Вы находитесь на странице: 1из 65

MODERN PRODUCTION TECHNOLOGIES

OOO -

2. Ammonia
Capacity and usage of ammonia
As Table II shows, ammonia is the second largest synthetic chemical product and, of all the products with which this work is concerned, it involves the greatest number of reaction steps in its production. Some of these process steps are also used in the production of methanol, hydrogen and carbon monoxide. For this reason it is logical to begin by considering ammonia and its production and to restrict the coverage of the other products and their production to their specific aspects only. As 85% of world nitrogen demand is for fertilizer, it might be expected that ammonia production should develop approximately in proportion to the growth of world population. In Fig. 2 it can be seen that that was, indeed, roughly the case in the mid 1980s; since then the rate of increase in ammonia production has been distinctly slower than that of world population. This has been mainly for economic reasons in the Third World and for ecological reasons in the industrialized countries, while the political changes in the former Eastern Bloc caused a dramatic slump in 1991/93. In recent years ammonia consumption has also been below production capacity and is expected to remain so, at least for this decade, as shown in Table III. The forecast from 1995 up to 2000 suggests that demand will again grow faster than capacity. For the end of this century Table HI shows an average world capacity utilization
Table II

rate well above 85%, a level never before achieved on a global scale. To cope with such a situation in the medium term, the efficiency of plants in eastern Europe and some of the Developing Countries will have to be improved. In addition, revamp projects may help to ease the tight supply situation, but eventually some new plants will be required as well.

Worldwide ammonia feedstock scenario


About 1.6% of the world's consumption of fossil feedstocks goes into the production of ammonia. All over the

world natural gas has grown markedly over the years at the expense of heavier feedstocks, as Table IV shows. If we exclude Asia, where almost all of the world's coal-based ammonia plants are located (mainly in China, where 66% of total ammonia production is made from coal and coke) as well as the majority of the naphtha-based installations, natural gas is the feedstock for 92.5% of the remaining capacity. Some of the reasons for preferring natural gas have already been mentioned. To summarize, there are three main factors. First, it is intrinsically the most hydrogen-rich of all feedstocks and so contributes a

Table III World Supply/Demand Balance (million t/a N)4 6


1985" Capacity Demand Capacity utilization rate needed to meet demand 118.4 92.0 1990" 122.5 96.4 1991" 122.5 98.2 I9924 1993s 124.5 93.1 126.7 90.4 1994s 116.3 93.4 1995' 117.5 95.2 1998' 2000' 121.5 104.4 124.4 108.6

77.7%

78.7%

80.2%

74.7% 71.3%

80.3%

81.0%

85.9% 87.3%

Fig. 2: World Population and Ammonia Production


7H

1-140

6-

5-

World Capacity for Major Basic Chemicals in 1995 (million t/a product)3
Sulphuric acid Ammonia Urea Ethylene Chlorine Soda Hydrogen Methanol
207 143 105 81 46 43 31 26

1920

1940

I960

1980

AMMONIA

Table IV Feedstocks for Ammonia Production as Percentage of World Capacity I9907


Natural gas Naphtha, LPG, refinery gas Fuel oil / vacuum residue Coke, coal, coke oven gas Water electrolysis, by-product hydrogen
77 6 3 13.5
0.5

greater proportion of the hydrogen in the synthesis gas than any other. Second, the heavier the feedstock the more complicated and, consequently, the more expensive is the process needed to convert it into synthesis gas. Third, it is the most widespread, easily available and easily deliverable feedstock and is, in consequence, usually favourably priced in relation to its technical advantages.

Carl Bosch was already well aware of the fact that the production of pure hydrogen/nitrogen synthesis gas is the largest single contributor to the total production cost of ammonia9. So> in contrast to the synthesis reaction, dramatic changes happened over the years in the technology of synthesis gas generation, which is the decisive factor of the production costs. Net energy consumption has been reduced progressively over the years, from around 88 GJ/t NHj in the days of cokebased water gas generators to around 28 GJ/t NH3 today, using

natural gas in a steam reforming unit There are three principal process routes for producing a pure ammonia synthesis gas with the 3:1 final H2:N2 mol ratio dictated by the stoichiometry of ammonia synthesis: Steam reforming of natural gas (and, to a minor extent, other light hydrocarbons) Partial oxidation of heavy hydrocarbons (e.g. vacuum residue or asphalt) Partial oxidation (or other gasification process) of coal.

Fig. 3: Ammonia Technology for Various Feedstocks


STEAM
REFORMING NATURAL GAS PARTIAL OXIDATION HEAVY FUEL OIL PARTIAL OXIDATION COAL coal

Principal process routes to ammonia


Based on the fundamental research work of Fritz Haber, BASF commissioned the world's first ammonia plant in Oppau, today a part of Ludwigshafen, Germany, in the autumn of 1913. It was Carl Bosch and his engineering team who, in an unprecedented effort, developed ammonia synthesis to technical operability using the promoted ironbased catalyst found by Mittasch and his co-workers8. Since then there has been no fundamental change in the synthesis reaction itself. Even today, nearly every plant in which ammonia is synthesized from its elements has the same basic configuration as this first plant. A hydrogennitrogen mixture reacts on the iron catalyst (today's formula differs little from the original) at elevated temperature in the range 400-500C (originally up to 600C), operating pressures above 100 bar. 3H, + N2 * 2NHj AH29g = -92.4 kj/mol (4)

heavy fuel oil

natural gas

The unconverted part of the synthesis gas is recirculated after removal of the ammonia formed and supplemented with fresh synthesis gas to compensate for the amount of nitrogen and hydrogen converted to ammonia.

NH3

MODERN PRODUCTION TECHNOLOGIES

How these different process routes compare can be seen schematically from Fig. 3. The steam reforming process for natural gas10"18 usually receives the gas under pressure. Any sulphur compounds it may contain are transformed by reaction with hydrogen on a cobalt-molybdenum catalyst into hydrogen sulphide, as in equation (5), which is then absorbed as zinc sulphide in a bed of zinc oxide pellets at a temperature between 300 and 400C, according to equation (6).
RSH + H2 + ZnO + RH ZnS + H2O (5) (6)

the reforming of the residual methane adiabatically (without any further heating). It also introduces just the right amount of nitrogen to give the correct stoichiometric H2 : N2 ratio for the eventual ammonia synthesis, and in fact it is this consideration that governs the amount of methane left after the primary reforming. The oxygen from the introduced air shows up as carbon monoxide and carbon dioxide. Secondary reforming can be considered as a. partial oxidation of the residual methane. The overall reaction shown in equation (9) does not, however, give any clue to the actual reactions taking place. 2CH4 + O2 (+ 4N2) 2CO + 4H2 (+ 4N2) AH298 = -71.4 kj/mol (9) The gas usually leaves the secondary reformer at 95M,000C and, depending on the process, with a methane content of 0.3 to 1.5% by volume. It is cooled down to about 350-400C, the waste heat removed being used to raise, and sometimes also to superheat, high-pressure steam. The carbon monoxide content is then reduced by the shift reaction of equation (3) to about 3%, using an iron-chromium catalyst. After further cooling with waste heat recovery, it is subjected to a second stage of shift conversion at 200C on a copper-zinc catalyst, which cute down the carbon monoxide content to 0.2-0.3%. The carbon dioxide is next removed by scrubbing with highly efficient solvents, which are regenerated and reused. But none of the processes can remove either the carbon monoxide or the carbon dioxide to the extremely low level needed to avoid damage to the ammonia synthesis catalyst. In a methanation reaction, represented by equation (10), which is the reverse of equation (7), and equation (11) using a nickel catalyst in the temperature range 300-350C, the level of these carbon oxides is reduced to less than 10 ppm.
CO + 3H2 CH4 + H2O AH29S = -206.3 kj/mol (10)
C0

The basic reactions involved in the reforming of methane, which is the main constituent of the natural gas, are represented by equations (7) and (3):
CH4 + H2O CO + 3H2 AH298= +206.3 kj/mol
CO

The synthesis gas is now finally ready to be compressed and introduced into the synthesis loop, which mostly operates between 150 and 300 bar. Steam reforming of LFG or naphtha17"19 is nowadays restricted to areas where access to natural gas is limited and where processing of heavy hydrocarbons is not feasible for economic or other reasons. Basically, steam reforming of these light hydrocarbon fractions is very much the same as of natural gas. For the primary reformer some parameters such as temperature and pressure are different, the steam surplus is higher, and special reforming catalysts have to be loaded in the furnace tubes. Equation (12) gives the overall reaction in a simplified form. Actually some methane is formed, for example as in equation (13), which may in turn be converted further.

(7) (3)

CA, + nHp - nCO + (2n+m)/2H2

(12)

CO2 + H, AH = -412 kj/mol

CnHm + (4n-m)/4H20- (4n+m)/8CH4 +(4n-m)/8CO2 (13) In partial oxidation21^29, heavy hydrocarbons react with oxygen according to equation (2). The reaction is non-catalytic and proceeds in an empty vessel lined with refractory alumina. The reactants (oil and oxygen), along with a minor amount of steam, are introduced through a nozzle at the top of the generator vessel. The nozzle consists of concentric pipes so that tine reactants are fed separately and react only after mixing at the burner tip or in the space below. The temperature measured in the generator is between 1,200 and 1,400C. Owing to insufficient mixing with the oxygen, about 2% of the total hydrocarbon feed is transformed into soot, which is removed from the gas by water scrubbing. The separation of soot from the washing water and its further treatment differ in the Shell and Texaco processes - the two commercially available partial oxidation concepts. After gas cooling by further waste heat recovery, carbonyl sulphide (COS) in the gas is hydrolysed to hydrogen sulphide. This and the hydrogen sulphide formed during gasification are removed along with

In addition the "Boudouard equilibrium", shown in equation (8), has to be considered as well.
2CO C02 + C AH298 = -172.5 kj/mol (8)

The required stoichiometric hydrogen/nitrogen ratio is attained by introducing air into the process in a most elegant manner. It is done by splitting the reforming into two sections. In the first (primary) reformer, reaction (7) takes place in furnace tubes packed with nickel catalyst. The intense heat needed for the endothermic reaction is supplied by natural gas burners in the furnace radiation box. The reaction is controlled to achieve only a partial conversion (ca. 65%), leaving about 14% methane in the effluent gas (dry basis) at temperatures of around 750-800C. The gas is then introduced into the secondary reformer - a refractory-lined vessel filled with nickel catalyst - where it is mixed with a controlled amount of air introduced through a "burner". This raises the temperature of the gas sufficiently to complete (as far as possible in an equilibrium system)

CH4 + 2H20 AH29g = -165.1 kj/mol (11)

AMMONIA

carbon dioxide by scrubbing with chilled methanol below -30C in the Rectisol5' process. Then, as in the steam reforming route, the gas undergoes the CO shift reaction. Because of the higher carbon monoxide content, much more reaction heat is produced, which makes it necessary to distribute the catalyst on several beds with intermediate cooling. The carbon dioxide formed in the shift conversion is removed in a second stage of the Rectisol unit, both stages of which have a common methanol regeneration system. The H2S-rich carbon dioxide fraction from the first stage of the regenerator is fed to a Claus plant, where elemental sulphur is produced. In the final purification, the gas is washed with liquid nitrogen, which absorbs the residual carbon monoxide, methane and a portion of the argon (which was introduced into the process in the oxygen feed). The conditions in this stage are set so that the stoichiometric nitrogen requirement is allowed to evaporate into the gas stream from the liquid nitrogen wash. The process needs, of course, an air separation plant to produce oxygen, usually around 98.5%+ pure, and to supply the liquid nitrogen. Coal-fed ammonia plants30*32 also use partial oxidation concepts in various embodiments, such as entrained or fluidized-bed systems, specially designed to handle the ash and slag remaining after gasification. Slag removal and handling is, indeed, the fundamental technical problem with coal-based systems. The gas purification steps are rather similar to those in the partial oxidation of heavy hydrocarbons. As Fig. 3 above showed, the comparison of the coal-based route with the other routes to ammonia was based on a partial oxidation process, which could be either the Shell33'34 or Texaco35-36 process. It was a valid choice, because the alternative processes - for example, the Lurgi Coal Gasification process37-38, are at least f the same complexity. In the Lurgi process, where the gasification temperatures are lower and less oxygen and more steam are used in a fluidized-bed technology, sizeable quantities of methane and a fair amount f organic by-products such as tar

and phenols are produced, to deal with which additional provisions are necessary. The Koppers-Totzek process32- 39 has been used in a number of installations. It operates at normal or only slightly increased pressure, which means that the synthesis gas subsequently requires a greater degree of compression; however, it contains far less of the by-product materials found in "Lurgi gas". Quite a number of other technologies have been tested in pilot pants and demo-units40-41. From the block diagrams in Fig. 3 it is obvious that the simplest and least elaborate process is the steam reforming route with natural gas as feedstock. Assuming, for the sake of simplicity, that the whole synthesis section, including compression and refrigeration, is one process step and shift conversion is another, in spite of the fact that it is performed in at least two catalyst beds with intermediate cooling in all three processes shown, an ammonia plant based on this steam reforming comprises seven basic stages. Including desulphurization, the steam reforming process uses eight different catalysts. The partial oxidation of heavy fuel oil involves 11 basic steps and only four or five different catalysts. In the coal-based ammonia plant there are thirteen basic steps with four or five different catalysts. The advances made in the last 15 years for these different production schemes for ammonia are discussed in the following sections, which are intended to give a picture, of course far from comprehensive, of modern ammonia production technology and to describe the present state of the art. In view of the enduring predominance of natural gas as the feedstock of choice for ammonia production, it is natural that the main emphasis is on steam reforming of natural gas for the synthesis gas generation.

and lighter hydrocarbons42. On account of the abundant supply of extremely cheap natural gas in the USA, it was there that this technology - greatly improved and developed further by ICI - first became widespread. Because natural gas was then not available in commercial quantities in other countries, it was a great advance when ICI developed a catalyst and a technology to process higher hydrocarbons (up to naphtha)43-44. This made it possible to use the process in new ammonia plants virtually anywhere in the world. Then, with the discovery of huge gas fields in the Netherlands, France and the North Sea and their development from the 1960s onwards, the European ammonia industry switched from the old coalbased processes to natural gas. As a result of the intensive exploration and development of petroleum resources in the subsequent years, enormous quantities of associated or pure natural gas subsequently became available in many parts of the world.

Reforming pressure, steam/ carbon ratio and heat flux


Because natural gas is usually already under elevated pressure, and because the reforming reaction entails an increase in the total volume, significant savings of compression energy are possible if the process is performed under elevated pressure. At the time of the advent in the mid 1960s of the 1,000-st/d single-train ammonia plant, pioneered by The M. W. Kellogg Co. (Kellogg), steam reformer operating pressures of about 30 bar were already commonplace and represented proven technology; but still there was considerable potential for improvement. Besides saving compression energy, there is another advantage in raising reforming pressure further. The higher the pressure, the higher will be the temperature level at which the heat of condensation of the surplus of steam remaining after the shift conversion is recovered. But there is also a serious disadvantage in using elevated pressure in the reforming process: on account of the volume increase in the reforming reaction, it exerts an unfavourable effect on the equilibrium in equation

Ammonia by steam reforming Reforming section


It is now 60 years since the first unit for catalytic steam reforming of hydrocarbons, based on research and development done by BASF, came on stream in Baywater, New Jersey, using refinery gas rich in methane

MODERN PRODUCTION TECHNOLOGIES

(7), which reduces the conversion of methane. An increase in the heating duty is necessary to compensate. This can be partially offset by lowering the steam-to-carbon ratio, which reduces the total mass of gas to be heated. On the other hand, a high steam-to-carbon ratio has a beneficial effect on the equilibrium in the reforming reaction and to some extent mitigates the negative effect of higher system pressure. The relationship between equilibrium methane concentration and temperature45, steam-to-carbon ratio S/C and reforming pressure are plotted in Figs 4 and 5. Apart from the primary reformer, there is also a considerable effect on the operation conditions of the secondary reformer. Figure 612 shows how primary reformer heat duty has to vary with pressure and steam-to-carbon ratio to keep the methane slip out of the secondary reformer constant. But there are three more important reasons for using a high steamto-carbon ratio. First, it prevents carbon deposition on the catalyst, which not only increases resistance to gas flow but also impairs the catalyst activity and, by lowering the rate of the endotherrnic reaction, can result in local overheating ("hot bands") and premature failure of the tube walls. Second, it provides the necessary steam for the shift conversion of carbon monoxide. And third, it reduces the risk of carburization damage to the tube material. In addition to the Boudouard reaction, which is a disproportionation of carbon monoxide according to equation (8), methane cracking - equation (14) - may also produce carbon.
CH C+ AH

occur. Contrary to this, the reactions according to equations (8), (14) and (15) are reversible and there is a dynamic equilibrium between carbon formation and removal. Thus, under typical steam reforming conditions, reactions (15) and (8) are carbon-re-

moving, whilst reaction (14) leads to carbon formation in the upper part of the tube1*. The mechanism of carbon formation and the determination of the risk areas in reformer operating conditions on the basis of relevant equilibrium data are discussed

Fig. 4: Methane Equilibrium vs Temperature at Various S/C ratios


45

40 35 30 25
E 20
pressure: 32 kg/cm2

i is !
Ill

10
5

650

700

750

800
temperature, C

850

900

Fig. 5: Methane Equilibrium vs Temperature at Various Pressures

= +74.9 kj/mol (14)

Besides that, it is possible for carbon monoxide to be reduced by hydrogen, yielding water and carbon, as shown in equation (15),
CO + H,* C + H2O AH' 298 -131.4 kj/mol (15)

With naphtha as steam reformer feed, irreversible pyrolysis (as in a steam cracker for producing ethylene) with the sequence naphtha - olefins - polymers - coke will

650

700

7SO

800

850

900

temperature, C

AMMONIA

Fig. 6: Primary Reformer Parameters for 0.3% CH4 Exit Secondary


U ~S <U 2

Table V
Comparison of Reforming Conditions in Modern Low-Energy Ammonia Plants with High- and Low-Duty Primary Reformers (1,800 t/d NH )
High duty Primary Reformer Steam ratio Exit temperature, C Outlet pressure, bar g Exit CH_, mol-% (dry basis) Reforming heat duty, GJ/t NH, Relative catalyst volume Secondary reformer Process air inlet temperature, C Exit temperature, C Exit CH,, slip, mol-% (dry basis) Ammonia plant Total energy consumption, GJ/t NH3 Low duty

5.55.04.51,500 (816)

E eu

30atm

1,600 1,700 1,800 1,900 (871) (927) (982) (1,038)

3.0 814 39.5 13.2 4.70 1.0

2.7 693 30.1 29.4 2.27 0.8

secondary reformer outlet temperature, F (C)

in some detail in previous publications46-50. Older designs used steam-tocarbon ratios of 3.5-4.0. Because surplus steam is surplus volume, it increases pressure drop through the catalyst - another reason for lower1 ing steam-to-carbon ratio. Overall, lowering steam surplus should result in considerable energy saving. Theoretically the minimum ratio needs to be only slightly over 1.0 to avoid cracking. Catalyst experts see the practical limits at 1.5-1.7 for the methane steam reforming12-45 and 2.2 for naphtha steam reforming. But to supply sufficient steam for the shift conversion it has to be at least 2.0 for stoichiometric reasons. And a safety margin is advisable in case of operational difficulties which might lead to a temporary loss of proper control of the steam-to-carbon ratio - occasional slugs of higher hydrocarbons in the feed, for example and to avoid hydrocarbon formation on the HT shift catalyst (a problem which will be discussed later). So, in actual operation, the steam-to-carbon ratio in a modern low-energy plant should be about 2.7 for a lowduty reformer and around 3.0 for a high-duty reformer. Since the tubular furnace and its associated "convection section" (flue gas heat recovery train) is the largest single item in an ammonia plant, and therefore the most expensive, some modern concepts have reduced its size by shifting some of the primary reformer duty to the secondary reformer. This is what is meant by a "low-duty" steam reformer. It is achieved by introducing more air into the secondary reformer, exceeding the quantity which is needed to produce an eventual 3:1 stoichioH2:N2 mixture, and then re-

600 1000 0.60

500 870 1.65

28

28

jecting the surplus nitrogen at a subsequent stage. A greater proportion of the total heat for the reforming reaction is thus supplied in an autothermal way and correspondingly less has to be provided as fuel for the primary reformer. An example for this approach is Brown & Root's Braun Purifier process, in which the excess of nitrogen is removed along with methane and argon in a cryogenic condensation system - the "Purifier" - downstream of the methanator. In the ICI AMV process, which uses the same principle, a non-stoichiometric synthesis gas (H2:N2 = 2.5:1) is passed through the loop. The surplus nitrogen is rejected during hydrogen recovery from a larger-than-normal volume of loop purge gas. The reduction in the reforming section can be considerable, as can be seen from Table V.

Secondary reformer operation


The most important part of a secondary reformer is the burner design. Burner failure is the most common cause of poor performance. Because the activity requirements for the catalysts are not very high and thermal and mechanical stability of the different brands offered is sufficient to achieve economic service life, no problems will usually arise from this side. In the last few years licensors and contractors have made a substantial effort to develop improved burner designs which could cope with more severe conditions, especially those encountered when

using pure oxygen instead of air, as in the combined reforming used in some cases for methanol or hydrogen production. Another factor which called for the redesign of burner configurations was the marked tendency to raise process air temperatures in modern low-energy concepts to 600C and above. For example, a few years ago Kellogg, which formerly used multiple-nozzle burners in its classical plants, developed and commercialized a new kyout for the burner and combustion chamber51'52. In this new design the air/steam mixture is introduced axially at high velocity at the upper end of a long cylindrical combustion chamber. The process gas coming from the primary reformer enters radially, also at the top end of the cylindrical chamber. This perpendicular flow pattern of the high-velocity air jet and the process gas stream induces intensive mixing and instantaneous start of combustion. A cold-flow Plexiglas model was used to study the flow pattern using air-CO2 mixtures as fluid. Additionally theoretical finite element modelling is used, based on the FLUENT software. Provisions have been made in the detailed design for the burner and refractory "plug" to be removable to allow easy entry to the vessel for catalyst loading. The reliability of the design has been proved in quite a number of plants, and it has even been successfully applied recently in a plant which utilizes enriched air. This kind of service may be of especial interest in retrofit applications.

MODERN PRODUCTION TECHNOLOGIES

For oxygen-operated secondary reformers in methanol plants, Kellogg offers a water-cooled burner53, whereas ICI uses, especially for its LCA and also for its LCM (methanol) concepts, a special geometry for the burner neck which ensures intense mixing of the gas and air before they enter the zone above the catalyst54-55. Cooling for the burner is provided by the oxidant (air and/or oxygen). In the development of the new burner, ICI used a Computational Fluid Dynamics (CFD) computer program56. Apart from the new version, ICI still offers its traditional ring burner in its licensing package. Both Haldor Tops0e AS (Tops0e) and Krupp-Uhde (Uhde) have evolved improved burner and combustion chamber designs with the aid of CFD. Tops0e's new CIS burner was developed with the aid of isothermal hydraulic and pilot plant testing15-57. In Uhde's new design the air is introduced from an external ring-shaped air header through nozzles which discharge it in a spiral path in the space between the hot face of the refractory lining and the central riser through which the incoming process gas is introduced58.

It should achieve the desired conversion of the hydrocarbon feed at the lowest possible tube wall temperature and the lowest possible pressure drop without forming carbon. It should have a reasonable lifetime without deactivation (lifetime should at least match turnaround intervals). It should withstand extraordinary conditions during start-up and shut-down. And, if possible: It should have the stability to withstand in situ regeneration procedures to redeem the effects of incidental poisoning or carbon deposition. Today this profile can be largely fulfilled by all leading catalyst producers. Two principal factors influence catalyst activity: chemical composition and surface area. But of no less importance for the performance are the heat transfer characteristics, which are governed by the size and shape of the particles. The active component of the primary reformer catalyst is nickel, which is finely dispersed over the support material as crystallites produced by reduction of nickel oxide. The support materials used nowadays are oc-alumina, calcium aluminate and magnesia-alumina spinel. Two different manufacturing methods are commonly used. In the older one, nickel is first precipitated, usually as its hydroxide, in the presence of a suitable dispersion component. After washing, drying and calcining to the oxide, the powder is mixed with an hydraulic cement, formed into particles and subjected to the appropriate treatment to get the full hydraulic bonding of the cement. The nickel oxide is thus evenly dispersed throughout the lattice structure of the cement support. The reduction process takes longer, and there is a greater degree of shrinkage than with the more modern type of catalyst, based on a prefabricated support. In this method, the support pellets are impregnated with a nickel salt solution and dried, after which they are cal10

cined to transform the nickel salt into the oxide. The surface area of the catalyst produced in this manner depends largely on the degree of firing of the support. With increasing temperature the surface will decrease but the mechanical stability will increase. Thus there is a trade-off between activity and strength. Alpha-alumina is now the predominant support in North America but is increasingly used in other parts of the world too. Calcium aluminate is still accepted worldwide on account of its natural alkalinity, which helps to suppress carbon deposition. It is therefore preferred for naphtha steam reforming catalysts. Catalysts intended for dealing with higher hydrocarbons are alkalized with potash - a development first introduced by ICI60. That is because cracking, shown in equation (16), is acid-catalysed.
CH -> x C + y/2 H,

(16)

Alkalization suppresses acidic spots on the surface of the catalyst, and will also promote reactions which remove carbon already deposited from the surface according to equations (17) and (18):
C + CO2 ^ 2CO AHM8 = +172.5 kj/mol (17) C + H2O CO + H2 AH298 = +131.4 kj/mol (18)

Reforming catalysts59
In the steam reforming section there have been two main causes of problems in the past: catalysts and reformer tubes. The performance of each influenced the other. Catalyst makers blamed contractors for using too high heat fluxes, which caused deterioration of the catalysts by sintering and chemical changes. For contractors and tube manufacturers, on the other hand, the culprit was the insufficient thermal stability of the catalyst. Of course, the reality was more complex and there were many and diverse causes for the problems experienced with both tubes and catalysts. There have been considerable improvements in both catalysts and in tube materials and manufacturing techniques over the last 15 years, and nowadays problems and failures in this area are more operational than technological in origin. The following are the criteria for a good primary reforming catalyst.

Therefore it is also a good practice when processing natural gas which contains some C4+ to have the first third of the tube, where the highest heat flux occurs, filled with alkalized reforming catalyst. A very good protection against carbon deposition when processing a somewhat "wet" natural gas containing higher hydrocarbons, whether continuously or only intermittently, is to pass the feedstock through an activated carbon vessel ahead of the desulphurization reactor to take them out. The adsorbent may be regenerated periodically by passing preheated feed gas through it; this is then mixed with fuel for the reforming furnace burners. This technique has been used in some ammonia plants

AMMONIA and has markedly prolonged the catalyst life. Magnesium aluminate, the preferred support of one catalyst supplier, has a larger specific surface area. But this material must be calcined to a higher temperature during manufacture of the support particles to ensure it contains no free magnesium oxide, which would hydrate to hydroxide at temperatures below 300C. This chemical change results in a volume increase, which would destroy the structure and impair the mechanical stability of the catalyst. The various support materials have different effects on potential carbon formation. This seems to go in parallel with the Lewis /Bransted acidity. On the basis of experimental comparisons the main commer, daily-used catalyst supports can be ranked as follows in decreasing order of carbon forming tendency (and thus in decreasing order of the allimportant minimum practical steam/ carbon ratio): Alpha-alumina > magnesium aluminate spinel > calcium aluminate > alkalized calcium aluminate49. The effect of the size and shape of the catalyst61 on heat transfer, and consequently performance, was mentioned above. Unlike most processes carried out under substantially adiabatic conditions, the endothermic steam reforming reaction has to be supplied continuously with heat as the gas passes through the catalyst; hence the need for the furnace. The strong dependency of the reaction rate on the surface temperature of the catalyst dearly underlines the need for effident heat transfer over the whole cross-section of the catalyst. However, the catalyst material itself is a very poor conductor and does not transfer heat to any significant extent. Therefore the main mechanism of heat transfer from the inner tube wall to the gas is convection, and its efficiency will depend on how well the gas flow is distributed in the catalyst bed. It is thus evident that the geometry of the catalyst parades is important. To make full use of the enhanced activity attainable by increasing the internal surface of the catalyst, the ne at transfer characteristics (shape) have to be improved, too. Obviously

Fig.

7: Performance of a 4-HoIe Reforming Catalyst in a Top-Fired Primary Reformer


net carbon formation without potash with potash -1,600

I
conventional ring (100% flowsheet) 4-hole (100% flowsheet) 4-hole( 120% flowsheet) 900 t/d ammonia plant steattKC = 3.3 reformer CH, slip 9.3 mol-% (dry)

- I.SOO

1,400

1,300
0.2 0.4 0.6 fraction down reformer tube 0.8

both are strongly interrelated. If a catalyst of higher activity with an improved heat transfer characteristic is installed in a given reformer and the firing conditions are kept as before, the peak skin-temperatures of the reformer tube will be reduced. If, on the other hand, the previous tube wall peak temperature is maintained, a higher throughput is possible. Figure 762 shows in a schematic manner the individual effect of surface and heat transfer characteristics. Various catalyst shapes have merefore been developed by the individual catalyst producers and have progressively replaced Raschig rings (which themselves once displaced simple tablets), especially in the high heat-flux zone in the upper third of

the tube: in the lower end of the tube there would be no significant difference between their performance and that of traditional sizes and shapes, apart from a certain reduction in the pressure drop. ICI, for example, has a four-hole type, UCI (United Catalysts Inc.) the "Wagon wheel" and Topsoe the "Sixshooter". Figures 8 and 9 show examples of such modern shaped steam reforming catalysts11-M. Important, too, of course, are the packing characteristics of the new forms. It is essential to avoid bridging, which is where - instead of packing down dosely - partides of catalyst become wedged together across the tube, leaving voids. This can lead to hot spots or, in extreme cases, to hot bands during operation.

Fig. 8: "Wagon Wheel" Primary Reforming Catalyst (UCI)

11

MODERN PRODUCTION TECHNOLOGIES

Fig. 9: Plain and "Six-Shooter" Type Primary Reforming Catalysts (Haldor Topsoe)

Some efforts have also been directed at minimizing the negative effect of alkali on the activity of nickel-containing catalysts. For example, UCI claims that the potassium in its new G-91 EW steam reforming catalyst formulation is incorporated in such a way that, while the catalyst retains excellent coking resistance, it has a significantly higher activity than other potassium-promoted catalysts on the market11. It is also reported that the catalyst can cope with occasional sulphur excursions in the feedstock without permanent damage of the kind that results in hot spots or an increase in pressure drop. In 1994 ICI introduced pre-reduced versions of its alkalized and non-alkalized steam reforming catalysts, which allows faster start-up and gives maximum activity in the top section of the tube11. Pre-reduced catalysts are now likewise available from Tops0e and other catalyst suppliers. The split-loading of reformer tubes with more than one type of catalyst mentioned above has now become very common. More and more plants are now using this approach and benefiting from better performance in respect of factors such as pressure drop at increased plant load, carbon formation potential, catalyst activity, catalyst cost and desired catalyst life. For example a reformer tube may be loaded with 15% alkali-free in pre-reduced form

(top section), 25% unreduced alkalipromoted (middle section) and 60% alkali-free unreduced (bottom section). The secondary reforming catalyst is traditionally made in the form of Raschig rings for the bulk and solid tablets for the top layer. In comparison with primary reforming catalyst, the nickel content is lower (5-10% against 15-25%). This, along with higher sintering rates due to the higher temperatures, accounts for the lower activity, which is only 5-10%

of that of primary reforming catalyst. For some time it was thought that there would be no advantage in using shaped catalyst in this adiabatic type of reforming and that it would probably also not be stable enough to withstand the mechanical and thermal shock of starting and stopping the process air flow. But catalyst manufacturers have meanwhile increased the mechanical stability of the shaped catalysts so much that, even in a secondary reformer, advantage can be taken of the higher activity resulting from the increased geometrical surface, allowing the catalyst volume to be reduced. This frees additional space for the gas-gas mixing zone above the catalyst bed - especially beneficial in a plant that is being operated at a much higher rate than its original design. Figure 10 gives an idea of how much it is now possible to reduce the volume of catalyst loaded in the secondary reformer. Some contractors and catalyst vendors use target bricks instead of inert balls to protect the top layer of the catalyst, while others have no protection at all. The catalyst lifetime is usually in excess of six years.

Reformer tubes
The reformer tubes have to operate under radier severe conditions (metal

Fig. 10: Reduced Secondary Reformer Loading Volumes for M. W. Kellogg Plants
Design Current practice

l"Alj03balls

lump, 2" balls or target brick

heat shield Cron I" rings

Ni on I.2"AI2O3 multi-hole ring Ni on AI2O3 % x %" rings Ni on AI2O3 1.2" multi-hole ring 2" AI2O3 balls

12

AMMONIA

temperatures up to 900C and internal pressures as high as 40 bar). At these high temperatures the tube material will undergo creep, which will lead in time to rupture. The time after which failure occurs is a function of stress and temperature. The "rupture" data are derived from laboratory tests of material samples having a standardized geometrical form. The convenient way to express these data is to plot the stress versus the Larson-Miller parameter, P: P = T {log t + KJ/1000 where T = material temperature t = time K = a material-dependent constant. Using these data the reformer tubes are usually designed for an expected lifetime of 100,000 hours. For practical purposes plots are often drawn of rupture stress versus temperature with the time to rupture as a parameter. An example is shown in Fig. 11. A698HK40 was the commonly accepted and widely used material for steam reformers up to the mid 1970s, when Pompey (now Le Manoir) came up with its Manaurite 36X. This alloy has a much higher nickel content than HK40 and somewhat more chromium too, and it is stabilized with niobium63"65. Its creep rupture behaviour is superior not only to HK40 but also to A297HP, which was also used in some instances as an advance on HK. Alloys comparable to Manaurite 36X are also offered by other vendors, for example G4852 from Schmidt & Clemens or Paralloy. The high creep-resistance of these materials, commonly referred to as HP modified, is attributable to heatstable complex chromium-niobium carbides. Generally the strength of these high-temperature alloys appears to be related to the low soluI l
Grade

Fig. 1 1 : Mean 100,000-Hour Stress Rupture Values vs Temperature for HK40 and Micro Alloy Material
45-

40353025-

H39WM

2015105850 875 900 925 temperature, C 950 975


1,000

bility of carbon in a high-nickel matrix. This leads to crystallization of carbides on the grain boundaries. It is assumed that these carbides hinder the inter-granular dislocations observed microscopically which, macroscopically, translate to the wellknown phenomenon of creep. On account of their superior high-temperature strength and stability, it is possible to design high-duty reformers with outlet pressures as high as 43 bar when using HP modified tube material. As mentioned previously, the endo-thermic reforming reaction - equation (7) - proceeds with an increase of volume; so, according to Le Chatelier's principle, a higher pressure will shift the equilibrium to the methane side of the equation. To compensate for this effect and to attain the same residual methane content, it is necessary to raise the temperature. Recently a new-generation tube material has emerged, called Micro Alloy64"67. This contains not only niobium but also titanium and zirconium (or lanthanum), which has improved creep rupture strength still further. The superior creep rupture strength compared to HK40 is seen

in Fig. 11. Table VI shows the chemical composition of the tube materials discussed. Manufacturing these advanced grades seems to be much more delicate from the point of view of quality assurance. Inevitably they are considerably more expensive than HK40. That is actually already the case for the HP modified and even more so with the Micro Alloy material. This, of course, adds another variable into the design optimization puzzle. The enhanced materials are equally useful in revamping older plants. Because they are inherently stronger, it is possible to retube the furnace with thinner tubes. Without altering the outside diameter, the inside diameter and, therefore, the capacity of the tube can be increased. Thus more catalyst can be accommodated within a tube of a certain outside diameter and, particularly when an optimized catalyst is used with improved packing and heat transfer properties, throughput is increased at (possibly) lower pressure drop than before retubing. Figure 1266 demonstrates this by comparing HK 40 with Micro Alloy material.

Table VI Chemical Composition of Reformer Tube Alloys


Ni
19-22 33-37 32-35 33-37

Cr
23-27 24-28 23-27 23-27

Mn

Si
0.5-2.0 2.5 Si. 5 1.5-2.0

Nb

Ti/Zr

Mo
<0.5 <0.5

C
0.35-0.45 0.35-0.75 0.35-0.45 0.4-0.5

P
<0.04 <0.04 <0.03 <0.03

S
<0.04 <0.04 <0.03 <0.03

A 698 HK 40 A 297 HP HP modified Hicro alloy

<2.0 <2.0

0.7-1.5

<0.5

13

MODERN PRODUCTION TECHNOLOGIES

Fig. 12: Benefits of Micro Alloy in Reformer Operation


RUPTURE STRENGTH WALL THICKNESS (MSW) CATALYST VOLUME

"HK40"

H39WM

"HK40"

H39WM

"HK40"

H39WM

Furthermore, the thinner the tube wall is at a given heat flux, the lower the radial temperature gradient in the tube wall will be. This reduces the thermal stresses which contribute to creep fatigue.

Furnace design and layout


Steam reformers usually operate close to their design limits, especially with respect to the tube material. The layout of a primary reformer furnace is, therefore, a complex task in which a number of parameters have to be well balanced. Two aspects have to be considered: the heat-consuming reaction of the hydrocarbon with the steam inside the tubes and the heat supply by radiation from outside. The object is to match both properly so as to achieve adequate conversion of the hydrocarbon at an economic steam/carbon ratio and at a reasonable tube wall temperature profile. The progress of the reforming reaction along the tube, which determines the amount of heat needed at each location and the amount of radiation supplied from the outside, results in a specific tube wall temperature profile. For the purpose of determining the requisite creep rupture strength to achieve a selected lifetime, the peak temperature is used. For design optimization, contractors and licensors have computerized mathematical models which take into account the many variables involved in the physical, chemical, geometrical and mechanical properties of the system. ICI, for example, was one

of the first to develop a very sophisticated and effective modelling of the primary reformer. Their program "REFORM"11-14 can simulate all major types of reformers (see below): top-fired, side-fired, terraced-wall, concentric round configurations, the exchanger reformer (GHR) and so on. The program is based on reaction kinetics, correlations with experimental heat transfer data, pressure drop functions, advanced furnace calculation methods, and a kinetic model of carbon formation68. The kinetic expression for the steam reforming reaction used in reformer models has to be based on the activity of the specific catalyst used. Methane reforming (equation 7) can be described as a first-order reaction. At low temperature, the molecular diffusion rate is higher than the reaction rate, so (at least in theory) full use can be made of the catalyst activity. Conversely, if the temperature is high, the conversion rate is governed by pore diffusion and the effectiveness of the catalyst is, in consequence, reduced. The reaction mechanism of methane reforming and the more complex naphtha reforming44 has been studied by various researchers and is discussed in detail elsewhere18. As mentioned earlier, naphtha seems to undergo a methanation reaction (equation 13) at 400-600C, and the methane is converted according to equation (7), but there are indications that intermediate products of naphtha degradation such as propane, ethane or ethylene react directly to
14

hydrogen and carbon dioxide. Experimentally-derived kinetic equations for various temperature and pressure ranges have been published69"75. Heat transfer within the tube, which occurs mainly by convection because the catalyst is a rather poor heat conductor, is modelled by correlating results gained in heat transfer measurements. Heat transfer on the outside of the tube is practically entirely radiant. Furnace calculations are usually done according to the method of Roesler76 using numerical methods which allow three-dimensional simulation. To create a furnace model in three dimensions is very important: it also takes into account, for example, the shielding effect by adjacent tubes, which may cause circumferential variations in the wall temperature of individual tubes..Another contributing factor to be modelled is the heat release pattern of the burner flame. Usually the number, diameter and length of the tubes are first tentatively selected on the basis of capacity and pressure drop considerations. These quantities will then be optimized in an iterative procedure, adjusting all the interdependent design features. For example, in a topfired furnace the distance between the rows of tubes has a strong influence on the peak wall temperature, but, of course, also on the size of the furnace and, consequently, on its investment cost. Tube pitch influences skin temperature, too, through the shielding effect mentioned above, which must also be factored in. Tube length is important for the radiant efficiency of the reformer box. On the other side of the optimization puzzle are the investment costs mainly influenced by the cost of the tubes - and the size of the furnace box. Designing for very high heat flux will reduce the dimensions and, consequently, the investment cost, but at the expense of operational reliability, whereas a rather more conservative furnace lay-out is more likely to give lengthy, trouble-free operation. Inevitably, though, the design of an industrial steam reforming furnace is a compromise between economics and technical feasibility. It is, for example, possible to design a reformer in which the tubes have an indefinite life, but it is hardly economic.

AMMONIA

Fig. 13: Typical Temperature Profiles for a Top-Fired Reformer


inlet a.pigtailsv

process gas

400

800

1,200

1,600

temperature, C

Normal practice is to design for a minimum tube life of 100,000 hours on the basis of creep rupture data. The sketch in Fig. 13 shows the principle of a top-fired furnace with the corresponding temperature profiles12. Simply from the point of view of

tube life, temperature profile A would be the best, because the maximum temperature is at the end of the tube where pressure is lowest. Temperature profile B demands thicker, more expensive tubes; on the other hand, the furnace box will be smaller. The maximum reaction and,

Fig. 14: Tube Wall Temperature and Heat Flux Profiles for Top-Fired and Side-Fired Configurations
top-fired reformer average heat flux: 75,000 W/m2

1 140

1,000 r

S
60

- 20

20

40 60 tube length, %

100

consequently, the highest demand for reaction heat and heat flux occurs in the upper part of the tube. In many modern top-fired furnaces the mean heat flux calculated for the inner tube wall surface is around 60,000 W/m2, although in some designs it can be as high as 75,000 W/m2. The maximum heat flux may be 1.4 to 2 times higher. In side-fired and terraced-wall furnaces (where mean fluxes are generally in the range 60-85,000 W/m2), the difference between mean and maximum flux is much smaller, as shown in Fig. 1415. When all is said and done, with a good reformer design (which is standard today for all contractors of repute and experience), there should be no problem in reaching the 100,000-hour tube life target. That this is possible even with standard HK40 tubes is shown by the example of an ammonia plant reformer with a rather low mean flux of 54,000 W/m2. After 110,000 hours of actual operation, the tubes were subjected to a metallurgical investigation, which showed that they could be expected to last for another 30,000 to 40,000 hours. But good design is only half the story. Just as important is the proper operation of the plant and the related equipment of the ammonia plants. Every year from all over the world there are still reports on premature tube failures, most of which can be attributed to maloperation or outright operator errors. It has been a constant topic at the annual AIChE Symposium on "Safety in Ammonia Plants and Related Facilities" over the years and right up to the present day. Among those influences which can definitely shorten the service life time of reformer tubes are catalyst poisoning, loss of steam, thermal shock, overfiring in fast start-up procedures, and condensation in hot tubes (for example, water carry-over from the steam system or condensation falling from the cold end of the tubes)77'79. Any loss of catalyst activity - by sulphur poisoning, for example reduces the rate of reaction and, therefore, the rate at which heat supplied in the furnace is absorbed. The tube wall temperature in the affected part of the tube therefore stabilizes at a higher temperature

15

MODERN PRODUCTION TECHNOLOGIES

than it should. This is observable as a so-called "hot band". Because, under normal circumstances, the hottest parts of the tubes are already operating close to the limit of the structural stability of the material, even running only slightly hotter than design has a serious effect on tube life. Hot bands, or sometimes only hot spots, may be experienced where the catalyst has not been loaded properly and there are as a result voids in the catalyst layer because of "bridging" of the pellets. Clearly, too, since it is the reacting gas which takes the heat from the tube wall, it is important that the gas flow through all the tubes is the same. To achieve this the catalyst has to be loaded very carefully so that there is a minimum of variation in the pressure drop (better than 3%) between the individual tubes. A tolerance of 10% could cause a temperature change of about 12C78. Problems like direct impingement of flames on the tubes or bowing of tubes should not happen in a welldesigned furnace. One important point not yet addressed- and, indeed, not so far well enough investigated and documented- is the effect of thermal cycling on tube life: to what extent is it influenced by the number of times a furnace is heated during start-ups and cooled during shutdowns? It is known, of course, that each time the furnace is heated up, the tube metal undergoes a certain amount of transformation, from a rather brittle state at low temperature to a much more ductile state at

high temperature in which, under the influence of pressure, it creeps. While creep is to some extent reversed during shut-down, this reversibility declines with operation time and the number of cycles. The mechanism is understood in a qualitative way but it is not clearly defined how many start-ups and shutdowns should be allowed for in designing for the 100,000-hour tube life. Experience is that thicker tubes (o.d./i.d. ratio > 1.35, e.g. 17 mm wall thickness for 100 mm i.d.) are more prone to fatigue induced by thermal cycling. In this respect, too, the enhanced tube materials are advantageous because they allow tubes with thinner walls to be used78. It should also be mentioned that it is nowadays general practice to machine the tubes on the inside to remove the 2-3 mm porous layer formed during centrifugal casting. This layer contributes nothing to the strengel of the tube and is susceptible to corrosion attack. Because the sound wall thickness is more dearly defined in these machined tubes, thinner tube walls can be allowed. For monitoring the condition of the tubes and assessing the remaining lifetime, a number of non-destructive techniques are available. It is generally inadvisable to rely on any single technique; it is much better to use several in parallel. Nonetheless, the results obtained have to be evaluated with some care and, sometimes, caution. The simplest method, which should always be included, is to keep track of the increase of tube diameter with time.

Reformer types and firing conditions


The arrangement of the burners is a distinguishing characteristic by which the reformer designs may be grouped (Fig. 1512). The overall thermal efficiency is not very different and can be - for example, in a topfired furnace - as high as 95%. But the split in heating duty between radiant and convection (heat recovery) sections will be different. Top-fired reformers are preferred for large capacities. It is possible to accommodate 600 to 1,000 tubes in the radiant box. Very large furnaces for methanol plants, where there is mostly no secondary reformer and the furnace carries the whole reforming duty, have been built by companies such as Kellogg, Jacobs (formerly Humphreys & Glasgow) and Uhde. HistoricaUy, Humphreys & Glasgow has constructed reformers with 896 4-inch (ID) tubes in 16 rows in a single radiant box. Top-fired reformers in world-scale ammonia plants usually have between 300 and 400 tubes. The system has several advantages: Firing occurs only on one level, so the number of burners in relation to the number of tubes is smaller than in side-fired systems. This simplifies distribution piping for fuel gas and preheated combustion air, which is nowadays standard in all efficient plants. The radiation efficiency is higher than in other designs. The burners are located close to the "cold" inlet of the feed/steam mixture, which is where the

Fig. 15: Typical Primary Reformer Burner Arrangements

1
top-fired bottom-fired side-fired (radiant wall) terraced wall

16

AMMONIA

strongest heating is needed. Heat fluxes of 125,000 W/m2 and more can be attained in this area. Less structural steel is needed in construction. However, the heat input is adjustable only to a limited degree. In side-fired reformers the burners are located in the wall and the box accommodates one or two rows of tubes, which receive their heat mainly by radiation from the walls of the furnace box. This is claimed to provide a very uniform heat distribution, which may additionally be adjusted by control of the individual burners. The larger number of burners makes fuel and preheated combustion air distribution more complicated and more expensive. As the height and width of the reformer are fixed by the radiation geometry of the tubes and furnace box walls, it is only possible to extend the reformer lengthwise to accommodate more tubes. This limits it to 100-150 tubes, after which multiple radiant boxes become necessary. Therefore this system seems to be more suited for smaller capacities or for low-severity reforming in a largersize plant, as is the case in the Braun Purifier process or in the ICI AMV process. The maximum wall temperature is at the tube outlet, whereas the maximum heat flux is at relatively low temperature. The radiation efficiency is smaller than in the top-fired system. Bottom-fired furnaces are not very common in modern ammonia plants. They have a rather constant heat flux profile along the tube with high metal temperatures on the outlet side. The terraced-wall type may be regarded as an intermediate between the side-fired and bottomfired species. The reformer has inclined walls with several terraces on which upward firing burners are installed. This unique burner positioning makes it possible to adjust the heat flux in each zone.

welded to a horizontal header located inside the fire box between the flue gas tunnels. With an equal number of tubes on both sides a central riser connects the "tube harp" to the water-cooled transfer line on top of the firebox (Fig. 1680). The tubes are suspended by spring hangers. This arrangement allows furnaces to be made rather compact and, because of the use of the partially pre-fabricated "harp modules", construction can be speeded up. Dealing with single tube failures, however, will be somewhat elaborate, because not only must the furnace be cut off from feed and fuel; it must also be cooled down because it is necessary to enter the furnace to isolate or to replace the defective tube. Nevertheless the concept has been incorporated worldwide in more than 140 plants and has proved to be very reliable, and it has been further improved in recent years. It is well suited for very large installations, as exemplified by a furnace for a 2,500 st/d methanol plant in Chile. Uhde has developed for its topfired furnace a unique arrangement for the reformer outlet which has performed excellently for more than

20 years in large furnaces. The tubes are rigidly connected to the header outside the fire box as shown in Fig. 1781-82. In the case of an individual tube failure the furnace does not need to be cooled down. With feed and fuel shut off, the outlet is cut at the welding seam and is sealed by welding a cap to the stub on the manifold. At the same time the inlet pigtail connection is sealed. The failed tube will be left in the firebox until the furnace is shut down completely for other reasons. Many top-fired reformers have outlet pigtails; examples are the Jacobs and ICI designs. The advantage claimed is that, when a failure occurs, the inlet and outlet pigtail can be nipped by means of a hydraulic clamp to isolate the plant without interrupting operation78. But this is a very delicate procedure from the point of view of safety, and it has lost much of its former attraction because failures are much less frequent on account of better tube materials and improved furnace designs. As the pigtails have to be made of a more malleable material like Incoloy 800H, which has lower strength than the tube material, this might introduce an additional possibility for failures. Side-fired and ter-

Fig. 16: The Kellogg Reformer


effluent chamber spring hangers

effluent chamber

, inlet pigtails

platform secondary reformer

outlet manifold

Special features of tubular reforming furnaces


Most contractors' processes have their own special reformer design features. Kellogg prefers a modular concept. The tubes of each row are
17

MODERN PRODUCTION TECHNOLOGIES

raced-wall reformers also use outlet pigtails in various forms.

Improving the efficiency of fired tubular reformers


In current reformers with normal tubes only 35-55% of the fired load is absorbed into the reaction zone. The actual value depends on the heat transfer rate, fuel composition, air preheat and, of course, on the reformer geometry. In side-fired fur-

naces a proportion of the radiant heat is usually lost into the convection section by a "viewing effect". With integral radiant/convection tubes, patented by Brown & Root83, the efficiency can be increased considerably and up to 65% of the fired load can be absorbed in the reforming reaction. For example, in an existing reformer, fuel consumption may be reduced by 20% while the capacity is increased by 20%. The upper part of the tubes is finned,

which makes it possible to achieve a very close temperature approach of about 200C between the inlet gas to the catalyst tubes and the flue gas leaving to the convection bank. A capital saving of 4.5% against a conventional furnace arrangement is claimed. The disadvantage is a higher pressure drop (the typical increase is 45 mm water), which requires tight bellows seals at the catalyst tube inlet11. Figure 18 shows how this modification is made in an existing terrace-wall reformer.

Fig. 17: Uhde Reformer Tube-Manifold Connection

Breaking away from the traditional fired furnace


It has already been mentioned that the tubular furnace and its associated convection bank (heat recovery section) is rather expensive and contributes substantially to the investment cost of the total ammonia plant. In some modern concepts its size has already been reduced by shifting some of the reforming duty to the secondary reformer, necessitating die use of a superstoichiometric amount of process air. An incidental benefit of reducing the primary reformer duty in this way is that, because less firing is required, less flue gas is produced and the NOx emission is reduced. Some companies - for example, Jacobs, Foster Wheeler and Lurgi have gone so far as to propose bypassing some of the natural gas round the primary reformer and feeding it straight to the secondary reformer, which then operates as a quasi-autothermal reformer. But there are additional reasons for breaking away further from the fired furnace concept. The temperature level of the flue gas from a traditional reformer is usually higher than 1,000C and the process gas at the outlet of the secondary reformer is also around 1,000C. From a thermodynamic point of view it is wasteful to squander heat of this grade in simply raising steam and preheating process air feed for the secondary reformer: the boiling temperature in the 125-bar main boiler on the secondary reformer outlet is only 325C, and the process air is usually preheated in the reformer heat recovery train only to about 500C. In their modern low-energy concepts the leading contractors have

furnace ceiling reformer tube

furnace bottom

gas conducting tube

outlet manifold

300

600

900

Rg. 1 8: Improving Efficiency of Terraced-Wall or Side-Fired Reformers with Partially-Finned Tubes


catalyst tube \VIr

m
{ 1 1 process 1 preheat coil

extended surface^

'

process jrgjL- preheat coii P^t

j C=*tauxiliary burners

^^ side wall burners

1
i

BEFORE

AFTER

-,

s to be removed

18

AMMONIA

already made improvements by raising the process air preheat temperature to 600-700C. But recycling the high-level process heat from the secondary reformer and making use of it for the primary reforming reaction is thermodynamically the better op'ion. Concepts which use this heat in exchanger reformers have been successfully developed and demonstrated in recent years.

Heat exchange reforming


Several years ago Kellogg84 had already proposed a low-energy variant in which the steam/feed gas mixture is split into three parallel streams. The largest stream, representing 47% of the total, is classically reformed in the catalyst tubes of the radiant section of the furnace. A further 12% is reformed in catalyst tubes located in the convection section, and the remaining 41% in an exchanger reformer, in which the catalyst tubes are heated by the hot process gas from the secondary reformer. Neglecting the delicate detailed mechanical engineering needed to deal with thermal expansions and tube sheet problems, the exchanger reformer can be viewed as a heat exchanger with catalyst inside the tubes. One step further than this is to eliminate the fired primary reforming furnace by using just a combination of an exchanger reformer and an autothermal (secondary) reformer. The hot gas effluent from the autothermal reformer is then the sole heat source. A surplus of air over the stoichiometric demand (about 50%) or oxygen-enriched air in the secondary reformer is required to balance the heat demand for the primary reforming reaction in the exchanger tubes. Such a concept was propagated by Chiyoda85-w. The first to come out with a real technical production unit for such an approach was Id with the Gas Heated Reformer (GHR)87-89. After extensive pilot plant research and detailed engineering efforts, ICI incorporated this reformer in the two LCA (Leadirig Concept Ammonia) units at Severnside, near Bristol, UK, each having a capacity of 450 t/d of ammonia. As Fig. 19 shows, the GHR contains a number of tube assemblies, each with a central open-ended

"bayonet" tube enclosed coaxially within an outer "scabbard" tube. The feed gas/steam mixture enters through the vessel top and passes via the boxed tube sheet down into the catalyst-filled annular space between the inner and outer tubes. From the blind bottom end of the scabbard tubes, the gas passes up through the central bayonet tubes. These are insulated on their inner surface to minimize heat exchange with the reacting gases in the annular space, because that would abstract heat transferred into the reacting gases from the hot effluent of the secondary reformer, which passes through the shell side of the GHR. To enhance heat transfer, the scabbard tubes are finned on the outside and are surrounded by "sheath" tubes. The two GHR units operating at Severnside each produce synthesis

gas for 450 t/d of ammonia. The pressure vessels are 12 m high and have an outside diameter of 2.2 m. They are refractory-lined and have an external water jacket. The tube bundles, which can be withdrawn as complete units, each contain 78 tubes. The tubes are charged with standard ICI 4-hole reforming catalyst (57-4M), 11.5 mm in diameter. An important element of the GHR is the tube sheet design. ICI has performed complex computer modelling of the transient states which occur in the tube sheet during startup, shut-down or sudden plant trips. From the company's eight years of actual operating experience, it is dear that the GHR is mechanically very stable under the thermal stresses induced by repeated start-ups and shut-downs. Detailed inspections have revealed no tendency for cracks or other defects, and after some

Fig. 19: ICI Gas-Heated Reformer


I methane and steam

primary effluent

cartridge

cooled secondary gas exit

scabbard tube

primary catalyst
fin

sheath tube bayonet tube hot secondary gas

refractory

19

MODERN PRODUCTION TECHNOLOGIES

minor modifications to the material and construction of the reformer tubes there is no problem with metal dusting. At Sevemside the GHRs operate under the conditions listed below90.
Inlet pressure, bar 41 Inlet temperature (tube side), C 425 Outlet temperature (to 2ary reformer), C 715 Inlet temperature (shell side), C 970 Outlet temperature (shell side), "C 540 Steam/carbon (overall) 2,5-2.7 Process air temperature, C 670 Methane (exit GHR tube), % 25 Methane (exit 2ary reformer), % 0.67

Fig. 20: Kellogg Reforming Exchanger System (KRES)

~500-SSOC

~660-680C -40 kg/cm'g natural gas plus steam

From these data and some assumptions it is possible to estimate the following:
Hydrocarbon conversion in GHR, % 34 Reformer heat duty, GJ/t NH3 3.9 (H2+CO)/N2 output secondary reformer 1.9 (corresponding to nearly 60% surplus over stoichiometric demand)

PROCESS HEATER

AUTOTHERMAL REFORMER

REFORMING EXCHANGER

Very recently the first contract for a commercial LCA plant has been concluded between Mississippi Chemical in the USA and Id's LCA licensee KTP1. Kellogg has also brought out a similar concept called KRES - the Kellogg Reforming Exchanger System - in which the exchanger reformer is partially by-passed by feeding about one third of the natural gas direct to the secondary reformer with oxygen-enriched air to make up the heat balance92-95. Figure 20 is a basic flowsheet of the concept. KRES was first demonstrated commercially at the Ocelot/ Methanex plant built in 1994 at Kitimat, British Columbia. The installation is used to increase Ocelof s ammonia production; its synthesis gas production capacity is equivalent to 330 t/d of ammonia. On the strength of that successful demonstration Kellogg has used a combination of KRES and the new KAAP synthesis (see page 31) as the basis of its new Ammonia 2000 Process (page 50), which can be designed for capacities of 1,850 t/d or more using only a single KRES. In contrast to the GHR of ICI, this exchanger reformer has only one tube sheet and the tubes are open at the lower end, where the reformed gas mixes with the hot effluent of the secondary reformer. The mixed gas

stream flows up on the shell side to provide the heat for the reformer tubes. Baffles are provided to create a cross-flow and achieve an effective and uniform heat transfer. Proper mixing of the two gas streams is essential for good performance. A multi-hole plate distributor is used to direct the flow of the autothermal secondary reformer effluent up towards the gas exiting the catalyst filled tubes. A computational fluid dynamics program (FLUENT) was used for the design of the mixer and baffle arrangement. The vessel is lined with castable materials in two layers and has a water jacket on the outside. The typical outside diameter of the tubes is 50 mm. A rather detailed description of the design was given in a recent presentation96. One of the prime advantages of exchanger reformers is the substantial reduction in maintenance costs resulting from the elimination of the fired reformer. This is largely because the tubes in the reforming exchanger are much less highly stressed. The heat flux in the reforming exchanger is significantly lower than in the fired reformer, and the potential for hot spots on the tubes is virtually eliminated. In a conventional fired reformer the pressure difference across the tube walls is about 35-40 bar, whereas in the exchanger reformer the maximum pressure difference is only about 3.4 bar. And at the lower end of the tubes, where the tube metal temperature is hottest, the pressure difference is at its lowest (below 1 bar). A key feature of the relatively
20

uncomplicated mechanical concept c KRES is its serviceability. The to] head of the reformer is a full-bor closure; loading and unloading c the catalysts is therefore easy be cause full access to the tube she and tube tops is gained by remov ing the top head of the exchange] The whole tube bundle can also b removed for inspection or repair; n cutting and rewelding of tubes i necessary. Figure 21 presents the perform ance profiles for various proces parameters of the KRES. Typical operating conditions fc KRES have been published96-97:
Catalyst tube inlet temperature, C Catalyst tube outlet temperature, C Shell-side inlet temperature, C* Shell-side outlet temperature, C Methane slip, tube side, mol-% (dry gas) "Temperature of mixed tube outlet and autothermal reformer outlet 5C 93 98 68
1.

The split between the reformin exchanger and the autothermal re former was varied from 25/75 t 30/70, the overall steam:carbo; ratio from 3.0 to 4.0, and the oxy gen content of the enriched air gc ing to the autothermal reformer fror 28 to 32 mol-%. Table V gives th composition of the effluents from th autothermal reformer and from th shell side of the KRES (mixed ga streams). Using these data along with som assumptions, the hydrocarbon cor version in the tubular reforming stej is calculated to be about 95% (27, based on total feed) and the hea duty 2.5 GJ/t NH3. The (H, + CO)

AMMONIA

Fig. 21: KRES Performance Profiles

10

20

30

40

50

60

70

80

90

100

percent of active tube length, top = 0

N2 ratio in the exchanger reformer outlet is about 2.5. A similar exchanger reformer concept called Potformer is offered by KTI (especially for hydrogen production plants). The catalyst tubes are suspended from a single tube sheet and are open at the lower end. Hot process gas from a fired reformer at a temperature in the range 900C down to 600C is mingled with the stream of exit gas from the potformer catalyst tubes in the proportion of about 2:1. The mixed gas stream supplies the heat for the exchange reforming11'98.

Another example of the exchanger reformer technique was developed in the former USSR and came into operation in 1986 in a hydrogen plant with the equivalent capacity of a 400-t/d ammonia plant at Grodno. It is now offered under licence by Brown & Root and Linde. This Tandem Reformer11-"-101 takes heat for the reforming reaction in the tubes entirely from the enthalpy of the secondary reformer effluent. As with the GHR of ICI, primary and secondary reforming operate in series; tube effluent is not mixed with secondary reformer outlet gas.

Table VII Example of Product Gas Compositions in KRES Operation


Autothermal reformer exit mol-%
H,

KRES exit (mixed streams) mol-%


51.55 9.84 11.67 26.18 0.46 0.3

S
A"'
C

",

44.47 9.89 11.23 33.78 0.59 0.03

Tops0e has also patented a design in which hot synthesis gas from the authothermal (secondary) reformer is cooled in a heat exchange reformer, supplying heat for the conversion of hydrocarbons. Because Tops0e developed the concept mainly for application in methanol plants, it was named HERMES (Heat Exchanger Reforming for Methanol Synthesis)11-15-102. A common feature of all the exchanger / autothermal reforming concepts is that little or no high-pressure steam is produced and the need for steam export is eliminated. Total plant concepts based on this technology produce medium-pressure steam and have to rely on imported electric power or auxiliary boilers to supply energy for their mechanical drives. In some cases the process steam requirement is supplied by feed gas saturation using low-level heat. Another important point is the achievable capacity. Whereas all the reputable licensors and contractors in the ammonia could probably build a traditional fired primary/ autothermal secondary reformer combination with a capacity of more around 2,000 t/d in one single line, up until now only one - Kellogg is able to offer an exchanger reformer/autothermal reformer concept as large as 1,850 t/d. An exchanger reformer concept which is even more progressive is Uhde's CAR (Combined Autothermal Reformer), which not only replaces the catalytic secondary reforming step by a non-catalytic partial oxidation but also combines this with the exchanger reformer in one single vessel. Uhde began developing it as long ago as 1982. After successful operation of a pilot plant a demonstration unit has been in operation since 1991, producing 13,000 m 3 /h of process gas at a pressure of 17 bar103-106. The steam/natural gas mixture is passed through the reforming catalyst in the tubes, as shown in Fig. 22. These tubes are heated externally by the hot gas returning from the partial oxidation section below the tubes. For the partial oxidation section a special design is necessary to create a vortex type flow pattern, which ensures intensive mixing with the oxygen or oxygen-enriched air.

21

MODERN PRODUCTION TECHNOLOGIES

The partial oxidation takes place at about 1,300C. The amount of the oxidant governs the residual methane content. A portion of the feed can be fed directly to the partial oxidation chamber. This is not necessary when producing ammonia synthesis gas, but will be used when producing gases with higher CO content for methanol or oxo synthesis. The ratio of primary to secondary feed and the steam-to-carbon ratio determine the CO/CO2 and H2/CO ratios in the CAR outlet.

Fully autothennal reforming


Although the concept of autothennal reforming is used in every secondary reformer, for the purposes of this section the term is taken to mean a process in which the entire-hydrocarbon conversion is performed in a single fixed-bed catalytic reactor, the reforming reaction heat being gener-

ated by internal combustion using oxygen. Unlike a secondary reformer, which is fed with partially converted gas from a fired primary reformer, the autothermal reformer is fed directly with hydrocarbon feedstock and steam. Owing to the higher heat of reaction to be supplied by the internal combustion (combustion temperatures 2,0009C and higher), the flow conditions, heat release characteristics and the risk of soot formation are very different from what they are in a normal secondary reformer. Special attention must therefore be given to the design of the burner and the reactor15. The burner has to provide proper mixing, giving a combustion zone which is a turbulent diffusion flame/ which is followed by a thermal zone in which homogeneous gas-phase reactions take place. In the fixed catalyst bed thermodynamic equilibrium is attained in the meth-

Fig. 22: Uhde Demonstration Combined Autothermal Reformer (CAR)


inlet
SECTION A-A

outlet sandwich type tube sheet

refractory lining

enveloping tube reformer tubes tube sheet

ane reforming reaction of equation (7) and the water gas shift reaction of equation (3). The combustion is a substoichiometric reaction with an overall oxygen-to-hydrocarbon ratio of 0.55-0.6. The process is especially suited for the production of methanol synthesis gas or hydrogen, but can also be applied to ammonia production, using enriched air instead of oxygen; this has been successfully done in the KRES concept of Kellogg. Primary reforming of natural gas is normally performed in fired tubular furnaces because it was formerly thought not to be technically feasible to preheat the process gas up to the extremely high temperatures needed to run the reforming reaction in an adiabatic reactor. But with the Rgate process gas heater10'107 this should be possible. The Rgate is similar to a steel works regenerator, except that it operates on much shorter cycles of only 2-3 minutes. It is made up of two refractory-lined vessels, each filled with a heat retaining mass and equipped with a burner. The vessels are operated alternately between a combustion phase for reheating and a heating phase for process gas. The hot gas line to the process has no valves; instead, the process is automatically controlled by valves on the inlet lines and on the flue gas line. On account of the short cycle time there is only a drop of 40C in the process gas passing to the adiabatic reforming reactor. So far, however, only pilot plant experience is available. Pre-reforming Pre-reforming108-110, introduced in recent years as a means of reducing the primary reformer heat duty, is actually an old gas works process in a new incarnation. Developed by British Gas111 and by Lurgi/ BASF112-"5, it was originally used to convert naphtha into town gas. In the more recent past it found application in CO/H2 plants with low steam and high carbon dioxide ratios in the reformer feed. The process is suited for hydrocarbons ranging from natural gas to heavy naphtha (200C). For natural gas the overall reaction will be endothermic, but for

water-cooled oxygen nozzles


sec,

feed nozzles

refractory iining

water-cooled start-up burner

water jacket

22

AMMONIA

Fig. 23: Typical Integration of a Prereformer

steam generation and superheat

Fig. 24: Basic Concept of Prereforming

inlet to reheat

inlet to prereformer

200

400

600

800

temperature, C

heavier feedstocks like naphtha it is exothermic or in heat balance. Because of the low operation temperature (around 500C), the catalyst readily picks up sulphur and acts as a guard to the primary reformer catalyst. Figure 23 shows how a prereformer is integrated into a steam reforming synthesis gas plant, and the basic concept of how a pre-reformer reduces the load on the primary reformer may be understood from Fig. 24, which compares the pre-reforming arrangement with the conventional configuration116. The mixed feed enters the fixed bed of the pre-reformer catalyst at about 530C instead of being fed to the primary reformer tubes at the same temperature. A temperature drop of about 60-70C occurs in the catalyst bed on account of the overall endo-thermic reaction. The steamto-carbon ratio ranges from 0.3 for natural gas to 1.5 for naphtha. Medium-grade heat is used to reheat the exit gas to the correct reformer entrance temperature. As a result, a smaller amount of high-level energy is needed in the main furnace to achieve the same net effect. This reduction in reformer duty corresponds to a fuel saving of about 510%. The "compensation heat" can be taken from a variety of sources, including flue gas, process gas after secondary reformer, gas turbine exhaust, etc. When it is taken from primary reformer flue gas, it substitutes combustion air preheat. In that case there may be no appreciable improvement in the overall energy consumption for the ammonia plant, as a study for a 1,350-t/d plant revealed110. But the heat duty for the tubular reforming came down from 250 to 200 GJ/h, allowing worthwhile savings in capital and maintenance (no combustion air fan, no elaborate air ducting to the burner, fewer reformer tubes, etc.), which are not out-weighed by the relatively small additional cost of the pre-reformer unit. Because pre-reforming actually converts higher hydrocarbons into the equivalent of natural gas, it is an elegant possibility for adapting a plant designed for natural gas to use naphtha feedstock instead, an option which is of interest in places with restricted natural gas supply, as in some locations in India117. Previously

23

MODERN PRODUCTION TECHNOLOGIES

published sources give good descriptions of the various possible configurations for the incorporation of a pre-reformer into an existing plant62- 116' n7, case studies118 and industrial experience119-m. A review of the developments in natural gas reforming technology for synthesis gas was presented at the 1993 AIChE Ammonia Safety Symposium in Orlando, Florida121.

Carbon monoxide shift conversion


The conversion of carbon monoxide and steam to carbon dioxide and hydrogen by the shift reaction is a very important step in the process sequence leading to ammonia122. A high conversion in this stage means that there will be a low residual carbon monoxide content. That is important for high process efficiency for two reasons. First, any residual carbon monoxide must be converted back to methane to prevent it from reaching and damaging the ammonia synthesis catalyst, which consumes some of the hydrogen. Second, the methane increases the rate of purge gas withdrawal that is required to prevent it from accumulating in the synthesis loop and thus the loss of unused hydrogen. The shift reaction, shown earlier in equation (3), does not involve a net change in volume and is therefore very little affected by the operation pressure. But it is exothermic, and for mermodynamic reasons it is necessary to conduct the reaction at the lowest possible temperatures at which it will proceed at a practical rate so as to achieve the minimum carbon monoxide content. In common with many such equilibria, a catalyst is used to lower the minimum operating temperature. Because most active shift catalysts are also very sensitive to overheating and poisoning, the conversion is traditionally done in two stages: the majority takes place in the first (HT) stage at an inlet temperature of around 350C with a rugged iron oxide-based catalyst, which can withstand the reaction heat and is insensitive to sulphur and chloride poisons in the process gas, and then the gas is cooled to 200-220C for the second (LT) stage, which uses a copper-zinc catalyst.

A surplus of steam helps move the equilibrium point in the desired direction, but for the sake of energyefficiency that should be kept low, since the condensation energy of the process condensate after the LT shift is recovered at a rather low and therefore not particularly useful temperature level. From Fig. 25 it can be seen that, given a sufficiently active LTS catalyst, an acceptably low equilibrium concentration of carbon monoxide can be attained at quite low steam-to-gas ratios123. As we have already seen, the primary reformer can be operated quite safely at the steam-to-carbon ratios which would result in these low steam-togas ratios in the LT shift. But unfortunately the HT shift stage is less tolerant.

metallic iron. In addition, thi Boudouard reaction of equation (8 becomes significant under these con ditions. Carbon accumulates in thi catalyst particles until they star breaking down, increasing the pres sure drop. In the absence of an ad equate amount of steam, carboi monoxide also reacts with the mag netite in the catalyst to form iroi carbide, as shown in equation (19-) Iron carbide itself is a very activi catalyst for the Fischer-Tropsch re action in which methane and highe hydrocarbons are formed, as showi in equation (20)mu5. 5Fe3O4 + 32CO 3Fe.C2 + 26CO2 (2x+l)H2 + *CO (19

(20

High-temperature shift (HTS)


A steam surplus is necessary in the HT shift stage to suppress undesirable side reactions. Cutting it back lowers the oxygen-to-carbon ratio in the HT shift to the point where the atmosphere is sufficiently reducing to reduce the magnetite (Fe3O4) in the classical HTS catalyst not just to FeO but at least partially as far as

Over-reduction and carbide for mation are obviously cumulative and so the formation of higher hy drocarbons gets worse as time goe on. The Sdchemie/UCI group ha introduced a copper-promoted ver sion of the classical HTS catalys which almost completely suppresse the reactions mentioned above a

Fig. 25: Equilibrium CO in LTS Converter Outlet Temperature

0.4-

composition of gas at entry, vo!-% (dry) H2 N2 CO : : : 59.94 20.47


3,11

0.3-

CO, CH.

15.91
0.26 0.32

0.2

8
o.i -

ISO

200

220

240

260

converter outlet temperature, C

24

AMMONIA

steam-to-gas ratios down to 0.4-0.5 125 an d CO/CO., ratios of around 2 (Fig- 26). For comparison, a classical catalyst in a conventional plant operating at 32-35 bar and a CO/CO2 ratio of 1.65-2.13 requires a steamto-carbon ratio of 3.0-3.3 for reliable operation. In the Braun process, where the CO/CO2 ratio is lower, at about 1.2, on account of the surplus air introduced into the secondary reformer, trouble-free operation is possible at a slightly lower steamto-carbon ratio of 2.8-2.9. Obviously the amount of overreduction is a function of the concentration balance between reducing components (CO f H2) and oxidizing components (CO," + H2O). It is assumed62 that the copper promotion enhances the reaction rate on the surface of the catalyst significantly, thereby lowering the local partial pressure of CO to the extent that over-reduction is prevented and rather low steam/gas values are possible without experiencing Fischer-Tropsch reactions. Copper-promoted HIS catalysts of similarly good performance are also available from other catalyst vendors, for example the ICI Katalco 71-3. Topsee has additionally developed an iron-free, copper-based HTS catalyst, allowing lower operation temperatures. This actually leads to the quite recently introduced concept of an intermediate-temperature shift conversion to replace the traditional HTS /LTS approach. That will be discussed later.

each other by zinc oxide and even smaller alumina crystals. This helps prevent sintering of the copper crystallites and resultant loss of activity in the normal temperature range127'128. Intensive research and development in this field has produced better LTS catalysts with improved activity, higher resistance to poisoning, low by-product formation and longer lifetime. Figure 27126 gives an example of the improvement in an activity test. These activity tests are important for comparing initial activity but provide no information on the long-term effectiveness of the catalyst, which is strongly influenced by its performance in the presence of

poisons. The main poisons are sulphur and chlorine compounds. Sulphur, in the form of hydrogen sulphide, usually comes from the natural gas, and occasionally when a new HTS catalyst has not been desulphurized long enough, whereas chlorine may enter with the process air at locations by the sea or close to factories making chlorinecontaining chemicals. Hydrogen sulphide is adsorbed at rather low partial pressure. As thermodynamically it reacts preferentially with zinc over copper, the adsorbed sulphur is rapidly transferred to the zinc oxide, converting it to zinc sulphide. This prevents quick sintering of the copper surface,

Fig. 26: Formation of Higher Hydrocarbons in HT Shift


140120"

,--"
G-3/C-I2-3 .-" conventional HTS x - X x" x

'

1
;
I
t. <U

80-

_s

.X

f 60Q

.s'
f

C rt

u"

40

-<

***

'

X "

20-

G-3C/C-I2-4 promoted HTS

o0. 0
0.5 1.0 1.5 2.0 on-stream time, years 2.5 3.0 3.5

Low-temperature shift (LTS)


LTS-catalyst is supplied in pellets consisting of a mixture of copper oxide, zinc oxide and alumina. The copper oxide has to be reduced in situ using hydrogen and a carrier gas (usually nitrogen), to form the fine copper crystallites on which the activity of the catalyst depends. The properties of the catalyst are, in fact, influenced far more by the formulation and manufacturing procedure than its chemical analysis. It makes ;^reat difference whether the components are just mixed physically together as oxides or are incorporated by co-precipitation126. The latter procedure assures a rather fine distribution and copper crystals of about 10'6 cm can be attained that way, which are well separated from

Fig. 27: Development of Activity of BASF LTS Catalysts


2.01

SV(dry) 10,000/h P 30 bar S/G 0.25 v/v COlnt3vol-%

1.01

equilibrium
0.0

ISO

190

200
C

210

220

230

25

MODERN PRODUCTION TECHNOLOGIES

and for that reason it is important to have a surplus of free available zinc oxide. Intensive studies have elucidated the mechanism129'130. When the sulphur-catching capability of the zinc oxide has been exhausted, the copper crystallites will sinter and lose their activity. The process moves in a rather sharp profile through the catalyst and can be followed by observing the temperature profile during operation and also by analysing the sulphur content along the length of the catalyst bed (Fig. 28). Because activity loss progresses as such a welldefined front through the catalyst bed instead of as a diffuse process affecting the whole catalyst mass, it is possible to use a guard bed ahead of the bulk catalyst. Formerly it was filled with zinc oxide, but nowadays it contains sacrificial LTS catalyst too. Changing this guard bed more frequently extends the life of the main catalyst charge considerably131. Poisoning by chlorides is more of a problem because of the low melting points and high water-solubilities of zinc and copper chlorides. They therefore migrate easily and become redistributed more diffusely through the catalyst mass. Like sulphur, chlorine deactivates LTS catalyst by accelerating sintering of the copper particles. In basic chemical composition and formulation, LTS catalyst is very

similar to methanol catalyst, and that creates a nasty problem, which is the formation of small quantities of methanol. It may be taken as a general rule that the higher the shift activity of an LTS catalyst the higher its activity for methanol formation will be, too. But the relative selectivity is higher for the more active catalyst One set of tests, for example, showed that the activity towards methanol formation increased by only 40% when the shift activity was raised by 90%. As high-activity LTS catalyst can be operated at 10-15C lower, no more methanol is formed than with the former low-activity catalyst132. In a consecutive reaction, amines (mainly methylamine) are formed from the methanol and traces of ammonia originating in the secondary reformer and the HTS. Both methanol and amine formation are higher with fresh catalyst and will decline strongly in the first 100 days of operation, after which the rate of decline is slower and parallels the loss of shift activity.

Intermediate-temperature shift (ITS)


Operation with intermediate temperature shift catalysts is a relatively new process concept and has so far been demonstrated only in a couple of ammonia plants, but it is expected to become a serious option for future ammonia plants. The ITS cata-

Fig. 28: Sulphur and Temperature Profiles for Used LTS Catalyst
100,000

10,000

1,000

100-

top

reactor height

bottom

lysts are based on a copper/zinc/ alumina formulation, optimized to perform efficiently in a wider temperature range than the standard LTS catalysts. The main advantage is that lower steam-to-carbon ratiof are possible in the reforming section because ITS catalyst does not catalyse Fischer-Tropsch reactions. In concepts where only one shift conversion step is used, followed by a PSA unit, the lower CO slip of 1-2% against 3% with iron HTS catalyst is tolerable. The ITS catalyst may be operated adiabatically, in the same way as an LTS catalyst bed, or isothermally, in which case the catalyst is cooled by an internal heat-exchange system using hot water circulation in the tubes. The catalyst volume in the isothermal system is lower than in the adiabatic one, but the catalyst must have a higher stability, and the neat method of keeping track of deactivation by monitoring the temperature profile is not applicable. Nevertheless, the low exit temperature provides more favourable equilibrium conditions and leads to a lower methane slip133. In a 1,350-t/d ammonia plant built for Gujarat State Fertilizer Co. (GSFC) at Baroda, India, Linde AG has used its spiral-wound reactor (originally introduced as a methanol synthesis converter- see page 82) for a one-stage isothermal shift conversion using copper-zinc-alumina modified LTS catalyst. It achieves a CO slip of only 0.7 mol-% (dry basis)134. The Linde spiral-wound reactor can remove very large reaction enthalpies. This is of especial interest for partial oxidation gases, which may have carbon monoxide contents of nearly 50 mol-%. One has already been installed in a partial oxidationbased 30,000-Nm3/h hydrogen plant, where it reduces the carbon monoxide content from 50 to 0.8 mol-% in a single step135. Recently Ammonia Casale (and also Jacobs in the past) has proposed using a radial-flow design for shift conversion reactors instead of the axial flow pattern and has developed a design for revamping existing shift converters in situ136. Although smaller, longer-lasting catalyst particles are used, which have a longer service life, this arrangement still presents a lower pressure drop, saving energy.

26

AMMONIA

Additional ways of achieving low residual CO concentrations


Some years ago Engelhard Industries developed the Selectoxo process, in which the level of carbon monoxide after the LT shift was reduced further by selective oxidat.jn on a palladium-based catalyst, using a stoichiometric amount of oxygen in the form of air137-138 (Equation 21).
CO + '/aO2 ^ CO2 AH298 = -283.2 kj/mol (21)

The process saved hydrogen, ',vhich would otherwise be needed in the methanation stage. Engelhard claimed an energy saving of 0.28 GJ/t NH3 at that time, though subsequent improvements in LT shift catalysts have to some extent diminished the potential saving. The reactor operated at a temperature between 40 and 135C, according to the requirement of the succeeding carbon dioxide removal system. Another process for removal of residual CO and carbon dioxide which can be installed between the carbon dioxide scrubbing system and methanation is methanolation139-14, proposed by Tops0e. It catalytically converts these impurities to methanol, which is recovered by water scrubbing and distillation.

Carbon dioxide removal


Considerable progress has been achieved in improving the efficiency of carbon dioxide removal systems in recent years. The first generation of single-train ammonia plants used scrubbing with an aqueous solution of monoethanolamine (MEA). This consumed about 5.8 GJ/t NH, of regeneration energy, or about 14% of the total energy consumption, which at that time was about 41.5 GJ/t NH3. In addition, the laden MEA solution was very corrosive, and this consideration placed limits on the nnissible carbon dioxide loadings. x ith the introduction of special inhibitors (Amine Guard I to IV from Ur >ion Carbide, now UOP) higher loadings could be used, lowering re genera-tion energy requirements to 88 Mj/kMol CO2 (equivalent to 2.5 G J/mt NH3)141.

The MEA process is a typical example of a chemical absorption solvent process, in which the acidic carbon dioxide reacts with a basecontaining solution to form compounds which can be reversibly decomposed by application of more or less heat to recover the carbon dioxide and restore the solvent to its original state. MEA forms a rather stable salt with carbon dioxide, and a considerable amount of energy is therefore needed in the stripper to decompose it. Modern energy-saving carbon dioxide removal processes using chemical solvents include BASF's Activated MDEA process (aMDEA), UOFs Benfield process, the Catacarb process of Eickmeyer & Associates, and the GiammarcoVetrocoke process. Physical absorption solvents, on the other hand, dissolve carbon dioxide under pressure without combining in a formal chemical sense. That allows the carbon dioxide to be recovered by flashing; no heat is required. UOP's Selexol process is an example142. Figure 29143 demonstrates very clearly the difference between a strong chemical solvent (MEA) and a pure physical solvent (Selexol) by their absorption isotherms. Typically, for a strong chemical solvent the loading is dependent on the partial pressure of the carbon dioxide only to a limited degree and approaches a saturation value, whereas for the physical solvent the loading is approximately proportional to the partial pressure, according to Henry's law.

The Benfield process144-146 is an example of a hot potassium carbonate process, others being the Catacarb process147 and the GiammarcoVetrocoke process148. They all use an aqueous solution of potassium carbonate along with an activator; in the case of Benfield it is usually diethanolamine. Because the solution is significantly corrosive, a corrosion inhibitor has to be used (Benfield's is vanadium-based), and its concentration must be controlled carefully. In the face of increasing energy consumption awareness, UOP has developed a number of Benfield process configurations. In many installations scrubbing is performed in a one-stage absorption/regeneration system and, with application of split flow, the residual carbon dioxide content can be reduced to 0.1%. Adding a two-stage regenerator brings this down to 0.05% CO2. In the "HiPure" concept (basically two separate absorption/regeneration circuits) a residual level of less than 50 ppm is possible. To improve the thermal efficiency a "LoHeat" version has been developed in which flash steam is recompressed with an injector or a mechanical vapour compressor (MVC). The MVC option has a lower heat requirement but at the expense of a mechanical energy demand. Values of 28-35 MJ/kmol CO2 are reported. Figure 30 shows the "LoHeat" version with a multistage flash/ejector system. Another version149, called BPS (Benfield Pressure Swing), involves a threestage absorber and a two-stage flash generator.

Fig. 29: CO2 Loading Characteristics of Various Solvents


80

10

12

j partial pressure, bar

27

MODERN PRODUCTION TECHNOLOGIES

Recently UOP has introduced a new activator, called ACT-1150-151 and also new high-performance packings for the Benfield carbon dioxide removal system. ACT-1 activator increases throughput and reduces thermal duty, and the high-performance packing allows a reduction in the column diameter (up to 10-15% for a given height) and/or packing height, so that - in total - the packing volume will come down by about 25%. ACT-1 can be used for upgrading the performance of an existing Benfield installation, in which case 0.5-1 wt-% is added to the solution to start with and ACT1 is used in place of the standard DEA make-up, so in the course of time DEA disappears from the system altogether. For new installations 1-3% ACT is applied without any DEA, with shorter columns, smaller reboUers, and lower solvent circulation. Exxon's Flexsorb HP is an effective new rate accelerator for conventional promoted potassium carbonate carbon dioxide removal systems. It is a sterically hindered amine. The Flexsorb HP system has a lower energy requirement and requires less solution make-up than conventional hot potash systems, and it has a good potential for debottlenecking existing units. So far it has been successfully used in several large ammonia plants144-152. British Gas has also introduced an accelerator for hot potash carbon dioxide removal systems. LRS-10, .as it is called, works by accelerating the rate of desorption of the carbon dioxide in the regenerator to give a leaner solution with a greater absorption capacity153'm. BASF's aMDEA process143- 15M6 uses an aqueous solution of methyldiethanolamine along with an amine activator. Corrosion inhibitors are unnecessary, and (unlike the MEA process) no solvent degradation is observed, so reclaiming facilities are not required. Thanks to the low vapour pressure of MDEA and the activator, there is virtually no solvent loss in normal operation. Carbon dioxide binds chemically to MDEA much less strongly than to MEA, and the solvent character is more like a hybrid of a strong chemical and a purely physical solvent. So, on account of the relative

weak binding forces between the solvent base and the carbon dioxide, as can be seen from Fig. 29, a substantial portion of the carbon dioxide content can be recovered just by depressurizing (flashing) and only a small proportion has to be recovered by steam stripping. There is a variety of configurations, ranging from the standard configuration with two-stage scrubbing using lean and semi-lean solution in one absorption column to one-stage scrubbing with just steam stripping. The standard configuration, which is normally used in modern low-energy ammonia plants, is given in Fig. 31. By increasing the activator concentration the character of the solvent may be shifted more to the chemi-

cal side and vice versa (Fig. 29). This makes the process very versatile and capable of being tailored to fit into a given situation, as would be the case in a revamp143. Table Vul demonstrates very well how different activator concentrations may influence the process design of an ammonia plant. The aMDEA process is also very well suited for revamping old MEA units by a simple solvent swap, keeping the existing plant configuration. This is made possible by judicious choice of the appropriate activator concentration. The first successful revamps of this kind were carried out in the United States16"63. The total number of solvent swaps in ammonia plants to date is 25164.

Fig. 30: Benfield LoHeat CO2 Removal Process


feed gas acid gas purified gas

BFW heater

Fig. 31: BASF Activated MDEA (aMDEA) Process (standard two-stage configuration)
treated gas
CO,

feed gas C$r=

28

AMMONIA

The Selexol process141165'168, developed by Allied Chemical and now owned by UOP, uses a mixture of glycol dimethylethers. A similar solvent is used in the Fluor Solvent process169 and in BASF's Sepasolv process156. These solvents are stable, non-corrosive, non-toxic and not very olatile. But they do have a rather high capacity for absorbing water, so the raw gas has to be rather dry. This is usually achieved by chilling and operating at low temperatures; for example, the Selexol process runs at around 5G For use in partial oxidation plants these solvents have the advantage that the solubility of hydrogen sulphide is about nine times higher than that of carbon dioxide. As the carbon dioxide is recovered merely by flashing, no heat is necessary, though a fair comparison with other processes must take into account the mechanical energy required to circulate the solvent. Before the solvent is recycled to the one-stage absorption column, it has

to be vacuumed and/or air-stripped to remove traces of carbon dioxide. Figure 32 presents a basic flowsheet. Some problems could arise when much higher hydrocarbons are formed in the high-temperature shift reactor because some of them have comparable solubilities to that of carbon dioxide in the solvent and may thus contaminate the carbon dioxide product.

Final purif ication of the synthesis gas


Methanation, the classic route
All stages considered so far in our tour through the ammonia process have been simply needed to produce a mixture of hydrogen and nitrogen (3:1 by volume in the conventional concepts) that is suitable for conversion into ammonia. On leaving the carbon dioxide removal system, the gas is almost (but not quite) ready to enter the synthesis loop - the

Table VIII Examples of aMDEA Process Options in a Modern Steam Reforming Ammonia Plant
(35 bar in the CO removal section)
aMDEA 02 (two-stage) Feed gas CO2 in feed, vol-% CO2 in lean gas, vol-% Solvent circulation (lean/semi-lean) Packing height (lean/semi-lean) Reboiler duty Specific energy, MJ/kmole CO2
100 18 O.I 102 (12/88) 180 (47/53) 98 28.S

aMDEA 03 (two-stage)
100 18 O.I 100 (12/88) 100 (52/48) 100 29

aMDEA 03 (one-stage)
100 18 O.I 43 52 250 73

Fig. 32: Selexol Physical CO2 Absorption Process


purified synthesis gas lean solution cooler

vent (to atmosphre)*

absorber

regenerator

raw synthesis gas

solution expander flash drums

s-29

ammonia plant proper. The gas still contains moisture and traces of both carbon dioxide and carbon monoxide, none of which can be allowed to enter the synthesis converter because they have a disastrous effect on the catalyst. In most processes the gas is reheated to about 300-330C and passed through a bed of nickelbased methanation catalyst, where the opposite of the primary reforming reactions take place to reconstitute methane (which is effectively inert under the conditions in the synthesis section) from the remaining traces of carbon oxides and a little of the precious hydrogen that has been so laboriously generated in the preceding stages (equations 10 and 11). Methanation is an entirely standard procedure and has not been the subject of any remarkable technological developments in recent years. The standard nickel catalysts contain 20-40% Ni on a support, normally based mainly on alumina with smaller amounts of magnesia and silica. Usually a life time of 10-15 years can be attained. Carry-over of potash solution from a hot potash carbon dioxide removal system will lead to serious damage by blocking the catalyst pores. In some cases original activity may be regained by in sz'hf washing of the catalyst with demineralized water or condensate. Small amounts of MDEA or methanol (from a Rectisol wash) are not considered to be serious. After cooling and separation of the condensed water, the gas still contains traces of carbon oxides (<10 ppm) and traces of water (<20ppm). In older plants this residual water used to be removed in the condensing ammonia in the synthesis loop, which meant that the fresh makeup synthesis gas had to be mixed with the recycle gas ahead of the ammonia condensation and separation. That was a disadvantage, because the make-up gas diluted the ammonia concentration of the loop gas and therefore reduced the effectiveness of the ammonia product removal procedure, resulting in either a higher recycle flow rate or increased refrigeration duty, either of which entailed an energy penalty. Most modern ammonia plants now avoid that problem by using molecular sieve driers in various configura-

MODERN PRODUCTION TECHNOLOGIES

tions for removing water. Some older plants have even been retrofitted with this technology during revamps.

Alternative concepts
Pressure Swing Adsorption (PSA) has also been used for synthesis gas purification in smaller plants. This process can be used to replace the LT shift, carbon dioxide removal and methanation stages and even the secondary reformer as well. It uses molecular sieves as adsorbents in a series of parallel-operated vessels running in staggered cyclic mode involving a periodic depressuring phase for regeneration. Because it is more commonly used in hydrogen rather than ammonia plants, the detailed description appears in Chapter 4 (see pages 109-111). KTI incorporated a UCC-designed PSA system in its PARC process170'm in a plant in Greece, using a special nitrogen flush to adjust the nitrogen for the synthesis loop in the absence of a secondary reformer. PSA is also a feature of ICI's LCA process plant at Severnside, and there it is proportionately larger in relation to the ammonia production because it has to remove the surplus nitrogen introduced in the air excess needed in the secondary reformer for the purpose of maintaining the heat balance. If product carbon dioxide is needed in this case for sale or for downstream synthesis (urea, for example), it can be recovered by scrubbing the process gas ahead of the PSA unit or (less efficiently) from the PSA offgas. For some time it was believed that PSA would only be an option for small-capacity plants because of the size and number of vessels needed. But Linde AG has shown that PSA can be economic at rather large capacities. It is currently constructing a 1,350-t/d ammonia plant in India according to the Linde Ammonia Concept (LAC)134-172-173, operating with a single-line 12vessel PSA system. Linde has also delivered a single-line 12-vessel PSA system for a hydrogen plant in Korea with a capacity of 150,000 Nm3/h135, equivalent to about 1,800 t/d NH3. Cryogenic purification may be used in process concepts with surplus air in the secondary reformer

to remove the over-stoichiometric amount of nitrogen. Brown & Roofs Braun ammonia process incorporates such a unit, called the Purifier, after the methanation stage174. The basic configuration of this unit, introduced as long ago as 1964, is rather simple. (Figure 33 is a basic flow diagram). There are only three major components: the feed /effluent exchanger, the rectifier column with built-in condenser, and a turboexpander. Additionally, the gas has to be dried by molecular sieves before entering the cold box to prevent dogging by moisture. At a temperature of about -185C, argon and methane are washed out from the synthesis gas as it flows upwards through the rectifier column by a descending flow of liquid nitrogen. The liquid nitrogen surplus containing the impurities is withdrawn from the bottom of the column and flashed in the exchanger. The cooling energy is supplied by expansion of the raw synthesis gas over the turboexpander and expanding the waste gas to the low-pressure level of the reformer fuel gas feed. The whole pressure loss in the synthesis gas is just 2 bar. There are two major advantages from the usage of a Purifier. First, the synthesis gas is almost inert-free (only a minor amount of argon is left) so that only a minimal purge from the synthesis loop is required.

Secondly, it separates the front-end and the synthesis loop, permitting the H/N ratio in the synthesis loop to be set independently from what is happening in the secondary reformer, unlike in the conventional processes, where the process air flow has to be controlled very carefully to meet the stoichiometric requirement. There is also no need for a purge gas hydrogen recovery unit, which is a common feature of conventional process configurations.

The ammonia synthesis section


The way in which the ammonia synthesis gas is produced, whether by steam reforming of light hydrocarbons or by partial oxidation of heavy oil fractions or coal, actually has no influence on the design and configuration of the ammonia synthesis section. Nevertheless, it is dealt with here rather than later because it forms part of the normal process sequence of a steam reforming plant, as of any other. Where necessary, reference will made back to this section from the chapters on other synthesis gas generation processes. The dominating factor is the position of the chemical equilibrium (Equation 4), which is more towards the left than the right side of the equation and thus rather unfavourable for the formation of ammonia

Fig. 33: Purifier arrangement

cold box enclosure

feed effluent exchangers rectifiervlx

30

AMMONIA

under conditions at which technical ammonia processes can operate from the point of view of catalyst performance, energy consumption and technical feasibility. 3H, + N2 ^ 2NH3 AH298 = -92.4 kj/mol (4) Because only a minor fraction of hie synthesis gas (usually 25-35%) can be converted to ammonia per pass, provision must be made to recycle a very large portion of unconverted synthesis gas after the product ammonia has been separated. The ammonia synthesis reaction entails a reduction in volume, so it will be favoured by higher pressure according to the Le Chatelier principle, and as it is an exothermic reaction, the lower the temperature the better the equilibrium concentration of ammonia will be. But from the point of view of energy consumption it is desirable to keep the pressure as low as possible. The energy needed for compression in modern low-energy versions of single-train ammonia plants still amounts to about 2.5 GJ/t NHy which corresponds to roughly 10% of the total energy consumption. But lower pressure leads to unfavourable equilibrium conditions, which must be compensated for by lowering the reaction temperature. With the standard iron-based ammonia synthesis catalyst, lowering the temperature will also reduce the reaction rate. Either a more active catalyst is needed or, if that is not available, the reaction time must be extended. That necessitates the use of a larger volume of catalyst. Figure 34175 illustrates this.

Fig. 34: Ammonia Equilibrium (N:H1=3:l)and Catalyst Volume

400C istant equilb. cone. 500C

50 100 150200250300
optimized catalyst volume with .standard ammonia catalyst

In the years since then only minor improvements have been made. Promoter concentrations, catalyst particle size and manufacturing procedures have been optimized, resulting in improved activity and longer service life. The most notable of these developments in this respect was the introduction of cobalt as an additional promoter by ICI178. It is only recently that we have seen the first radical change in ammonia synthesis catalyst composition: the new ruthenium-based catalyst now being commercialized by Kellogg178-m.

50 100 150200 250 300


pressure, bar

Ruthenium-based ammonia synthesis catalyst


In 1979 BP disclosed to Kellogg a new ammonium catalyst composed of ruthenium on a graphite support. In October 1990, after a 10-year test programme, Kellogg started commercialization of the Kellogg Advanced Ammonia Process (KAAP)180, based on this ruthenium catalyst, which is claimed to be 10-20 times more active man the traditional iron catalyst. According to the patent181, the catalyst is prepared by subliming ruthenium-carbonyl Ru3(CO)12 onto a carbon-containing support which is impregnated with rubidium nitrate. The catalyst has a considerably higher surface area than a conventional catalyst and, according to the patent example, it should contain 5% Ru and 10% Rb by weight. Not much is known on how far the reaction temperature can be lowered, nor on life-expectancy and price. Engelhard Industries, experienced in precious metal catalyst preparation, is toll-manufacturing this new catalyst for Kellogg. The first commercial installation was a retrofit project for the Meihanex ammonia plant in Kitimat, Canada, where a KAAP reactor is placed as an "after-burner" downstream of the existing horizontal converter to boost the ammonia concentration from about 15% to 21%182. The project was intended to increase the capacity by 40% (+250 t/d). The plant was commissioned in November 1992, and it was the first time in history that ammonia was synthesized on a commercial scale using a non-iron catalyst. Kellogg has been awarded additional revamp

The classical iron-based ammonia synthesis catalyst


With the standard commercial ammonia catalyst, it is technically possible to operate with inlet temperatures of around 350C and synthesis pressures of 40 to 50 bar. In this c ase, however, the required catalyst >lume is hugely increased, and to Accommodate it very large reactors would be necessary. As the reaction temperature cannot be lowered below 350C, ammonia concentrations would be rather sr nall, and the increased energy

demand for recycle and for ammonia recovery from such a large volume of gas would nearly offset the saving resulting from the lower synthesis pressure. To attain a production of 1,500 t/d at 50 bar with the classical catalyst needs six times the normal volume of catalyst. Even so, Kellogg has proposed a process based on classical magnetite synthesis catalyst operating at 30-40 bar. Because the ammonia concentration in the converter outlet gas is very low, ammonia has to be separated by water scrubbing. It is recovered by distillation and is produced as liquid at -33C from an energetically integrated multistage refrigeration unit176-177. This is, indeed, a very interesting approach, but the full energetic benefit of operation at low temperature could only be achieved with a catalyst capable of operating adiabaticaUy between 250 and 350C and thus producing a higher concentration of ammonia in the reactor outlet. It is remarkable that since 1913, when the first commercial ammonia plant started at BASF in Oppau (now a part of the city of Ludwigshafen), the activated iron catalyst developed by Alwin Mittasch and his co-workers has changed very little. This should probably be attributed to their comprehensive research work: in more than 20,000 experiments they tested 3,000 different compositions, including almost every element from the Periodic Table.
31

MODERN PRODUCTION TECHNOLOGIES

projects in which the KAAP synthesis loop wiE be used which are due for commissioning in 1996 and 1997. As mentioned on page 20, Kellogg has announced its new Ammonia 2000 process, which combines the KRES exchanger reformer with the KAAP synthesis loop, and it is building two 1,850-t/d plants based on the KAAP concept in Trinidad; they are due to come on stream in 1998. In a grass-roots installation, Kellogg fills the first of the four beds of its new vertical converter, which uses radial flow and cooling by interbed heat exchangers, with traditional iron catalyst because of the high rate at which ammonia is formed at the low inlet concentration. The other three beds are charged with ruthenium catalyst. At a synthesis pressure of 90 bar (or lower)94-95-182 the ammonia concentration at the converter outlet is in excess of 20%. University research in Italy183' on the activity of bimetallic ruthenium clusters incorporating nickel and osmium as components proved the superiority of the simple ruthenium cluster. The chance for ruthenium to displace the classical magnetite.catalyst in new plants will depend on whether the advantages claimed are sufficient to offset the increased cost of using a platinum-group metal. Ruthenium is a member of the platinum metal group (PMG). Its current annual production is 450,000 troy ounces (about 14 tonnes); consumption is estimated to be about 340,000 troy ounces (about 10 tonnes). Of known world PMG production in 1990, South Africa had a share of 75% and the former Soviet Union 19% (based on its exports). Hitherto the price of ruthenium has been around one sixth that of platinum. Some properties of ruthenium described in the literature184 suggest that care is needed when handling the catalyst during loading, unloading and maintenance. Finely-dispersed metallic ruthenium produced by reduction can bind a rather large amount of hydrogen and, when exposed to air, can cause an explosive reaction with atmospheric oxygen. Under these conditions some toxic ruthenium tetroxide, RuO4 which has a low boiling point, might evaporate. Because of its high price

Fig. 35: Effect of H/N Ratio on

Activity of Ru and Fe3O4 Catalysts


600

S. R. Tennison185 and K. Aika and Tamaru186, who have themselves contributed much to this subject.

3*400 0

I3,790kPa promoted ruthenium on

Reaction mechanism, structure and surface chemistry of ammonia synthesis catalyst


An immense amount of research work has been dedicated to the ammonia synthesis reaction over the years. It is probably the most investigated heterogeneous catalytic reaction of all. It is now well established that the dissociative adsorption of nitrogen on magnetite catalyst is the rate-determining step. Outstanding contributions in the elucidation of the now commonly accepted reaction mechanism have been made by such workers as Brill, Boudart, Ertl and Somorjai, whose work is reviewed and summarized in excellent accounts of the state of the knowledge in two recent books edited respectively by J. R. Jennings187 and A. Nielsen186. But one cannot discuss the advance of knowledge of ammonia synthesis, be it the theory or the practice, without mentioning Hador Topsee. With excellent coworkers like Anders Nielsen and many others, he combined scientific research with practical application in catalyst manufacture, technology development and process engineering to the benefit of the entire industry. An impressive series of very good papers was presented in the symposium "Ammonia Synthesis and Beyond"188 celebrating his 80th birthday and A. Nielsen's 50 years of catalyst research.

8,620 kpa -

200
magnetite
2 3 H/N ratio 4

it is necessary to recover and recycle the ruthenium content of spent catalyst. Both that procedure and the supply of new catalyst are currently undertaken by only one supplier in the world. Besides having a substantially higher volumetric activity, the promoted ruthenium catalyst works best at a lower than normal H2:N2 ratio (Fig. 35185), is less susceptible to ammonia self-inhibition (Fig. 36) and has excellent low-pressure activity. But, like the classical magnetite catalyst, ruthenium catalysts are sensitive to oxygen-containing impurities (moisture and carbon oxides) in the synthesis gas. Further improvements seem to be possible, but it is not very likely that the ultimate activity of this type of catalyst will allow the synthesis loop to be operated at the same pressure as the synthesis gas plant. Detailed reviews of research on non-iron catalysts are given by Fig. 36: Ammonia Inhibition
of Ru and Fe3O4

Catalyst particle size and shape


The first generation of single-stream ammonia plants, mainly developed by Kellogg, used axial-flow quenchtype converters. At the prevailing operating pressure of around 150 bar, which represented about the maximum that could be achieved by the centrifugal compressor designs of the day, the standard grain size was 6-10 mm at space velocities of about 10-15000 h'1. It had been well known for some time that there is an approximately linear inverse relation between the grain size of the catalyst and its activity.

Catalysts
promoted ruthenium on carbon, H/N =1.5:1

8 12 16 ammonia, %

20

32

Fig. 37: Relation of Pressure Prop and Catalyst Particle Size


(bed depth: 7m; pressure: 27.1 M Pa; temperature: 450C)

3-6 mm

15,000 20,000 25,000 30,000 space velocity, Mm3 nv3 Iv'

But the pressure drop increases so markedly as the grain size is reduced (Fig. 37189) that fine catalyst was precluded from use in an axialflow configuration. It was Haldor Tops0e's company which, about 25 years ago, introduced the radial flow concept into industrial ammonia synthesis190' m. With this geometry the gas velocities are only about one tenth of those in an axial-flow converter, calculated for the empty bed. So it is possible to use small-sized catalyst particles and nevertheless have a rather low pressure drop. The radial concept makes optimal use of the converter volume. Kellogg chose another geometrical approach with its horizontal cross-flow converter, introduced in 1971192'193. The first generation of both designs used quench cooling for reaction control. Two factors are responsible for the lower activity of larger catalyst particles. Firstly, the larger grain size etards movement of the ammonia formed from the interior of the particle into the bulk gas stream, because this occurs only by slow diffusion through the pore system. Slow ammonia diffusion inhibits the reaction rate. Secondly, the reduction of an individual catalyst particle starts from the outer side and pro1 'eds to the interior of the grain. The x ater formed by removing the oxySen from the iron oxide in the interior of the grains will pass over already reduced catalyst on its way to the outer surface of the particle. This induces some recrystallization.

The effect is considerable: going from a particle size of 1 mm to one of 8 mm, the inner surface will decrease from 11-16 to 3-8 m2/g189. As a consequence of the fabrication process, the particles of an ammonia synthesis catalyst are traditionally irregularly shaped. That is an advantage because the overall activity of an individual particle is actually better than for a more cubic or spherical particle, and there is a greater degree of perpendicular intermixing, which is beneficial from the point of view of both mass flow and heat dissipation. Set against this, the slight improvement in pressure drop obtained by using a more regular shape is negligible. There are various ways to take advantage of the higher activity of a small-grain catalyst when designing a synthesis loop, depending on the chosen optimization strategy. If the pressure is kept constant, the volume of the catalyst required is considerably reduced. If, on the other hand, the same volume of catalyst is used, the exit ammonia concentration is increased, which reduces the amount of recycle and refrigeration duty. Alternatively, or in addition, the synthesis pressure may be lowered. In revamp projects the objective is often to boost capacity by installing a new converter or modify the converter internals to accommodate small-grained catalyst. Modern low-energy designs for large ammonia plants tend to put more empha-

sis on high conversion rates than on pressure reduction, because the latter can be complicated by other factors such as pipe and exchanger dimensions in the loop.

Converter design
The apparent rate of ammonia formation is greatest in a pure hydrogen/nitrogen gas mixture and appears to fall as the concentration of the product increases, because the retrograde (dissociation) reaction increases in proportion to the concentration of ammonia until, at equilibrium, it balances the rate of the synthesis reaction. It is also temperaturedependent. In the lower end of the temperature range the effect is kinetic: a rise in temperature increases the rate just as it accelerates any sort of chemical reaction. Above about 450-550C, however, the kinetic effect begins to be outweighed by the adverse effect of the temperature on the position of the equilibrium. The curves shown in Fig. 38 represent the relation between the ammonia concentration and temperature (at 200 bar and 11 vol-% inerts in the synthesis gas) for various fixed rates of ammonia formation. It dearly shows that there is always a definite temperature at which the reaction rate attains a maximum for any chosen ammonia concentration. Joining these points will result in a line giving, for each NH3 concentration, the temperature for the maximum ammonia formation rate. To

Fig. 38: Reaction Rate in Ammonia Formation


r=m3NH3/m3catalyst/s

500
temperature, C

600

650

33

MODERN PRODUCTION TECHNOLOGIES

maintain the maximum rate, the temperature must decrease as the ammonia concentration increases. For optimal catalyst usage the reactor temperature profile (after an initial adiabatic heating zone in the first part of the catalyst) should follow this line. In the early days converters were always compared to this "ideal" for optimum use of highpressure vessel volume. Today the objective is radier to maximize heat recovery (at the highest possible level) and to minimize investment costs for the total synthesis loop. The design of an ammonia converter is a demanding chemical engineering task. To calculate the parameters for the design, including the number and dimensions of the catalyst beds, temperature profiles, ammonia concentrations, gas compositions, pressure drops, and so on, a suitable mathematical model is required. In principle, two differential equations will describe the steady state behaviour of the reaction in the converter. The first gives a concentration-location relationship within the catalyst bed for the reactants and the ammonia. It reflects the reaction kinetic expression. The second models the temperature-position relationship for the synthesis gas, catalyst and vessel internals. The form of this equation is specific to the type of the converter. The Temkin-Pyshew equation (equation 22)194 - or, better, one of its improved modifications - may be used to describe the reaction rate. Other equations are suitable, too194-199.

ing compounds may be regarded as a temporary poison, severe exposure for an extended period of time may lead to permanent damage, probably as a result of recrystalHzation caused by some oxidation/reduction cycle on the catalyst surface. How well the different kinetic expressions fit with reality depends to some extent on the chosen reaction conditions and to some extent on catalyst type, particularly its promoter concentration.195 If we disregard earlier concepts in which cold synthesis gas is heated up by passing it, usually co-currentiy for tiie best approach to the ideal line, in "cooling tubes" running through the catalyst mass195-200-202, direct cooling in a quench converter was for some time the favourite on account of its relatively simple design. The quench-cooling principle was also retained in the first radialFig. 39: Kellogg Axial-Flow Quench Converter
outlet

-*,
3 .

0W2

(22)

These equations more or less model the intrinsic reaction, free from mass transfer restrictions. They are usually derived from measurements with very fine catalyst particles. For practical application, correction factors (pore effectiveness factors) have to be included to allow for the larger particles used in industrial ammonia converters, for which transport phenomena cannot be ignored198-199. Additional terms may be included to take into consideration the influence of oxygencontaining impurities in the synthesis gas198. Although oxygen-contain-

inlet

flow and cross-flow configurations. In the quench converter, a fraction only of the recycle enters the first catalyst layer at about 400C The catalyst volume of the bed is chosen so that the gas will leave it at about 500C, which is the limit suggested by most of the catalyst vendors. Before entering the next catalyst bed, the gas temperature is "quenched" by injection of cooler (150-200C) recycle gas. The same thing is done at the subsequent beds. In this way the reaction profile describes a zigzag path around our ideal line. One disadvantage is that not all the recycle gas will pass over the whole catalyst volume. In addition, the injection of colder gas lowers the grade of the heat that ultimately can be recovered at the outlet of the converter. After abstraction of the heat necessary to raise the inlet gas to the first bed to the initiation temperature, what remains is at a level suited only for boiler feed water preheat. Fig. 39 is an illustration of this converter concept (the actual design is the familiar Kellogg axialflow quench converter193). Similar designs have been used by Grande Paroisse, Casale and others. A special design in this respect was the ICI lozenge converter with direct injection of the quench gas into the catalyst bed203-204. The quench converter concept, which can be applied up to very large capacities, was almost exclusively used in the early generation of single-stream ammonia plants commissioned in the 1960s and 1970s. Many, indeed, are still in use today, although they are being progressively replaced as the plants are revamped. Because of growing energy-consciousness, the quench type has actually become obsolete for new plants. In an indirect-cooled converter the exchanger for preheating the inlet gas is positioned after the first bed. In this configuration gas leaving the last bed (and also intermediate beds, if there are any) is then at a high enough temperature level to raise HP steam. This is a definite gain in exergy, which improves the overall efficiency. Also, as the total quantity of the recycle gas passes through all the catalyst beds, the total catalyst volume will be lower under comparable conditions, usually

34

Fig. 40: Ammonia Conversion/ Temperature Profiles for Quench- and Indirect-Cooled Converters (Topsoe Series 100 and 200)
20-

Fig. 4l: Topsoe Series 200 Converters


pressure shell ~~ indirect quench gas inlet interbed heat exchanger

8
cd

I st catalyst bed

l 10
S

annulus round catalyst bed 2nd catalyst bed

360

400 440 480 temperature, C

lower heat exchanger

allowing the number of catalyst beds to be reduced. This basic concept had already been introduced by C F Braun (today Brown & Root)205 in the late 1960s, before the first "energy crisis" and at a time when the quench converter reigned supreme. Today all modern, low-energy designs use this approach, although design and lay-out differ considerably. Figure 40 compares the temperature patterns for the quenchand indirect-cooled converters. Topsee, with the Series 200191-206 (Fig. 41) and Kellogg, with its Horizontal Converter (Fig. 42)193-207, use two catalyst layers with an intermediate heat exchanger for the feed, all in one vessel, suited for capacities of at least 1,800 t/d. Uhde (Fig. 43)208 uses three catalyst beds. The first two, along with an inlet/outlet exchanger, are accommodated in one vessel. Then a waste heat boiler generating HP steam

With lower heat exchanger

gas outlet cold bypass

pressure shell

interbed heat exchanger 1st catalyst bed annulus round catalyst bed 2nd catalyst bed cold bypass pipe

main gas inlet Without lower heat exchanger

main gas inlet / 777* gas outlet cold bypass

cools the gas before it enters the second vessel containing the third catalyst bed, which discharges through a second HP boiler. Recently, Tops0e has also introduced a two-vessel, three-catalyst bed concept, known as S-250. Brown & Root, whose ammonia process used a two-vessel system for many years, now favours three catalyst beds in separate vessels, with an inlet/outlet exchanger after the first and HP steam boilers after the other two. To withstand the high outlet temperature level (450-500C) needed for the HP boilers, special designs must be employed to keep the outlet nozzle cool. Brown & Root has devised a good solution by directly coupling the boiler to the converter (Fig. 44209). In most converter designs the pressure shell is kept below 250C by means of insulation or cooler gas flow. Today, with excellent steels available and proper welding procedures applied, it is possible to run converters with pressure vessel walls at 400C in compliance with the present Nelson210 diagram. An example is the Brown & Root converter, in which inlet gas for the second converter at 400C and with an ammonia content of 10% NH3 is in direct contact with the wall. Special care has to be given to welds regarding the quality of preheating and post-weld heat treatment to avoid crack formation, which has occurred in a few instances under these operation conditions (see Material questions on pages 60-61). The distinctive characteristic of Ammonia Casale's ACAR converter (Ammonia Casale Axial-Radial)211-212 is a mixed flow pattern. In each

Fig. 42: Kellogg Horizontal Ammonia Converter with Indirect Cooling


inlet

nnn / m nnrii/nmiiifn/jiiiiniitn/)

iuuitn)/>!!ii>ii>t

outlet

111 iiinnini ii)iiiii}(iiiinii!inii/r!itir)>/i


interbed heat exchanger

35

MODERN PRODUCTION TECHNOLOGIES

Fig. 43: Uhde Dual-Vessel Converter System


gas inlet start-up gas

first bed

third bed

second bed

gas inlet gas outlet gas outlet

catalyst layer the gas flows through the top zone predominantly axially but traverses the lower part in the radial direction. This design avoids the need for special sealing of the top end of the catalyst bed, which is needed with purely radial designs to prevent synthesis gas tracking through empty space resulting from

catalyst settling. The small axial-flow section has actually no discernible negative influence on the pressure drop. All in all, the mixed-flow concept simplifies the detailed mechanical design very much. The individual beds are held in separate cartridges within the basket assembly and can easily be detached in most models.

Fig. 44: Braun Adiabatic Converter with Close-Coupled Heat Exchanger

The whole concept is very flexible, with configurations for both quench cooling and indirect cooling using internal heat exchangers, and it fits well into existing converter shells, needing little downtime to install The record of revamps using the Casale concept is quite remarkable in number and performance. The ICI tube-cooled converter, introduced a few years ago, is mainly used in methanol plants (six units so far) and also in the two LCA ammonia plants213 at Severnside. This concept is actually a modern variation of the Tennessee Valley Authority (TVA) ammonia converter used about 40 years ago. It is a vertical axial-flow design in which the reactor temperature is controlled by cooler feed gas flowing through a multitude of heat exchanger tubes running parallel to the axis of the reactor through a single body of catalyst. The flow in the tubes is in countercurrent to the gas stream in the catalyst bed. A more detailed description is given in Chapter 3 on page 76. The radial-flow converter of the Kellogg KAAP synthesis loop is a vertical vessel with four catalyst beds (as already mentioned). Intermediate cooling is provided by three internal heat exchangers. Figure 45 is a simplified sketch of the KAAP converter and synthesis loop.

Operating pressure
In principle, it is possible to calculate the minimum amount of mechanical work needed in the synthesis loop. If plots of the kWh needed for feed gas compression, recycle and refrigeration are superimposed, the result will be a more or less flat minimum201-2U-2I5. But one should interpret such diagrams with caution. In the first place, this kind of plot will be strongly influenced by the catalyst activity and thus by any factor that affects it, such as grain size and choice of minimum possi ble temperatures for the catalyst beds, and especially the equilibrium temperature of the last bed216. Secondly, it only considers mechanical energy; the recovered heat of reaction, its energy level and its impact on the total energy balance of the plant are quite ignored. Thirdly, the temperature attainable by air- 01

downcomei

gas outlet

36

AMMONIA

Fig. 45: Kellogg Advanced Ammonia Process (KAAP)


-r synthesis gas make-up W PGRUj> recycle Muel

^jj BFW anc QO./ f *LJ?

~90 bar ~20+% NH

water-cooling will affect the refrigeration duty. And fourthly, the type of front end can profoundly alter the result. The front-end pressure determines the suction pressure of the synthesis gas machine. In a plant based on partial oxidation operating .t 75 bar, less than half as much energy is needed to compress the synthesis gas to 180 bar as in a steam reforming plant (suction pressure 25 bar). So the problem of choosing the best synthesis pressure seems to be complex, because not only does the whole energy balance have to be examined but also the lechanical design and, consequently, uie investment costs associated with it. In this latter respect, the costs for the energy, (in other words, the feedstock) have to be weighed against investment. In the earliest single-train ammonia plants, built in the mid-1960s with capacities of around 600 t/d, the maximum attainable pressure as restricted to about 150 bar by the technical limitations of the centrifugal compressor. Unless the total volumetric gas flow has a reasonable relationship to the passage width of the last impeller and the pressure ratio, excessive pressure losses would occur in the diffusers Between the compressor wheels, rening the machine extremely inefv vive. Even with the manufacturig capabilities of today, a minimum How of 400 m 3 /h (at temperature and pressure) is needed, which translates to about 140 bar for a 450l /d plant. Of course, at the current

plant sizes of 1,200-1,800 t/d this constraint is no longer of importance. Such plants usually operate with loop pressures of 170-190 bar. These pressure differences are not reflected in the total energy consumption.

Purge gas management


To keep down the concentration of the inerts in the loop recycle (argon and other atmospheric .inert gases taken in with the process air in the secondary reformer; methane slip from the secondary reformer along with the methane regenerated in the methanator), it is necessary continuously to withdraw a small purge gas stream from the loop. In the older, less energy-efficient plants this gas was usually added directly to the reformer fuel. That at least utilized most of its intrinsic energy value, but it meant that all die effort and additional energy put into generating its hydrogen content and purifying it as synthesis gas was thrown away. The purge gas problem was, however, all but avoided altogether in processes with cryogenic final purification stages: partial oxidation plants, which used the liquid nitrogen byproduct from the oxygen unit to wash the synthesis gas, the Braun Purifier process and the version of the Foster-Wheeler AM2 process equipped with a cold box (refer to page 47). The most capital-intensive purge gas treating method would consist of a second loop, operating on a slightly lower pressure. As this loop
37

would operate on a very high inert level, only a very small purge and no recompression would be needed. Up to 75% of the hydrogen from the purge of the first loop could be recovered. This system has the advantage that the nitrogen is recovered, too, but it is just too expensive for use in modern plants and revamps. Cryogenic technology217 is rather simple and offered by the major contractors specializing in cryogenic processes: Air Products, L'Air Liquide, Linde, and Costain Engineering (formerly Petrocarbon). The Costain process, for example, operates as high as 80 bar and recovers 90-94% of the hydrogen at 90% purity and 30% of the nitrogen of the purge gas. Ahead of the cold box the purge gas is scrubbed with water to remove remaining ammonia and is then dried by molecular sieves. Separation is performed by condensing the nitrogen and the inerts, leaving a hydrogen-rich phase. The equipment basically consists of a multi-stream exchanger and a separator. The Permea Prism Separator, introduced by Monsanto in 1979218-219, makes use of selective permeation through a membrane composed of an asymmetric polymer substrate and a polymer coating of high permeability. Components of the gas first dissolve into the membrane and diffuse through it and are evolved again from the opposite surface, which is under lower pressure. Since different gases travel at different rates, they are separated. The membranes are shaped into hollow fibres (diameter 0.5 mm) to provide maximum surface area. The fibres are contained in vertically-mounted pressure shells like a shell-and-tube heat exchanger. One module has many thousands of hollow fibres sealed at the upper end and embedded in a tube sheet at the lower end. Modules are arranged in series to achieve the desired degree of separation, and these banks of modules are themselves arranged in parallel to provide the necessary capacity. With decreasing hydrogen differential pressure the permeation rate becomes slow. For this reason a second (sometimes a third) bank of separators will operate at a lower pressure on the permeate side. For

MODERN PRODUCTION TECHNOLOGIES

example, in a two-stage version, 65% of the hydrogen recovered from the first bank would be returned to the suction side of the second compressor stage, while the remaining 35% from the second bank would be returned to the suction side of the first compressor stage. The total hydrogen recovery is 90%. The great appeal of this technology lies in its simplicity, easy operability and low maintenance. In addition, the gas does not have to be dried and traces of ammonia (200 ppm) can be tolerated. Membrane technology is also offered by other licensors; an example is the POLYSEP Membrane System of UOP220. In addition to the systems based on hollow fibres, membrane modules have been developed in which the membrane is in the form of a sheet wrapped around a perforated central tube using spacers to separate the layers. The raw gas flows in an axial direction in the high-pressure spacer and the permeate is withdrawn in the low-pressure spacer. Linde, for example, offers such a module under the name SEPAREX221'222. The well-known pressure swing processes (PSA)220-w of UOP (formerly Union Carbide), developed under the name HYSIV but now called POLYBED PSA, and of Linde and other companies are also well suited for purge gas treatment. A PSA process based on carboncontaining absorbents originating from Bergbau-Forschung224 is offered by Costain. The reversible formation of metal hydrides is a new development of Air Products and Chemicals225 226. It is based on the selective and reversible absorption capabilities of special metal alloys (LaNij, FeTi, Mgpi). In pilot plant operation 90-93% recovery with a purity of 99% was demonstrated. The metal components were formed into pellets with a binder.

Fig. 46: Block Flow Diagram ani Gas Temperature Profile for a Steam-Reforming Ammonia Plant
block diagram
I )
200 i 100 300

gas temperature, C
400 i 500 600 i 700 800 i 900

1,000 i ^

natural gas feed

i
Idesulphurizationl process 1 L. ' steam t t" . r, . 1 ] j primary reformer^ [l combustion 1 air process air >i ^-(second, reformer | fuel /"5\ HP steam w 'O' superheated
CO ^

1 COj removal |

L,

sy ngas [ methanation ] y .com aressor "

~^1 JiA
L

|| T

^ \ rfrigrt on *~T

y ^n 1 NH loop
r

^A Jk

HP steam to superheater product NH3 | fc.


lue

1 [ H2 recovery |

>

i
certainly had a revolutionary effect on the economics of ammonia production, making possible an immense growth in world capacity in the subsequent years. The basic reaction sequence has not changed since then. Figure 46 shows it for a modern plant together with the temperature levels of the individual sections.

respect to the mass flow, and energy management was handled within the separate process sections, which were often sited separately because they usually consisted of several parallel units. The breakthrough was the advent of the single-train ammonia plant. To use a single-train for large capacities (no parallel lines) and to be as far as possible energetically self-sufficient (no energy import) through a high degree of energy integration (with process steps in surplus supplying those in deficit) was the design philosophy for the new steam reforming ammonia plants pioneered by Kellogg and some others in the mid 1960s. It

Machinery drives
High-level surplus energy is available from the flue gas and the process gas streams of various sections while there is a need for heat in other places such as the process steam for the reforming reaction and

Table IX Main Energy Sources and Sinks in the Steam Reforming Ammonia Process
Process section Reforming Shift conversion CO2 removal Methanation Synthesis Machinery Unavoidable ioss Balance Originating Primary reforming duty Hue gas Process gas Heat of reaction Heat for solvent regeneration Heat of reaction Heat of reaction Drivers Stack and general (Auxiliary boiler or import) (Export) Contribution Demand Surplus Surplus Surplus Demand Surplus Surplus Demand Demand Deficit Surplus

Energy recovery and machinery drive type


The individual process steps so far described have to be assembled together to make a complete ammonia plant. In the old days this was more or less just a combination with

38

in the solvent regenerator of the carbon dioxide removal unit. Because a considerable amount of mechanical energy is needed to drive compressors, pumps and fans, it seemed most appropriate to use steam turbine drives, since plenty of steam could be generated from waste heat. ^s the temperature level was high i ough to raise HP steam of 100 bar, it was possible to use the process steam first to generate mechanical energy in the synthesis gas compressor turbine before extracting it at the pressure level of the primary reforming section. Table IX lists all the significant energy sources and sinks within the process. The earlier plants were in deficit, t-jid they needed an auxiliary boiler, which was integrated in the flue gas duct. This situation was partially caused by inadequate waste heat recovery and low efficiency in some of the energy consumers. Typically, the furnace flue gas was discharged up the stack at unnecessarily high temperatures because there was no dr pre-heat, and too much htat was rejected from the synthesis loop, while the efficiency of the mechani-

cal drivers was low and the heat demand in the carbon dioxide removal unit regenerator was high. In the first and also the second generation of plants built to this concept, maximum use was made of direct steam turbine drives, not only for the major machines such as synthesis gas, air and refrigeration compressors, but even for relatively small pumps and fans. The outcome was a rather complex steam system; Fig. 47 is a simplified example. Small wonder that ammonia plants were facetiously described as sophisticated power stations making ammonia as a by-product: they were, after all, producing more steam than ammonia. Even today, the most modern plants still produce about three times as much. And, indeed, in recent years electrical drives have swung back into favour for the smaller items of equipment. There are two penalties for using small turbine drives. They are rather expensive to make, and they cost more to maintain than electric motors. Condensation types were too elaborate in such cases, so back-pressure versions with very low steam

outlet pressure (4-5 bar) were more usual. This concept was only acceptable if the low-level steam could be utilized either for very heat-thirsty carbon dioxide removal systems such as MEA or in some other production plant at the location. Secondly, the efficiency of small turbines is substantially lower, at around 50% or less, than that of a turbine of, say, 10-12 MW (ca 75%). So in some modern plants where there is no use for low-grade steam but some electric power is available at the site, turbine drivers are used only for the main machines - the synthesis gas and air compressors and possibly also the refrigeration compressor and all other mechanical equipment is driven electrically. Often the plant includes one additional large turbine driving an electric generator, which supplies the electric motors in the plant while the surplus is exported. There is, of course, a loss involved in transforming mechanical energy into electricity in the generator and then back again into mechanical energy in the drive motors, but this only about 10%, so in comparison with the efficiency

Fig. 47: Typical Steam System of a Steam Reforming Plant


l03bar,520C
superheater (prim. rf. flue gas) turbine driver for syngas compressor

I.
auxiliary boiler economizer jr-1 (prim. rf. flue gas) steam generation behind methanation

-42bar,400C 21 bar, 320C

Bifl'H d

--------

6 bar, 200C

] 0.15 bar

c waste heat boiler downstream HT shift d economizer process gas before compressor 6 economizer process gas after LT shift f economizer process gas after methanation

Turbine drivers for g flue gas and combustion air h boiler feed water pumps j process air compressor k COj wash solution pump I ammonia gas recompression

steam

39

MODERN PRODUCTION TECHNOLOGIES

difference between small and large turbines, there is an overall effidency gain. In most modern plants total energy demand (feed/fuel/power) has been drastically reduced. On the demand side important savings have been achieved in the carbon dioxide removal section by switching from old, heat-thirsty processes like MEA scrubbing to low-energy processes like the newer versions of the Benfield process or aMDEA. In the synthesis loop the mechanical energy needed for feed compression, refrigeration and recycle has been reduced, and throughout the process catalyst volumes and geometry have been optimized for maximum activity and minimum pressure drop. On the supply side, available energy has been increased by greater heat recovery, and the combined effect of that and the savings on the demand side have pushed the energy balance into surplus. Because there is no longer an auxiliary boiler, there is nothing in the plant that can be turned down to bring the energy situation into perfect balance; therefore the overall savings have not, in fact, translated into an actual reduction in gross energy input to the plant (in the form of natural gas); they can only be realized by exporting steam or power, and it is only the net energy consumption that has been reduced. But under favourable circumstances this situation can be used in a very advantageous way. If there is a substantial outlet on the site for export steam, it can be very economic (depending on the price of natural gas and the value assigned to steam) to increase the steam export deliberately by using additional fuel, because the net energy consumption of the plant is simultaneously reduced (Table X). It is only possible to reduce the gross energy demand - that is, to reduce the natural gas input to the plant - by reducing fuel consumption, because the feedstock requirement is, to all intents and purposes, stoichiometric. So the only way is to cut the firing in the reforming furnace, which means that the amount of reaction taking place there will be reduced. That can be done by shifting some of the reforming duty from the primary reformer to the secondary reformer by operating the

Table X Increase of Plant Efficiency by Steam Export (GJ/t NH3) Plant


Natural gas Electric power Steam export Total energy
B

Difference
+5.5 0
-6.4 -0.9

27.1 I.I 28.2

32.6 I.I -6.4 27.3

latter with a surplus of air, though then it becomes necessary to remove surplus nitrogen from the synthesis gas later on. A more radical step in this direction was done in ICI's GasHeated Reformer (GHR) system, which is the heart of its LCA process, and more recently with the KRES of Kellogg and the Tandem Reformer (now marketed by Brown & Root). Even more advanced is the Combined Autothermal Reformer (CAR) concept of Uhde. But none of these designs necessarily achieves any significant improvement over the net energy consumption of the most advanced conventional concepts under the best conditions. For the cases in which export of steam and/or power is welcome there is the very elegant possibility of integrating a gas turbine into the process to drive the air compressor. The hot exhaust of 500-550C contains well enough oxygen to serve as preheated combustion air for firing the primary reformer. The gas turbine does not even have to be particularly efficient, because any heat left in the exhaust gas down to the flue gas temperature of 150C is put to good use in the furnace. Thus an overall efficiency of about 90% can be achieved. It is now standard practice to use a hydraulic turbine to recover energy when solvent from the absorber of the carbon dioxide removal unit is depressurized. And because natural gas is usually available at 25-50 bar at battery limit whereas the furnace burners require a pressure of only about 3 bar to operate satisfactorily, it is worth while to install a fuel gas expander to provide power for rotary equipment: up to 1 MW can be saved this way.

Boiler design
The economic efficiency of a plant is not merely a matter of low energy consumption figures but also of on40

stream time. In this context a high priority must be accorded to the quality and design of the waste heat boilers227. The operating conditions are more severe than those normall} encountered in power plants; for, 01 account of the high pressure on both sides, the heat transfer rates and, therefore, the thermal stresses induced are much higher. After the secondary reformer, the process gas has to be brought down from around 1,000C to about 350C for the high-temperature shift (HTS). In earlier-generation plants two boilers were usually installed in serie; with a by-pass around the second to control the inlet temperature for the HTS. Common practice was to use a water-tube design, and the famous Kellogg bayonet-tube boiler, so familiar to all ammonia people, found application in more than 100 plants. (Because of size limitations, two of these were used in parallel. * To assure adequate natural (convective) water circulation, with this type of boiler the steam drum had to be placed at rather high elevation and was connected to the steam generator by a considerable number of downcomers for the feed water and risers for the steam-water mixture. In contrast, fire-tube boilers arr much better suited for natural circu lation and the steam drum can sit piggyback-fashion, right on top of the boiler. This makes it feasible to provide each boiler with its own separate steam drum, which allows greater flexibility in the plot plan. Nevertheless, water-tube boilers continued to be considered obligatory for 100-bar steam in these plants fo; a long time because of the less com plex and more predictable stress pattern. In a fire-tube boiler, the inlet tube sheet and the tube sheet welds are exposed to the extreme temperature of the reformed gas, which creates rather large temperature gradients and therefore high expansive stress. A positive point in favour of

AMMONIA

the fire-tube design, however, is that debris in the feed water (mainly magnetite particles spalling from the water-side surface of the tube walls) can collect at the bottom of the horizontally-mounted vessel without creating difficulties and can be removed >asily by blow down. Water-tube v 'ilers, especially bayonet-types, are very vulnerable in this respect, since the deposits may collect in the lowest and most intensively heated part of the tube. In an extreme case of scaling, this may restrict the water flow to the point where boiling occurs irregularly (film boiling) and there is a risk of overheating and abe failure. The key factor which allowed the use of fire-tube boilers after the secondary reformer was the development of thin tube sheet designs. Thick tube sheets in this kind of service are too rigid and have too big a temperature gradient, and the resultant high stress on the tube-totubesheet welds leads to cracks. The nherent flexibility of thin tube sheets helps to disperse stresses and reduces the risk of fatigue failure of the tube-to-tubesheet welds and tubesheet-to-shell welds. Common to all the thin tubesheet designs is that the tubesheet is only 25-30 mm thick The hot inlet channel and the tubesheet are shielded by a refracory layer and the tube inlets are protected by inlet ferrules.

In the so-called membrane type designs of Uhde and Steinmller228 the tubesheet is anchored to and supported by the tubes to withstand the differential pressure. This imposes some restriction on the tube length. A different arrangement was developed by Babcock-Borsig229-230 in which the tubesheet is reinforced by stiffening plates on the back side (Fig. 48). Both concepts have fullpenetration tube-to-tubesheet welds for the tubes to avoid crevice corrosion. In another thin tubesheet version developed by Struthers227, membrane stress is avoided by making the tubesheet-to-shell connection flexible. The synthesis loop boiler on the exit of the converter is also a very important piece of equipment. In some modern plants not equipped with an auxiliary boiler it supplies nearly half of the total steam generation. It may generate as much as 1.5 tonnes of steam per tonne of ammonia, equivalent to about 90% of the reaction heat. Fire-tube versions are also currently used, including Babcock-rBorsig's thin tubesheet design. However, the conditions and stress patterns are somewhat different from those in the boiler after the secondary reformer, where the pressure is greater on the steam side. In the synthesis loop boiler the opposite is the case, with the result that the tubes are subjected to longitudi-

nal compression instead of being under tension, as in the main boiler. Several failures in this application have been reported231 so far, and though some of them may have been at least partly related to fabrication problems, they do raise fundamental doubts over whether this concept is appropriate for use as the synthesis loop boiler. More generally accepted are thick tubesheet concepts in various configurations. Proven U-tube type designs are available from Uhde232, Balcke-Drr233 *", Kellogg235 and Babcock-Borsig236. A special advantage of this type is that the tube bundle, being rigidly mounted at one end only, is essentially free of stress. The main tubesheet is protected in some way against damage by the hot gas. In one type the hot gas is introduced into an inlet chamber into which the hot ends of the boiler tubes, which project beyond the outer face of the main tubesheet and are fitted with long ferrules, are welded. The cold ends of the tubes are flush with the tubesheet in the normal way. Thus only cooled gas comes into contact with the outer surface of the tubesheet. The principle is shown in Fig. 49. Uhde has designed a variation on this basic concept, which is principally intended for use as the first synthesis loop boiler, in which the tubesheet has cooling channels

Fig. 48: Reinforced Tubesheet of Borsig Boiler

shell supporting ring

_ tube
before welding after welding (softener I plate TIG welded root pass anchor

41

MODERN PRODUCTION TECHNOLOGIES

Fig. 49: Conventional Synthesis Loop HP Steam Boiler with Thick Tube Sheet
steam outlet

Fig. 50: Babcock-Borsig Hot/Cold Synthesis Loop Boiler

O OO O OO

o oo o oo o oo oo
O 00 O 00

SECTION A-A: Q hot shank cold shank

o o o o o

blowdown

gas in

gas out

outlet

through which incoming boiler feed water circulates. An interesting and unique approach is Babcock-Borsig's patented Hot/Cold Tubesheet236. The hot and cold ends of the tubes are arranged alternately, so that a hot shank is always next to a cold shank and vice versa (Fig. 50). The advantage is that the tubesheet and the hot end tube wall temperature inside the tubesheet can be kept below 380C, which avoids nitriding and allows ferritic ferrules to be used. The entire tubesheet surface is shielded by a plenum chamber into which the cooled gas issues from the cold ends of boiler tubes, which end flush with the outer surface of the main

tubesheet. The hot ends, fitted with ferrules, extend about 50 mm beyond the outer face of the main tubesheet and into a "dummy" (nonload-bearing) tubesheet, which forms the dividing wall between the cooled-gas plenum chamber and a specially-designed inlet chamber. Another very successful design is the horizontal synthesis loop waste heat boiler of Balcke-Drr AG237. This boiler is a straight-tube design with thick tubesheets on both ends. In this concept, too, the hot ends of the tubes, equipped with Incploy ferrules, protrude above the hot-side tubesheet to end on a thin dummy tubesheet enclosing a gas inlet chamber. A particular feature is the welding of die tubes to the 6 mm Inconel cladding of the tubesheet. This weld is mainly for sealing purposes; the
42

mechanical tube-to-tubesheet joint which accommodates the stress is provided by hydraulic expansion of the tubes over the entire thicknes of the tubesheet238. Two shallow circumferential grooves in the tube hole in the tubesheet provide the anchoring needed to withstand the high axial forces to which the tubes are subjected in service and to prevent stress on the tube-to-tubesheet weld. The hydraulic expansion provides redundant gas-tightness.

Compressors
Centrifugal compressors239 for synthesis gas compression have become much more reliable over the years, though they are not spectacularly more efficient than they used to be. Some vendors claim efficiencies of

AMMONIA

nearly 80%, but in actual plant operation 75% is about as much can be expected. The turn-down capability can be brought down to 70% by changing the surge characteristic, though with a slight sacrifice of efficiency. Axial compressors are under disc ission for the process air compressor but seem to have found no widespread acceptance up to now. The optimum speed ranges for turbines of the size used in ammonia plants (5-7,000 rpm) differ considerably from those desirable for a centrifugal compressor (around 15,000 rpm), so the usual speed of about 11-12,000 rpm is a compromise. A gearbox would overcome this dilemma, but that would incur an additional efficiency loss. On account of the impeller dimensions in the low-pressure casing of the process air compressor there is already a gearbox between the high- and lowpressure casings. An interesting development is dry (oil-less) gas seals for turbocompressors. Development began as far back as 1969 and the first commercial application was in a natural gas compressor in 1973. Since then this technique has been widely used in refinery and natural gas installations, especially in off-shore service240"242. It is only quite recently that it has gained acceptance in the ammonia industry, where several compressors for synthesis gas and refrigeration duty equipped with dry seals have been successfully placed in service. Dry seals are now also in use in methanol plant synthesis gas compressors. Nitrogen is used as an inert fluid for the seal and the sealing is achieved at the radial interace of rotating and stationary rings n an unique and ingenious way. During operation the seal is not absolutely tight; some of the seal gas flows back to the suction side to be recompressed and a small amount from the suction side may go to the atmospheric level and will be sent to the flare on account of its content f combustibles. During stops, when e shaft is not rotating, the seal ring is pressed tight against the seat by means of a spring and the differential gas pressure. Dry gas seals in combination with oil-lubricated bearings (dry/wet) have the advantage that a much smaller oil system is

Table XI Development of the Energy Consumption of Ammonia Plants


Year Plant Feed Fuel, reformer Fuel, aux. boiler Export Total GJ/mt NH3 GJ/mt NH3 GJ/mt NH3 GJ/mt NH3 GJ/mt NH3 1966 A 23.9 13.0 2.6 39.5 1973 B 23.32 9.21 5.02 37.55 1977 C 23.48 7.35 3.06 33.89 1980 D 23.36 5.6S 1.17 30.18 1991 E 22.65 5.90 0.55 28.00

needed and that there is no contact between oil and the gas. A new development, already in commercial service but not so far used in ammonia plants, is magnetic bearings240'243. Magnetic bearings promise a wider temperature range, are less prone to wear, suffer less vibration due to imbalance and require less maintenance. A combination of magnetic bearings and dry seals (dry/dry) could totally eliminate the oil system.

Modern low-energj steam reforming ammonia plants


So far we have mainly considered the different process steps and the progress made with them in recent years. The previous section briefly discussed the energy-integration of these steps into a complete ammonia plant, and this section will review how all these elements are brought together in the major modern low-energy ammonia processes that have been or are being realized in actual plants or are at least offered on the market. Although the introduction of the single-train integrated large plant concept in the 1960s revolutionized the energy-economics of ammonia production, it is perhaps surprising to see just how substantial are the further improvements that have been made since then. Table XI shows how the specific energy consumption of ammonia plants has been progressively reduced over the years, using figures from actual plants. These examples show the general trend. There are, of course, exceptions. C F Braun (now Brown & Root) in fact anticipated the current need for low-energy processes and brought the first Braun Purifier ammonia process plant on stream as early as the end of the 1960s. It did not, of course, achieve the consump43

tion figures that can be attained today, but by the early 1980s the company was building plants consuming only 28.5 GJ per tonne of ammonia. The difference between plants A and E is 11.5 GJ per tonne of ammonia - a reduction of 29%. These savings can be apportioned as follows: 29% in the main compressors, 43% in the carbon dioxide removal system, 8% by introducing air preheat, 6% by lowering the steam-tocarbon ratio, and 6% by hydrogen recovery, the remainder being attributable to measures such as replacing small turbines by electric drives, generating higher-pressure steam and using a fuel gas expander. From the review of Table XI the question may come up, what is the theoretical minimum energy consumption for ammonia production via steam reforming of natural gas? Based on pure methane, we may formulate the following stoichiometric equation16: CH4 + 0.3035 O2 + 1.131 N2 + 1.393 Up - CO2 + 2.262 NH3 (23) mm = -86 kj/mol, AF298 = -101 kj/mol So from a mere thermodynamic point of view, in an ideal engine or fuel cell, heat and power should be obtained from this reaction. But because there is a high degree of irreversibility in the real process, a considerable amount of energy is necessary to produce the ammonia from methane, air and water. The stoichiometric quantity of methane derived from the foregoing equation is 583 Nm3 per tonne NHy corresponding to 20.9 GJ (LHV) per tonne of ammonia, which with some reason could be taken as minimum value. Of course, if one assumes full recovery of the reaction heat, then the minimum would be the heating

MODERN PRODUCTION TECHNOLOGIES

value of ammonia, which is 18.6 G] (LEV) per tonne NH3. An interesting analysis of the thermodynamics, the energy balances and the efficiency of the steam reforming ammonia process using me concept of "exergy" was made by I. Dybkjser201. From the exergies in total energy input and in ammonia product, he calculates a thermodynamic efficiency of 66% for a balanced plant consuming 29.3 GJ/t NH.,. It has become common practice to measure modern ammonia concepts above all by their total net energy consumption, which is the balance of energy input and energy output (export of steam or power) per tonne of ammonia. Yet these comparisons need some caution in interpretation; without precise knowledge of design and evaluation criteria, misleading conclusions may be drawn. In many cases, too, the degree of accuracy of such figures is overestimated. But, with these restrictions in mind, such figures may be rather useful. Here are some examples of how energy consumption values may be changed by plant variables201:

Assuming for the base case of a water-cooled ammonia plant a cooling water temperature of 20C, a rise by 10C would mean an increase of the energy consumption by 0.7 GJ/t NH3. Using natural gas containing 10% carbon dioxide (natural gas with up to 20% may be used in a steam reforming unit) would add 0.2 GI/NH 3 more. A nitrogen content in the natural gas is marginally beneficial: the saving resulting from a 10% nitrogen content is about 0.1 GJ/t NH3. However, the energy consumption increases slightly with the amount of higher hydrocarbons. The delivery temperature of the product ammonia is important. In relation to the case in which liquid ammonia is produced at ambient temperature, producing it as vapour at 3 bar would save 0.6 GJ/t NHy while delivering it as liquid at -33C would need an extra 0.3 GJ/t. Other important factors are the thermodynamic efficiency at which imported or exported power are converted to heat values and the state of imported or exported steam.

the following descriptions of different processes. The number of contractors or licensors having their own proprietary processes is actually quite small244. The following are the main ones. In principle they can be classified into three main categories: advanced con ventional processes with high-duty primary reforming and stoichiometric process air in the secondary reformer; processes with reduced primary reformer duty and surplus process air; and processes without a fired primary reformer, of which the LCA process of ICI is currently the only operating example in a grass-roots facility. However, this grouping is of little actual practical use, beyond the fact that it gives some idea of the extent of flue gas emissions. Even that is of questionable value, because site effects may govern the individual plant lay-out and, consequently, the specific amount of flue gas.

Kellogg LEAP concept24"47


Though Kellogg was the originator of the large-scale integrated singletrain ammonia plant, it was not the

These remarks should be thoroughly kept in mind when reading

Fig. 51: M. W. Kellogg's Low-Energy Process cooling


'"^ HT shift converter heat recovery

natural ,_ gas <3h-l 1 feed f- ] feed gas compressor process steam

LT shift converter heat recovery condensate to BFW system

-cooling
1

cooling JjCDijT, refrigeration I I r3r~~~ compressor

methanator feed preheat CO product to urea plant

syn. gas coo ing

Uf

jdryer

cooling I ling ammonia product

toatfnosphere T synthesis gas compressor 1 to condensate stripper

absorber semi-lean pump

lean pump

44

AMMONIA

to develop an avowedly "lowenergy" design. That honour went to C F Braun (now Brown & Root), whose Purifier process was remarkably adventurous for its time. Indeed, Kellogg's first low-energy flowsheet - known as LEAP (LowEnergy Ammonia Process) - was a furly conservative optimization of aie traditional process. Recently, however, Kellogg has evolved a much more radical design for a large plant embodying improvements of ideas previously proposed only for fairly small-scale installations (see page 50). One of the main features of the LEAP scheme is the degree of heat recovery from the reformer convection section. The temperatures of the mixed feed entering the reformer tubes and of the process air entering the secondary reformer are raised to the maximum extent possible with current metallurgy. This allows the reformer firing to be lowered and, conversely, the reforming pressure to be increased somewhat, which saves on compression costs. An important contribution comes from Kellogg's proprietary cross-flow horizontal converter, operating with small-sized catalyst, low inlet ammonia concentration and high conversion. Low-energy carbon dioxide removal systems such as aMDEA or

t j rs t

Selexol (the latter with lower product carbon dioxide recovery) also contribute to energy optimization. As with other advanced concepts, only a part of the secondary reformer waste heat is used for steam generation and the remainder is used for superheating. With only minimal export, a specific energy consumption of 27.9 GJ/t NH3can be achieved, but this could probably be brought down to about 27 GJ/mt NH3 were a use to be found for larger quantities of export steam. Kellogg also restricts the use of turbine drivers to the big machines: the process air and synthesis gas compressor (and possibly, also, the refrigeration compressor). If site conditions favour it, a turbine to drive an electric generator will be included in order to supply power for the plant and export needs. Obviously, additional savings may result from sophisticated design of the heat recovery equipment: for example, the "unitized chiller". Especially in the detailed design and engineering, Kellogg has gained vast experience in executing over 140 large single-train ammonia projects, a number unmatched by any other contractor. This may have a beneficial effect on investment costs and plant reliability. Figure 51248 is a simplified flow sheet of the process. Fig. 52: Uhde Synthesis Loop

Uhde's low-ene technology81'm'

9 25

ammona -

Uhde has been in the ammonia plant engineering business since 1928. Continuous development efforts over the last 20 years aimed at improving process efficiency, plant reliability and - especially - overall costs resulted in Uhde 's low-energy process. A 1,800-t/d plant based on this technology was constructed for BASF in Antwerp, Belgium, and commissioned in 1991251. The process is also nearer to the conventional design philosophy. Using Uhde's well-proven reformer concept with its proprietary outlet header (page 17) and improved tube metallurgy, the pressure could be raised by 5 bar to 40 bar at the outlet. The mixed feed temperature was raised to 580C and process air preheat to 600C, with only slightly over the stoichiometric amount just to compensate for the nitrogen loss in the hydrogen recovery from the loop purge. The overall steam-tocarbon ratio has been reduced from 3.5 to 3.0. Once again, process heat from the secondary reformer outlet gas is partly used to generate highpressure steam, partly recovered by HP steam superheating. (For the Antwerp plant only 45% of the heat goes into steam generation208. For

45

MODERN PRODUCTION TECHNOLOGIES

carbon dioxide removal Uhde uses the BASF aMDEA process. The synthesis loop is a three-bed radial-flow converter system with two beds in the first high-pressure shell and one in the second. The outlets of the two vessels are connected to HP steam boilers of Uhde's proprietary design, raising about 1.5 tonne steam per tonne of ammonia, which is nearly 50% of the plant's steam demand. Other features include an optimized arrangement of water cooler and hot/cold recycle gas exchanger, in which nearly 80% of the ammonia produced is condensed and separated; in that way refrigeration energy is saved. Figure 52 is the basic loop configuration. For the refrigeration section Uhde prefers the use of screw compressors, which are regarded as the most economical choice in terms of energy consumption and investment costs; in addition, they can be very flexibly adjusted to plant load. Following the general trend, only the air and synthesis gas compressors and (if site conditions favour its inclu-

sion) a power generator are driven by steam turbines, resulting in a less complicated steam system. HP steam is generated at 125 bar and superheated to 530C; the MP steam header is usually kept at 48 bar. The net energy consumption is 28 GJ/t NH3. With inclusion of a gas turbine and a large surplus steam export, Uhde engineers expect values below 27 GJ/t NH/50.

known S-200 converter (with two radial catalyst beds and an intermediate exchanger), in combination with the usual HP steam boiler, to which a third radial catalyst bed in a separate vessel is added, followed by a second HP boiler. An energj consumption of 29.3 GJ/t NH3has been realized in 1989, but for newer concepts values of 27.9 GJ/t NH,110 are given. Brown & Root253 Characteristic of this low-energy process is the reduced primary reformer heat duty, which is achieved by shifting a proportion of the reforming reaction to the secondary reformer and using about 150% of the stoichiometric air flow. The excess of nitrogen is removed after the methanation stage by means of a cryogenic condensation step (the Purifier)174'254, which also removes all the methane remaining in the synthesis gas and part of the argon. That results in a purer than usual synthesis gas, requiring only mini-

Tops0e low-energy process201-252


Like Kellogg and Uhde, Tops0e has also reduced the energy consumption of a basically traditional process concept by applying systematic analysis and intensive process engineering. Special contributions are the low steam-to-carbon ratio of 2.8, made possible by a special HT shift catalyst development, the use of a more effective carbon dioxide removal system, and Tops0e's S-250 configuration191 for the ammonia synthesis section, similar to the Uhde concept. It consists of the wellFig. 53: The Braun Purifier Process

gas turbine air compressor primary reformer

absorber

air

steam

methanator

water

high-pressure separator

waste gas to fuel

let-down drum

46

AMMONIA

mal purge from the loop. The synthesis loop comprises Braun adiabatic converters with hot wall design, (see page 35): each catalyst bed (of which three are now used)209 is placed in a separate vessel and, of course, there is an inlet/outlet heat exchanger and high-pressure steam generation. The reformer heat duty may be reduced by as much as 50% in comparison with the classical process using only the stoichiometric amount of air in the secondary reformer. The smaller furnace will produce less flue gas and consequently less waste heat, which makes it easier to design an energybalanced plant with no export. Under the mild reforming conditions the steam/carbon ratio does not need to be as high as in a conventionally-operated reformer; it can, in fact, be reduced to 2.7 without adverse effects on the HT shift stage because the synthesis gas has a higher carbon dioxide content and therefore - as explained earlier (page 25) - a greater "oxidative potential". To suit the lower demand for high-pressure steam in a balanced plant, only a part of the process heat from the secondary reformer is used to raise steam; the rest is used to superheat it. Along with this, Brown & Root is following the general tendency to apply more electric drivers instead of turbines; in some plants only the air compressor and synthesis gas machine are turbine-driven. The concept shows great flexibility255 for design options. It is, for example, possible to go all out for the lowest natural gas consumption, even at the cost of importing some electric power. Alternatively, it can be arranged to export energy to improve the overall efficiency (net consumption) even further. In the latter case it is advantageous to incorporate a gas turbine to drive the process air compressor. The exhaust of the gas turbine is at about 500C and contains 16-17% oxygen. It can therefore serve as preheated combustion air for the primary reformer, a oncept which is analogous to a ombined-cycle power plant. Of course, an electric generator can be incorporated as well: this covers the power demand of the plant and, if the site conditions favour it, allows the surplus energy to be exported as electricity instead of as steam.

The Brown & Root process can attain 28 GJ/t NH3 in a balanced plant; with steam export and some yet unrealized potential improvements, a value of 27 GJ/t NH3 or slightly below seems to be feasible. Fig. 53 presents a simplified process flowsheet of the Brown & Root process253. Id's AMV process256 26 This concept also operates with surplus air (about 20%) to the secondary reformer. The consequently smaller primary reformer operates at a steam/carbon ratio of 2.8. The surplus of nitrogen is allowed to go right to the synthesis loop, which operates at the very low pressure of 80 bar using an unusually large catalyst volume and the cobalt-enhanced formulation261. The prototype is operating at Nitrogen Products (formerly CIL) in Canada. There is only one steam turbine, which drives the air compressor; all other machines have electric motors. Make-up synthesis gas and loop recycle are compressed in separate compressor sets. Half of the process steam is supplied by feed gas saturation. The feed gas compressor is located upstream of the methanator to make use of the compression heat in warming the rather cold gas coming from the Selexol carbon dioxide scrubber. Obviously the low synthesis pressure and the partial supply of the process steam by saturation has no marked effect on the overall efficiency, which, at 28.5 GJ/m NHj, is at the level of competitive designs.

Concept (page 52). Low-energy concepts of the more conventional approach, are the LEAD process of Jacobs265, not yet built but claiming 29.3 GJ/t NH3, and the Exxon Chemical low-energy ammonia process266-267, which is used in one plant in Canada with a consumption of 29 GJ/t NH3.

Economic plant size


During the 1980s about 110 new ammonia plants went on stream throughout the world268). Neglecting a few very small ones, the average capacity of the so-called "world-scale plant" was 1,120 t/d. The principle of the "economy of scale", long-established and well known in manufacturing processes, is as valid for chemical production plants as for any other. For ammonia plants of the size mentioned a scale exponent of 0.7 is widely used as a rule of thumb: Cost2 = costj x (capacity2/capacity1)a7 There is no doubt that, within reasonable limits, the specific capital investment per tonne of ammonia is lower for larger than for smaller capacities. Additional factors in the production cost which favour larger plants are other fixed costs, such as labour, maintenance and overhead. But the simple formulaic approach has to be handled with care because of some constraints. First, it can only be valid as long as one scales up or down a fixed process configuration. Secondly, dimensions of equipment for very large-capacity plants may lead to disproportionate price increases; for example, there may be fewer vendors, transport to the site may become very complicated, or special materials and fabrication procedures may become necessary. At the other end of the scale, with the smaller-size equipment for very small plants, savings may result from modular construction and preassembly and precommissioning of units in vendors' shops. So scale factors should differ for the very low and the very large capacity ranges. Undoubtedly, though, specific investment should decrease if the capacity of the world-scale plant is increased. And, although the arguments given above suggest that there

Other concepts
The Foster Wheeler AM2 process262'264 also belongs to the group which shifts load from the primary to the secondary reformer using an excess of process air and needs a cryogenic unit for removal of the surplus nitrogen. Energy consumption for that process, which so far has not been commercialized, is calculated as low as 29.3 GJ/t NH3. Jacobs's MDF process (32.8 GJ/t NH3)201 uses only a primary reformer and HT shift followed by a PSA purification. This process needs an air separation unit and is similar to the PARC process of KTI (page 48) and the Linde Ammonia
47

MODERN PRODUCTION TECHNOLOGIES

is a capacity maximum beyond which the specific investment will begin to rise again, it seems from present experience that a capacity of 2,000 t/d is not beyond the optimum.

A chance for small plants


But these world-scale ammonia plants suffer from one inherent disadvantage: They require an extremely large commitment of capital and feedstock and, in the case of the latter, there may sometimes be doubts about long-term availability and reliability of supply. Another unpleasant feature is the lack of flexibility, which makes it difficult to respond to fluctuating demand by turning down the production rate without losing efficiency or stressing equipment. The high integration complicates start-up and shut-down, thus drawing out the time needed and leading to considerable expenditure of natural gas without producing ammonia. Large-scale equipment and machinery need more elaborate maintenance and turn-around schedules. For these reasons, and on account of their generally simpler operation, smaller-capacity units have their attractions, so long as they are not associated with unacceptable penalties in respect of specific investment cost and efficiency. As already mentioned, skidmounting, pre-assembly and pretesting of process units are possible when dimensions shrink. This mighfc indeed, change the picture considerably. Furthermore, small dimensions will more easily allow differences in technological approach in some sections, stepping away from the traditional established concept. This is exactly what ICI did with its LCA (Leading Concept Ammonia) process, a concept for the 500 t/d range claiming to match closely the process economics of a world-scale plant. But also a new, possibly less costly technology should have its own economics of scale; there should be a certain capacity at which a minimum of specific investment is reached. KTI is a licensor for the LCA process and its engineers came to just that conclusion170, but their plot of the specific investment shows the (rather flat) minimum for the classical world-scale plant at around 1,500

t/d and for the LCA around 500 t/d, yet with the surprising peculiarity that the specific investment at this point should be lower for the LCA than for the classical worldscale plant, which implies that producing 1,500 t/d in three LCA units should be cheaper than in one classical world-scale plant of 1,500 t/d. (The abscissa of the plot representing the investment cost per tonne has no scale, so absolute values are not available for comparison.) A similar advantage was figured out by ICI in a more general article on small-sized plants269. It is quite natural that some competitors in the ammonia contracting business have a different view regarding the competitiveness of small plants270. But theoretically it cannot be ruled out that with new, much cheaper technologies a scenario for the investment cost as described is possible. It is, after all, well known that cost estimates for plants are a most difficult and delicate subject, which may lead to endless discussions among experts. Certainly, with limited local demand and difficult logistic conditions sometimes combined with capital shortage, there will be a future market for small-capacity plants, especially in developing countries.

duct and, optionally, in the synthesis loop. The surplus over the process steam requirement drives an electrical generator providing a substantial part of the plant's electric demand. A special proprietary part is the PSA version (UCC), in which nitrogen flushing is used to enhance hydrogen recovery and introduce the stoichiometric nitrogen demand. In this concept no less than four sections of the classical process sequence (secondary reforming, LT shift, carbon dioxide scrubbing and methanation) are replaced by the fully-automatic, high-efficiency PSA system. The overall efficiency ranges from 29.3 to 31.8 GJ/t NHy depending on plant specifics. If carbon dioxide recovery is needed a carbon dioxide scrubbing unit has to be added ahead of the PSA unit and another 0.8 GJ/t NH3 is needed.

ICI LCA process


This process is a radical departure from the philosophy of the highly integrated large plant which has been so successful for more than 25 years. After intensive research and development, including a pilot plant for the Gas Heated Reformer (GHR) at the Billingham Site, ICI started to build two identical 450-t/d plants together with a common utilities section at its Severnside site near Bristol, in the west of England. The installation came on stream in 1988 and cost about 60 million87-273-274. Figure 54 is a flow diagram for a core unit, and Figure 55 shows how the core unit connects up with the common utilities section170. For feed preheat, superheated steam is used (450C) and the process air is heated to 670C in a separate fired heater. The heat of the primary reforming in the GHR is supplied by the hot effluent of the secondary reformer. The partially-reformed gas exits the tubes at about 700C, so there is a rather high methane slip. This needs an over-stoichiometric air supply (similar to the Brown & Root process, but even higher). This feature is necessary to dose the energy balance between the two linked process steps. Because electric drivers are used for the synthesis gas compressor and air compressor, no 100-bar steam is raised. The CO shift is conducted in a

PARC ammonia process of KTI


An early approach to economic small-size ammonia plants was made by KTI with the PARC ammonia process170-171'271'272. The process combines the following well-proven steps: Air separation to produce nitrogen Steam reforming HT shift Pressure swing adsorption Loop. A Rankine cycle can be included to generate electric power from the heat of the HTS. It was originally proposed that this should use trichorofluoromethane as working fluid, though presumably in the light of current strictures on CFCs some alternative would be necessary today. Steam is raised downstream of the primary reformer in the flue gas
48

AMMONIA

Fig. 54: ICI's LCA Process (Core Unit) - 450 t/d


process air
r rvmtMr QC c r\i

to air
S^T\

fuel gas
T
r

fue|

< ~ n

> "-r

^ZjWz^
i0Jfro^sT^^ 6IOkmol/h s-rf . ^
bar A *, 26S''C desaturator
j'' *

250C

_ "i

recovery

shift

methanator I 30C M gas dryers

2-stage flash cooling (one stage shown)

Fig. 55: Arrangement of LCA Core Units and Utilities Section at Severnside
feed

I
desulphurization

~I
utilities boiler turboalternator IP steam export > electricity I* LP steam

saturation
reforming

process air heater process air compressor

shift conversion

desatu ration
PSA

t air
condensate

off-gas fuel

j recovery

purge gas methanation and drying compression circulator saturated steam synthesis loop ammonia vapour
refrigeration

cooling water system


demin. plant bfw

make-up water , liquid ammonia -> ammonia vapour

L.
49

MODERN PRODUCTION TECHNOLOGIES

single stage using a special copperbased shift catalyst operating at 265C in an isothermal reactor. The heat of reaction is used for feed gas saturation to provide process steam. With this catalyst and the low temperature level, there is no danger of Fischer-Tropsch side-reactions, and the process can therefore be run at a steam/carbon ratio of only 2.5. A pressure swing absorption (PSA) unit removes inerts and surplus nitrogen; separate treating of the synthesis loop purge for hydrogen recovery is unnecessary because this is recycled to the PSA. Carbon dioxide is removed from the off-gas of the PSA using BASF's aMDEA process at 3.5 bar. The synthesis reactor operates at a pressure of 80 bar, using ICI's cobalt-promoted catalyst. Not much has been disclosed about its internals, so it is just speculation that the concept might be ICI's Tubular Cooled Converter (TCC), also used for methanol synthesis, in which the inlet gas is heated in tubes surrounded by the catalyst. The gas enters at 225C and leaves at 380C. A report on the first two years of operating experience given at the AIChE Ammonia Safety Symposium in 199189 confirmed expectations on operabity and reliability, especially of the GHR. In the early period of operation some catalyst breakage occurred as result of differential expansion between bayonet and scabbard tube. Subsequent modification eliminated this problem. Material specification is a key part in the GHR design. Operating experience has confirmed the original choice of material. Some minor corrosion observed is of no importance and does not lead to failures or outages, An update on experience and further details were presented at subsequent Ammonia Safety Symposia275-276. As designed for the conditions at Severnside, the flowsheet energy consumption corresponds to 29.3 GJ/t NH3. ICI has reported89 that, on account of underperformance of the air compressor and PSA and gas supply for nitrogen oxides abatement in the nitric acid plant, an efficiency loss of 0.8 GJ/t NH3 was encountered in the core unit. A further penalty of 1.7 GJ/t NH, resulted in the utility section from having to

Table XII Typical Figures for the LCA Process


Natural gas (feed + fuel), GJ/t NH3 Import power (5 MW), Gj/t NH3 Total energy consumption, GJ/t NH3 Liquid ammonia product (14 bar, 30C),t/d CO2 (purity >99%), t/d

27.0 3.1
30.1

457 617

run the electric generator at a very low load because of the contemporary poor export possibilities at the Severnside location. The sustained overall plant energy consumption at Severnside was therefore 31.8 GJ/t NH3. Table XII shows typical data for the LCA process given by the licensee KTF. The LCA process was, according to ICI's own statements, definitely tailored according to the "small is beautiful" philosophy. But it is likely that the GHR could be built for single-line capacities considerably larger than 500 t/d NHj without changing the present mechanical configuration or the existing mode of operation. Of course, it should also be possible to by-pass the GHR by introducing a smaller proportion of the feed directly into the secondary reformer an idea that is a feature of the KRES but not a proprietary one, having been proposed years ago in the BYAS process of Jacobs and in the Foster-Wheeler AM2 process (page 47). Because in that case a lower proportion of the heat content of the secondary reformer exit gas is used in the tubular reforming, the shellside exit temperature of the GHR will rise, presenting an opportunity to raise some steam. In a larger-scale plant, that could allow an advantageous partial switch from electric to turbine-driven machinery. Elsewhere in the process, neither the isothermal shift conversion nor the PSA system imposes any particular limits on substantial capacity increases, and the tubular synthesis converter can be scaled up considerably. Where pure carbon dioxide is required as a by-product, it would probably be advantageous in a large plant to place the recovery system in the process gas stream ahead of the PSA unit instead of in the lowpressure PSA off-gas stream. So far, however, ICI has not released any
50

plans to scale the process up to higher capacities, and it is significant that the first commercial LCA project to be announced - a plant for Mississippi Chemical Corporation at Yazoo City, Mississippi, which will be built by KTI - is of the standard size91. That a process so explicitly targeted at developing world applications has found its first customer in a market dominated by the traditional philosophy of the economy of scale is not merely a surprise; it is a vindication of its design philosophy. It can probably be expected that, with some modifications, especially for higher capacities, the total energy consumption can also be brought down to an eventual level of about 28GJ/tNH 3 . Jacobs now also specializes in the design and construction of small plants in the 300-500 t/d range. A description of its "PLUS-Technology" in a recent company brochure, asserts a natural gas consumption figure as low as 26.8 GJ/tonne NH,.

Newest large-plant concepts


Kellogg's Ammonia 2000
After the successful commercialization of the KAAP synthesis loop in the Methanex ammonia plant retrofit at Kitimat in Canada, Kellogg currently has two other KAAP retrofits under construction, one in the USA (due to open this year) and the other in Australia (scheduled for start-up in 1997). Two grass-roots ammonia projects for American proprietors in Trinidad, each of 1,850 t/d capacity, include a full KAAP synthesis loop. Both plants will be commissioned in 1998. Kellogg was also able to prove the operational feasibility and performance of the KRES concept in the Kitimat plant. Based on these favourable experiences", Kellogg has combined its KAAP synthesis loop with its KRES exchanger reformer/ autothermal reformer combination into Ammonia 2000 - an optimized integrated concept for grass-roots ammonia plants. In contrast to the ICI concept described above, Kellogg's Ammonia 2000 concept is intended for use in world-scale single-train plants in the 1,800 t/d range94'95. Kellogg has

AMMONIA

executed very detailed engineering studies with project quality for two potential clients in the United States. In the new process, the traditional direct-fired primary reformer, which has been the workhorse of ammonia plants since the early 1950s, is completely eliminated and the classical iron-based catalyst is now irgely replaced by the new ruthenium-based catalyst. Ammonia 2000 uses energy more efficiently by recovering it at a higher temperature level. For example, up to 50% of the available heat in the reformed gas effluent is recovered in reforming the natural gas feed. The overall energy consumption of the complete process is as low as 27.2 GJ/t NHj. There are also savings in site area and structural steel and, with the dramatic technological change in the traditionally high-cost areas of primary reforming and the synthesis loop, the capital cost of Ammonia 2000 is about 10% lower than that of a conventional design. Figure 56 is a flow diagram of Ammonia 2000. Desulphurized natu-

ral gas is mixed with steam and then split into two streams in the approximate proportion of 2:1. These are heated separately in a fired heater. The smaller of the two enters the exchanger reformer at 500-550C, while the remainder is passed directly to the autothermal reformer at 600-640C. The exchanger reformer and the autothermal reformer use conventional nickel-based primary and secondary reforming catalysts respectively. To satisfy both the stoichiometry and thermodynamics of the reforming step, the autothermal reformer is fed with enriched air (30% O2), preheated to the same temperature as the gas/steam feed mixture. The required heat for the endothermic reaction in the tubes of the exchanger reformer comes from the gases on the shell side, comprising a mixture of the effluent from the autothermal reformer and the gases emerging from the reformer tubes, as shown in Fig. 21 and previously described on page 20). The outlet (shell-side) gas pressure is 40 bar. The high- and lowtemperature shift, carbon dioxide re-

moval and methanation sections are essentially the same as in a conventional plant. To raise the synthesis gas to the low synthesis pressure of around 90 bar only a single-casing compressor is required. As described previously (page 36), the KAAP converter is a vertical 4-bed radial-flow design with interbed heat exchangers. The low synthesis pressure and temperature allow a hot-wall design constructed in mild steel to be used. The first bed is charged with conventional iron-based catalyst for bulk conversion and the other beds with Kellogg's high-activity ruthenium-based catalyst. Because of its superior lowtemperature performance, this allows an exit ammonia concentration in excess of 20% to be attained, in spite of the low operating pressure. Other advantages of Ammonia 2000 are reduced operator demands (mainly on account of the elimination of the reforming furnace) and easier and less costly maintenance. The process scores from the environmental point of view, too: atmospheric emissions of both nitrogen

Fig. 56: Ammonia 2000 (Integrated KRES/KAAP Process)

feed sulphur

process steam

I 1
methanator

.*.

---f ,
_ -jm

.sC_.
process heater

autothermal reformer

reforming exchanger

to condensate stripper

dl

7er

purge gas to fuel

51

MODERN PRODUCTION TECHNOLOGIES

oxides and carbon dioxide are dramatically reduced because there is no reforming furnace.

LAC - the Linde Ammonia Concept


This recently-announced concept173-277/27B consists essentially of a hydrogen plant with only a PSA unit to purify the synthesis gas, a standard cryogenic nitrogen unit, and an ammonia synthesis loop. In principle it is thus rather similar to the KTI PARC process (page 48), but Linde is aiming its process at worldscale projects. The first commercial LAC plant is a 1,350-t/d unit in India at the Gujarat State Fertilizer Co.'s Baroda works. Figure 57 compares LAC with a conventional ammonia process in block form. LAC has three fewer catalytic steps than the conventional process and, because the synthesis gas entering the loop is essentially inerts-free, there is no need for a purge gas hydrogen recovery unit, either. The single isothermal shift conversion stage uses the Linde spiral-wound reactor, which has been used successfully for methanol synthesis and hydrognation in ten plants around the world. The special geometric arrangement of the spiral-wound cooling tubes in the catalyst bed provides excellent heat

transfer characteristics, and the operating temperature can be finely adjusted simply by controlling the pressure of the steam generated in the tubes. The 12-bed PSA system is Linde's own technology. In the loop a Casale three-bed converter with two interbed heat exchangers is used. As in Id's LCA process, pure carbon dioxide can be recovered by scrubbing the off-gas from the PSA unit, for which Linde, too, uses the BASF aMDEA process. According to Linde, the process should consume about 28.5 GJ/t NH3 or, with inclusion of pure CO2 recovery, 29.3 GJ/t NH3.

Ammonia bi oxidation of heavy hydrocarbon fractions


For lack of economic incentive, much less optimization and development work has been dedicated to this field in comparison with steam reforming. Consequently, less has been published on the subject32-279. As synthesis gas compression and ammonia synthesis are basically the same as described in the steam reforming section, this section will be concerned with gasification, shift conversion and purification only. Because pure oxygen (98.5+%) is usually needed for the partial oxi-

dation reaction and a source of pure nitrogen is needed to cover the stoichiometric demand for the ammonia synthesis, it is essential to include an air separation unit. This does, however, provide an incidental benefit, in that it allows a liquid nitrogen wash system to be used as the final stage of the synthesis gas purification train, giving an extremely pure hydrogen/nitrogen mixture with virtually no inerts, so there is no need for any kind of purge gas recovery treatment. Today highly efficient air separation systems are available from several contractor/licensors. As there have been neither any fundamentally new developments nor any new features relating to the partial oxidation processes specifically for ammonia production, this section refers mostly to the general literature of gasification rather than that of ammonia technology.

Partial oxidation processes


The basic principles of partial oxidation have been discussed in Chapter 1 and earlier in this chapter in the general discussion of routes to synthesis gas production. The two commercially-proven processes, the Texaco synthesis gas generation process (TSGP) and the Shell gasifi-

Fig. 57: Comparison of LAC and Conventional Ammonia Process Schemes


Conventional Ammonia Plant
CO,
* NG
desuiphuri primary secondary zation reformer reformer
NH3

removal

coz-

methanation

ammonia synthesis

air k

Linde Ammonia Concept (LAC)


NG
primary reformer isotherm, shift

PSA

air

52

AMMONIA

cation process (SGP), are rather similar and differ mainly in respect of the burner nozzle for introduction of feedstock and oxygen, the method used for cooling the raw gas after the reaction, and the soot recovery and waste water management. Both processes are operated under el3 /ated pressure, which may be as high as 80 bar.

Shell Gasification Process (SGP)


In Shell's process25'28- 28, preheated oxygen and heavy hydrocarbon feedstock are discharged through a nozzle into an empty reactor vessel, vhich is lined with high-alumina brickwork. The pressure shell is equipped with a special system to monitor the skin temperature for sarly detection of any defects in the reactor lining. Oil enters through a central jet in the burner nozzle. A substantial pressure drop (up to 80 bars) is needed to ensure that the iil is thoroughly atomized by the jxygen, which is fed in through the annulus between the central jet and the outer case of the burner nozzle. Steam is added as moderator for the reaction. The combustion temperature is in the range 1,200-1,400CC The raw synthesis gas contains soot, formed because of insufficient mixing of the reactants, and ash. A vaste heat boiler, which produces iOO-bar steam, cools the gas down

to 340C. The boiler is of a special design and has so far been used in 135 installations world-wide. Immediately after this boiler the gas is cooled further in an economizer, which preheats boiler feed water. Figure 58 is a simplified flow diagram of the SGP281. Soot is removed from the product gas in a two-stage water wash. About 95% is removed by a direct spray in a quench pipe and the remainder by countercurrent scrubbing in a packed tower. The older units used a light oil fraction to extract the carbon from the aqueous soot suspension, forming small pebbles, which were strained off by vibrating sieves. The pebbles could be mixed into the gasifier feedstock or were burnt separately in a steam boiler. In a further development/ naphtha was used as extraction medium, also forming pellets, which were treated as described. However, after the pellets had been mixed with the main feedstock, the naphtha (which is a very expensive fraction) was distilled off and recycled to the extraction. Because the process produces a surplus of water, some waste water run-down is inevitable and it is necessary as well to remove the ash formed in the reaction. (If the soot were 100% recycled in the way described, there would be a considerable build-up of ash in the sootscrubbing cycle282"285. The run-down

water from the sieves contains traces of naphtha and ash. First, the naphtha is stripped out and returned to the extraction, then the ash is separated and dewatered in a filter press. The filter cake, which may contain 25-30% of vanadium, is sold to the metallurgical industry. This clean-up is sufficient for the recycle-water to the soot-scrubbing circuit. But any surplus of water formed in the process must be discharged, and before that can be done it must be treated further. Dissolved gases such as hydrogen sulphide, carbon dioxide and ammonia are removed by stripping and are sent to an incinerator. Final cleaning takes place in a flocculation-sedimentation system for trace metal-containing ash and in a biological treatment. The vanadium- and nickel-containing paste is sold to the metallurgical industry285. A number of factors have prompted updating of the described extraction methods, especially when using heavy residues such as those from Visbreaking, which yield sufficient vanadium (about 200 t/a for a 1,800-t/d ammonia plant) to be worth recovering for metallurgical use280. In the new separation technique, the carbon/water slurry is directly filtered on an automatic filter. The moist filter cake is subjected to controlled oxidation in a multiple-hearth furnace. A vanadium concentrate containing about 75% V2O5

Fig. 58: The Shell Gasification Process (SGP)


steam naphtha -*

oxygen

feedwater heavy oil

waste water

a - preheater b - reactor c - waste heat boiler d - scrubber e - quench f - extractor g - vibrating screen h - flash tank i - soot-oil tank j - homogenizer k - naphtha stripper I - naphtha recycle tank

53

MODERN PRODUCTION TECHNOLOGIES

is obtained. The process needs no additional fuel and is less expensive to install than the extraction process.

Texaco Syngas Generation Process (TSGP)


In the Texaco process20-23- M1- m preheated reactants also enter an empty refractory-lined reactor through a (water-cooled) burner with concentric nozzles and mix at the burner tip. However, in contrast to the Shell arrangement, oxygen enters through the centre pipe and the oil is fed through the outer annulus. In this process, too, steam is used as a moderator. The raw synthesis gas contains about 1-2% of soot based on the feed and may have, for example, the following composition (dry basis): 47% Hy 46% CO, 5% CO2 and 0.5% CH4. When the process is used for generating ammonia synthesis gas, the raw gas is cooled from its initial temperature of about 1,400C by direct quench with water from the soot removal cycle. Figure 59286 gives a simplified flowsheet of this process, also showing the soot removal and recovery system. After the quench comes a twostage soot scrubbing system, consisting of a venturi-scrubber followed by a packed tower for counter-current

scrubbing. The soot can be extracted from the water with naphtha because of its more lipophilic surface character. The soot-naphtha suspension is separated in a settler from the nearly carbon-free grey water, which is recycled to the quench and scrubbing section. The soot-naphtha phase is mixed with feed oil, after which the naphtha is distilled off in a stripping column for recycle to the extraction step. The soot-containing oil from the bottom is then fed back to the Texaco synthesis gas generator. Because water is formed in the gasification, and also to prevent accumulation of ash or slag in the water circuit, some water has to be discharged. In a combined chemical and physical treatment step sulphides, cyanides and the suspended solids (ash) are removed. A concentrated sludge separated in a settlerfilter combination has to be disposed of. The water from the filter is stripped of ammonia and, after pH adjustment, is sent to a biological treatment unit There is also a version of the Texaco process with a waste heat boiler instead of the quench, but it is normally used for synthesis gas generation for methanol or oxoalcohol production. For ammonia and hydrogen production Texaco

licensors are of the opinion that the quench is advantageous because it provides most of the steam needed in the subsequent shit" conversion.

Shift conversion
In Texaco's standard quench route to ammonia the shift conversion is performed on a Co-Mo-alumina catalyst287"292, which operates in the temperature range 230-500C in a sulphided form; the sulphidation i< effected autogenously by the hydro gen sulphide content of the gas. In principle, the classic iron-chromium catalyst can be used, too, but it is less active under these conditions and, in the presence of the sulphur content of the gas and the rather high steam content needed to convert its high carbon monoxide content, it tends to undergo hydro thermal aging, leading to acceleratec mechanical disintegration. The cobalt-moly catalyst combines high activity with long service life. In the Shell concept, with its waste heat boiler for cooling the raw synthesis gas, sulphur compounds are removed ahead of the shift conversion by scrubbing with cold methanol in the Rectisol proces; (described later on page 55). Hydrogen sulphide and carbon dioxide are

Fig. 59: Ammonia Process Based on Texaco Gasification


""/S6" ^ Lreactor condensate
compression
' synthesis >. ammonia

crude gas

oil feed
-k

ammonia to storage
separator
>0 CW

steam

carbon separation section naphtha, H a S,COJcw soot-free water i upW

8
naphtha h
'i

tofla

.
soot oil

soot-free water

condensate

54

AMMONIA

both soluble in the scrubbing medium, and some effort is needed to obtain from the solvent regeneration stage an acid gas rich enough in hydrogen sulphide to be suitable for processing in a Claus sulphur recovery unit. For this reason some advantage can be seen in placing this ; ep ahead of the shift conversion, because in that case the hydrogen sulphide is more concentrated in the raw gas and not yet diluted by the huge quantity of carbon dioxide formed in the shift conversion. In this process configuration, the traditional iron-chromium shift catalyst is used. The Co-Mo-alumina catalyst annot be used in this application jecause it needs sulphur for its performance. Because of the high initial carbon monoxide concentration, which causes a considerable adiabatic temperature rise, several catalyst beds with intermediate heat removal have to be used for thermodynamic reasons to attain a sufficiently low reidual carbon monoxide content. vYith the cobalt-molybdenum cata-

lyst, 1.3% CO (on dry basis) is attainable.

CO2/H2S removal
The Rectisol chilled methanol wash process293"296 seems to be the dominating choice in partial oxidation plants and is the only one which will be discussed here. Other routes using aqueous amine solutions or the Selexol process are far less commonly employed297-298. The Rectisol process, invented by Lurgi and developed further in collaboration with Linde, takes advantage of the fact that methanol is a cheap and easily available solvent in which hydrogen sulphide, carbon dioxide and carbonyl sulphide (COS) are readily soluble at low temperature. On account of the different solubilities of these substances (at 30C, for example, the Henry absorption coefficient per mol for hydrogen sulphide is about six times higher than for carbon dioxide), it is possible to design the process to produce sulphur-free synthesis gas

(<0.1 ppmv) with a low residual carbon dioxide content of about 10 ppmv, pure sulphur-free carbon dioxide product and an H2S-rich carbon dioxide fraction to be fed to a Claus plant for sulphur recovery or to be directly used for sulphuric acid production. The Rectisol process is very versatile and, according to the gasification process being used, different configurations are possible. When used with the Shell partial oxidation process (Fig. 58, page 53) a first scrubbing stage, which removes the sulphur compounds, is positioned upstream of the shift conversion. The second stage, removing carbon dioxide, is located downstream of the shift reactors. A different arrangement, used with the quench-type Texaco partial oxidation process, has both separation stages after the shift conversion. It is shown in Fig. 60299. After separation of the surplus steam remaining after the shift conversion as process condensate, the gas is cooled in a chiller by evaporating ammonia. More condensate is

Fig. 60: Rectisol Acid Gas Removal as Used with the Texaco Gasification Process
raw gas-

- methanol

gas from N2 wash

55

MODERN PRODUCTION TECHNOLOGIES

formed and removed, and then the cold gas passes through the hydrogen sulphide absorber, where it is scrubbed with sulphur-free methanol, preladen with carbon dioxide. Next it enters the carbon dioxide absorber, where its carbon dioxide content is removed in different sections: in the top section, through which it passes last, it is washed with practically pure methanol from the hot regeneration. Additional sections use flash-regenerated methanol, and interstage cooling is provided to remove the heat of absorption. Hydrogen sulphide enrichment in the hot regeneration is achieved by a special design involving a combination of flashing and reabsorption.

Final purification of the synthesis gas


Final purification, mainly to remove the residual carbon monoxide content, which may be anything from

1.5% to 4%, depending on the shift conversion lay-out, is usually performed in a liquid nitrogen wash300-301, where the synthesis gas is scrubbed with liquid nitrogen at about -190C. The nitrogen is liquefied in a refrigeration cycle through compression, cooling and expansion and flows in a tray column in countercurrent to the cooled synthesis gas. Traces of water and carbon dioxide have to be removed carefully by molecular sieves or freezing out in reversible exchangers to avoid blocking and damage of the system. It is important to keep a continuous check on the gas for traces of nitric oxide and hydrocarbons, because accumulation of these materials by condensation may cause explosions. For energy optimization the refrigeration system of the air separation plant and the liquid nitrogen wash may be integrated with each other. Figure 61 gives the basic configuration302.

It has also been proposed to use methanation instead of a liquid nitrogen wash. A three-stage shift was proposed for this case to reach about 0.6 to 0.7% residual methane cortent using an improved cobal molybdenum catalyst. No informtion is available on practical experience with this approach. In spite of the upstream desulphurization, the use of a classical copper-containing low-temperature shift catalyst in a Shell process-based ammonia concept was considered too delicate because of the occasional presence of trace ; of hydrogen sulphide.

Sulphur management
The HjS-rich carbon dioxide fraction from the Rectisol unit may be either fed to a Claus plant to recover elemental sulphur or used to produce sulphuric acid. The first non-catalytic step of the Claus process303 is .partial combustion of the hydroger, sulphide to sulphur dioxide, which then reacts with the remaining hydrogen sulphide on an alumina catalyst to yield elemental sulphur (equation 24).
SO

Fig. 61: Liquid Nitrogen Wash Unit


trim valve nitrogen

3S + 2H2O AH29B = -142.8 kj/mol (24)

synthesis gas

residua! gas

pretreatment

synthesis gas

H 75 mol%
N2 25mol%

Because this reaction is an equilibrium, the catalytic stage is carried out in two or three stages and sulphur is removed after each in a condenser. With the inclusion of one of the many supplementary tail gas treatment processes, 99.7% or greater recovery can be achieved.

heat exchanger

Complete ammonia plants based on partial oxidation


Usually the degree of energy integration is lower than in the steam reforming process because, in the absence of a large fired furnace, there is no large amount of hot flue gas and consequently less waste heat is available. So in this process route a separate auxiliary boiler is usuall necessary to provide steam for me chanical energy and power generation. Nevertheless, in modern concepts considerable efforts have been made to bring the energy consumption down. Whereas older plant concepts had values of around 38 GJ/t NHy for a concept with the traditional use of 98.5%+ oxygen quit'1

wash refrigeration valve


N a control valve

wash column

-Cxd-*
56

AMMONIA

recently a figure of 33.5 GJ/t NH3 vvas claimed in a commercial bid. There are not really any complete process concepts for ammonia processes based on partial oxidation comparable to those described earlier for steam reforming ammonia processes because the various process stages >f these plants have been developed by companies from outside the ammonia world. Parts of any complete ammonia plant will therefore be licensed from their specialist developers. Lurgi, for example, has constructed two very large-capacity ammonia plants using Shell's gasification process, its own proprietary version of the Rectisol process, Linde's air separation process, Linde's liquid nitrogen wash, and Tops0e's ammonia reactor and loop design. Linde, on the other hand, has used for its projects the Texaco synthesis gas generation process, Tops0e's ammonia reactor and loop design, and its own Rectisol, liquid nitrogen wash and air separation technology. Figure 62 gives a simplified flow sheet for the Linde concept304-305 of a partial oxidation ammonia plant using the Texaco synthesis gas generation process. The figures in lozenge-shaped frames refer to the compositions of process streams given in Table XI. Feedstock is a heavy oil fraction

Table XIII

Composition of Process Streams in the Partial Oxidation Ammonia Plant in Fig. 62


l
CO CO,
43.0 47.7 6.9 O.I 1.1 1.2
60.8 1.5 35.8 O.I 0.8 1.0 96.3 2.3
0.9 0.1 98.8
O.I O.I O.I

75
100

N H S/COS
2

0.2 1.2

67.4 2.5 30.0

25

CH4/Ar

with 84.6% C, 10.0% H, 4.5% S and 0.1% ash. In the interests of lowering investment cost and energy consumption, it has been recommended to use air or enriched air instead of pure oxygen. Topsee306 proposed the use of enriched air (43%) and methanation instead of liquid nitrogen wash. For CO2/H2S removal Selexol is used. The shift reaction proceeds over a sulphur-resistant catalyst in a threebed configuration, bringing the residual carbon monoxide content down to 0.55%. For the loop a Series 200 converter is chosen. The partial oxidation step can be designed according to either the Texaco or the Shell process. For the overall consumption a value of 34.8 GJ/mt NH3 is claimed. Foster Wheeler264-307 has proposed the use of highly preheated air in a

Texaco synthesis gas generator operating at 70 bar. The synthesis gas purification train comprises soot scrubbing followed by shift conversion, acid gas removal and methanation. The gas is dried by molecular sieves and finally fed to a cryogenic unit, which removes the surplus nitrogen and residual traces of methane, argon and carbon monoxide. The rejected nitrogen is heated up and expanded in a turbine, which helps to drive the air compressor. A conventional air separation plant is based on the fractional distillation of oxygen and nitrogen, but the difference in their boiling points is only 13C. In the cryogenic unit of the Foster-Wheeler process a lesser quantity of nitrogen (because the stoichoimetrically needed proportion remains in the gas) is separated from hydrogen at a much higher boiling

Fig. 62: Integrated Ammonia Plant Using Texaco Gasification Process

HP

tail gas partial oxidation

CO. Rectisol scrubbing unit cold box ammonia loop

CO shift conversion

a - air separation unit b soot extraction c - CO2 absorption d - methanol/H2O distillation e - stripper f - hot regenerator g - refrigerant h - dryer i - liquid nitrogen scrubber j - syn. gas compressor k - NH3 reactor

57

MODERN PRODUCTION TECHNOLOGIES

point differential (57C). This gives, according to Foster Wheeler, a considerable saving in capital investment and energy consumption against the traditional approach with air separation and liquid nitrogen wash. A figure of 35.6-37.6 GJ/t NH3 is claimed for heavy oil feedstock and 31.4-32.7 GJ/t for natural gas as feedstock307.

Table XIV Coal Gasification Processes in Commercial Operation and Their Capability for Handling Different Types of Coal
Pressure range, bar Lurgi (dry) British Gas/Lurgi Winkler/HTW Koppers-Totzek Shell Texaco Dow 20-100 20-70 1-10 20-40 20-40 20-40 10-20 Temp., C 300-600 450 1050 1800 1500 1350 1450/1000 Anthracite

Bituminous Lignite coal


+ /++ ++ + ++ ++ ++ ++ ++ -/ +
-M-

Ash >30% Xes no yes no no no no

+ /-H+ /++ + + + + +

++ ++

Coal-based ammonia plants


There are no differences in principle between the shift conversion, final synthesis gas purification and ammonia loop of an ammonia plant based on coal gasification and those of one based on partial oxidation of heavy hydrocarbons. This section therefore considers only the gasification step.

Fig. 63: Koppers-Totzek Gasifier


raw gas

coal

Gasification processes
Coal gasification processes30-41'308-310 in commercial operation are shown in Table XIV309. So far only the Koppers-Totzek, Texaco and Lurgi gasifiers have been used commercially in ammonia production. But the rather successful demonstration of the Shell process in other applications justifies its inclusion here as a potential candidate for ammonia production processes. A review of additional technology developments was presented at Ute Symposium Ammonia from Coal organized by TVA in 1979. The Koppers-Totzek process39'311-312, used in several ammonia plants in China, India and South Africa, operates at essentially atmospheric pressure. Figure 6330* is a sketch of die gasifier. Dried coal, ground to dust size, is fed to the two (in some cases four) burners of the gasifier. Oxygen is introduced immediately ahead of the burners and, together with a small amount of steam, the mixture enters the reaction zone at high velocity through the burner mouth. Coreflametemperatures may be as high as 2,000G The raw synthesis gas passes out through the top of the generator at 1,500-1,600C and through a waste heat boiler. It is cooled further in a soot scrubber. The soot-water suspension formed at that point is fed to a settler from which water and

rotary air lock

concentrated carbon slurry are separately recycled to the process. Any surplus of water has to be treated further before discharge. The combustion chamber has a refractory lining and, additionally, a low-pressure water-jacket for cooling. Most of the ash leaves the reactor at the bottom in liquid form and is solidified into granules by water quenching. The maximum gas production capacity of one generator is about 50,000 Nm3/ h. Typical gas composition is about 8-12% COy 59-62% CO and 26-29% H,. To overcome the disadvantage of its atmospheric-pressure operation Krupp-Koppers (originally also with participation of Shell) developed a version which can operate at 25-30 bar. This PRENFLOW (Pressurized Entrained Flow) process has so far been demonstrated on a pilot scale only313.
58

The Texaco coal gasification prois very sircar t0 tne Texaco partial oxidation process for heavy hydrocarbons. An aqueous pulp containing 60-70% coal is fed by a reciprocating pump to the burner, where oxygen flowing through a central pipe is added at the burner tip. For ammonia and hydrogen production, quench-cooling is applied, whereas, for methanol and oxo synthesis gas production, a waste heat boiler version is offered. In the waste heat boiler variation the raw gas flows down from the reaction space through a restriction into a radiation cooler to reduce the gas temperature below the solidification range of the slag, which is then collected in a water sump to be discharged through a pressure lock. After the radiation cooler the gas passes through a waste heat boiler and is then treated for carbon re
Cess308,314-317

MririVINI/"V

Fig. 64: Lurgi Dry Bottom Gasifier


coal

tar recycle jacket steam

moval in a Ventura scrubber and a packed column. Carbon is separated from the scrubbing water in a settler, which also removes some ash fines. In the Lurgi Dry Gasifier309-318-320, the gasification is performed with steam and oxygen in a moving bed operating usually at 25-30 bar, though pressure up to 100 bar is feasible in principle. Figure 64309 is a basic sketch. Crushed coal enters at the top of the gasifier via a lockhopper and is evenly distributed over the cross-section of the coal bed surface by a distribution disc equipped with scraper arms. The coal actually moves downward very slowly in the bed. Ash with less than 1% coal is removed at the bottom of the gasifier using a revolving grid with slots, through which steam and oxygen are introduced. In the hot lower section of the bed the temperature is around 1,000C, whereas in the top section, from which the raw gas leaves, it is only 600C. The whole reaction vessel is cooled by a water jacket. A modification of the process is the British Gas/Lurgi Slagging Gasifier309'm 321, in which ash residues are removed as molten slag. On account of the relatively low gasification temperatures - a feature which saves oxygen in comparison

with other processes - an increased amount of impurities are present in the raw gas - for example, tar, phenols and some higher hydrocarbons. Additionally the raw gas contains a rather high concentration of methane, usually around 10%. For these reasons, apart from the removal of particulates as coal dust and ash fines, a rather elaborate purification sequence310 is necessary to arrive at a sufficiently pure synthesis gas. The methane has to be removed in most cases further down stream, too, either by a cryogenic operation or by a pressure swing adsorption unit. In some cases the methane is then separately converted to synthesis gas in a small steam reformer. Unlike the Texaco process, the Shell Coal Gasification Process (SCGP)33'34'309'322-324 has departed completely from the concept of the earlier fuel oil partial oxidation process, successfully proven in more than 30 years. Compared to the latter, the flow pattern is reversed, the reaction takes place from bottom upwards, with the gas leaving at the top. The outside appearance has some resemblance to the KoppersTotzek gasifier with the integrated waste heat boiler unit. (Fig. 65309). Much development work has been devoted to the liquid slag withdrawal system and the nitrogenaided dry coal feeding system. To guard against small amounts of entrained slag in a semi-solid state sticking to the surfaces of the waste heat boiler, the hot gas is quenched by recycling cold raw gas.

Particulates are removed by water scrubbing in a subsequent Venturi scrubber and a packed column. The gas generator vessel is refractorylined. The gasifier of Dow Chemical's coal gasification process325'326 also has some common features with the Koppers-Totzek Process, but with the difference that coal is fed into the reaction zone as an aqueous slurry. The burners are positioned opposite to each other, and further up in the flow additional coal is introduced to a secondary reaction zone. The gas leaves at the top of the vessel and the slag, which is withdrawn from the reaction zone at the bottom in liquid form, is solidified and ground under pressure. A big installation with a gas production equivalent to about 1,500 t/d ammonia supplies a combined-cycle power generation scheme.

Complete ammonia plant concepts


As with heavy fuel oil, little noticeable work has been done on complete ammonia plant concepts based on coal. The traditional leading ammonia contractors have to rely on proprietary processes licensed from different companies, which similarly tend not to have any ammonia technology of their own. Therefore, even today, a coal-based ammonia plant will be at best a combination of technologies of different licensors put together by an experienced contractor. Compared to a steam reforming plant, the degree

Fig. 65: Shell Coal Gasification Process (SCGP)


coal dust quench gas

oxygen (steam)

y
s|ag

oxygen (steam)

water

59

MODERN PRODUCTION TECHNOLOGIES

of integration is considerably lower; usually the power generation facilities are more or less separate. So it is difficult to identify specific ammonia processes for the individual contractors. Lurgi33 offers a concept which includes, for the reasons explained above, steam reforming of the methane component of the raw gas in addition to the shift conversion, Rectisol unit and liquid nitrogen wash. The gasification needs 32-34 GJ/t NHy power and steam generation consumes 18-22 GJ/t NHy resulting in a total energy consumption of 50-56 GJ/t NH3. For the (unpressurized) Koppers-Totzek30 route a figure of 51.5 GJ/t NH3 is reported. The more elaborate gas cleaning in the Lurgi concept may account for the fact that the advantage of pressure gasification is not reflected more strongly in the consumption figures. be Industries commissioned a 1,000-t/d ammonia plant in 1984 using Texaco's coal gasification process327- 32S. An energy consumption of 44.3 GJ/t NHj is stated, which is lower than the normally quoted figure of 48.5 GJ/t NH3 for ammonia processes based on coal41. A recent feasibility study by Shell for its coal gasification process arrived at even lower figures33. But as no more data representing real plant experience are available, it is safer to stay on the conservative side and use higher figures.

Material questions
It is not proposed to present here a wide-ranging review (examples of which appear in the literature329- 33) or a detailed evaluation of materials used in the construction of ammonia plants, but merely to discuss briefly some special problems that have been encountered. To some degree, these points are also relevant to hydrogen and methanol plants, hi several steps of the ammonia process, especially in the synthesis section, the pressure shells of the reaction vessels are in contact with hydrogen at elevated pressure and temperature. Under certain conditions chemical hydrogen attack will occur2io,:,332 jn tjjig phenomenon, hydrogen diffuses into the steel to react with the carbon, which

is responsible for the strength of the material, to form methane which, on account of its higher molecular volume, cannot escape. The pressure that builds up causes cavity growth along the grain boundaries, transforming the steel from a ductile to a brittle state. This may finally lead to rupture of the affected pipe or vessel, usually without any significant prior deformation. This phenomenon was already recognized and in principle understood by Carl Bosch and his workers when they were developing the first commercial ammonia process'. Special alloy components which can react with the carbon to form stable carbides (for example, molybdenum, chromium, tungsten and others) enhance the resistance of steel against this sort of attack. The rate at which the material degrades depends on the pressure of the trapped methane, the creep rate of the material and its grain structure. The areas most prone to attack are those which have the greatest tendency to contain unstable carbides - for example, welding seams333. The type of carbides and their activity strongly depends on the quality of post-weld heat treatment (PWHT). There is already a risk of attack at quite moderate temperatures (around 20QQ and at hydrogen partial pressures as low as 7 bar. Extensive research work and careful investigation have made it possible largely to prevent hydrogen attack in modern ammonia plants by proper selection of hydrogen-tolerant alloys with the right content of metals which form stable carbides. Of fundamental significance in this respect was the work of Nelson334, who produced curves for the stability of various steels as a function of temperature and hydrogen partial pressure. An extensive survey, still valid today, was made by Class335-336. On account of newer experience gained in industrial applications, the original Nelson curve has been revised several times. So, for example, 0.25 and 0.5% moly steels are treated now with respect to their hydrogen resistance as ordinary non-alloyed carbon steels. A related phenomenon is physical hydrogen attack, which may happen alongside the problem just described. It occurs when adsorbed
60

molecular hydrogen dissociates at higher temperatures into atomic hydrogen, which can diffuse through the material structure. Anywhere that two hydrogen atoms recombint into a molecule in the material structure (at second phase particles or material defects like dislocations), an internal stress is set up within the material. This effect causes progressive deterioration of the material, reducing its toughness. Eventually the affected vessel will crack and, ultimately, rupture. The phenomenon is also referred to as hydrogen embrittlement. It is most likely to happen in welds with no proper PWHT. Holding a weld for a prolonged period at elevated temperature (an operation known as soaking) will allow the majority of any included hydrogen to diffuse out of the material. But this may not be efficient enough if there was any moisture present during the original welding operation, as when wet electrodes or hygroscopic fluxes have been used, because traces of atomic hydrogen are formed by thermal decomposition of water under the very intense welding conditions. Very critical in that respect are dissimilar welds, such as welds between ferritic and austenitic steels337, where the formation of martensite, which is sensitive to hydrogen attack, may increase the risk of brittle fracture. A problem specific to ammonia synthesis is nitriding, which occurs on the surface of steel in the presence of ammonia at temperatures above 300C338. With unalloyed steels, the nitride layer will grow with time to a thickness of several mifiimetres. Austenitic steel, used for the converter baskets, develops very thin but hard and brittle nitride layers. It has been reported339-340 that a few ammonia converters in which the pressure shell is in contact with ammonia-containing synthesis gas at 400C have developed circumferential cracks in the welded seam which ran round the entire circumference to a depth nearly approaching the wall thickness in parts. In one case the problem was discovered only when the converter developed a leak. This was a very surprising fact, because the operating conditions were well below the Nelson curve for the particular shell material used

AMMONIA

(2'/4Cr-lMo). An elaborate investigation341 came to the conclusion that hydrogen attack had occurred by a special mechanism at lower temperatures than predicted by the Nelson chart. Surface nitriding proceeds along microcracks in the weld and transforms carbides into nitrides or carbo-nitrides, yielding free active carbon, which will then be hydrided to methane. High residual welding stress and internal pressure are essential to promote propagation of the cracks. The propagation of the cracks is slow, since it depends on the diffusion of the nitrogen into the steel and on the transformation of the carbides. A rival theory340-MZ attributed the damage to physical hydrogen attack resulting from the use of agglomerated hygroscopic flux in combination with insufficient heat soaking. But that argument does not explain the fact that the severity of crack formation increases with higher ammonia concentration in the synthesis gas. However, no problems were experienced with the same converter design in other cases where non-hygroscopic (fused) flux was used and a more conservative pressure vessel code as adopted, leading to less stress on account of thicker walls. Metal dusting343"347 is a corrosion phenomenon which has come into the limelight in the last few years with the introduction of the new exchanger reformer technology and the operation of steam superheaters in the hot process gas stream downstream of the secondary reformer waste heat boiler. In both cases the gas has to pass a critical temperature range. Metal dusting occurs at 500-800C on iron-nickel- and cobaltbased alloys with gases containing carbon monoxide and with a carbon activity much greater than 1. As the name implies, the affected material disintegrates into fine metal and metal oxide particles mixed with carbon. The problem may manifest itself as pitting or general attack. A possible mechanism based on the work Grabke348 and Hochmann349 is hown in Fig. 66s*7. Extensive reearch and review of industrial experience reveals that the problem may be overcome by the use of materials like Inconel 601, 601H, 625 and similar alloys. Corrosion by hydrogen sulphide

Fig. 66: Mechanism of Metal Dusting

gas system

CO -t- H2 = C + H20
P 1
a

/-rt

.P

' LJ

= g-AG'/RT
'H 2 Q

penetration of the gas phase at defects in the oxide layer

for the actual gas composition

"2 = 10
H20

-P

I02 10' 10
carbon i steels

carfaurization of the matrix and formation of stable and unstable carbides

10-' IO2

coke deposition
austenitic steels

io-3
ICH

decomposition of the carbides to graphite and metal

io-s
10-'

CO + H2 ;? C + H2O

IO-5 ICH

IO'3

IO-2
P

10-'

10

10'

CO'K/ H20

disintegration of metal particles and graphite

in partial oxidation plants can be controlled by use of austenitic steel, but special care to ensure proper stress relief of welds is advisable to avoid stress corrosion cracking caused by traces of chloride sometimes present in the feed oil. Stress corrosion by liquid ammonia350'351 occurs at ambient temperature under pressure as well as in atmospheric storage tanks at -33C. Measures needed to avoid it include maintaining a certain water content in the ammonia and excluding even
Table XV Relative Investment Costs for Steam-Reforming Ammonia Plants (Standardized to 1,000 t/d)
Year Investment at time Investment in 1992 value 1967 1982 1992
100 100

traces of oxygen and carbon dioxide. Welds in pressurized vessels must be properly stress-relieved, and in atmospheric tanks it is important to select the right welding electrodes, the right plate thickness and the right weld geometry.

Modernization of older plants (revamping)


Production cost factors
So far we have reviewed how the various process sections have contributed to the reductions in total energy consumption that have been achieved over the years and what modern plants look like. Investment costs, which are the other great determinant of the economics in ammonia production, have only been considered in passing. The processes have had to become more sophisticated in order to achieve the energy savings dis-

27 90

63
102

61

MODERN PRODUCTION TECHNOLOGIES

cussed. Has this made the plants more expensive in terms of real costs, after inflation is allowed for? As Table XV demonstrates, the straightforward answer is: Not as much as one might expect. (Note, however, that the actual plants to which the figures in Table XV apply were not of the same capacity, and they have been standardized to 1,000 t/d using the 0.7 scaling exponent.) The reason for this moderate increase in real terms is due to the so-called "learning effect", which leads to more optimized mechanical designs (e.g accommodation of different exchanger functions in a single pressure shell, optimizing plot plan to minimize piping, etc.) and in the progress made in vessel and machinery fabrication techniques. For the other important cost factor - the natural gas price - the development was worse. As a consequence of the two oil crises the price increased by a factor of 5 over its 1967 value.

Pressure for revamping


Another crucial point is the market situation. During the last decade nitrogen fertilizer production has not been a very profitable business for most of the producers of the Western world, and some companies have left the business or at least thought about getting out. To some extent, the tendency of state-controlled companies to export at any price has contributed to the deterioration of the market. Additionally, some over-capacity has aggravated the situation and, on top of that, consumption in western industrialized nations has slumped on account of environmental concern and agricultural production cut-backs in response to surplus. As a consequence of this not very bright picture, only a few new ammonia plants have been built in the last few years. However, many producers have looked for possibilities to "revamp" or modernize their older, less efficient plants so that they can stay competitive. Most revamp projects could be combined with a moderate capacity increase because some of the original equipment was oversized and only specific bottlenecks needed to be eliminated, not entail-

ing excessive costs. As the market possibilities for a company do not generally increase in steps of 1,000 or 1,500 t/d but slowly and continuously, such a moderate capacity addition will involve less risk and will be more economical than building a whole new plant352. A recent example353 illustrates the economic advantage of such an approach. A steam reforming plant based on natural gas was revamped with an associated capacity increase from 1,000 t/d to 1,200 t/d, to reduce its energy consumption from 38.5 to 35.3 GJ/t NH3. The investment was $14 million. This corresponds to a specific investment of US $70,000 per t/d NH3 for the added capacity. In a new plant of 1,200 t/d, this value would be about $140,000 per t/d. The following main changes were involved in this project: shifting primary reformer load to the secondary reformer; lowering the steam/carbon ratio; adding a booster compressor to increase the capacity of the air compressor; installing a new converter for the synthesis loop. Quite a few additional items were altered, too.

1. Making maximum use of additional capacity available in oversized equipment. 2. Relieving the burden on overtaxed units by shifting process duties to oversized ones, reducing flow or changing process conditions. 3. As far as possible, avoiding replacement of existing equipment where improvements can be made by simple modifications. 4. Keeping the number of additionally required equipment items to a minimum352. One factor which can strongly influence the economic success of the whole revamp project is the downtime needed to incorporate the changes. Even in an old plant, which may be fully written off, the costs of lost production from down-time can be significant. For that reason, retrofits have to be planned carefully to fit the normal turn-around schedule. As far as possible, additional equipment items should be already premounted during operation so that they can just be tied in during the scheduled shut-down.

Strategy for revamping


The first step has to be a careful audit of the status quo of the plant. This should be done as a joint effort of the technical plant management and a competent ammonia contractor or a very capable consultant (with a long-term specific expertise in ammonia technology). In this undertaking it is important to tune the plant to its optimum condition to produce an actual baseline flowsheet, from which any proposed improvement can be measured. Next the objective or objectives must be carefully defined. Is it to be energy saving and/or a capacity increase and/or improvement of the plant's reliability (which also, of course, improves the production economics)? Then the optimization criteria have to be fixed: the return on investment required and the parameters needed, which are the anticipated prices for natural gas and other utilities352. That accomplished, the ammonia contractor (or consultant) should then prepare a detailed study using the following guidelines:
62

Revamp options
The state of the art has been described in considerable detail in the preceding parts of this article. Many features and approaches have been given which, properly combined by the skilled process engineer, will result in a low-energy ammonia plant consuming about 28 GJ/t NHj. Of course, one can argue and philosophize endlessly in a more general way on the various possibilities with their associated energy, capacity and cost effects and on how well they fit together in a retrofit. It would definitely be beyond the scope of this monograph to go through an exhaustive list of individual modification options354; indeed, it would in part be a repetition of facts and features already discussed. Long-standing readers of Nitrogen will be aware of earlier reviews on this subject207-226-355-356. Just a few special possibilities should be mentioned. In cases where the primary reformer is identified as the bottleneck, several possibilities can be considered to shift load off the primary reformer. Several of them entail increasing the duty on

AMMONIA

the secondary reformer and using a corresponding excess of process air over the stoichiometric requirement. The principle has been extensively explained earlier (pages 9,19-22). Art elegant variant of this principle is the BYAS process proposed a number of years ago by Jacobs226'352, in which a large portion of the natural gas is by-passed around the primary reformer and fed directly to the secondary reformer. This by-pass concept is also followed by FosterWheeler in the AM2 process263'307, in which the use of a large excess of air (up to 200%) allows the overall steam/carbon ratio to be brought down to 1.0-1.6. Another method is to perform a part of the primary reforming in a pre-reformer using low level heat116'118-120. Alternative measures to enlarge the reforming capacity are to make use of the pro-cess heat of the secondary reformer in an exchanger reformer such as Id's GHR357 (see page 19) or Kellogg's KRES96 (page 20) or to install a parallel CAR unit. Another option is to install a parallel autothermal reformer, as proposed by Lurgi and Tops0e. In the last few years an impressive number of former axial-flow ammonia synthesis converters have been retrofitted successfully to a radial configuration using the ACAR technology of Ammonia Casale358-359. Since 1986 Casale has to date revamped more than 90 ammonia converters with a total capacity exceeding 100,000 t/d, mostly by insitu modernization using the axialradial concept. In a few cases complete new converters were installed. A promising loop revamp option for the future will be the KAAP process of Kellogg180'182. Of the total world capacity, 68% is accounted for by plants more than 10 years old and more than 40% is in plants which are 15 years old and more352. This might suggest that there is a big potential for revamp projects in the future, even in plants that have already undergone past revamps.

Table XVI Ammonia Production Cost from Various Feedstocks in 1996 in North West Europe (1,800 t/d, new plant)
Feedstock Process Feedstock price Total energy consumption Feedstock & energy costs Other cash costs Total cash costs Capital-related costs Total cost Investment (excl. off-sites) S/million Btu million Btu/t $/t NH3 $/t NH3 $/t NH3 $/t NH3 $/mt NH3 million $ Natural gas Steam reforming
3.0 27 81 34.1 115.1 72.4 187.5 250

Vacuum residue Partial oxidation


2.1 36 75.6 45.5 121. 1 104.2 225.3 360

Coal Partial oxidation


1.9 45.5 86.5 68.2 154.6 173.7 328.3 600

Investment includes: contractor price for turnkey installation of plant and storage facility; owner costs for process definition, bid evaluation, project excution and supervision; a limited number of spare parts; but does not include off-sites site costs. For capital-related costs a debt/equity ratio of 60:40 is assumed. With 6% depreciation, 8% interest on debts and 16% ROI on equity, total capital-related charges are 1 7.2% on investment

using today's best technological standards for each process. From this it is obvious that at present there is no chance for the other feedstocks to compete against steam reforming of natural gas. Only under very special circumstances in co-operation with a refinery, for example - partial oxidation of heavy oil fractions might be economically justified. (It should be noted, however, that the average energy consumption of the steam reforming plants presently in operation is noticeably higher than the example of the modern low-energy concept used in Table XVI). The combined cost of feedstock and energy for a steam reforming plant - both are natural gas - is the principal determinant of the overall production costs. Even with the most modern concepts, gas costs represent (in Western Europe) more than 70% of cash costs and as much as 45% of total costs. For many West Euro-

pean plants with effective consumption of 33 million Btu/tonne NH3 and more and with historically lower investment costs, the percentage is even higher. The price of gas - and, by extension, the price of ammonia - is to a greater or lesser extent linked to the price of crude oil, which is well illustrated by the plot of crude and ammonia prices in the last decade given in Fig. 67. It is assumed that ammonia will continue to follow this trend in the future with rising oil prices. A crucial link between gas and oil pricing which has to be considered is the "the interfuel relationship". The market structure is being distorted by the need for cleaner and less polluting fuels resulting from increasing environmental awareness. In this respect, natural gas is so advantageous in relation to other fossil fuels that demand could well be pushed up in the coming years. The received opinion is that, al-

Fig. 67: World Ammonia and Energy Costs 1980-i 991


160
ammonia - f.o.b. US Gulf

40

140120crude oil-Arab light

-35 - 30 - 25 -20

100-

Economics of ammonia production


Table XVI gives an estimate for ammonia production costs in northwest Europe for different feedstocks

8060

- 15

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991

10

63

MODERN PRODUCTION TECHNOLOGIES

Percentage of World Natural Gas Consumption and Reserves Iupward pressure on prices is necesfor Various Geographical Areas in 1 994 sary to make that happen. The fore% of world reserves (l4MO"Nm3) Former Soviet Union (Russian Federation Arab Gulf North America Western Europe (inc. North Sea) South and Central America Africa Asia (inc. Australia + New Zealand)
39.7 34.1 32.0 6.3 3.8 3.8 6.8 7.1

Table XVII

though gas supply can be increased, cast is for higher production costs in Western Europe and the USA. In the low gas cost areas such as the Arab Gulf, Trinidad and Indonesia, competing usages for natural gas are not expected to grow to any great extent, and in such locations feedstock costs for the ammonia producers are expected to rise only moderately. A very good account of natural gas development was presented recently360. Table XVII361 shows the proven reserves of natural gas by location. The states of the former USSR have the biggest share, followed by the Middle East. Because it is not quite clear at present how gas will be valued internally in the Former Soviet Union (probably $1.6-2.3 per million Btu6-362), which is still a major ammonia exporter, this area not been included in Fig. 68, which compares production costs of various geographic areas. The investment costs used in Fig. 68 include the contractor price (depending on efficiency and location) for turnkey installation of plant and storage facility; owner costs for process definition, bid evaluation, project execution and supervision; a limited number of spare parts; incremental off-site costs necessary for the ammonia plant at a developed chemical or refinery industry site. For the capital-related costs a debt/equity ratio of 60:40 is assumed. With 6% depreciation, 8% interest on debts and 16% ROI on equity, total capital-related charges are 17.2% on investment. In the very long term, coal has prospects - at least, from the world reserves and consumption rate (shown in Table XVIII) one might expect it to. Even though investment and consumption figures are lower for partial oxidation plants designed to feed on heavy fuel oil, it is questionable whether there will be any growth in this area. With the objective of getting the most out of the barrel, the refiners wfll undoubtedly have plenty of heavy residues, but up to now, with the aid of various coking processes, they have always been able to get rid of them. However, environmental concern about the sulphur in the petroleum coke

% of world consumption (2 f 36- 10' Mm3)


27.1 18.4) 6.4 34 14.4 3.6 5.2

to.i

Table XVIII. World Reserves of Fossil Raw Materials 1994 361


Coal (tonnes) Reserves Consumption, per year Expected life time (years) 1043.10' 4438.10' 235 Mineral oil (tonnes) 137.10' Natural gas (Mm') 141.1012 2136.10' 66

3! 72.10'
43

Fig. 68: Ammonia Costs at Various Locations (capacity 600,000 t/a)

200 -

capital-related costs cash costs

i80 J

160 -

140 -

120 -

100 -

80 -

60 -

40 -

20 -

Arab Gulf Trinidad plane nat. gas, $/millionBtu feed + fuel, million Btu/t NH3 investment, $ million new 0.5 29 270 new 1.2 28 285

US Gulf new 2.0 28 260

US Gulf existing 2.0 33 -

NW Europe new 3.0 27 270

NW Europe existing 3.0 33

64

ould cause some problems in the future for that method of disposal. That could give a fillip to the partial oxidation route to synthesis gas, since any sulphur present in the feedstock is converted to hydrogen sulphide and is removed in the Rectisol scrubbing process and converted into the elemental form in a Claus unit. Nevertheless, when looking in Table XVIII at the world resources of fossil raw materials and their rate of exhaustion by present consumption, it is obvious that natural gas can continue as the preferred feedstock, at least into the medium term. Although natural gas offers the most economic route to ammonia, to stay competitive it is vital to continue with the effort to optimize mechanical design (to reduce investment costs) and to improve reliability (to boost on-stream factor and, therefore, to cut fixed costs), especially in highly industrialized areas, where natural gas is at its most expensive.

Environmental aspects of ammonia production


From the point of view of its overall environmental impact - air, water and ground pollution, noise generation, and materials and energy consumption - today's ammonia plant is actually rather clean technology, with low emissions, little in the way of unnecessary energy or materials consumption, and relatively few "cross-media" dilemmas, where action to improve one environmental effect worsens another. The principal concern at the moment is atmospheric emissions of acid-forming nitrogen oxides (NOx)363-365 These all originate from the combustion of fuel in primary reforming furnaces, fired heaters and auxiliary boilers; the ammonia process itself does not give rise to any nitrogen oxides. A fair and proper evaluation of the atmospheric nitrogen oxides contribution of an ammonia plant must take into account additional NOX contributions from the external generation of imported electric power and must equally allow an NOX credit for exported surplus power. There are two sources of nitrogen oxides in combustion gases: fuel

NOx arises from the oxidation of any chemically-bound nitrogen that may be present in components of the fuel, while thermal NOx is formed from elemental nitrogen present in the combustion air and, occasionally, as a minor constituent of the fuel gas. The formation of nitrogen oxides is mainly influenced by the peak flame temperature, the amount of available oxygen and the residence time in the combustion zone. Because of the low overall nitrogen content of most fuel gases, thermal NOx-formation from the nitrogen in the combustion air is the main source of NOX, but if ammoniacontaining purge and flash gases are fired as fuel, the nitrogen oxides level will rise considerably, as about 50% of their ammonia content will be converted to NOx in the combustion. Locally varying legislation either limits the total amount of NOx emitted (USA) or the NOX concentration of emitted exhaust gas, as in several EU member states. The present limit in Germany for new furnaces with a heat duty up to 300 MW is 200 mg NOx/m3, calculated as NO2, at 3% oxygen in the flue gas. (This value corresponds to 97.5 ppmv). In the USA, the latest amendment to the Clean Air Act classifies NOx sources into five categories and sets the limits on total emissions accordingly. The five categories are classified by geographical location and air quality. Each is also assigned a baseline emission threshold ranging from 100 t/a (marginal area) to 10 t/a (extreme area) and a "significance level" ranging from 10 t/a to 100 t/a for new emission sources. A 1,350-t/d (1,500 st/d) ammonia plant with 200 mg/m3 NOX concentration in the flue gas would release about 210 mt NO x /a and would thus be classified as a significant emitter anywhere in the United States. All modern low-energy steam reforming ammonia concepts have already contributed remarkably to NOx reduction by saving fuel, which cuts back the amount of flue gas. Modern plants use only about 20% of their total energy demand as fuel, whereas in older units this used to be 45% or more. One possible means of reducing NOX emission in those plants is to limit the combustion air preheat temperature to 250-300C.
65

But that reduces the thermal efficiency of the plant and would virtually rule out the use of a gas turbine drive on the air compressor, where the very hot exhaust is fed to the reformer as combustion air - a concept incorporated in some modern plants. Air preheat is one example of a cross-media effect, where improving the energy-efficiency increases emissions to the air. With appropriate measures, including so-called low-NOx burners, flue gas NOx levels of 150 mg/m 3 are a realistic practical performance objective for new plants today, even when the combustion air is at 350C. Scrubbing of purge and flash gases well below 100 ppm NH3 is an obvious and very necessary expedient when they are used as fuel. Developments in the design of natural gas burners have been aimed at reducing the peak flue gas temperature and restricting the concentration of free oxygen in the hottest part of the combustion zone. These low-NOx burners366-367 are based on staged air addition, staged fuel addition, or internal flue gas recirculation. It is believed that values of 100 mg/Nm3 could be realized. Figure 69 is an example of a stagedair burner. In principle, the amount of flue gas per tonne of ammonia could serve as a guideline for NOx generation because, when multiplied by the NOX concentration, it gives the total amount of NOX per tonne of ammonia. In this way, for example, plant concepts in which the primary reforming duty has been reduced by running the secondary reforming with an air excess could have an advantage over those with the classical high-duty reformer in combination with a stoichiometricallyoperated secondary reformer. But this picture is oversimplified, because the specific amount of flue gas may be very much increased when, for example, a gas turbine used for the process air compressor drive delivers its hot exhaust as preheated combustion air. Also, designs aimed at maximizing steam export will have a higher specific flue gas volume. The reformer configuration is also of some importance: in a top-fired reformer the individual burners have a higher thermal load than in a side-fired furnace because

MODERN PRODUCTION TECHNOLOGIES

Fig. 69: Staged-Air Low-NOx Burner


fuel combustion air

tertiary air

there are fewer of them, and this could lead to higher NOX concentrations. On the other hand, plants which use low-calorific waste gases to fuel their reformers may achieve much lower NOx values on account of the lower flame temperatures. So it is possible for local site-specific factors to far outweigh the foregoing considerations. In partial oxidation plants the main focus for attention on NOx emissions is the auxiliary boiler for power and steam generation. Frequently, also, there is the additional problem of sulphur dioxide emissions (not a problem in steam reforming of natural gas) which may range from <0.1 to 2 mg SO2/Nm3 or higher, depending on the sulphur content of the fuel oil used363. In principle it is possible to

achieve even lower NOx-vaIues by installation of a selective catalytic reduction (SCR) DeNOx system. A recent development, the Shell Lateral Flow Reactor, has an unusually wide operating temperature range (400C down to 150C). This allows flexibility in the positioning of the SCR catalyst in the flue gas duct of the reformer furnace according to the particular disposition of the heat recovery system. So far no economic data are available. To achieve very low NOx values, say below 50 mg NOx/Nm3, is not without problems, however, because at such a low concentration it becomes rather difficult to control the ammonia addition properly. Power plants, for example, usually do not attempt to reach the lowest possible NOx levels in their SCR installations so as to avoid ammonia emissions and am66

monium sulphate formation. A general requirement to install DeNOx units in European ammonia plants would seriously affect the competitiveness of domestic ammonia production against imports not produced under such restriction. Besides, it would really only be a cosmetic action because, alongside electric power generation and especially motor traffic, the ammonia industry is a trivial source of NOx emissions. Worldwide NOX emission from human activities was about 110 x 106 t/a in 1990 (30% North America, 20% Europe)368. With an ammonia production of around 120 x 106 t/a and assuming a rather high average value of 1.5 kg NOX per tonne NHy the contribution from ammonia would be only 0.16%. All of the carbon contained in the fossil raw material is converted to carbon dioxide. The carbon dioxide formed in the process gas is usually recovered as pure carbon dioxide and, in most cases, is used immediately for urea production and only rejected to the atmosphere in minor amounts. The carbon dioxide in the flue gas from the combustion of fuel is normally released into the atmosphere. Modern steam reforming plants based on natural gas produce 1.15 to 1.30 tonnes of carbon dioxide per tonne of ammonia from the process gas and release 0.5-0.35 t CO2/t NH, with the flue gas, depending on the degree of air reforming in the secondary reformer. Concepts with exchanger reformers have about the same total energy consumption and if all the energy comes from fossil material, they will consequently release the same amount of carbon dioxide as the traditional plants. They will only be able to claim a truly lower atmospheric carbon dioxide contribution if they import electricity generated from hydroelectric or nuclear power. Urea production requires 1.29 tonnes of carbon dioxide for every tonne of ammonia converted. In partial oxidation plants somewhere between 2 and 2.6 tonnes of carbon dioxide have to be removed from the process gas, depending on the C/H ratio of the feedstock. A substantial surplus therefore has either to be emitted or used for some other purpose besides urea production.

AMMONIA

In modem steam reforming plants there are virtually no polluting liquid effluents because nowadays the process condensate is completely recycled. It is first stripped with process steam and then, after polishing with ion-exchangers, it is reused as boiler feed water. It is also possible to recycle process condensate in partial oxidation plants, but there is additional waste water from the gasification section, which has to be purified to rather low levels of BOD5 and very minimal concentrations of Ni and V282. To harmonize European regulations concerning emission value permits for industrial production, the EU Commission has proposed a Council Directive Concerning Integrated Pollution and Prevention Control (IPPC Directive). A key element of the Directive is the concept of Best Available Techniques (BAT). As defined in article 2(11):
BAT means the most effective and advanced stage in the development of activities and their method of operation which indicate the practical suitability of particular techniques for providing the principal basis for emission limit values designed to prevent, and where that is not possible, generally to reduce emissions and the impact on the environment as a whole.

The IPPC Directive requests an information exchange between national environmental authorities, industry and non-governmental orjganizations with the aim to prepare ~'PC BAT Reference Documents (BREF), which could provide a basis for emission limit value regulations to be imposed by the national governments of the EU member states. Currently a BREF for ammonia is being prepared to serve as a pilot project for this exercise. In 1995 the European Fertilizer Manufacturers' Association (EFMA) published a booklet "Production of Ammonia" in its series "Best Available Techniques for Pollution Prevention and Control in the European Fertilizer Industry"364.

Future perspectives
To give a perspective for the future of the ammonia technology is not an easy task. An extrapolation of the

present state of the art and knowledge is of little help in foreseeing fundamental new developments. Only 15% of all ammonia consumed today is used for chemical and technical purposes; the other 85% goes into the manufacture of fertilizers. It thus obvious that the future of the ammonia industry and its technology is very closely bound up with future fertilizer needs and the pattern of world supply. In the Western World we are faced with a growing concern about nitrate pollution, and it is true that in some countries nitrogen in the ground water may be becoming a problem. This is sometimes used as an argument against the use of "inorganic" or synthetic fertilizers produced by industry and in favour of more "organic" or "natural" fertilizers (biomass). But it is a misconception to believe that "organic" nitrogen will solve this problem. Degradation of biomass and manure will also lead to nitrate run-off. The most effective solution to this problem is likely to lie in better management of the timing of the fertilizer applications, which requires a better understanding of the biological nitrogen cycle, rather than just cutting fertilizer nitrogen usage. Indeed, in the Developing Countries there is a very clear-cut need to increase fertilizer application to feed their growing populations, and this will definitely be much more important to their governments than the environmental debate currently exercising the collective minds of the Western countries. In any case, the supply of "organic" fertilizer is far short of what would be required to satisfy demand. So, for the foreseeable future, a widespread "bio-agriculture" using a low input of "inorganic" nitrogen is unlikely. Now, what are the options to produce nitrogen fertilizers by alternative routes not involving the severe temperatures and pressures that are inherent in the conventional ammonia process? The first option would be biological nitrogen fixation. Certain bacteria and blue-green algae are able to absorb atmospheric nitrogen, by themselves or in symbiosis with a host plant, and transform it into organic nitrogen compounds with the ammonium ion as
67

an intermediate. A well known example is the symbiotic relationship between legumes and Rhizobium bacteria, which settle in their root nodules. The enzyme nitrogenase practically performs an ammonia synthesis. Energy for this process is supplied by oxidation of photosynthesis products from the host plant. For synthesis of the nitrogenase in the cell, the so-called NIF gene is responsible. Genetic engineering research on biological nitrogen fixation is pursuing various routes: (a) modifying Rhizobium to broaden its spectrum of host plants; (b) transferring the NIF gene to other bacteria which already have a broader spectrum of host plants but have no natural nitrogen-fixing ability; (c) engineering soil bacteria to assimilate nitrogen to ammonia and to release it to the soil; and (d) inserting the NIF gene directly into plants369. One important point to consider, especially with the last option of constructing a nitrogen-fixing plant, is the energy balance of the nitrogen fixation process. Because of die low efficiency, a considerable amount of the photosynthesis product would be consumed to supply the energy needed. This would consequently lead to a reduction in yield, which is estimated by some researchers to be as high as 18%37D. It is difficult to hazard a guess on how long it might take to realize any of the possibilities mentioned, and even experts can make completely wrong statements in their very own field. Lord Kelvin, President of the Royal Society from 1890 to 1895, was a master of that peculiar art, categorically declaring at various times that radio has no future, X-rays would prove to be a hoax, and heavier-than-air flying machines were impossible! (And on the last subject there is another famous example: "Man will not fly for 50 years" - Wilbur Wright, 1901190.) But the general opinion is that it might take another 15-20 years. In addition, there is still a big question about the political, social and psychological acceptance of the use of genetic engineering to such an extent.

MODERN PRODUCTION TECHNOLOGIES

The possibility of converting atmospheric nitrogen into ammonia in homogeneous solution using metalorganic complexes was first raised around 1966370'3n. The prospects for this route are not judged to be very promising in terms of energy consumption and also with respect to the cost of these very sophisticated catalyst systems. Photochemical methods of producing ammonia at ambient temperature and atmospheric pressure in the presence of a catalyst have been reported, but the yields are far too low to be economically attractive. So, for the foreseeable future, we have to rely on the conventional ammonia synthesis reaction, combining hydrogen and nitrogen using a catalyst at elevated temperature and pressure in a recycle process, as conceived in the laboratory by Fritz Haber and made operable on an industrial scale by C. Bosch, and on the use of the known routes to produce the hydrogen and nitrogen needed (which are in fact, what consumes all the energy).

In agreement with many ammonia experts one can sum up the prospects by the following broad predictions: Natural gas will remain the preferred feedstock for at least the next 10-15 years, which can be assumed from the world energy supply and demand balance shown in Table XVIQ (see page 64). Coal gasification will not play a major role in ammonia production in that period. The present ammonia technology will not change fundamentally, at least in the next 10-15 years. Even if there are radical, unforeseeable developments, they will take time to develop up to commercial introduction. With the traditional concepts, the margins of additional improvements have become rather small after years of intensive research and development. Thus minor improvements of individual steps, catalysts and equipment might be expected.

There is unlikely to be any further significant reduction in the energy consumption of the natural gas-based steam reforming ammonia process; figures between 27 and 28 GJ/t NH, are already dose to the theoretical minimum, which is only 23% lower. In the next 10-15 years the bulk of ammonia production will still be produced in world-scale plants of 1,000-2,000 t/d NH3. Small-capacity plants will be limited to locations where special logistical, financial or feedstock conditions favour them. New developments in ammonia technology will mainly reduce investment costs and increase operational reliability. Smaller integrated process units (e.g. exchanger reformer, CAR) contribute to this reduction and give additional savings by simplifying piping and instrumentation. Reliability may be improved by advances in catalyst and equipment quality and by improved instrumentation and computer control.

68

Вам также может понравиться