STEAM REFORMING

Abstract

A process for the steam reforming of hydrocarbons comprises partially oxidising a feedgas comprising a hydrocarbon feedstock with an oxygen-containing gas in the presence of steam to form a partially oxidised hydrocarbon gas mixture at a temperature >1200 C. and passing the resultant partially oxidised hydrocarbon gas mixture through a bed of steam reforming catalyst, wherein the bed comprises a first layer and a second layer, each layer comprising a catalytically active metal on an oxidic support wherein the oxidic support for the first layer is a zirconia.

Claims

1. A bed of steam reforming catalyst, wherein the bed comprises a first layer and a second layer, each layer comprising a catalytically active metal on an oxidic support wherein the catalytically active metal in the first layer is selected from the group consisting of platinum, palladium, iridium, ruthenium or rhodium, the oxidic support for the first layer is a zirconia, the catalytically active metal in the second layer is nickel and the oxidic support for the second layer is a refractory support selected from the group consisting of zirconia, alumina, calcium aluminate, magnesium aluminate, titania, magnesia and a mixture thereof.

2. A bed of steam reforming catalyst according to claim 1 further comprising a layer of zirconia balls, pellets or tiles placed on top of the first layer of reforming catalyst.

3. A bed of steam reforming catalyst according to claim 1 wherein at least one of the first layer and the second layer reforming catalysts are particulate, or comprise one or more monolithic supports.

4. A bed of steam reforming catalyst according to claim 1 wherein the first layer comprises rhodium on the zirconia and the second layer comprises nickel on an alumina or magnesium- or calcium-aluminate.

5. A bed of steam reforming catalyst according to claim 1 wherein the first layer is between 5 and 25% of the thickness of the bed.

6. A bed of steam reforming catalyst according to claim 1 wherein the first layer is comprised of a rhodium impregnated zirconia catalyst.

7. A bed of steam reforming catalyst according to claim 1 wherein the first layer is comprised of a 0.05 to 0.5% wt rhodium impregnated zirconia catalyst.

8. A bed of steam reforming catalyst according to claim 1 wherein the second layer is comprised of a catalyst consisting of nickel on alumina or calcium aluminate.

9. A bed of steam reforming catalyst according to claim 1 wherein the zirconia of the oxidic support for the first layer is a magnesia-, calcia-, lanthana-, yttria- or ceria-stabilised zirconia.

10. A bed of steam reforming catalyst according to claim 1 wherein the zirconia of the oxidic support for the first layer is a magnesia-, calcia-, lanthana-, yttria- or ceria-stabilised cubic zirconia.

11. A bed of steam reforming catalyst according to claim 1 wherein the zirconia of the oxidic support for the first layer is a yttria-stabilised cubic zirconia.

12. A bed of steam reforming catalyst according to claim 1 wherein the first layer has a catalytic activity higher than that of the second layer.

13. A bed of steam reforming catalyst according to claim 1 wherein the second layer has a catalytic activity which is between 50 and 95% of the catalytic activity of the first layer.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The invention is further illustrated by reference to the examples below and to FIGS. 1 and 2, which depict calculated vapour pressure of support species or catalytically active metal species at a range of temperatures above 1200 C.

DETAILED DESCRIPTION OF THE INVENTION

[0016] The feedgas may comprise a desulphurized hydrocarbon feedstock such as methane, natural gas or naphtha with a boiling point up to 200 C. which may be pre-heated to about 400-650 C., or may be a pre- or primary reformed gas stream comprising unreacted hydrocarbon, steam, hydrogen and carbon oxides. The latter are preferable as it may be desirable to ensure that the feed to the secondary reformer or autothermal reformer contains no hydrocarbons higher than methane and also contains a significant amount of hydrogen, as these factors reduce the risk of carbon/soot formation above/on the steam reforming catalyst.

[0017] The oxygen containing gas may be substantially pure oxygen, air or an oxygen enriched air. The amount of oxygen used preferably provides an an oxygen:carbon molar ratio in the range 0.4 to 0.7:1. Steam is preferably present at a steam:carbon ratio between 0.5 and 2, preferably 0.5-1.5, most preferably 0.5-1. Steam may be provided by adding it directly to the combustion zone or by mixing, either with the feedgas or the oxygen-containing gas. Alternatively, if the feedgas is a pre- or primary-reformed gas, no additional steam may be necessary. Furthermore, if the feedgas contains hydrogen, its combustion with oxygen will generate steam under the reaction conditions.

[0018] The hydrocarbon in the feedgas is partially oxidised by the oxygen in the oxygen-containing gas in the combustion zone of a suitable reformer such as a secondary or autothermal reformer.

[0019] The partially combusted hydrocarbon/steam mixture then passes from the combustion zone to the surface of the first layer of steam reforming catalyst at a temperature in the range 1200-1500 C. When the partially combusted feedgas/steam mixture, e.g. a partially combusted pre- or primary reformed gas, contacts the steam reforming catalyst, the steam reforming reaction, which is endothermic, cools the gas and the surface of the catalyst. The surface temperature of the catalyst may therefore, depending upon the conditions and activity of the catalyst, be in the range about 900-1400 C. In particular, we have found alumina or calcium aluminate volatilisation to occur when the catalyst temperature is above about 1200 C. Thus, preferably the first layer reduces the catalyst temperature below about 1200 C., more preferably below about 1100 C.

[0020] In the present invention, the first layer catalyst support is a zirconia. Preferred zirconia supports are stabilised zirconias, such as magnesia-, calcia-, lanthana-, yttria- or ceria-stabilised zirconias, which are most preferably in the cubic form. Such zirconias are known and are commercially available. Yttria-stabilised cubic zirconia is most preferred, e.g. a 16% wt yttria-stabilised cubic zirconia. Typically such stabilised zirconias have been fired to temperatures above 1200 C. We have found zirconia-containing supports to have lower volatility than supports comprising alumina or magnesium- or calcium-aluminate and so the presence in the first layer of alumina or magnesium- or calcium-aluminate is undesirable.

[0021] The catalytically active metal in the first layer of steam reforming catalyst may be nickel or another metal suitable for catalysing the steam reforming reaction such as cobalt, platinum, palladium, iridium, ruthenium or rhodium. Zirconia-supported nickel catalysts may contain nickel in an amount between 2 and 15%, preferably 3 and 8% by weight. Thus in one embodiment, the first layer comprises nickel on zirconia, preferably 3-8% by weight nickel on a stabilised zirconia.

[0022] Preferably, the catalytically active metal in the first layer of steam reforming catalyst forms compounds having a lower vapour pressure than the metal in the second layer. Where the second metal is nickel, catalytically active metals that form lower vapour pressure compounds under comparative reforming conditions include platinum and rhodium.

[0023] To further reduce the possibility of volatilisation of the first, i.e. upper, layer of catalyst, it desirably has a higher activity than the second layer, i.e. the catalytic activity of the first layer is higher relative to that of the second. By increasing the catalytic activity of the first layer, the endothermic steam reforming reactions take place to a greater extent and thereby act to cool the gas stream passing through the bed more rapidly than in the case where the layer of increased activity is absent. Preferably the catalytic activity of the second layer of steam reforming catalyst is between 1 and 95% of the catalytic activity of that of the first layer, more preferably between 50 and 95%.

[0024] Increased catalytic activity may be achieved by providing the first layer of steam reforming catalyst with a catalytically active metal with a higher activity per gram for the steam reforming reaction than the catalytically active metal of the second layer, i.e. the nickel may be replaced by sufficient amount of a more active catalyst for the steam reforming reaction. Such more catalytically-active metals include platinum, palladium, iridium, ruthenium and rhodium. Rhodium is particularly preferred as it also forms compounds with a lower vapour pressure than nickel under typical reaction conditions. Suitably active rhodium catalysts comprise 0.01-1.00% preferably 0.05 to 0.5%, more preferably 0.1 to 0.25% Rh by weight.

[0025] A particularly preferred first layer catalyst therefore consists of a rhodium impregnated zirconia catalyst, particularly a 0.05 to 0.5% wt rhodium-impregnated stabilised zirconia.

[0026] The second, i.e. lower, layer preferably comprises a nickel catalyst on a suitable refractory support. The refractory catalyst support for the second layer may also comprise zirconia, but for reasons of cost may alternatively comprise alumina, calcium aluminate, titania or magnesia or mixtures thereof. More preferably, the second layer catalyst consists of nickel on alumina or calcium aluminate.

[0027] The first and/or second layer of reforming catalysts may be particulate, in the form of shaped units such as pellets, rings or extrudates, which may be lobed or fluted. Alternatively, the first and/or second layer may comprise one or more monolithic supports such as a metal, ceramic foam or honeycomb supporting the catalytically-active metal. For example, the first and second layers may both comprise shaped units or may comprise a layer of shaped units over or under one or more monoliths. Preferably the first layer is a particulate catalyst, more preferably 4-hole cylinder, particularly one that is a lobed or fluted to provide a higher geometric surface area than a similarly sized solid cylinder without increasing pressure drop through the layer.

[0028] Steam reforming catalysts are typically made using impregnation methods well known to those skilled in the art of catalyst manufacture. For example nickel or rhodium may be provided in the first layer steam reforming catalyst by impregnation of the zirconia support with a solution of a suitable nickel or rhodium compound, for example an aqueous solution of metal acetate or nitrate, followed by heating in air to convert the compound to nickel or rhodium oxide. The nickel or rhodium oxide may then be reduced to elemental form by treatment with a reducing gas such as hydrogen at elevated temperature, although it is generally more convenient to install the catalyst in the un-reduced oxidic form and perform the reduction in-situ by reaction with a reducing gas (hydrogen and/or carbon monoxide). For example, a rhodium catalyst may be prepared by impregnating a stabilised cubic zirconia with an aqueous solution of rhodium nitrate, if necessary separating the impregnated material from the solution, drying and calcining to 400-500 C. The rhodium oxide, Rh.sub.2O.sub.3, is subsequently reduced in-situ. In a preferred embodiment, the rhodium is provided on the support as a so-called eggshell catalyst in which the rhodium is concentrated in the surface layers of the catalyst support rather than being distributed evenly throughout the support. This provides a more efficient use of the rhodium, which is expensive, compared to e.g. nickel.

[0029] If desired a layer of zirconia balls, pellets or tiles may be placed on top of the first layer of reforming catalyst to protect the surface of the steam reforming catalyst from irregularities in the combusting gas flow. A benefit of providing this layer is to prevent disturbance of the surface of the bed.

[0030] The catalytic activity of the first layer of catalyst may be further enhanced by providing the catalyst in a form having a higher geometric surface area (GSA) than the second layer. The geometric surface area of a catalyst may be calculated from its support dimensions. Increasing the GSA has the effect of increasing the surface area of catalytically active metal and thereby increasing the potential for the steam reforming reaction within the first layer. Where the steam reforming catalyst layers are particulate, this may be achieved by providing the first layer with particles of a smaller cross-sectional area than those of the second layer, or preferably by providing the first layer with particles of the same cross-sectional area, but having one or more holes, flutes or lobes therein. This latter method has the advantage that the pressure drop through the bed of steam reforming catalyst may be maintained at an acceptable value while increasing the activity of the first layer. Where a higher GSA material is provided as the first catalyst layer, preferably an inert layer of zirconia balls, pellets or, tiles is provided on top to protect the surface of the steam reforming catalyst from the combusting gas flow.

[0031] Thus in one embodiment the first layer consists of a particulate catalyst, preferably 0.05 to 0.5% wt Rh on a stabilised zirconia, in the form of a 4-holed pellet with a GSA of about 420 m.sup.2 per cubic metre, over a particulate 5-20% wt Ni on alumina catalyst in the form of a 4-holed pellet of the same GSA.

[0032] In a further embodiment, the first layer comprises a particulate catalyst, preferably 0.05 to 0.5% wt Rh on a stabilised zirconia in the form of a 4-holed pellet with a GSA of about 420 m.sup.2 per cubic metre over a monolithic alumina honeycomb supporting 5-20% wt Ni having a lower GSA.

[0033] The thickness of the bed of the steam reforming catalyst will depend upon the activity of the catalytically active metals, the conditions under which it is operated and whether the feedgas is a hydrocarbon/steam mixture or a pre- or primary-reformed gas. The thickness of the bed of steam reforming catalyst may be in the range 1-10 metres, preferably 3-5 metres with the first layer preferably comprising between 1 and 50%, preferably 3 and 25%, more preferably 5 and 15% of the thickness of the bed.

[0034] It will be understood by those skilled in the art that it may be useful to graduate the activity of the steam reforming catalyst through the bed. Therefore the second layer may comprise two or more successive layers of a steam reforming catalyst, the third or further layers having a lower catalytic activity than that preceding it.

EXAMPLES

[0035] Although the invention is illustrated and described herein with reference to specific embodiments, the invention is not intended to be limited to the details shown. Rather, various modifications may be made in the details within the scope and range of equivalents of the claims and without departing from the invention.

Example 1

Calculated Example

[0036] Thermodynamic calculations on vapour pressure of different Ni and Al species above 1200 C. indicate that the alumina and nickel are reacting with the reformed gas. The reaction is predominantly with the steam and this produces hydroxides as shown below.

Al.sub.2O.sub.3(s)+3H.sub.2O(g)2Al(OH).sub.3(g)

Ni(s)+2H.sub.2O(g)Ni(OH).sub.2(g)+H.sub.2 (g)

[0037] There are other potential reactions forming other sub-oxides and hydroxides and the predicted vapour pressure of the most volatile aluminium species at 1,250 C. is 6.210.sup.6 atm and for the most volatile nickel species 1.110.sup.6 atm under conditions typically found in an autothermal or secondary reformer. Based on certain assumptions about the partial steam pressure and temperature of the surface of the steam reforming catalyst, this could, in a reformer designed for a 4,000 mtpd methanol plant where the autothermal reformer processes 8,000 mtpd of reformed gas, vaporise and condense approximately 4.4 kg/day of alumina from the top of the catalyst bed. Over a 4-year catalyst life this could amount to as much as 6.2 te. In the hypothetical 4,000 mtpd methanol plant, the autothermal reformer would be expected to contain of the order of 80 te of catalyst, thus 8% of the catalyst may move in the lifetime of the catalyst charge.

[0038] Tests on alumina-supported catalysts in full-scale reformers have confirmed a close correlation of the plant data to the thermodynamic data, which supports, the theoretical figures.

[0039] Furthermore, the effective vapour pressures of Pt and Rh are lower than the effective vapour pressure of Ni under simulated reforming conditions. Under simulated secondary reformer conditions (1 bar, 5% methane in nitrogen and steam, with a steam to carbon ratio of 3.0) nickel, platinum or rhodium catalysts on the same alumina support material were heated until the temperature reached 1200 C. When the temperature approached 1200 C., the reforming activity, as measured by methane slippage, was observed to decline with the nickel-containing catalyst but was maintained for the platinum and rhodium catalysts. Analysis showed that nickel had volatilised and migrated downstream to cooler parts of the reactor.

[0040] FIG. 1 depicts the calculated vapour pressure of species formed with alumina and zirconia supports over a range of temperatures above 1200 C. at conditions typically found in an autothermal or secondary reformer. It shows that zirconia is considerably less volatile than alumina. FIG. 2 depicts the calculated vapour pressures of species formed with nickel and rhodium catalysts over a range of temperatures above 1200 C. It shows that rhodium is considerably less volatile than nickel.

[0041] Thus rhodium on stabilised zirconia catalysts by comparison have vapour pressures orders of magnitude lower than alumina and nickel. Based on this, the amount of catalyst vaporisation and transfer downstream will very small. In the case calculated above, over a 4 yr period 6,200 Kg of Ni/Al.sub.2O.sub.3 catalyst could be vaporised. Using Rh/ZrO.sub.2 the amount of material vaporised would be reduced to the order of 0.06 Kg over 4 years. As the active metal is also refractory, there is no loss of the active component either, resulting in a catalyst that retains activity at the top of the bed for longer.

[0042] Furthermore, rhodium exhibits a specific activity/g between 2 and 3 times that of nickel. With a 2 to 3 fold increase in specific activity, the gas temperature can be reduced from the inlet temperature of 1,250 C. to less than 1,100 C. in the layer of the rhodium/zirconia catalyst before entering the main bed of standard nickel on alumina. A reduction in the maximum gas temperature flowing over nickel/alumina catalyst from 1,250 C. to 1,100 C. will reduce the saturated vapour pressure by a factor of 4, giving a substantial decrease in the transfer of alumina down the bed.

[0043] Calculations have been performed on three cases looking at the amount of alumina that may be transported in an autothermal reformer. The calculations were based on the hypothetical 4,000 mtpd methanol plant autothermal reformer operated at the same assumed partial steam pressure and catalyst surface temperature already discussed above.

[0044] The first case, not according to the present invention, is an autothermal reformer charged with a bed of alumina lumps over a bed of large low activity cylinders on top of standard single hole ring catalyst. The catalysts are all standard nickel on alumina, although the large cylinders have 50% less nickel making them lower activity. This bed would vaporise 4.4 kg/day of alumina, principally from the alumina lumps at the top of the bed and condense 4.1 kg/day of this back into the bed, 3.1 kg/day of which would condense out in the top 0.5 m of the large cylindrical catalyst. This would be the most problematical part of the bed, slowly blocking up and increasing resistance to flow. The rest of the alumina would be spread throughout the remaining 3.5 m of catalyst. 0.3 kg/day of alumina vapour would pass downstream to cause fouling of the waste heat boiler or other heat transfer surfaces.

[0045] The second case, also not according to the present invention, is the same autothermal reformer charged with the same bed except that the alumina lumps have been removed. This bed would vaporise 1.9 kg/day of alumina from the first 0.3 m of cylindrical catalyst and condense 1.6 kg/day of this back into the bed, 0.7 kg/day of which would condense out in a band 0.3 m deep near the top of the bed. Whilst this is a large improvement on the case above, it still represents a substantial transfer of catalyst over a 4 year catalyst cycle. The rest of the alumina would be spread throughout the remaining 3.5 m of catalyst. 0.3 kg/day of alumina vapour would pass downstream to cause fouling of the waste heat boiler or other heat transfer surfaces.

[0046] In the third case, according to the present invention, a bed is charged with an upper layer of a rhodium-impregnated zirconia in the form of a lobed 4-hole cylinder on top of standard Ni-alumina single hole ring catalyst. In this case, enough activity is installed in the bed to reduce the gas temperature to 1,100 C. before it enters the bed of standard nickel on alumina ring catalyst. This bed would vaporise 0.7 kg/day of alumina from the top 0.2 m of the nickel on alumina rings and condense 0.4 kg/day of this back into the bed spread throughout the remaining perhaps 3.5 m of catalyst. This is a substantial improvement on the cases above and this bed would probably not exhibit any real pressure drop increase over a 4 yr period. 0.3 kg of alumina vapour would pass downstream to cause fouling of the waste heat boiler or other heat transfer surfaces.

Example 2

Comparison of the Mass Loss from Alumina and Zirconia Supported Rh Catalysts

[0047] At an industrial scale, a primary reformed natural gas stream was fed to a secondary reformer at where it was subjected to partial oxidation in a combustion zone with an oxygen stream fed via burner apparatus disposed near the top of the reformer and passed downwards from the combustion zone to a bed of steam reforming catalyst. The oxygen:gas ratio was 0.48-0.49.

[0048] The bed of steam reforming catalyst was 4 metres thick and comprised 4-hole cylindrical pellets. The top 10% (0.4m) of the bed comprised pellets of either (a) rhodium on alumina or (b) rhodium on stabilised zirconia. The lower 90% of the bed comprised a standard nickel on alumina catalyst. The steam reforming catalysts were provided to the reformer in oxide form and reduced in-situ. The reforming process was operated over an extended period for both rhodium catalysts and the weight loss measured for pellets from throughout the first top layer.

[0049] The average weight loss was then calculated.

[0050] The catalysts pellets were as follows

TABLE-US-00001 Hole Catalyst Diameter Length diameter GSA (as supplied to reformer) (nm) (nm) (nm) (m.sup.2m.sup.3) (a) 0.15% wt Rh.sub.2O.sub.3/Al.sub.2O.sub.3 11 15 3 530 (b) 0.15% wt Rh.sub.2O.sub.3/16% wt 11 12 3 560 Y.sub.2O.sub.3/ZrO.sub.4

[0051] The average process conditions and weight loss/pellet in grams/day are as follows;

TABLE-US-00002 Primary reformer Primary Secondary Secondary Time Flow Reformer Reformer Reformer Mass On- Natural Steam: Inlet Exit Loss/ line Gas carbon Pressure Temp. Temp pellet Catalyst (days) Nm.sup.3hr.sup.1 ratio (atm) ( C.) ( C.) g/day Rh/Al.sub.2O.sub.3 647 5407 3 39 668 990 1.4 10.sup.3 Rh/Y.sub.2O.sub.3/ZrO.sub.4 90 6250 3 39 681 1014 1.6 10.sup.4

[0052] The results demonstrate a lower erosion rate for the stabilised zirconia-supported catalyst, despite the harsher temperature conditions and higher GSA.