The Haber-Bosch Heritage:The Ammonia Production Technology

Haber Bosch Mittasch

Max Appl

50th Anniversary of the IFA Technical ConferenceSeptember 25 26th 1997, Sevilla, Spain

Introduction

Based on the fundamental research work of FritzHaber, Carl Bosch and his engineering team devel-oped the ammonia synthesis to technical operabilityusing the promoted iron-based catalyst found byAlwin Mittasch and co-workers. Since then there hasbeen no fundamental change in the synthesis reactionitself. Even today every plant has the same basic con-figuration as this first plant. A hydrogen-nitrogen mix-ture reacts on the iron catalyst (todays formula dif-fers little from the original) at elevated temperaturein the range 400 500 C (originally up to 600 C),operating pressures above 100 bar, and the uncon-verted part of the synthesis gas is recirculated afterremoval of the ammonia formed and supplementedwith fresh synthesis gas to compensate for the amountof nitrogen and hydrogen converted to ammonia.

3H2 + N2 2NH3 (1)H0298 = 92.4 kJ/mole D F0298 = 32.8 kJ/mole

Of course, progress made in mechanical and chemi-cal engineering and increased theoretical knowledgehave led to improvements in efficiency, converterdesign and energy recovery in the synthesis section,but really dramatic changes happened over the yearsin the technology of synthesis gas generation.

As the synthesis is the very heart of every ammoniaproduction and is also from an historical point of viewthe most interesting section, it is probably appropri-ate to start our review with this section.

The synthesis

The ammonia equilibrium and therecycle conceptThe reaction proceeds with a reduction in volume andis also exothermic, so the equilibrium concentrationsof ammonia are higher at high pressure and low tem-perature, but at the turn of the last century a quan-titative knowledge of chemical equilibrium was notavailable, and this might explain why early experi-ments aimed at the ammonia synthesis were unsuc-cessful. A famous victim of the lack of thermodynamicdata was Wilhelm Ostwald. He offered in 1900 BASFa process in which nitrogen and hydrogen were passedover heated iron wire at atmospheric pressure, claim-ing several percent of ammonia, a concentration

which was far beyond equilibrium. BASF found thereason for his erroneous data and irony of history he withdraw his patent application, not knowinghow important that application could have been laterwhen indeed iron became the basis of the commer-cial ammonia synthesis catalyst.

First systematic measurements were made by Haberin 1904/05 but they yielded too high figures as a con-sequence of problems with exact analysis of the lowconcentrations values attained at atmospheric pres-sure and 1000 C using iron for catalysis. As this fig-ures did not comply with the Heat Theorem, W.Nernst made own measurements at 75 bar, which wereactually the first experiments at elevated pressure.From the results he concluded that a technical pro-cess, which he probably anticipated as a once-throughprocess, should not be feasible as the much higherpressures needed in this case seemed to be beyond thetechnical possibilities of the time. Haber continuedwith his investigations now also including pressureexperiments.

From the more reliable equilibrium data now avail-able it was obvious that at normal pressure the reac-tion temperature should be kept well below 300 Cin order to obtain even a small percentage of ammo-nia. For this temperature level no catalyst was

2

Figure 1: Equilibrium conversion and space time yieldfor NH3 and SO3 production

available. By increasing the pressure for example to75 bar the equilibrium conditions were improved, butwith catalysts active at 600 C only low ammonia con-centrations were attained. So Haber concluded thatmuch higher pressures had to be employed and that,perhaps more importantly, a recycle process had tobe used, an actually new process concept at that time,and thus he overcame his collegues preoccupationwhich resulted from the unfavorable equilibrium con-centrations and the concept of a once-through process.

The amount of ammonia formed in a single pass ofthe synthesis gas over the catalyst is indeed much toosmall to be of interest for an economic production.Haber therefore recycled the unconverted synthesisgas: after separating the ammonia formed by conden-sation under synthesis pressure and supplementing itwith fresh synthesis gas to make up for the portionwhich was converted to ammonia, the gas was recir-culated by means of a circulation compressor to thecatalyst containing reactor.

Habers recycle idea changed the static conception ofprocess engineering in favor of a more dynamicapproach. For the first time reaction kinetics wereconsidered as well as the thermodynamics of thesystem. In addition to the chemical equilibrium Haberrecognized that for the technical realization reactionrate was a determining factor. Instead of simple yieldin a once-through process he concentrated on spacetime yield. Figure 1 illustrates this consideration ofequilibrium concentration in combination with spacetime yield by a comparison of the ammonia synthe-sis with the SO2 oxidation process.

Also anticipated by him was the preheat of the syn-thesis gas to reaction by heat exchange with the hoteffluent gas from the reactor. In 1908 Haberapproached the BASF to find support for his workand to discuss the possibilities for the realisation ofa technical process. Early in 1909 he discovered infinely distributed osmium a catalyst which yielded8 Vol% of ammonia at 175 bar and 600 C. A success-ful demonstration in April 1909 of a small labscaleammonia plant convinced the representatives ofBASF and the companys board decided to pursue thetechnical development of this process with all avail-able resources.

In BASF then Carl Bosch, entrusted with extraordi-nary authority, became project leader and succeededtogether with a team of dedicated and very able

co-workers to develop in an unprecendented efforta commercial process in less than five years. The pro-duction facilities for 30 t/d were erected on a new sitenear the village Oppau (now a part of the city of Lud-wigshafen), the first production was in September1913 and full capacity was reached in 1914.

The ammonia catalyst

In BASF Alwin Mittasch was responsible for the cat-alyst search. Osmium, used by Haber showed excel-lent catalytic activity but was difficult to handle, themain disadvantage, however, was that the worldsstock of this rare material was only a few kilograms.Mittasch started a systematic screening program, covering nearly all elements of the periodic table.Until 1910 more that 2500 different formulas weretested in 6500 runs. For these experiments specialsmall test reactors containing easily removable car-tridges holding about 2 g of catalyst were developed.

In November 1909 a sample of magnetite from a placein Sweden showed exceptionally good yields, whichwas surprising because other magnetite types weretotal failures. Mittasch concluded that certain impur-ities in this Gallivara magnetite were important for

3

Figure 2: Ammonia equilibrium and catalyst volume

its good performance. So he investigated the influenceof various individual additives, which in todays ter-minology are called promoters. By 1911 the catalystproblem had been solved. Iron with a few percent alu-mina and a pinch of potassium yielded a catalyst withacceptable reproducibility and performance and tol-erable lifetime.

But the research program was continued until 1922to be certain about the optimum composition. Theonly additional result was that the further addition ofcalcium gave a certain improvement. All magnetitebased catalysts on the market today have a similarcomposition to that of the original BASF catalyst.Also the catalyst preparation remained practically thesame: Melting natural magnetite from Sweden withaddition of the various promoters, cooling the melt,breaking the solidified melt into small particles fol-lowed by screening to obtain a fraction with suitableparticle size.

From these early days until today an enormousamount of academic research was dedicated to elu-cidate the mechanism of the synthesis, to study themicrostructure of the catalyst and to explain the effectof the promoters. Besides the scientific interest therewas of course some hope to find an improved cata-lyst, which could operate at far lower temperaturesand thus at lower pressures saving compressionenergy, which is in a modern plant still 300 kWh/tNH3. In principle one can operate with the classicmagnetite catalyst at 35 45 bar in the temperaturerange of 350 to 450 C, but needing a trainload of cat-alyst about 450 m3 (1300 t) for a plant of 1350 t/dNH3 to achieve very low ammonia concentrationswhich would require removal by water-scrubbinginstead of condensation by refrigeration. M. W. Kel-logg proposed such a process in the early 1980s, butdidnt succeed with commercialization. For a real lowpressure catalyst operating at front end pressure toneed no compression, an operation temperature wellbelow 300 C would be required. To illustrate this sit-uation figure 2 shows ammonia equilibrium and cat-alyst volume.

With the modern spectroscopic tools of Surface Sci-ence rather detailed information on the reactionmechanism at the catalyst surface was obtained.Kinetics of nitrogen and hydrogen adsorption anddesorption were investigated and adsorbed interme-diate species could be identified. The results allowedto explain, for the most part, the mechanism of ammo-

nia synthesis in the pressure range of industrial inter-est. This success has many fathers, outstanding con-tributions were made by Brill, Ertl, Somorjai, Bou-dard, Nielsen, Scholtze, Schlgl and many others. Therate determining step is the dissociative adsorptionof the nitrogen at the catalyst surface and the mostactive sites are the crystal faces (111) and (211), whichis probably caused that these are the only surfaceswhich expose C7 sites, which means iron atoms withseven nearest neighbors.

The primary function of the Al2O3 is to prevent sin-tering by acting as a spacer between the small ironplatelets and it may in part also contribute to stabi-lize the Fe(111) facets. The promoting effect of thepotassium is probably based on two factors. One is thelowering of the activation energy of the dissociativeadsorption of nitrogen by an electronic charge trans-fer effect from potassium to iron which increases thenitrogen bond strength to the iron and weakens thenitrogen-nitrogen bond. The other factor consists inreducing the adsorption energy of ammonia thus eas-ing the desorption of the formed ammonia whichavoids blocking the surface and hindering the nitro-gen adsorption.

Commonly the term ammonia catalyst is used for theoxidic form consisting of magnetite and the promot-ers. Actually this is only the catalyst precursor, whichis transformed into the active catalyst consisting of a -iron and the promoters by reduction with synthesisgas. In the 1980s pre-reduced ammonia catalysts foundacceptance in the market as they avoid the relativelylong in-situ reduction which causes additional down-time and considerable feedstock consumption with-out production. These catalysts are reduced at thevendors facilities and subsequently passivated at tem-peratures around 100 C using nitrogen with a smallamount of air.

A notable improvement of the magnetite system wasthe introduction of cobalt as an additional componentby ICI in 1984. The cobalt enhanced formula was firstused in an ammonia plant in Canada using ICICatalcos AMV process with a synthesis pressure of90 bar. With similar kinetic characteristics, the vol-umetric activity is about two times higher than thatof the standard iron catalyst.

In October 1990 Kellogg commercialized the KelloggAdvanced Ammonia Process using a catalyst com-posed of ruthenium on a graphite support, which is

4

claimed to be 10 20 times as active as the traditionaliron catalyst. According to original patents asignedto BP, the new catalyst is prepared by subliming ruthe-nium-carbonyl Ru3(CO)12 onto the carbon-contain-ing support which is impregnated with rubidiumnitrate. The catalyst has a considerably higher surfacethan the conventional catalyst and, according to thepatent example, it should contain 5% Ru and 10%Rb by weight. This catalyst works best at a lower thanstoichiometric H/N ratio of the feed gas and it is alsoless susceptible to selfinhibition by NH3 and has anexcellent low pressure activity.

The potential of ruthenium to displace iron in newplants will depend on whether the benefits of its useare sufficient to compensate the higher costs. In com-mon with the iron catalyst it will also be poisoned byoxygen compounds. Even with some further poten-tial improvements it seems unlikely to reach an activ-ity level which is sufficiently high at low temperatureto allow an operation of the ammonia synthesis loopat the pressure level of the syngas generation.

The ammonia converter and thesynthesis loop configuration

With the catalyst at hand, the next step was to con-struct somewhat larger test reactors for catalystcharges of about 1 kg. Surprisingly, these reactors rup-tured after only 80 hours. Further studies showed thatthe internal surface had totally lost its tensile strength.This phenomenon had apparently propagated fromthe inner surface outward until the residual unaffectedmaterial was so thin that rupture occurred.

With the aid of microscopic investigations by thin sec-tion technique Bosch found the explanation. Decar-bonization of the carbon steel had occurred, but, sur-prisingly, the result was not soft iron but rather a hardand embrittled material. Hydrogen diffusing into thesteel caused decarbonization by methane formation.This methane, entrapped under high pressure withinthe structure of the material, led to crack formationon the grain boundaries which finally resulted inembrittlement. Systematic laboratory investigationsand material tests demonstrated that all carbon steelswill be attacked by hydrogen at high temperatures andthat the destruction is just a matter of time.

Boschs unconventional solution to the embrittlementproblem was to use a carbon steel pressure shell witha soft iron liner. To prevent the hydrogen which hadpenetrated this liner from attacking the pressure shell,measures had to be taken to release it safely to nor-mal pressure. This was achieved by providing smallchannels on the outer side of the liner which was intight contact with the inner wall of the pressure shelland by drilling small holes, later known as Bosch-Holes, through the pressure shell, through whichhydrogen could escape to the atmosphere. Theseholes had no effect on the strength of the shell andthe resulting losses of hydrogen were negligible. Fig-ure 3 gives a sketch of such a pilot plant converter.

Bosch did not content himself with his liner/hole con-cept but looked further for alternative solutions forthe embrittlement problem. He intitiated in the late1920s research in the steel industry to develop steelsresistant to hydrogen under pressure. Special alloycomponents as for example molybdenum, chromiumtungsten and others form stable carbides and enhancethe resistance of steel against this sort of attack con-siderably. This problem and the related physicalhydrogen attack is not restricted to the synthesis buthas to be considered carefully also in the synthesis gasproduction section because of the temperatures and

5

Figure 3: First pilot plant converter with soft ironlining and external heating

hydrogen partial pressures involved there. Extensiveresearch and careful evaluation of operation experi-ences have made it possible to prevent largely hydro-gen attack in modern ammonia plants by proper selec-tion of hydrogen-tolerant alloys with the right con-tent of metals which form stable carbides. Offundamental significance in this respect was the workof Nelson, who produced curves for the stability ofvarious steels as a function of operation temperatureand hydrogen partial pressure.

In the small reactors heat losses predominated andcontinuous direct external heating by gas was neces-sary and this led to deterioration of the pressure shellsafter short operation times even without hydrogenattack. With increasing converter dimensions in thecommercial plant heating was only necessary for startup.

Bosch developed an internal heating by the so-calledinversed flame, introducing at the top of the reactora small amount of air, igniting with an electricallyheated wire. Later this was replaced by an electricresistance heater. Subsequently introduced flushingwith nitrogen as shown in figure 4 and later with cold

synthesis gas kept the pressure vessel walls cool andrendered the liner-hole concept redundant.

Subsequent reactor designs in the technical plantincluded internal heat exchangers and later the cat-alyst was placed in separate tubes which were cooledby the feed gas. Another improvement was the intro-duction of an externally insulated catalyst basket.Because of the low concentrations aqueous ammoniawas separated from the loop by water scrubbing. Con-verters with catalyst tubes had a better temperaturecontrol and this led together with an increased pres-sure to higher ammonia concentrations which nowallowed from 1926 onwards the direct production ofliquid ammonia. In 1942 the first quench converterwas installed and this design gradually has replacedthen the converters with the catalyst tubes.

Soon after the first world war development startedalso in other countries, partly on basis of BASFs pio-neering work. Luigi Casale built 1920 the first plantin Italy, and based on developments by M. G. Claudethe first French plant started to produce in 1922. Boththe Casale and the Claude process operated underextreme high pressure. In contrast to this Uhde con-structed a plant based on coke oven gas, operatingunder extreme low pressure. (Mont Cenis process).Futher developments were by G. Fauser who workedtogether with Montecatini. During the 1920s severalplants were built in the USA, some based on Euro-pean some on American Technology. The successfulUS company was Nitrogen Engineering Corporation(NEC), the predecessor of Chemico.

Mechanical design was now already rather advancedbut for the process design of converter and loop sofar empirical data in form of charts were used as nosuitable mathematical expressions for the reactionkinetics were at hand. When better experimental datafor the reaction kinetics and other process variablesbecame available in the 1940s and 1950s lay-out ofconverters received a better quantitative chemicalengineering basis. Figure 5 shows reaction rate ofammonia formation and equilibrium. When the tem-perature is increased (under otherwise constant con-ditions), the reaction rate increases to a maximum, todecrease with further temperature increase andbecomes zero when reaching equilibrium tempera-ture. Joining these points will result in a line giving,for each NH3 concentration, the temperature for themaximum rate. This curve runs about parallel to the

6

Figure 4: Converter with pressure shell cooling bynitrogen

equilibrium line and at a about 30 50 C lower tem-perature. To maintain the maximum ammonia forma-tion rate, the reaction temperature must decrease asthe ammonia concentration increases.

For optimal catalyst usage the reactor temperatureprofile (after a initial adiabatic heating zone in the

first part of the catalyst) should follow this ideal line.For a long time converters were always compared tothis ideal for optimum use of high-pressure vesselvolume. Today the objective is rather to maximizeheat recovery (at the highest possible level) and tominimize investment costs for the total synthesis loop.In any case it is necessary to remove the heat of reac-tion as the conversion proceeds to keep the temper-ature at an optimal level. For the removal of the reaction two principal configurations are possible:

Tubular converters have cooling tubes within the cat-alyst bed through which the cooling medium, usuallycooler feed gas, flows co-currently or counter-cur-rently to the gas flow in the catalyst bed. Alternativelythe catalyst can be placed within tubes with the cool-ing medium flowing on the outside. The tube cooledconverters dominated until the early fifties, but arelargely outdated today. Well known examples werethe TVA converter (counter-current) and theNEC/Chemico design (co-current, with best approx-imation to the maximum rate curve). An interestingrevival of this principle is the ICI tube cooled con-verter used in the LCA process and also for metha-nol production.

In the multi-bed converters the catalyst volume isdivided into several beds in which the reaction pro-ceeds adiabatically. Between the individual catalystlayers heat is removed either by injection of coldersynthesis gas (quench converters) or by indirect cool-ing with synthesis gas or via boiler feed water heat-ing or steam raising (indirectly cooled multi-bed con-verter).

7

Figure 5 : Reaction rate of ammonia formation

Figure 6: Quench converter

In the quench converters only a fraction of the recy-cle gas enters the first catalyst layer at about 400 C.The catalyst volume of the bed is chosen so that thegas will leave it at around 500 C. Before entering thenext catalyst bed, the gas temperature is ,,quenchedby injection of cooler (125 200 C) recycle gas. Thesame thing is done at subsequent beds. In this way thereaction profile describes a zig-zag path around themaximum reaction rate line. A schematic drawing ofa quench converter together with its tempera-ture/location and temperature/ammonia concentra-tion profile is presented in figure 6. The catalyst bedsmay be separated by grids designed as mixing devicesfor main gas flow and quenchgas (cold shot), or be justdefined by the location of cold gas injection tubes asfor example in the ICI lozenge converter.

A disadvantage is that not all of the recycle gas willpass over the whole catalyst volume with the conse-quence that a considerable amount of the ammoniaformation occurs at higher ammonia concentrationand therefore at reduced reaction rate. This meansthat a larger catalyst volume will be needed comparedto an indirect cooled multi-bed converter. On theother hand, no extra space is required for inter-bedheat exchangers, so that the total volume will remainabout the same as for the indirect cooled variant.

As the quench concept was well suited for large capac-ity converters it had a triumphant success in the earlygeneration of large single stream ammonia plants con-structed in the 1960s and 1970s. Mechanical simplic-ity and very good temperature control contributed tothe widespread acceptance.

Multibed converters with indirect cooling. In convert-ers of this category the cooling between the individ-ual beds is effected by indirect heat exchange with acooling medium, which may be cooler synthesis gasand/or boiler feed water warming and steam raising.The heat exchanger may be installed together with thecatalyst beds inside one single pressure shell but anattractive alternative, too, preferentially for largecapacities, is to accommodate the individual catalystbeds in separate vessels and have separate heatexchangers. This approach is especially chosen whenusing the reaction heat for raising high pressure steam.The indirect cooling principle is applied today inalmost all large new ammonia plants, and also inrevamps an increasing number of quench convertersare modified to the indirect cooling mode.

Axial flow through the catalyst in the converters asexclusively used until the early 1970s face a generalproblem: With increasing capacity the depth of thecatalyst beds will increase, as for technical and eco-nomical reasons it is not possible to enlarge the pres-sure vessel diameter above a certain size. In order to compensate for the increasing pressure drop axial flow converters with usual space velocities of 10 15000 h-1 have to use relatively large catalyst par-ticles and a particle size of 6 10 mm has become stan-dard. But this grain size has compared to finer cata-lyst a considerably lower activity, which decreasesapproximately in a linerar inverse relation. Two fac-tors are responsible for the lower activity of the largerparticles. Firstly, the larger grain size retards onaccount of the longer pores the diffusion from theinterior to the bulk gas stream and this will inhibit thedissociative nitrogen adsorption and by this the reac-tion rate. Secondly, the reduction of an individual cat-alyst particle starts from the outside and proceeds tothe interior. The water formed by removing the oxy-gen from the iron oxide in the interior of the grainswill pass over already reduced catalysts on its way tothe outer surface of the particle. This induces somerecrystallization leading to the lower activity. Theeffect is considerable: going from a partide size of 1mm to one of 8 mm, the inner surface will decreasefrom 1116 to 3 8 m2/g.

8

Figure 7: Topsoe Series 200 indirect cooled converter(radial flow)

Haldor Topses company solvedthe dilemma with the pressuredrop and small catalyst particleswith a radial flow pattern, usinga grain size of 1,5 3 mm (Fig-ure 7). M.W. Kellogg choseanother approach with its hori-zontal crossflow converter (Fig-ure 8). The catalyst beds arearranged side by side in a car-tridge which can be removed forcatalyst loading and unloading

through a full-bore closure of the horizontal pressureshell.

Today each new world-size ammonia plant employsthe indirect cooling concept raising high pressuresteam up to 125 bar. Generally after the first bed aninlet-outlet heat-exchanger is placed and after the sec-ond or further beds the reaction heat is used to raisehigh pressure steam.

Brown and Root (formerly C. F. Braun) or Uhde (Figure 9) accommodate the cat-alyst in several vessels. Figure 9 is a simplified flowsheet of Uhdes synthesis loop. Actually the conceptof separate vessels for the catalyst beds, with heatexchange after the first and waste heat boiler after the

9

Figure 8: Indirect Cooled Horizontal Converter of M. W. Kellogg

Figure 9: Uhdes synthesis loop with two pressure vessels and three catalyst beds

second (nowadays they use also a third one followedby a boiler, too) was already introduced by C. F. Braunat time when most plants still used quench convert-ers.

The Ammonia Casale ACAR Converter has a mixedflow pattern. In each catalyst layer the gas flowsthrough the top zone predominantly axially but tra-verses the lower part in radial direction. This simpli-fies the design by avoiding special sealing of the topend of the bed to prevent by-passing.

Today computerized mathematical models are usedfor converter and loop lay-out. In principle, thesemodels use two differential equations whichdescribe the steady state behavior of the reaction inthe converter. The first gives a concentration-locationrelationship within the catalyst bed for the reactantsand the ammonia. It reflects the reaction kineticexpression. The second models the temperature-posi-tion relationship for the synthesis gas, catalyst andvessel internals. The form of this equation is specificto the type of the converter.

The kinetics of the intrinsic reaction, that means thereaction on the catalyst surface without any masstransport restrictions, are derived from measurements

on very fine catalyst particles. The first useful expres-sions for engineering purposes to describe the reac-tion rate was the Temkin-Pyshew equation, proposedin 1940. It was widely applied, but today there areimproved versions and other equations available.Additional terms are included to model the influenceof oxygen-containing impurities on the reaction rate.Although oxygen-containing compounds may beregarded as a temporary poison, severe exposure foran extended period of time leads to permanent dam-age. For practical application these equations have tobe modified to make allowance for transport phenom-ena (heat and mass transfer), and this is done by so-called pore effectiveness factors.

10

Figure 10: Simplified flow sheet of a coke-based ammonia plant

Synthesis Gas Preparation

The classical route based on coke

The pilot plant experiments at BASF for the ammo-nia synthesis were based on hydrogen from the chlo-rine-alkali electrolysis. When the capacity of this gassource was exhausted, water gas served as an indepen-dent hydrogen feedstock using the cryogenic Linde-Frnkl process for the separation. In this process car-bon monoxide is condensed out of the water gas at 200 C and 25 bar. Nitrogen was provided by an airseparation unit and nitrogen was also used in an indi-rect liquid nitrogen circulation system in the cryogenichydrogen separation. The residual content of 1.5 %CO in the hydrogen was removed by conversion tosodium formate in a gas scrubber operated with a l0%sodium hydroxide solution and at 230 C and 200 bar.

The initial operation of the commercial plant commis-sioned in September 1913 was based on hydrogen andnitrogen produced by this cryogenic separation, butafter a few months on line, it became apparent thatthe Linde refrigeration process was not reliable andeconomic enough for the production on large scale.A new catalytic process, the shift conversion, wasintroduced. In this reaction, found by W. Wild inBASF already in 1912, the gas is passed together witha surplus of steam over an iron oxide/ chromium oxidecatalyst at about 350 to 450 C. The carbon monox-ide reacts with water to form hydrogen and carbondioxide. The use of the shift reaction permitted a greatsimplification of the synthesis gas preparation. Insteadof using the refrigeration processes, producer gas (amixture of 60% nitrogen and 40% carbon monoxide)was generated by reacting air with red hot coke andmixed with the parallel generated water gas suppliedby the alternating air blowing and steaming processand this mixture was converted in the shift reactionto yield a gas consisting of hydrogen, nitrogen, car-bon dioxide and a small amount of residual carbonmonoxide. The carbon dioxide could then be removedsatisfactorily by water scrubbing at 25 bar. Theremoval of the residual carbon monoxide by scrub-bing with hot caustic soda solution with formation ofsodium formate used in the initial cryogenic route wascorrosive and troublesome. It could now be replacedby copper liquor scrubbing. Water gas productionfrom lignite started in 1926 in Leuna using a processdeveloped by Winkler. This process, in which coal isgasified continuously with oxygen and steam in a

fluidized bed, was a spin-off of the research work onthe removal of sulfur from ammonia synthesis gas.Figure 10 is a simplified flow sheet of a coke basedHaber-Bosch plant as it was operated in the 1930s and1940s at BASF and elsewhere. In the 1950s BASFdeveloped and introduced continuously operatedwater gas generators using oxygen or oxygen enrichedair from which the slag could withdrawn in liquidform.

A new age with hydrocarbonsThe plants continued to be based on coal for synthe-sis gas generation until the 1950s. With growing avail-ability of cheap hydrocarbon feedstocks and novel costsaving gasification processes a new age dawned in theammonia industry. The development started in theUSA where steam reforming was introduced, a pro-cess, originally developed in the 1930s by BASF andgreatly improved by ICI which extended it also tonaphtha. Before natural gas became available in largequantities in Europe, too, partial oxidation of heavyoil fractions was used in several plants, with processtechnology developed by Texaco (1940) and Shell(1950). After several oil crisis coal gasificationresearch and development was resumed with theresult that for this route a few technically proven pro-cesses are available today.

The chemical reaction of water, oxygen, air or anycombination of these reactants with fossil feedstocksis generally described as gasification. In a simplifiedway it can be viewed as the reduction of water bymeans of carbon and carbon monoxide. It yields a gasmixture made up of carbon monoxide and hydrogenin various proportions along with carbon dioxide and,where air is introduced, some nitrogen.

[CHx] + H2O CO + H2 + x/2H2 D H > 0 (2)[CHx] + 1/2 O2 CO + x/2H2 D H < 0 (3)

Reaction (2) is endothermic and needs an externalsource of energy supply, whereas reaction (3) is exo-thermic and can be carried out adiabatically. For theinitial carbon dioxide content in the raw gas from thegasification the shift reaction equilibrium is respon-sible which at the high temperature is rather on theCO side.

CO + H2O CO2 + H2 D H0298 = 41,2 kJ/mol (4)

11

This shift reaction, in which actually CO reduceswater to yield additional hydrogen, is favored by lowtemperature and is therefore purposely made to pro-ceed on a catalyst in a separate step at a temperaturelower than the preceding gas generation step.

With coke the reaction (2) corresponds to the non-catalytic classic water gas process. With light hydro-carbons reaction (2) is called steam reforming and ismade to proceed over a nickel catalyst. The reaction(3) is commonly called partial oxidation and in prin-ciple applicable for any fossil feedstock, from coal tonatural gas. As can be seen from the stoichiometricequation, the hydrogen contributed by the feedstockitself increases with its hydrogen content, whichranges from a minimum of [CH0.1] in coke to a maximum of CH4 in methane.

Syngas preparation via steamreformingThe steam reforming process is restricted to lighthydrocarbons ranging from natural gas (methane) tolight naphtha. For higher hydrocarbons, such as fueloil or vacuum residue this technology is not applicableon account of impurities as sulfur and heavy metalswhich would poison the sensitive nickel catalyst. Inaddition cracking reactions are more likely to occuron the catalyst, depositing carbon which might blockthe catalysts pores and also restrict the gas flow. Asthe nickel catalysts are highly sensitive to sulfur com-pounds, these catalysts poisons have to be removedprior to the reforming reaction. For this purpose anyorganic sulfur compounds contained in the hydrocar-bon feedstock are first hydrogenated on a cobalt-molybdenum catalyst to hydrocarbon and hydrogensulfide, which is then absorbed with zinc oxide to formzinc sulfide.

RSH + H2 fi H2S + RH (5)H2S +ZnO fi ZnS + H2O (6)

For ammonia production the steam reforming is per-formed in two steps: First the hydrocarbon /steammixture is passed through high-alloyed nickel-chro-mium tubes filled with a catalyst containing finely dis-persed nickel on a carrier. The heat needed for theendothermic reaction is supplied by gas burners in afurnace box. The reaction in this primary reformer iscontrolled to achieve only a partial conversion ofaround 65% , leaving about 14% methane (dry basis)content in the effluent gas at a temperature of 750 to

800 C. The gas is then introduced into the so-calledsecondary reformer a refractory lined vessel alsowith a nickel catalyst where it is mixed with a con-trolled amount of air introduced through a burner.This raises the temperature sufficiently to completethe reforming of the residual methane adiabatically.It also introduces the right amount of nitrogen toachieve the correct stoichiometric ratio in the finalsynthesis gas. The overall reaction in the secondaryreformer may be described as some sort of a partialoxidation, but the stoichiometric equation (7) doesnot give a clue to the actual reactions taking place.

2CH4 + O2 (+4N2) 2CO + 4H2 (+4N2) D H0298 = 71,4 kJ/mol

(7)

The gas leaves the secondary reformer at 950 to1000 C and a methane content of 0,3 to 1.5 %. It iscooled down to 350 400 C using the removed heatfor high pressure steam generation. In the first steamreforming based plants the shift conversion used onlythe classical chromium-iron catalyst achievingaround 2 % residual CO. For CO2 removal in thisearly plants the traditional water scrubbing wasapplied and the final purification was still performedby copper liquor. In the early 1960s copper-zinc-alu-mina catalysts became available for a second conver-sion step at temperatures of about 200 C, wherebythe residual CO concentration could be lowered to0.2 0,3 %. This allowed to eliminate the copper liq-uor scrubbing, removing the residual concentrationsof CO and CO2 by methanation. In this highly exo-thermic reaction which is performed at about 300 Con a nickel catalyst, hydrogen reacts with carbonmonoxide to methane and water; it is the reverse ofthe steam reforming reaction of methane (equation8 and 9).

CO + 3H2 CH4 + H2O D H0298 = 206.3 kJ/mol (8)CO2 + 4H2 CH4 + 2H2O D H0298 = 165,1 kJ/mol(9)

With aqueous monoethanolamine (MEA) a new sol-vent for CO2 removal was introduced in 1943. Thisprocess has been used extensively in many ammoniaplants until hot potash and other solvents with lowerheat requirement were developed. The plants withcapacities up to 300 t/d used reciprocating compres-sors for compression.

As natural gas is usually delivered under elevated pres-sure and because the reforming reaction entails an

12

increase in total volume, significant savings of com-pression energy are possible if the process is performedunder higher pressure. But there is also a disadvan-tage in raising the pressure level of reforming as theequilibrium is shifted to lower conversions, which canbe compensated by higher temperatures. As all theheat in the primary reformer has to be transferredthrough the tube wall, the wall temperatures will riseand approach the material limits. Originally HK 40tubes with a content of 20% nickel and 25% chro-mium were commonly used. With new grades as HPmodified with higher nickel content andstabilized with niobium and the recentlyintroduced Micro Alloys which addition-ally contain titanium and zirkoniumhigher wall temperatures and thushigher pressures up to 44 bar in the pri-mary reformer have become possible.The steam surplus applied in thereformer could thus also be reduced froma steam to carbon ratio of 4 and higher toabout 3 or slightly below, and this wasassisted by improved catalysts withenhanced activity and better heat trans-fer characteristic. For naphtha reforminga higher steam surplus is necessary.

Fancy catalyst shapes as wagon wheels, six-shooters,shamrock or four-hole have replaced the old Raschigrings. The stability of the standard catalyst supportsas calcium aluminate, magesium aluminate and a -alumina has been improved and it has become awidely accepted pratice to install in the first third ofthe catalyst tube where the bulk of the reforming reac-tion takes place, a potassium promoted catalyst whichwas developed by ICI originally for naphtha steamreforming in order to prevent carbon deposition bycracking reactions. From the various primary reformerdesigns the top fired concept with a single radiationbox dominates in the larger plants, the side-fireddesign in which only 2 tube rows can be accommo-dated in the radiation box, allows only a linear exten-sion and additional fire boxes connected to a commonflue gas duct. The secondary reformers have beenoptimized regarding hydrodynamics and burnerdesign using computational fluid dynamics. Figure 11shows an example of a top-fired reforming furnacetogether with the secondary reformer.

The reduction of the steam-to-carbon ratio was abigger problem for the HT shift than in the reformerstep, as the gas mixture became a higher oxidativepotential and tended to over-reduce the iron-oxidefrom magnetite to FeO and in extreme cases partiallyto metallic iron. Under these conditions the Boudu-ard reaction will become significant and carbon accu-mulation in the catalyst particles leads to breaking.In addition the Fischer-Tropsch reaction leads to theformation of methane and higher hydrocarbons. Cop-per promotions of the iron catalyst suppresses theseside reaction. The nasty problem of methanol andamine formation in the LT shift is largely solved by

13

Figure 11: Top-fired primary reformer and secondaryreformer (Uhde design)

Figure 12: CO2 Loading characteristics of various solvents

improved formulations of the copper/zinc/alumina,and a new development is the intermediate temper-ature shift catalyst, operated quasi isothermal in atubular reactor, for example in the ICI LCA ammo-nia process or the Linde ammonia process (LAC).

Large progress in the CO2 removal systems was madein the last decade. The original MEA systems had aheat consumption for solvent regeneration over 200kJ/kmol, a corrosion inhibitor system called amineguard III brought it down to about 120 kJ/kmol, butthis is still nearly 5 times as high as the most advancedsystem, the BASF aMDEA Process, which uses anaqueous solution of monomethyl-diethanolaminetogether with a special promotor which enhances themass transfer. Other low energy systems are the Benfield LoHeat Process, which is a hot potash systemor the Selexol Process, which uses a mixture of gly-col dimethylethers, a pure physical solvent. In phys-ical solvents, a prominent example was water in theold plants, the solubility of the CO2 is according toHenrys law direct proportional to the CO2 partialpressure and regeneration can be achieved by flash-ing, without application of heat.

In contrast to this the MEA is a chemical solvent, thesolubility is only slightly dependent on the CO2partial pressure and approaches a saturation value.MEA forms a stable salt with the carbon dioxide anda high amount of heat is required in the stripper todecompose it. BASFs aMDEA Process is about inbetween, the characteristic can be adjusted in a flex-ible way by the concentration of the activator, so thatthe major part of the dissolved carbon dioxide can bereleased by simple flashing and only a smaller propor-tion has to be stripped out by heat. Figure 12 showsCO2 loading characteristics of various solvents.

The tubular steam reformer has become a very reliableapparatus and the former problems with tube and trans-fer line failures and catalyst difficulties are largely his-tory. But the tubular furnace and its associated convec-tion bank is a rather expensive item and contributes sub-stantially to the investment cost of the total ammoniaplant. So in some modern concepts the size was reducedby shifting some of the load to the secondary reformernecessitating an overstoichiometric amount of processair. The surplus of nitrogen introduced in this way, canbe removed downstream by the use of a cryogenic unit.C.F. Braun was the first contractor which introducedthis concept in the so-called Purifier Process. Some con-tractors have gone so far to by-pass some of the natu-

ral gas around the tubular reformer and feeding itdirectly to the secondary reformer which likewise needssurplus of process air or oxygen enriched air.

But there are additional reasons for breaking away fur-ther from the fired furnace concept. The temperaturelevel of the flue gas from a traditional reformer is usu-ally higher than 1000 C and the process gas at the out-let of the secondary reformer is also around 1000 C.It is thus from a thermodynamic point of view waste-ful to use this high temperature level simply to raise andsuperheat high pressure steam. The boiling point of HPsteam is only 325 C and the first heat exchanger in theflue gas duct preheats process air in the conservativeplants to only 500 C (600 700 C in more moderninstallations). Recycling high-level heat from the sec-ondary reformer and making use of it for the primaryreforming reaction is thermodynamically the betteroption. Concepts which use this heat in an exchangerreformer have been successfully developed and com-mercially demonstrated. The first to come out with thisconcept in a real production plant was ICI with its GHR(Gas Heated Reformer). The hot process gas from thesecondary reformer is the sole heat source. A surplusof process air of around 50% is needed in the secon-

14

Figure 13: ICI Gas-Heated Reformer

dary reformer to achieve a closed heat balance. Figure13 is a simplfied drawing of the ICI Gas-HeatedReformer.

Quite recently ICI has come out with a modifieddesign, the AGHR, with the A standing foradvanced. The bayonet tubes are replaced by normaltubes attached to a bottom tube sheet using a specialpacking which allows some expansion. Thus the del-icate double tubesheet is now eliminated.

In the Kellogg Exchanger Reformer System, abbre-viated KRES, the gas flow pattern is different. Thetubes are open at the lower end and the reformed gasmixes with the hotter effluent of the secondaryreformer. The mixed gas stream flows up-ward on theshell side to heat the reformer tubes. Thus primaryreforming and secondary reforming reaction proceedin parallel in contrast to the ICI concept where thetwo reactions proceed in series. The Kellogg processuses enriched air. The complete elimination of thefired tubular furnace leads to a drastic reduction ofNOx emission, because there is only flue gas frommuch smaller fired heaters required for feed and process air preheat. An even more progressiveexchanger reformer presently operating in a demo-plant is Uhdes CAR (Combined autothermalreformer) which not only replaces the catalytic sec-ondary reforming step by a non catalytic partial oxy-

dation step but also combines this with the exchangerreformer in one single vessel.

Syngas from heavy oil fractions viapartial oxidation

In partial oxidation heavy oil fractions react accord-ing to equation (2) with an amount of oxygen insuf-ficient for total combustion . The reaction is non-cat-alytic and proceeds in an empty vessel lined with alu-mina refractory. The reactants, oil and oxygen, alongwith a minor amount of steam, are introduced througha nozzle at the top of the generator vessel. The noz-zle consists of concentric pipes so that the reactantsare fed separately and react only after mixing at theburner tip in the space below. The temperature in thegenerator is between 1200 and 1400 C. Owing to theinsufficient mixing with oxygen, about 2% of the totalhydrocarbon feed is transformed into soot, which isremoved by water scrubbing. The separation of thesoot from the water and its further treatment differsin the Shell and the Texaco Process the two commer-cially available partial oxidation concepts. The gas-ification pressure can be as high as 80 bar.

After gas cooling by further waste heat recovery, thehydrogen sulfide formed during gasification isremoved along with carbon dioxide by scrubbing withchilled methanol below 30 C in the Rectisol pro-

15

Figure 14: Ammonia syngas by partial oxidation of heavy hydrocarbons (Texaco)

cess. Then, as in the steam reforming route, the gasundergoes the CO shift reaction. Because of thehigher carbon monoxide content much more reactionheat is produced, which makes it necessary to distrib-ute the catalyst on several beds with intermediatecooling. The carbon dioxide formed in the shift con-version is removed in a second stage of the Rectisolunit; both have a common methanol regenerationsystem. The H2S-rich carbon dioxide fraction from thefirst stage of the regenerator is fed to a Claus plant,where elemental sulfur is produced. In the final pur-ification, the gas is washed with liquid nitrogen, whichabsorbs the residual carbon monoxide, methane anda portion of the argon (which was introduced into theprocess in the oxygen feed). The conditions in thisstage are set so that the stoichiometric nitrogenrequirement is allowed to evaporate into the gasstream from the liquid nitrogen wash. The processneeds, of course, an air separation plant to produceoxygen, usually around 98.5% pure, and to supply theliquid nitrogen. Figure 14 is a simplified flowsheet ofsynthesis gas preparation by partial oxidation of heavyfuel oil using the Texaco Syngas Generation Process.The Shell process uses of a waste heat boiler for rawgas cooling whereas Texaco prefers for ammoniaplants a water quench for this purpose which has theadvantage that this intro-duces the steam for thesubsequent shift conver-sion which differentfrom Shell is performedwithout prior removal ofthe sulfur compoundsusing a sulfur tolerantshift catalyst.

Besides some optimiza-tions there are no funda-mental new develop-ments in the individualprocess steps. Some pro-posed changes in the pro-cess sequence, for exam-ple methanation insteadof liquid nitrogen wash,or the use of air insteadof pure oxygen are not realized so far. Thoughother CO2 removalsystems as Selexol orPurisol (N-Methylpyrrol-idon ) and alternative

sulfur recovery processes are suitable too, Rectisoland Claus Process remain the preferred options.

Synthesis gas by coal gasificationThere is no chance for a wide-spread use of coal as feed-stock for ammonia in the near future, but a few remarksshould be made regarding the present status of coal gas-ification technology. Proven gasification processes arethe Texaco Process, the Koppers-Totzek Process, andthe Lurgi Coal Gasification. The Shell gasification, notyet in use for ammonia production , but successfullyapplied for other productions is an option , too. Texacosconcept is very similar to its partial oxydation processfor heavy fuel oil feeding a 70% coal-water paste intothe generator. Koppers-Totzek is an entrained flow con-cept , too, but feeding coal dust. In the Lurgi process,the coarse grounded coal is gasified in a moving bedat comparably low temperature using higher quantitiesof steam as the others. Shells process differs consid-erably from its oil gasification process in flow patternand feeds coal dust. Texaco, Lurgi, and Shell operateunder pressure, whereas the Koppers-Totzek gasifieris under atmospheric pressure, but a pressure version,called PRENFLOW is presently tested in a demo-plant. Continuous slag removal either in solid or mol-

16

Figure 15: Ammonia plant temperature profile

ten form is, indeed, the fundamental technical problemwith coal-based systems and the technical solutions dif-fer considerably. Gas cooling is achieved by quench andor waste heat boiler, entrained coal dust is removed bywater scrubbing. The following process steps for shiftconversion, CO2 removal and final purification arelargely the same as in partial oxdiation of heavy fueloil.

Energy integration and ammonia plant concept

The integrated steam reformingammonia plant

In the old days an ammonia plant was more or less justa combination with respect to mass flow and energymanagement was handled within the separate processsections, which were often sited separately, as theyusually consisted of several parallel units. A revolu-tionary break-through came in the mid of the 1960swith the steam reforming ammonia plants. The newimpulses came more from the engineering and con-tractor companies than from the ammonia plantindustry itself. Engineering contractors have beenworking since the thirties in the oil refining sector. Thegrowing oil demand stimulated the development ofmachinery, vessel and pipe fabrication, instrumenta-tion and energy utilization leading to single-train unitsof considerable size.

By applying the experience gained in this field it waspossible to create within a few years in the mid 1960sthe modern large-scale ammonia concept. To use asingle-train for large capacities (no parallel lines) andto be as far as possible energetically self-sufficient (noenergy import) through a high degree of energy inte-gration (with process steps with surplus supplyingthose with deficit) was the design philosophy for thenew steam reforming ammonia plants pioneered byM. W. Kellogg and some others. It certainly had alsoa revolutionary effect on the economics of ammoniaproduction, making possible an immense growth inworld capacity in the subsequent years. The basic

17

Table 1: Main energy sources and sinks in the steam reforming ammonia Process

Process section Originating Contribution

Reforming Primary reforming duty DemandFlue gas SurplusProcess gas Surplus

Shift conversion Heat of reaction SurplusCO2 removal Heat for solvent regeneration DemandMethanation Heat of reaction SurplusSynthesis Heat of reaction SurplusMachinery Drivers DemandUnavoidable loss Stack and general DemandBalance (Auxiliary boiler or import) Deficit

(Export) Surplus

reaction sequence has not changed since then. Figure15 shows the process sections and the relevant gastemperature levels in a steam reforming ammoniaplant.

High-level surplus energy is available from the fluegas and the process gas streams of various sections,while there is a need for heat in other places such asthe process steam for the reforming reaction and inthe solvent regenerator of the carbon dioxide removalunit (Table 1). Because a considerable amount ofmechanical energy is needed to drive compressors,pumps and fans, it seemed most appropriate to usesteam turbine drives, since plenty of steam could begenerated from waste heat. As the temperature levelwas high enough to raise HP steam of 100 bar, it waspossible to use the process steam first to generatemechanical energy in a turbine to drive the synthe-sis gas compressor before extracting it at the pressurelevel of the primary reforming section.

The earlier plants were in deficit, and they needed anauxiliary boiler, which was integrated in the flue gasduct. This situation was partially caused by inadequatewaste heat recovery and low efficiency in some of theenergy consumers. Typically, the furnace flue gas wasdischarged up the stack at unnecessarily high temper-atures because there was no combustion air pre-heatand too much heat was rejected from the synthesisloop, while the efficiency of the mechanical drivers

was low and the heat demand in the carbon dioxideremoval unit regenerator was high.

A very important feature of this new concept was theuse of a centrifugal compressor for synthesis gas com-pression and loop recycle. One advantage of the cen-trifugal compressors is that they can handle very largevolumes which allows also for the compression dutiesa single line approach. The lower energetic efficiencycompared to the reciprocating compressors of whichin the past several had to be used in parallel is morethan compensated by the lower investment and theeasy energy integration. In the first and also the sec-ond generation of plants built to this concept, max-imum use was made of direct steam turbine drives notonly for the major machines such as synthesis gas, airand refrigeration compressors but even for relativelysmall pumps and fans. The outcome was a rather com-plex steam system and one may be tempted todescribe an ammonia plant as a sophisticated powerstation making ammonia as a by-product. The plantsproduce more steam than ammonia, even today, themost modern plants still produce about three timesas much. In recent years electrical drives have swungback into favor for the smaller machines.

In most modern plants total energy demand(feed/fuel/power) has been drastically reduced. Onthe demand side important savings have beenachieved in the carbon dioxide removal section byswitching from old, heat-thirsty processes like MEA

18

Figure 16: Simplifiedflow sheet of amodern steam reforming ammoniaplant (C.F. BraunPurifier Process)

scrubbing to low-energy processes like the newer ver-sions of the Benfield process or aMDEA. Fuel is saved by air preheat and feed by hydrogen recoveryfrom the purge gas of the synloop by cryogenic, mem-brane or pressure swing adsorption technology. In thesynthesis loop the mechanical energy needed for feedcompression, refrigeration and recycle has beenreduced, and throughout the process catalyst volumesand geometry have been optimized for maximumactivity and minimum pressure drop.

On the supply side, available energy has beenincreased by greater heat recovery, and the combinedeffect of that and the savings on the demand side havepushed the energy balance into surplus. Because thereis no longer an auxiliary boiler, there is nothing in theplant that can be turned down to bring the energy sit-uation into perfect balance; therefore the overall sav-ings have not, in fact, translated into an actual reduc-tion in gross energy input to the plant (in the form ofnatural gas); they can only be realized by exportingsteam or power, and it is only the net energy consump-tion that has been reduced. But under favorable cir-cumstances this situation can be used in a very advan-tageous way. If there is a substantial outlet on the sitefor export steam, it can be very economic (depend-ing on the price of natural gas and the value assignedto steam) to increase the steam export deliberatelyby using additional fuel, because the net energy con-sumption of the plant is simultaneously reduced).

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 feedstockrequirement is stoichiometric. So the only way is to cutthe firing in the reforming furnace by shifting reform-ing duty to the secondary reformer, as we had alreadydiscussed earlier or to choose a more radical aproachby the use of an exchanger reformer instead of thefired furnace: ICIs Gas-Heated Reformer (GHR)system, the KRES of M. W. Kellogg and the TandemReformer (now marketed by Brown & Root), or theeven more advanced Combined AutothermalReformer (CAR) of Uhde. But none of these designsnecessarily achieves any significant improvement overthe net energy consumption of the most advanced con-ventional concepts under the best conditions.

For the cases in which export of steam and/or poweris welcome there is the very elegant possibility of inte-grating a gas turbine into the process to drive the aircompressor. The hot exhaust of 500 550 C containswell enough oxygen to serve as preheated combustionair for firing the primary reformer. The gas turbinedoes not even have to be particularly efficient,because any heat left in the exhaust gas down to theflue gas temperature level of 150 C is used in the fur-nace. Thus an overall efficiency of about 90 % can beachieved.

19

Boiler makers provide today largely reliable designsfor high-duty waste heat boilers after secondaryreformer and in the synthesis loop, in which up to 1.5t steam/t NH3 are produced, corresponding roughlyto a recovery of 90% of the reaction enthalpy of thesynthesis. Centrifugal compressors have become muchmore reliable, though their efficiency has notincreased spectacularly in recent years. Someimprovements were made in turn-down capability inimproving the surge characteristic. New developmentsare dry seals instead of oil seals and another poten-tial improvement, already successfully introduced innon-ammonia applications, is the magnetic bearing.

Although the introduction of the single-train inte-grated large plant concept in the 1960s revolutionizedthe energy-economics of ammonia production, it issurprising that since then the total consumption hasbeen reduced by about 30%, from roughly 40 to 28GJ/t. An example of a modern plant shows Figure 16.

From this enormous reduction in energy consumptionthe question may come up, what is the theoretical min-imum energy consumption for ammonia productionvia steam reforming of natural gas. Based on puremethane, we may formulate the following stoichio-metric equation:

CH4 + 0.3035 O2 + 1.131 N2 + 1. 393 H2O fiCO2 + 2.262 NH3

(10)

D H0298 = 86 kJ/mol; D F0298 = 101 kJ/mol

So from a mere thermodynamic point of view, in anideal engine or fuel cell heat and power should beobtained from this reaction. But because there is ahigh degree of irreversibility in the real process a con-siderable amount of energy is necessary to producethe ammonia from methane, air and water. The stoi-chiometric quantity of methane derived from the fore-going equation is 583 Nm3 per mt NH3, which corre-sponds to 20.9 GJ (LHV) per tonne of ammonia, whichwith some reason could be taken as minimum value.Of course, if one assumes full recovery of the reactionheat, then the minimum would be the heating value ofammonia, which is 18.6 GJ (LHV) per mt NH3.

Energy and exergy anal-ysis (First and SecondLaw of Thermodynamicsrespectively) identify theprocess steps in whichthe biggest losses occur.The biggest energy loss isin the turbines and com-pressors, whereas theexergy loss is greatest inthe reforming section,almost 70%. Based onexergy the thermody-namic efficiency for theammonia productionbased on steam reform-ing of natural gas isalmost 70%.

It has become rather common to measure modernammonia concepts above all by their energy consump-tion. Yet these comparisons need some caution ininterpretation; without a precise knowledge of designbases, physical state of the produced ammonia andstate of the utilities used, e. g. cooling water temper-ature, nitrogen content in natural gas, or conversionfactors used for evaluating imported or exportedsteam and power, misleading conclusions may bedrawn. In many cases, too, the degree of accuracy ofsuch figures is overestimated.

The best energy consumption values for ammoniaplants using steam reforming of natural gas are around28 GJ/tNH3. Industrial figures reported for plants withhigh-duty primary reforming and stoichometric pro-cess air and for those with reduced primary reform-ing and excess air show practical no difference.

20

Figure 17: Flow diagram of ICIs LCA Ammonia Process (Core unit) for 450 mtpd

New steam reforming ammonia process configurations

The ICI Leading Concept Ammonia Process LCA a radical break-away from the philosophy of thehighly integrated large plant which has been so suc-cessful for more than 25 years had its industrialdebut in 1988 at ICIs own location in Severside, Eng-land. The process consists of a core unit with all theessential process steps (Figure 17) and a separate util-ity unit which comprises utility boiler and electric gen-erator, CO2 recovery, cooling water system, demiwater and boiler feed water conditioning and ammo-nia refrigeration. Feed gas is purified in a hydro-desulfurization oper-ating at lower than usual temperatures and passes asaturator to supply a part of the process steam, thebalance is injected as steam. Heated in an inlet/out-let exchanger to 425 C the mixed feed enters the ICIGas Heated Reformer (GHR) at 41 bar, passing to thesecondary reformer at 715 C. The shell side entrancetemperature of the GHR (secondary reformer exit)is 970 C falling to 540 C at the exit of the GHR.Methane levels exit GHR and secondary reformer are25 % and 0.67% respectively (dry basis). Overallsteam to carbon ratio is 2.5 to 2.7. The gas, cooleddown to 265 C in the inlet/outlet exchanger, entersa single stage shift conversion, using a special copper-zinc-alumina based catalyst operating in quasi-isother-mal fashion in a reactor with cooling tubes, circulat-ing hot water, whereby the absorbed heat is used forthe feed gas saturation as described above. CO2removal and further purification is effected by a PSASystem, followed by methanation and drying. The syn-thesis operates at 82 bar in a proprietary tubular con-verter loaded with a cobalt enhanced formula of theclassical iron catalyst. Purge gas is recycled to the PSAunit and pure CO2 is recovered from the PSA wastegas by an aMDEA wash. Very little steam is gener-ated in the synloop, and from waste gases and somenatural gas in an utility boiler in the utility section (60bar) and all drivers are electric. The original inten-tion was to design a small capacity ammonia plantwhich can compete with modern large capacity plantsin consumption and specific investment, and toachieve with lower energy integration a higher flex-ibility for start up and reduced load operation, need-ing a minimal staffing. The basic plant features (GHR,isothermal shift and synthesis) can principally beapplied for larger capacities, too. The flow sheetenergy consumption is 29.3 GJ/t NH3.

In the context of the LCA process some discussion onthe economics of scale came up. Within the same sortof process configuration specific investment will bereduced by increasing capacity, at least to a pointwhere limitations for equipment size and transportmight play a role and specific investment would thenincrease again after having reached a minimum. Inany case for the traditional modern steam reformingammonia plant, a capacity of 2000 t/d is not beyondthe optimum. On the other hand it cannot be excludedthat concepts as the LCA with no elaborate steamsystem and a modular and prefabrication constructionmay come close to the specific investment of worldsize plants, but with regard to the other fixed costs,e. g. staffing, some question marks remain.

Kellogg has combined the ruthenium catalyst basedsynthesis loop (KAAP) with its exchanger reformersystem ( KRES) to an optimized integrated ammo-nia plant concept (Ammonia 2000) intended for theuse in world-scale single-train plants in the 1850 t/drange. Desulfurized gas is mixed with steam and thensplit into two streams in approximate proportion 2 : 1.These streams are separately heated in a fired heater.The smaller of the two enters the exchanger reformerat 550 550 C, while the remainder is passed directlyto the autothermal reformer at 600 640 C. Theexchanger reformer and the autothermal reformer useconventional nickel-based primary and secondaryreforming catalysts respectively. To satisfy the stoi-chiometry and the heat balance, the autothermalreformer is fed with enriched air (30% O2). Therequired heat for the endothermic reaction in thetubes of the exchanger reformer comes from the gaseson the shell side, comprising a mixture of the efflu-ent from the autothermal reformer and the gas emerg-ing from the tubes. The shell side gas leaves the ves-sel with 40 bar. The synthesis proceeds at about 90 barin a 4-bed radial-flow converter (hot wall design) withinter-bed exchangers. The first bed is charged withconventional iron-based catalyst for bulk conversionand the other beds with Kelloggs high activity ruthe-nium-based catalyst, allowing to attain an exit ammo-nia concentration in excess of 20%. The other pro-cess steps are more along the traditional lines. Theoverall energy claimed for this process can be as 1owas 27.2 GJ/t NH3.

Another recently launched process is the LindeAmmonia Concept (LAC) which consists essentiallyof a hydrogen plant with only a PSA unit to purify thesynthesis gas, a standard cryogenic nitrogen unit and

21

an ammonia synthesis loop. The concept is similar toKTIs PARC process for small capacities. The firstproject with a capacity of 1350 t/d is presently exe-cuted in India. The single isothermal shift conversionuses Lindes spiral-wound reactor, which has been suc-cessfully used for methanol plants and hydrogenationin ten plants around the world. In the loop a Casalethree bed converter with two interbed exchangers isused. As in ICIs LCA process, pure carbon dioxidecan be recovered by scrubbing the off gas from thePSA unit, for which Linde also uses the BASFaMDEA process. The process consumes about28.5 GJ/t NH3, or, with inclusion of pure CO2 recov-ery 29.3 GJ/t NH3.

The status of ammonia plants basedon heavy fuel oil and coal

For lack of economic incentive, not much optimiza-tion and development work has been dedicated in thelast few years to the field of partial oxidation of higherhydrocarbon fractions. The gasification of these plantsusually does not consist of a single line. Compared to

a steam reformer furnace there are more productioninteruptions because of periodic burner changes andcleaning operations in the gasification units. For thisreason most installations have a standby unit. In addi-tion to that the maximum capacity of single gas gene-rator corresponds only to 1000-1100 t/d of ammonia.Therefore world size ammonia plants have 3 4 par-tial oxidation generators. Generally the degree ofenergy integration is lower than in the steam reform-ing process because, in the absence of a large fired fur-nace, there is no large amount of hot flue gas and con-sequently less waste heat is available. So in this pro-cess route a separate auxiliary boiler is usuallynecessary to provide steam for mechanical energy andpower generation. Nevertheless, in modern conceptssome efforts have been made to bring the energy con-sumption down. Whereas older plant concepts hadvalues of around 38 GJ/t NH3, for a concept with thetraditional use of 98.5%+ oxygen quite recently a fig-ure of 33.5 GJ/t NH3 was claimed in a commercial bid.

To reduce investment cost and energy consumptionit has been recommended to use air or enriched air

22

Table 2: Comparison of ammonia production 1940 1990

Multi-line plant Modern Single TrainBASF 1940 1991

Feedstock Coke Natural gasCapacity t/d 800 1800Plot size m3 35 000 18 000Steel t 30 000 13 000Personal 1800 100Investment (1990) Mio DM 1000 300Energy consumption GJ/t NH3 88 28

Table 3: Ammonia production cost from various feedstock in 1996 in NW Europe (1800 t/d, new plant)

Feedstock Natural gas Vacuum residue CoalProcess Steam Partial Partial

Reforming Oxidation Oxidation

Feedstock price DM/GJ 4.3 3.0 2.7Total energy consumption GJ/t NH3 28.5 38 48.5Feedstock & energy costs DM/t NH3 123 114 131Other cash costs DM/t NH 50 65 100Total cash costs DM/t NH3 173 179 231Capital-related costs DM/t NH3 100 143 260Total cost DM/t NH3 273 322 491Investment Mio DM 350 500 900

For capital-related costs a debt/equity ratio of 60 : 40 is assumed. With 6 % depreciation, 8 % interest on debts and16 % ROI on equity, total capital-related charges are 17.2 % on investment.

instead of pure oxygen. Topse proposed the use ofenriched air (43 %) and methanation instead of liq-uid nitrogen wash. For CO2/H2S removal Selexol isapplied. The shift reaction proceeds over a sulfur-resistant catalyst in a three-bed configuration, bring-ing the residual carbon monoxide content down to0.55%. For the loop a Series 200 converter is chosen.The partial oxidation step can be designed accordingto either the Texaco or the Shell process. An overallconsumption of 34.8 GJ/mt NH3 is stated.

Foster Wheeler suggests the use of highly preheatedair in a Texaco generator operating at 70 bar. The gaspurification train comprises soot scrubbing followedby shift conversion, acid gas removal and methana-tion. The gas is dried by molecular sieves and finallyfed to a cryogenic unit to remove the surplus nitro-gen and residual methane, argon and carbon monox-ide traces. The rejected nitrogen is expanded in a tur-bine, which helps to drive the air compressor. A spe-cial design consideration was the following:Conventional air separation uses fractional distilla-tion of oxygen and nitrogen at a difference in boilingpoints of only 13 C. In the cryogenic unit of the Fos-ter-Wheeler process a lesser quantity of nitrogen (because the stoichoimetrically needed proportionremains in the gas) is separated from hydrogen at amuch higher boiling point differential (57 C). Thisshould save capital investment and energy consump-tion against the traditional approach. A figure of35.6 37.6 GJ/mt NH3 is given for heavy oil feedstock.

For coal based plants the economic incentive forextensive R&D is even lower than with fuel oil. Themajor part of coal fed ammonia plants -most of themof rather small size are located in China and use stillthe water gas route. A few ammonia plants based onmore modern coal gasification processes as the Tex-aco Process, the Koppers-Totzek Process, and theLurgi Coal Gasification are of larger size and oper-ate in South Africa, India and Japan. Also in coalbased ammonia plants the gas generation consists ofseveral lines. Depending on the gasification processthe maximum capacity of a single gasifier correspondsto an ammonia production between 500 t/d (Koppers-Totzek, Texaco) and 800 t/d (Lurgi gasifier). Regard-ing the degree of energy integration the situation isat best as in the partial oxidation of fuel oils, but inany case much lower than in a steam reforming plant.Lurgis moving bed gasification produces a gas witha rather high content of methane, which after separ-ation in the cryogenic step is processed in a small

steam reforming unit. Shift conversion, Rectisol unit,liquid nitrogen wash are the other essential steps inthe synthesis gas preparation. The gasification needs32 34 GJ/t NH3, power and steam generation con-sumes 18 22 GJ/t NH3, resulting in a total energyconsumption of 50-56 GJ/t NH3. For the Koppers-Tot-zek route a figure of 51.5 GJ/t is reported. Ube Indus-tries commissioned a 1000 t/d ammonia plant in 1984using Texacos coal gasification process. An energyconsumption of 45.5 GJ/t NH3 is stated, which is lowerthan the normally quoted figure of 48.5 GJ/t NH3 forthis technology.

Economics of ammonia production

The enormous technical and economical progressmade from the old plants using coke and water gastechnology to the modern steam reforming ammoniaplant with natural gas feedstock may be seen from thetable 2. Table 3 gives an estimate for ammonia pro-duction in 1996 cost in northwest Europe for differ-ent feedstocks using todays best and proven techno-logical standards for each process.

From table 3 it is obvious that at present there is nochance for the other feedstocks to compete againststeam reforming of natural gas. Only under very spe-cial circumstances in cooperation with a refinery, forexample partial oxidation of heavy oil fractionsmight be economically justified. It should be noted,however, that the average energy consumption of thesteam reforming plants presently in operation isnoticeable higher than the example of the modernlow-energy concept used in table 3.

The combined cost of feedstock and energy for asteam reforming plant both are natural gas is theprincipal determinant of the overall production costs.The price of gas and, by extension, the price ofammonia is to a greater or lesser extent linked tothe price of crude oil. The present interfuel relation-ship between gas and oil pricing might be distortedby the need for cleaner and less polluting fuels result-ing from increasing environmental awareness. In thisrespect, natural gas is so advantageous in relation toother fossil fuels that demand could well be pushedup in the coming years.

23

The received opinion is that, although gas supply canbe increased, upward pressure on prices is necessaryto make that happen. The forecast is for higher pro-duction costs in Western Europe and the USA. In thelow gas cost areas such as the Arab GuIf, Trinidad andIndonesia, competing usages for natural gas are notexpected to grow to any great extent, and in such loca-tions feedstock costs for the ammonia producers areexpected to rise only moderately. The biggest shareof the proven world reserves of natural gas have for-mer USSR, followed by the Middle East.

In the very long term, coal has prospects which mightbe drawn from consumption and world reserves of fos-sil feedstocks. At present consumption rates coal willcover the demand for 235 years, natural gas for 66years and oil for 43 years. But at least for the mediumterm natural gas can continue as preferred feedstock.

Future perspectives for theammonia production technology

Albeit world population and thus the demand of fer-tilizers is increasing 87% of the ammonia produc-tion is consumed in this sector building of newammonia plants did not keep up adequately with this.Of course, the main increase of demand is in thedeveloping countries, but in most cases there is notsufficient capital available for the investments needed.In the industrial countries with sufficient food sup-ply the fertilizer consumption is at best stagnant forecological concerns presently exercising the collectiveminds in the Western World. The sometimes fromenvironmentalists propagated ecological agricultureis no alternative as manure and biomass are not suf-ficient in effect and quantity to supply the necessarynitrogen and in addition they have the same problemwith nitrate run-off.

Direct biological fixation is presently restricted to thelegumes by their symbiotic relationship with the Rhi-zobium bacteria, which settle in the root nodules ofthe plants. Intensive genetic engineering research hasprovided so far a lot of insights in the mechanism ofthis biological fixation but a real breakthrough forpractical agricultural application has not yet hap-pened. The enzyme nitrogenase practically performsan ammonia synthesis in the bacteria, and for the syn-thesis of the nitrogenase the so-called NIF gene isresponsible. One option, for example, would be tobroaden the host spectrum of the Rhizobium bacte-ria by genetic manipulation. Other possibilities are totransfer the NIF gene to other bacteria which havea broader host spectrum but have no own nitrogen fix-ation ability or to insert the NIF gene directly intoplants. One important point to consider especiallywith the option of constructing a nitrogen fixing plant,is the energy balance of the plant. Because of the lowefficiency, a considerable amount of the photosynthe-sis product would be consumed to supply the energyneeded. This would consequently lead to a reductionin yield, which is estimated by some researchers to beas high as 18%.

The possibility of converting atmospheric nitrogeninto ammonia in homogeneous solution using metal-organic complexes was first raised around 1966. Theprospects for this route are not judged to be verypromising in terms of energy consumption and alsowith respect to the cost of these very sophisticated cat-

24

alyst systems. Photochemical methods of producingammonia at ambient temperature and atmosphericpressure in the presence of a catalyst have beenreported, but the yields are far too low to be econom-ically attractive.

So for the foreseeable future we have to rely on theconventional ammonia synthesis reaction, combininghydrogen and nitrogen using a catalyst at elevatedtemperature and pressure in a recycle process, as con-ceived in laboratory by Fritz Haber and made oper-able on an industrial-scale by C. Bosch, and on the useof the known routes to produce the hydrogen andnitrogen needed (which are in fact, what consumes allthe energy).

In conclusion it is possible to sum up the prospects bythe following broad predictions:

Natural gas will remain the preferred feedstock forat least the next 15 years. Coal gasification will notplay a major role in ammonia production in thatperiod.

The present ammonia technology will not changefundamentally, at least in the next 15 years. Evenif there are radical, unforeseeable developments,they will take time to develop to commercial intro-duction. With the available concepts, the marginsof additional improvements have become rathersmall after years of intensive research and devel-opment. Thus only minor improvements of individ-ual steps, catalysts and equipment might beexpected.

A further significant reduction in the energy con-sumption of the natural gas-based steam reform-ing ammonia process is unlikely; figures between27 and 28 GJ/t NH3 are already close to theoreti-cal minimum.

In the medium term the bulk of ammonia produc-tion will still be produced in world-scale plants of1,000 2,000 t/d NH3. Small capacity plants will belimited to locations where special logistical, finan-cial or feedstock conditions favor them.

New developments in ammonia technology willmainly reduce investment costs and increase oper-ational reliability. Smaller integrated process units(e. g. exchanger reformer, CAR) contribute to thisreduction and give additional savings by simplify-

ing piping and instrumentation. Improved reliabil-ity may result from advances in catalyst and equip-ment quality and from improved instrumentationand computer control.

It is very likely that genetic engineering will suc-ceed in modifying some classical crops for biolog-ical nitrogen fixation and that application in largescale will occur predominantly in areas with stillstrongly growing population to secure the increas-ing food demand. This development may be pushedby the fact that compared to the classical fertilizerroute less capital and less energy would be needed.This may happen within the next 20 years, but timeestimates are always risky. (A famous example:Man will not fly for 50 years, Wilbur Wright1901). But even with the introduction of this newapproach, traditional ammonia synthesis will con-tinue to operate in parallel, because it might be nec-essary to supplement the biological nitrogen fixa-tion with classical fertilizers. In addition, the exist-ing ammonia plants represent a considerable capitalinvestment and a great number of them may reli-ably operate for at least another 20 30 years froma mere technical point of view.

References:

A. V. Slack, G. Russel James, Ammonia, Part I-III, MarcelDecker, New York, 1073, 1974, 1977

J. R. Jennings (Ed.), Catalytic Ammonia Synthesis, PlenumPress, New York London, 1991

A.Nielsen (Ed.), Ammonia Catalysis and Manufacture, Sprin-ger-Verlag, Berlin Heidelberg New York, 1995

S. A. Topham, The History of the Catalytic Synthesis of Ammo-nia in Catalysis (ed. R. Anderson and M. Boudart), Volume 7,p. 1 50, Springer-Verlag, Berlin Heidelberg New York, 1980

M. Appl, Ammonia, Methanol, Hydrogen, Carbon Monoxide Modern Production Technologies, Ed. Alexander More, BritishSulphur Publishing, 1977, ISBN 1 873 387 26

M. Appl, Ammonia Synthesis and the Development of Cataly-tic and High Pressure Processes, Dr. H. L. Roy Memorial Lec-ture, IIChE Conference Hyderabad (Dec 1986); Indian Chemi-cal Engineer XXIX (1), 2-29 (1987)

M. Appl, The Haber-Bosch Process and the Development ofChemical Engineering in A Century of Chemical Engineering(ed. W. Furter), Plenum Publishing Corporation, 20-51 (1982)

M.Appl, A Brief History of Ammonia Production from the EarlyDays to the Present, Nitrogen (100), 47-58, Mar-Apr 1976)

L.Connock, Ammonia and Methanol from Coal, Nitrogen (226)47-56 (Mar-Apr 1997)

25