited by the tube materials.

Table 6.1.5 shows the effect of changes in temperature, pressure, and the steam/carbon ratio.
The degree of conversion is measured by the remaining methane in the reformer out- let,
known as the methane leakage.

Shift Conversion. In contrast to reforming, shift conversion is favored by low temperature. The
gas from the reformer is reacted over iron oxide catalyst at 600 to 700°F (315 to 370°C), with
the limit set by the low-temperature activity of the catalyst. In wet scrubbing plants using a
methanator, it is necessary to remove CO to a much lower level to avoid excessive
temperatures in the methanator. In those plants the gas is cooled again and reacted further
over a copper-based catalyst at 400 to 500°F (205 to 260°C). Table 6.1.6 shows the effect of
temperature and the steam/carbon ratio on the CO remaining after shift conversion.

Alternative Processes

Partial Oxidation. Partial oxidation (POX) reacts hydrocarbon feed with oxygen at high
temperatures to produce a mixture of hydrogen and carbon monoxide. Since the high
temperature takes the place of a catalyst, POX is not limited to the light, clean feedstocks
required for steam reforming. Partial oxidation is high in capital cost, and for light feeds it has
been generally replaced by steam reforming. However, for heavier feedstocks it remains the
only feasible method.

In the past, POX was considered for hydrogen production because of expected short- ages of
light feeds. It can also be attractive as a disposal method for heavy, high-sulfur streams, such
as asphalt or petroleum coke, which sometimes are difficult to dispose of.

Consuming all a refinery’s asphalt or coke by POX would produce more hydrogen than is likely
to be required. Because of this and the economies of scale required to make POX economic,
hydrogen may be more attractive if produced as a by-product, with electricity as the primary
product.

Figure 6.1.4 is a block flow diagram of a unit to produce electricity from asphalt, with hydrogen
as a by-product. Besides being high in carbon, the asphalt contains large amounts of sulfur,
nitrogen, nickel, and vanadium (Table 6.1.7). Much of the cost of the plant is associated with
dealing with these components.

The asphalt is first gasified with oxygen in an empty refractory-lined chamber to pro- duce a
mixture of CO, CO2, and H2. Because of the high temperature, methane production is minimal.
Gas leaving the gasifier is first quenched in water to remove solids, which include metals (as
ash) and soot. Metals are removed by settling and filtration, and the soot is recycled to the
gasifier. The gas is further cooled and H2S is removed by scrubbing with a selective solvent.
Sulfur removal is complicated by the fact that a significant amount of carbonyl sulfide (COS) is
formed in the gasifier. This must be hydrolyzed to H2S, or a sol- vent that can remove COS
must be used.

Hydrogen processing in this system depends on how much of the gas is to be recovered as
hydrogen and how much is to be used as fuel. Where hydrogen production is a relatively small
part of the total gas stream, a membrane unit can be used to withdraw a hydrogen- rich
stream. This is then purified in a PSA unit. In the case where maximum hydrogen is required,
the entire gas stream may be shifted to convert CO to H2, and a PSA unit used on the total
stream.

Catalytic Partial Oxidation. Also known as autothermal reforming, catalytic partial oxidation
reacts oxygen with a light feedstock, passing the resulting hot mixture over a reforming
catalyst. Since a catalyst is used, temperatures can be lower than in noncatalytic partial
oxidation, which reduces the oxygen demand.
Feedstock composition requirements are similar to those for steam reforming: light
hydrocarbons from refinery gas to naphtha may be used. The oxygen substitutes for much of
the steam in preventing coking, so a lower steam/carbon ratio can be used. Since a large
excess of steam is not required, catalytic POX produces more CO and less hydrogen than steam
reforming. Because of this it is suited to processes where CO is desired, for exam- ple, as
synthesis gas for chemical feedstocks. Partial oxidation requires an oxygen plant, which
increases costs. In hydrogen plants, it is therefore used mainly in special cases such as
debottlenecking steam reforming plants, or where oxygen is already available on-site.

By-product Recovery

Carbon dioxide and steam are the major by-products from hydrogen manufacture.

Carbon Dioxide. Where there is a market for CO2, recovery can be very attractive. Historically
the major use has been in the food industry, with recent growth being for injection in
enhanced oil recovery. A substantial amount of CO2 is available from hydrogen plants: a plant
making 10 million SCFD [11,000 (N) m3/h] of hydrogen from natural gas vents 2.5 million SCFD
or 145 tons/day (132 MT/day) of CO2.

Recovery of CO2 is easiest in older plants using wet scrubbing. These produce a con-

centrated CO2 stream which needs only final purification to remove traces of H2, CO, and CH4,
followed by compression.

More recent plants, using PSA, can use a vacuum-swing adsorption (VSA) system for CO2
recovery (Fig. 6.1.5). Tail gas from the PSA system is compressed and fed to the VSA system,
which uses a separate set of adsorber vessels. By using vacuum regeneration, the system can
split the tail gas into a CO2 product stream, a hydrogen-rich stream which is recycled to the
reformer, and a nitrogen-rich reject stream.

Besides recovering CO2, the VSA system increases overall hydrogen production by recovering
hydrogen which would otherwise have been lost in the tail gas.

A wet scrubbing system can also be installed upstream of a PSA unit to recover CO2. This can
also be used in a revamp to increase capacity by reducing the load on the PSA system.

Steam. Most hydrogen plants generate steam, mainly for use as process steam with the excess
available for export. A typical 50 million SCFD [56,000 (N) m3/h] unit based on PSA will export
between 70,000 and 160,000 lb/h (between 30 and 70 MT/h), depending on configuration.
Plants with air preheat are at the lower end of the steam production range, while steam export
can be further increased by adding auxiliary burners between the radiant and convection
sections.
Catalysts

Hydrogen plants are one of the most extensive users of catalysts in the refinery. Catalytic
operations include hydrogenation, steam reforming, shift conversion, and methanation.7

Sulfur and halogen removal are actually done by reaction with solid adsorbents, but they are
included here for completeness.

Reforming. Because of the high temperatures and heat load of the reforming reaction,
reforming catalyst is used inside the radiant tubes of a reforming furnace. The catalyst is
subject to severe operating conditions: up to 1600°F (870°C), with typical pressure drops of 40
lb/in2 (2.8 bar). To withstand these conditions, the carrier is generally an alumina ceramic,
although some older formulations use calcium aluminate.

The active agent in reforming catalyst is nickel, and normally the reaction is controlled by both
diffusion and heat transfer. The catalyst is therefore made in rings to provide increased mass
and heat transfer at minimum pressure drop. To further increase heat trans- fer, most catalyst
vendors now offer specially shaped catalyst.

Even with a high-strength carrier, catalyst life is limited as much by physical break- down as by
deactivation. Thermal cycling is especially hard on the catalyst: When the tubes are heated,
they expand and the catalyst tends to settle in the tube; then when the tube cools and the
tube contracts, the catalyst is crushed. This can cause voids to form in the tubes, leading to hot
spots and ultimately to ruptured tubes.

The main poisons are sulfur and chlorides, which are present in small quantities in most
feedstocks. Sulfur poisoning is theoretically reversible, and the catalyst can often be restored
to near full activity by steaming. However, in practice the deactivation may cause the catalyst
to overheat and coke, to the point that it must be replaced.

Chlorides are an irreversible poison: The chlorine combines with the nickel to form nickel
chloride, which is volatile. The nickel migrates and recrystallizes, reducing the cat- alyst
activity.

The catalyst is also sensitive to poisoning by heavy metals and arsenic, although these are
rarely present in feedstocks.

The catalyst is supplied as nickel in the oxide form. During start-up the catalyst is heat- ed in a
stream of inert gas, then steam. When the catalyst is near the normal operating temperature,
hydrogen or a light hydrocarbon is added to reduce the nickel oxide to metallic nickel.
Steaming the catalyst will oxidize the nickel, but most catalysts can readily be rere- duced.

Shift Conversion. The second important reaction in a steam reforming plant is the shift
conversion reaction

The equilibrium is dependent on temperature, with low temperatures favoring high con-
versions.
Two basic types of shift catalyst are used in steam reforming plants: iron/chrome high-
temperature shift catalysts and copper/zinc low-temperature shift catalysts.

High-Temperature Shift. High-temperature shift catalyst operates in the range of 600 to

800°F (315 to 430°C). It consists primarily of magnetite, Fe3O4, with chrome oxide, Cr2O3,
added as a stabilizer. The catalyst is supplied in the form of Fe2O3 and CrO3, and must be
reduced. This can be done by the hydrogen and carbon monoxide in the shift feed gas, and
occurs naturally as part of the start-up procedure.

If the steam/carbon ratio of the feed is too low, the reducing environment is too strong and
the catalyst can be reduced further, to metallic iron. This is a problem, since metallic iron will
catalyze Fischer-Tropsch reactions and form hydrocarbons. In older wet scrubbing plants this
was rarely a problem, since the steam/carbon ratio of the process gas was in the range of 5 to
6, too high for iron formation. In some newer plants with steam/carbon ratios below 3, the
shift catalyst is slowly converted to iron, with the result that significant amounts of
hydrocarbons are formed over the high-temperature shift catalyst.

To slow down (but not eliminate) overreduction, the catalyst can be doped with copper, which
acts by accelerating the conversion of CO. It increases activity at lower tempera- tures, but also
makes the catalyst sensitive to poisoning by sulfur and chlorides.

High-temperature shift catalyst is very durable. In its basic form it is not sensitive to most
poisons and has high mechanical strength. It is subject to thermal sintering, howev- er, and
once it has operated at a particular temperature, it loses its activity at lower tem- peratures.

Low-Temperature Shift. Low-temperature (LT) shift catalyst operates with a typical inlet
temperature of 400 to 450°F (205 to 230°C). Because of the lower temperature, the reaction
equilibrium is better and outlet CO is lower.

Low-temperature shift catalyst is economic primarily in wet scrubbing plants, which use a
methanator for final purification. The main advantage of the additional conversion is not the
extra hydrogen that is produced, but the lower CO residual. This reduces the tem- perature
rise (and hydrogen loss) across the methanator.

PSA-based plants generally do not use LT shift, since any unconverted CO will be recovered as
reformer fuel. Since an LT shift increases hydrogen production for a fixed reformer size, it may
be used in revamps to increase production.

Low-temperature shift catalyst is sensitive to poisoning by sulfur and chlorides. It is also

mechanically fragile and sensitive to liquid water, which can cause softening of the catalyst
followed by crusting or plugging.

The catalyst is supplied as copper oxide on a zinc oxide carrier, and the copper must be
reduced by heating it in a stream of inert gas with measured quantities of hydrogen. Reduction
is strongly exothermic and must be closely monitored.

Methanation. In wet scrubbing plants, final hydrogen purification is by methanation, which

converts CO and CO2 to CH4. The active agent is nickel, on an alumina carrier.
The catalyst has a long life, as it operates under clean conditions and is not exposed to poisons.
The main source of deactivation is plugging from carryover of CO2 removal solu- tions.

The most severe hazard is overtemperature, from high levels of CO or CO2. This can result
from breakdown of the CO2 removal equipment or from exchanger tube leaks which quench
the shift reaction. The results of breakthrough can be severe, since the methanation reaction
produces a temperature rise of 125°F per 1 percent of CO, or 60°F per 1 percent of CO2. While
the normal operating temperature in a methanator is approximately 600°F (315°C), it is
possible to reach 1300°F (700°C) in cases of major breakthrough.

Feed Purification. Long catalyst life in modern hydrogen plants is attributable to a great extent
to effective feed purification, particularly sulfur and chloride removal. A typical natural gas or
other light hydrocarbon feedstock contains traces of H2S and organic sulfur. Refinery gas may
contain organic chlorides from a catalytic reforming unit.

To remove these it is necessary to hydrogenate the feed to convert the organic sulfur to H2S,
which is then reacted with zinc oxide: organic chlorides are converted to HCl and reacted with
an alkali metal adsorbent. Purification is done at approximately 700°F (370°C), since this results
in best use of the zinc oxide as well as ensures complete hydro- genation.

Coking of Reforming Catalyst. Coking of the reformer catalyst is the most characteristic
problem in a hydrogen plant. While it may be similar in appearance to the fouling of heater
tubes found in other units, additional precautions are necessary here. A major reason for the
high reliability of modern units is the reduction in catalyst coking. This is due to advances in
catalyst technology and in reformer design.

While light, methane-rich streams such as natural gas or light refinery gas are the most
common feeds to hydrogen plants, there is often a requirement to process a variety of heav-
ier feedstocks, including LPG and naphtha, because of seasonal variations in feedstock price,
an interruptible supply of natural gas, or turnarounds in a gas-producing unit. Feedstock
variations may also be inadvertent, for example, changes in refinery offgas com- position from
another unit.

When heavier feedstocks are used in a hydrogen plant, the primary concern is coking of the
reformer catalyst. There will also generally be a small capacity reduction due to the additional
carbon in the feedstock and additional steam required. This increases the load on the shift and
CO2 removal section of the plant. The size of this effect will depend on the feedstocks used
and on the actual plant. Coking, however, is of more immediate concern, since it can prevent
the plant from operating.

Coking is most likely about one-third the way down the tube, where both temperature and
hydrocarbon content are high enough. In this region, hydrocarbons can crack and polymerize
faster than the coke is removed by reaction with steam or hydrogen. Once the catalyst is
deactivated, temperature increases further and coking accelerates. Farther down the tube,
where the hydrocarbon/hydrogen (HC/H2) ratio is lower, there is less risk of cok- ing. Coking
depends to a large extent on the balance between catalyst activity and heat input; more active
catalyst produces more hydrogen at lower temperature, reducing the risk of coking. Uniform
heat input is especially important in this region of the catalyst tube, since any catalyst voids or
variations in catalyst activity can produce localized hot spots, leading to coke formation or tube
failure.

Coke formation results in hot spots in catalyst tubes and can produce characteristic pat- terns
known as giraffe necking or tiger tailing. It increases pressure drop, reduces conver- sion of
methane, and can cause tube failure. Coking may be partially alleviated by increasing the
steam/hydrocarbon ratio to change the reaction conditions, but the most effective solution is
to replace the reformer catalyst with one designed for heavier feeds.

In addition to the reforming and shift reactions over reforming catalyst, a number of side
reactions occur. Most of these include the production or removal of carbon. Carbon is
continuously formed on the catalyst, but ordinarily reacts with steam faster than it can build
up. Heavier feeds produce carbon at a much higher rate. Unless the process condi- tions or the
catalyst is changed, the carbon can accumulate.

Standard methane reforming catalyst uses nickel on an alpha-alumina ceramic carrier. The
alumina is acidic, which promotes hydrocarbon cracking and can form coke with heav- ier
feeds. Some catalyst formulations use a magnesia/alumina spinel which is more neutral than
alpha-alumina. This reduces cracking on the carrier and allows somewhat heavier feed- stocks
to be used—typically into the LPG range. The drawbacks to this approach include dif- ficulty in
reducing the catalyst unless there is a supply of hydrogen in the reducing gas, and the possible
damage to the catalyst by hydration of the catalyst during start-up.

Further resistance to coking can be achieved by adding an alkali promoter, typically some form
of potash (KOH), to the catalyst. Besides reducing the acidity of the carrier, the promoter
catalyzes the reaction of steam and carbon. While carbon continues to be formed, it is
removed faster than it can build up. This approach can be used with naphtha feed- stocks up
to a boiling point of 350°F (about 180°C).

Under the conditions in a fired reformer, potash is volatile, and it is incorporated into the
catalyst as a more complex compound which slowly hydrolyzes to release KOH. The promoted
catalyst is used only in the top half of the catalyst tubes, since this is where the hydrocarbon
content, and the possibility of coking, is the highest. In addition, this keeps the potash out of
the hottest part of the tube, reducing potash migration.

Alkalized catalyst allows the use of a wide range of feedstocks, but it does have draw- backs. In
addition to possible potash migration, which can be minimized by proper design and
operation, the catalyst is somewhat less active than conventional catalyst.

Prereforming. Another option to reduce coking in steam reformers is to use a prereformer.

This uses a fixed bed of very active catalyst, operating at a lower temperature, upstream of the
fired reformer1 (Fig. 6.1.6). Inlet temperatures are selected so that there is minimal risk of
coking. Gas leaving the prereformer contains only steam, hydrogen, carbon oxides, and
methane. This allows a standard methane catalyst to be used in the fired reformer. This
approach has been used with feedstocks up to light kerosene. The drawback to this approach
is the need for a separate prereformer reactor and a more complicated preheat train.
Since the gas leaving the prereformer poses reduced risk of coking, this also makes the fired
reformer more “forgiving.” Variations in catalyst activity and heat flux in the primary reformer
become less critical.

Besides its use for feedstock flexibility, a prereformer can be used to reduce the fuel
consumption and steam production of the reformer. Since the prereformer outlet gas does not
contain heavier hydrocarbons, it can be reheated to a higher temperature than the orig- inal
feedstock without the risk of carbon formation. The higher preheat temperature reduces the
radiant duty and fuel consumption as well as steam production.

Reformer Design

Equipment Configuration. Designs for steam reforming furnaces must deal with the problems
caused by the extremely high process temperatures. These include thermal expansion,
cracking, and overheating. The high temperatures also mean the use of exotic alloys; as an
example, a common tube material is HP-45, which contains 25 percent chrome and 35 percent
nickel, with the element niobium added to stabilize the grain structure.

Tube expansion at reforming temperatures is approximately 10 in (250 mm) for a typ- ical 40-ft
(12-m) tube. This expansion is taken up at the cold end of the tube by connect- ing the tubes to
the inlet header with long, relatively flexible tubes called pigtails. A counterweight system is
used to support the tube and ensure that the tube is kept in con- stant tension to prevent
bowing.

The combination of light feedstock and the good thermal conductivity of hydrogen allow the
use of high flux rates, typically above 20,000 Btu/(h · t2) [63,000 W/m2]. This in turn requires
that heat flux be very uniform to avoid hot spots. In larger furnaces, firing is from both sides of
the tube, and measures are taken to ensure that heat flux is relatively uniform over the length
of the tube. This may be done by using a radiant wall design such as a terrace wall unit, or
positioning the flame next to the coldest part of the tube in down- fired units.

Since catalyst is packed into the tubes, many multiple passes are used to keep pressure drop to
a manageable level. There are several hundred parallel passes in a large furnace. Careful
packing of the catalyst into the tubes ensures even flow distribution.

Several reformer configurations have evolved to deal with these factors: Terrace Wall, side-
fired, down-fired, and bottom-fired furnaces are used. These designs are summarized below.

Terrace Wall. Foster Wheeler’s Terrace Wall reformer was developed to handle the high
temperatures and high heat fluxes used in steam reforming. This design uses a long, relatively
narrow firebox, with the tubes in a single row down the center (Figs.

6.1.7 to 6.1.9). The burners are located in terraces along the sides and fire upward against
sloping, refractory-lined walls. Generally two terraces are used. The hot refractory then
radiates heat to the tubes, resulting in a very uniform, controlled heat distribution. This helps
to avoid localized overheating and carbon laydown. The process gas flow is downward, and the
flue gas flow upward. The convection section is located above the radiant section. Larger
furnaces often use two radiant cells, located side by side and sharing a common convection
section.

The radiant wall design provides uniform heat flux and is resistant to localized overheat- ing,
even in the event of catalyst coking. The vertical stacking of the furnace, with the con- vection
section located above the radiant section, results in smaller plot area for most sizes.

The updraft arrangement minimizes power required for fans, and the furnace can be designed
to operate in natural draft, without fans.

Side-Fired. This design is similar to the Terrace Wall furnace, with burners located at multiple
levels (often six levels). Special burners are used to direct the flames back against the walls. It
is possible to get additional control of firing from the larger number of burners.

Down-Fired. This design uses burners located on the roof of the furnace, firing downward (Fig.
6.1.10). Multiple rows of tubes are used, alternating with rows of burners. Special burners are
used to ensure the proper flame pattern. This is required in order to get good heat distribution
along the length of the tube. Both process gas and flue gas flow is downward.

The multiple rows allow reduced cost for extremely large units, as is required in large
methanol or ammonia plants. The convection section is located at grade; this allows good fan
access and more stable fan mounting but increases the plot area required. Fewer (but larger)
burners are required.

Cylindrical. The furnace is in the shape of a vertical cylinder, with burners located in the center
of the floor. The tubes are arranged in a ring around the burners. Ample spacing between the
tubes allows radiation to be reflected from the furnace walls and to reach the backs of the
tubes, in order to provide good heat distribution. Both process gas and flue gas flow is upward.

This design is used for smaller units, with an upper plant size limit of 5 to 10 million SCFD [5500
to 11,000 (N) m3/h] of hydrogen with a single reformer. Since the hot end of the tubes is at
the top, the tubes can be anchored at the top and expand downward. The counterweights or
spring hangers used on larger units are not necessary, reducing the cost of the furnace. These
units are generally shop-fabricated. Size is therefore limited by shipping restrictions.

Plant Operation

Several operations are characteristic of hydrogen plants. They include loading catalyst into the
reformer tubes, measuring tube metal temperatures, and pinching off of catalyst tubes.

Catalyst Loading. The goal of catalyst loading is to fill the 40-ft-long (12-m-long) tubes
completely, without voids and without fracturing any of the catalyst rings. Early reformers
were loaded by filling the tubes with water and dumping the catalyst in. This was discontinued
after it was found that on start-up, water trapped inside the catalyst turned to steam and
fractured the rings.

Traditionally, loading has been done by first loading the catalyst into cloth tubes, known as
socks, and then lowering the socks into the tubes. By manipulating the rope, the catalyst is
dumped, falling only a few inches. This is a slow process, requiring vibration of the tubes to
eliminate voids and careful measurement of the tube pressure drop and vol- ume loaded into
each tube to ensure consistency.

Catalyst Tube Temperature Measurement. As hydrogen plant technology has matured,

competitive pressures have made it necessary to operate plants closer to their limits, including
temperature limits on the catalyst tubes. To avoid tube failures, many plants now monitor
tube metal temperatures each day, or even each shift. Optical(actually infrared) pyrometers
are used to measure catalyst tube temperatures, since thermocouples do not survive the 1700
to 1800°F (930 to 980°C) conditions.

Besides measuring metal temperatures, it is important to identify variations in temper- ature

which may indicate catalyst problems. Catalyst deactivation will raise the tube tem- perature,
as it becomes necessary to fire harder to reach the same conversion. Poisoning also often
causes variations in catalyst activity, leading to hot spots on the tubes and dis- tinctive
patterns, known as tiger tailing or giraffe necking. Catalyst breakup from thermal cycling can
also cause similar patterns, as well as hot tubes from plugging.

Whether one is measuring temperature or identifying patterns, accurate readings require a

clear view of the catalyst tubes, preferably from a direction perpendicular to the metal surface.
The Terrace Wall or side-fired furnaces provide an advantage in this case, since most furnaces
include viewing ports to allow measurement of temperature on virtu- ally all tubes. The
multiple tube rows common in down-fired furnaces require viewing from the end of the tube
rows, making accurate measurement difficult.

Tube Damage and Pinching. The life of the catalyst tubes depends to a large extent on the
condition of the catalyst, which in turn is subject to damage by poisoning or mechanical stress.
Poisoning is possible from sulfur or chlorides, in either the feedstock or low-quality steam,
while mechanical stress is usually from thermal cycling. The metal tubes have a higher
coefficient of thermal expansion than does the catalyst. As the tubes heat up, they expand and
the catalyst settles farther down the tube. When the tube cools, it contracts and the catalyst is
fractured. After a number of cycles, the catalyst can break up, plugging the tube or forming
voids.

Breakup from thermal cycling can be aggravated by high pressure drop in the catalyst tubes. A
smaller tube diameter can reduce furnace cost, since catalyst volume and tube weights are
reduced for a given tube surface area. However, pressure drop is increased at smaller
diameters, leading to greater stress on the catalyst. During process upsets it becomes easier to
exceed the crush strength of the catalyst, and the catalyst fractures.

As the condition of the catalyst worsens, hot spots can develop in tubes and the tube can
rupture. Shutting down the furnace at this point to repair the tubes would lead to lost
production as well as extra heating/cooling cycles. Individual tubes can be isolated on line to
seal them off and continue operation. This is done by pinching the pigtails shut with a hydraulic
clamp while the unit is operating. Many operators shut off hydrocarbon feed while this is
done, keeping steam flowing to the tubes.

The tubes themselves are also subject to damage from thermal cycling. As the tubes heat up,
the outside and hotter part of the tube wall expands more than the inner portion, leading to
high stress levels. The metal will creep in operation, normalizing the stress. The process is then
reversed when the tube cools. Continued cycling can lead to cracks.

INTEGRATION INTO THE MODERN REFINERY

Purification

A wide variety of processes are used to purify hydrogen streams.4,5 Since the streams are
available at a wide variety of compositions, flows, and pressures, the best method of purifi-
cation will vary.

Wet Scrubbing. Wet scrubbing systems, particularly amine or potassium carbonate systems,
are used for removal of acid gases such as H2S or CO2. Most depend on chemical reaction and
can be designed for a wide range of pressures and capacities. They were once widely used to
remove CO2 in steam reforming plants, but have generally been replaced by PSA units except
where CO2 is to be recovered. They are still used to remove H2S and CO2 in partial oxidation
plants.

Wet scrubbing systems remove only acid gases or heavy hydrocarbons, but not methane or
other light gases, hence have little influence on product purity. Therefore, wet scrubbing
systems are most often used as a pretreatment step, or where a hydrogen-rich stream is to be
desulfurized for use as fuel gas.

PSA. Pressure-swing adsorption uses beds of solid adsorbent to separate impure hydrogen
streams into a very pure high-pressure product stream and a low-pressure tail gas stream
containing the impurities and some of the hydrogen. The beds are then regenerated by
depressuring and purging (Figs. 6.1.11 and 6.1.12). Part of the hydrogen—typically 10 to 20
percent—is lost into the tail gas.

The cost of the system is relatively insensitive to capacity. This makes PSA more eco- nomic at
larger capacities, while membrane units tend to be favored for smaller plants.

PSA is generally the first choice for steam reforming plants because of its combination of high
purity, moderate cost, and ease of integration into the hydrogen plant. It is also often used for
purification of refinery offgases, where it competes with membrane systems.
Turndown is simple to about 30 percent of flow, where it is limited by the accuracy of flow
measurement. Systems can be designed to go somewhat lower by adding low-range
transmitters. Reliability is very high.

It is not generally economic to design a PSA unit to process both synthesis gas from steam
reforming and hydrogen/hydrocarbon gas. Doing so causes problems with both the fuel
balance and the adsorbents. Tail gas from the steam reforming unit consists largely of CO2 and
is returned at low pressure to the reformer furnace as fuel. The plant fuel balance requires
that the tail gas from the hydrocarbon PSA be compressed into the fuel system. Combining
these two units would result in too much fuel gas to supply the reformer fur- nace, with much
of the CO2 from the synthesis gas being compressed into the refinery fuel system. In addition,
the adsorbents for the two systems are different, and combining them would affect the
hydrogen recovery.

Membranes. Membrane units separate gases by taking advantage of the difference in rates of
diffusion through membranes. Gases which diffuse faster (including hydrogen) become the
permeate stream and are available at low pressure. The slower gases become the
nonpermeate and leave the unit at close to feed pressure.

Membrane units contain no moving parts or switch valves and have potentially very high
reliability. The major threat is from components in the gas (such as aromatics), which attack
the membranes, or from liquids, which plug them.

Membranes are fabricated in relatively small modules; for larger capacity more mod- ules are
added. Cost is therefore virtually linear with capacity, making them more com- petitive at
lower capacities.

Design of membrane systems involves a trade-off between pressure drop (or diffusion rate)
and surface area, as well as between product purity and recovery. As the surface area is
increased, the recovery of fast components increases; however, more of the slow com-
ponents are recovered, which lowers the purity. Operating them at turndown changes the
relationship between diffusion rate and surface area; modules may be taken out of service to
keep conditions constant.

Cryogenic Separation. Cryogenic separation units operate by cooling the gas and condensing
some of or all the gas stream. Depending on the product purity required, separation may be by
simple flashing or by distillation. Cryogenic units tend to be more expensive than other
processes, especially in smaller sizes. This is so partly because of the feed pretreatment
required to remove compounds which would freeze, such as water or CO2. They are therefore
used either in very large sizes or where they offer a unique advantage, such as the ability to
separate a variety of products from a single feed stream. One example is the separation of
light olefins from an FCC gas.

Hydrogen recovery is in the range of 95 percent, with purity above 98 percent obtain- able.
Once the material has been condensed, additional fractionation is relatively cheap.
Feedstocks

The best feedstocks for steam reforming are light, saturated, and low in sulfur; this includes
natural gas, refinery gas, LPG, and light naphtha. These feeds can be converted to hydrogen at
high thermal efficiency and low capital cost.

Natural Gas. Natural gas is the most common hydrogen plant feed, since it meets all the
requirements for reformer feed and is low in cost. A typical pipeline natural gas (Table 6.1.8)
contains over 90 percent C1 and C2, with only a few percent of C3 and heavier hydrocarbons.
It may contain traces of CO2, with often significant amounts of N2. The N2 will affect the purity
of the product hydrogen: It can be removed in the PSA unit if required, but at increased cost.

Purification of natural gas, before reforming, is usually relatively simple. Traces of sul- fur must
be removed to avoid poisoning the reformer catalyst, but the sulfur content is low and
generally consists of H2S plus some mercaptans. Zinc oxide, often in combination with
hydrogenation, is usually adequate.

Refinery Gas. Light refinery gas, containing a substantial amount of hydrogen, can be an
attractive steam reformer feedstock. Since it is produced as a by-product, it may be available
at low cost. Processing of refinery gas will depend on its composition, particularly the levels of
olefins and of propane and heavier hydrocarbons.

Olefins can cause problems by forming coke in the reformer. They are converted to sat- urated
compounds in the hydrogenator, giving off heat. This can be a problem if the olefin
concentration is higher than about 5 percent, since the hydrogenator will overheat. A recy- cle
system can be installed to cool the reactor, but this is expensive and wastes heat.

Heavier hydrocarbons in refinery gas can also form coke, either on the primary reformer
catalyst or in the preheater. If there is more than a few percent of C3 and higher compounds, a
promoted reformer catalyst should be considered, to avoid carbon deposits. When hydrogen
content is greater than 50 vol % and the gas is at adequate pressure, another option is
hydrogen recovery, using a membrane or pressure-swing adsorption unit.

The tail gas or reject gas, which will still contain a substantial amount of hydrogen, can then be
used as steam reformer feedstock.

Refinery gas from different sources varies in suitability as hydrogen plant feed. Catalytic
reformer offgas, as shown in Table 6.1.9, for example, is saturated and very low in sulfur and
often has a high hydrogen content. This makes excellent steam reformer feed- stock. It can
contain small amounts of chlorides. These will poison the reformer catalyst and must be
removed.
The unsaturated gas from an FCC or coker, on the other hand, is much less desirable. Besides
olefins, this gas contains substantial amounts of sulfur, which must be removed before the gas
is used as feedstock. These gases are also generally unsuitable for direct hydrogen recovery,
since the hydrogen content is usually too low.

Hydrotreater offgas lies in the middle of the range. It is saturated, so it is readily used as
hydrogen plant feed. The content of hydrogen and heavier hydrocarbons depends to a large
extent on the upstream pressure. Sulfur removal will generally be required.

The process scheme shown in Fig. 6.1.13 uses three different refinery gas streams to produce
hydrogen. First, high-pressure hydrocracker purge gas is purified in a membrane unit. Product
hydrogen from the membrane is available at medium pressure and is com- bined with
medium-pressure offgas, which is first purified in a PSA unit. Finally, low-pres- sure offgas is
compressed, mixed with reject gases from the membrane and PSA units, and used as steam
reformer feed. The system also includes a recycle loop to moderate the tem- perature rise
across the hydrogenator from the saturation of olefins.

Liquid Feeds. Liquid feeds, either LPG or naphtha, can be attractive feedstocks where prices
are favorable. Naphtha is typically valued as low-octane motor gasoline, but at some locations
there is an excess of light straight-run naphtha, and it is available cheaply. Liquid feeds can also
provide backup feed, if there is a risk of natural gas curtailments.

The feed handling system needs to include a surge drum, feed pump, and vaporizer, usually
steam-heated. This will be followed by further heating, before desulfurization. The sulfur in
liquid feeds will be in the form of mercaptans, thiophenes, or heavier compounds. These
compounds are stable and will not be removed by zinc oxide; therefore a hydro- genator will
be required. As with refinery gas, olefins must also be hydrogenated if they are present.

The reformer will generally use a potash-promoted catalyst to avoid coke buildup from
cracking of the heavier feed. If LPG is to be used only occasionally, it is often possible to use a
methane-type catalyst at a higher steam/carbon ratio to avoid coking. Naphtha will require a
promoted catalyst unless a prereformer is used.

HEAT RECOVERY

In selecting a heat recovery system for a new plant, a number of factors must be balanced:
environmental regulations, capital cost, operating cost, and reliability. The relative impor-
tance of these will vary from project to project.2

The environmental regulations with the greatest impact on plant design are typically NOx
limitations. Other impacts such as SOx or water emissions are minimal, because low-sulfur fuel
is typically used and there are few emissions other than flue gas. The choice of heat recovery
system can have a major effect on NOx production, since both the amount of fuel fired and the
flame temperature will be affected. Preheating combustion air will reduce fir- ing, but since
NOx formation is strongly influenced by flame temperature, there will be an overall increase in
NOx formation. Other methods of reducing firing, such as prereforming or heat-exchange
reforming, do not affect the flame temperature and will therefore reduce NOx production. Any
of these methods can also be useful if there is a limit on the total amount of fuel fired, such as
when a plant is to be expanded under an existing permit.

Capital cost and operability will generally favor steam generation. This is the simplest

scheme, and it is favored wherever the additional steam can be used (Table 6.1.10). No
additional catalysts are necessary, and if a Terrace Wall or a side-fired reformer is used, it is
possible to build the reformer as a natural-draft unit. This eliminates the forced- and induced-
draft fans and further improves reliability. In cases where steam has little value, air preheat,
prereforming, or heat-exchange reforming will be favored, although capital cost will be
increased with these options.

Prereforming

In prereforming, use of a highly active catalyst allows reforming to occur at a lower tem-
perature.1

The reformer feedstock is mixed with steam and passed over the prereforming catalyst, as is
shown in Fig. 6.1.6. As reforming proceeds, the temperature falls. The gas is then reheated and
sent to the primary reformer. For feedstocks heavier than methane, heat of cracking will tend
to raise the temperature and can result in a temperature rise for liquid feeds or heavy refinery
gases. The technology is well proved, and the catalyst has been used for other applications in
naphtha reforming. Other than the reactors, the only addi- tional equipment required is a
preheat coil in the reformer.

On the other hand, only a limited amount of heat can be recovered, since the reactor inlet
temperature is limited to about 930°F (500°C) to avoid cracking of the feedstock. Much of the
savings in energy comes from the ability to reheat the feed to a high temper- ature. Since the
prereformer outlet contains no hydrocarbons heavier than methane, there is little risk of
cracking.

The high-activity catalyst is also sensitive to deactivation, and provision must be made to allow
catalyst changeout during operation.

Heat-Exchange Reforming

The process gas leaving the reformer can be used as a heat source for additional reform- ing.
Reforming catalyst is packed in the tubes of a heat exchanger, and the primary reformer outlet
gas flows in the shell. Various arrangements are used to cope with tube expansion, such as the
one shown in Fig. 6.1.14. Here the hot gas from the primary mixes with the gas leaving the
open-ended catalyst tubes and then flows along the outside of the catalyst tubes. An
advantage of the heat-exchange reformer is that it can reach higher tem- peratures and
recover more heat than the prereformer, although at higher equipment cost. The temperature
in the heat-exchange reformer is lower than that in the primary. The steam/carbon ratio in the
heat-exchange reformer can be increased to correct this which affects the reforming
equilibrium. This also shifts the reforming heat load to a lower tem-

perature, improving the heat balance.

The main effect of the heat-exchange reformer is to reduce the fuel demand and steam
generation. Table 6.1.11 shows this reduction: from 159,000 lb/h (72 MT/h) with the pri- mary
reformer alone, to 77,000 lb/h (35 MT/h) with the addition of the heat-exchange reformer. By
combining the heat-exchange reformer with air preheat, there is a further reduction in the
steam generation and fuel demand for the plant: Export steam is reduced to 21,000 lb/h.

ECONOMICS

Process Route

Capital costs for hydrogen production are illustrated in Fig. 6.1.15, which compares costs for
purification, steam reforming, and partial oxidation. Where hydrogen is already avail- able in
sufficient quantity, it is cheapest to merely purify it as required. In most cases this is not
sufficient, and it is necessary to manufacture it.

Figure 6.1.15 illustrates why steam reforming is favored over partial oxidation. For light
feedstocks, capital costs for the inside battery limit (ISBL) plants are similar for steam
reforming or partial oxidation. However, when the cost of oxygen is included, the cost for
partial oxidation (POX) rises substantially. Naphtha reforming is slightly higher in capital cost
than reforming of natural gas. Feedstock cost will depend on the value of the naphtha; where
the naphtha is valued as motor gasoline, as in Fig. 6.1.15, it cannot com- pete with natural gas.
Where there is a surplus of low-octane naphtha, it may be valued at fuel cost or even below; in
this case steam reforming of naphtha can be attractive.

For partial oxidation of residual fuel, a substantial amount of equipment is required to handle
the soot, ash, and sulfur (Fig. 6.1.4). The cost for this additional equipment, as well as the
additional oxygen required, means that heavy oil must be much cheaper than natu-

FIGURE 6.1.15 Production cost of different cost process routes. PSA = pressure-swing
adsorption; SMR = steam-methane reforming; POX = par- tial oxidation ral gas to justify partial
oxidation. Alternatively, partial oxidation may be used as a way to dispose of a stream such as
petroleum coke or asphalt, which is considered a waste prod- uct.

Capital Cost

Where capacity, feedstock, and method of heat recovery are known for a steam reforming
plant, a reasonable estimate may be made of capital cost, typically to an accuracy of ±30 per-
cent. For a 50 million SCFD[56,000 (N) m3/h] hydrogen plant, based on natural gas feed and
using steam generation for heat recovery, capital cost is approximately $30 million.

This assumes a battery limit process unit, including the equipment shown in Fig. 6.1.3, on the
U.S. Gulf Coast constructed in second quarter 2002 through mechanical comple-

tion. It also assumes that the site is free of above- and below-ground obstructions. It does not
include the cost of land, taxes, permits, warehouse parts, escalation, catalyst, and sup- port
facilities.

Make or Buy

In recent years refiners have been presented with a viable alternative to building their own
hydrogen plant. It is possible to buy hydrogen like a utility “over the fence” from one of the
major industrial gas companies. These companies typically have experience producing and
selling many industrial gases such as hydrogen, nitrogen, and oxygen, and several have
pipeline networks which can provide additional reliability and economies of scale.

Revamps

Changes to refinery product or feedstock slates often require incremental increases in

hydrogen production. Especially with older plants, it may often be feasible to get this extra
capacity by revamping an existing unit.

Bottlenecks to increased capacity typically fall into one or more of the following areas: the
reformer, CO2 removal, or hydraulics/compression. Heat transfer is typically less of a problem,
and changes here are often integrated with improving the plant hydraulics.

A number of the developments already mentioned can be used to increase capacity. For
example, improved catalyst tube metallurgy can allow operation at higher reformer tem-
peratures and lower steam/carbon ratio. The reduced steam flow then allows operation at
higher hydrogen production rates.

In wet scrubbing plants, a change in solution composition or tower internals can allow more
gas through the same towers.

Increasing capacity of an existing PSA unit may also be attractive, but here the situa- tion is
more complex. The economics of PSA systems depend on moving gas from one absorber to
another during the depressuring/repressuring steps. Because of this they are highly integrated,
and a revamp may involve changing adsorbent, modifying cycles, adding beds, or increasing
the size of existing piping and valves.
One option that has not proved viable is the replacement of wet scrubbing systems with PSA
units. Because of the loss of hydrogen into the PSA tail gas, this actually would reduce the
capacity of the plant.

REFERENCES

1. B. J. Cromarty, K. Chlapik, and D. J. Ciancio, “The Application of Pre-reforming Technology in

the Production of Hydrogen,” NPRA Annual Meeting, San Antonio, Tex., March 1993.

2. J. D. Fleshman, Chem. Eng. Prog., 89(10), 20 (1993).

3. A. Fuderer, “Catalytic Steam Reforming of Hydrocarbons,” U.S. patent 4,337,170, June

29, 1982.

4. M. H. Hiller, G. Q. Miller, and J. J. Lacatena, “Hydrogen for Hydroprocessing

Operations,” NPRA Annual Meeting, San Antonio, Tex., March 1987.

5. G. Q. Miller and J. Stoecker, “Selection of a Hydrogen Separation Process,” NPRA

Annual Meeting, San Francisco, March 1989.

6. Physical and Thermodynamic Properties of Elements and Compounds, United

Catalysts, Inc., Louisville, Ky.

7. M. V. Twigg, Catalyst Handbook, 2d ed., Wolfe Publishing, London, 1989.

8. G. J. Lambert, W. J. A. H. Schoeber, and H. J. A. Van Helden, “The Hydrogen Balance in

Refineries,” Foster Wheeler Heavy Oil Processing and Hydrogen Conference, Noordwijk, The
Netherl