TECHNOLOGY Enrichment can decrease NO x , SO x formation

Satish Tamhankar, Raghu Menon, Ting Chou
BOC Group Inc.
Murray Hill, N.J.

Ram Ramachandran
BOC Gases
Murray Hill, N.J.

Randall Hull
BOC Gases-Americas
Murray Hill, N.J.

Richard Watson
BOC Gases-Europe
Surrey, England

One of the key issues associated with fluid catalytic cracking (FCC) unit operations and debottlenecking options is the effects of process variables and operational modes on regenerator-related NOx and SOx emissions.

This second of two articles on FCC oxygen enrichment explains regenerator NOx-formation pathways and summarizes the effects of variations in the regenerator environment, including regenerator gas and solids compositions, on these pathways. Also reported are results indicating that operation with oxygen enrichment can improve the effectiveness of SOx-transfer additives.

The first article in this series described advances in oxygen-production technology and presented results from materials-compatibility studies (OGJ, Feb. 26, p. 54).

Background

Considerable knowledge has been generated in the refining industry regarding emissions of sulfur and nitrogen oxides (SOx and NOx) from FCC boiler systems. There is, however, little published information regarding the effects of variations in regenerator gas compositions profiles on these emissions. In addition, no overall framework exists to facilitate the understanding of NOx emissions under various operating modes.

Gas composition profiles strongly affect regenerator emissions. The effects of composition profiles have been investigated in depth, and are reported to be strong in combustion systems. Operating conditions and mechanisms, however, can be significantly different in these systems.

Besides the obvious differences between operation in partial and complete CO combustion modes, the nature of the regenerator environment varies from one zone to another. For example, high oxygen levels are present in the immediate vicinity of the regenerator air grid. Oxygen concentrations decline rapidly, however, moving from the distributor to the exit of the dense phase.1

At only one fourth the distance from the air distributor to the top of the dense phase, oxygen partial pressure can be expected to decrease to about 10 vol % for partial-combustion operation with air. Also, the presence of highly reactive hydrocarbons significantly reduces the free oxygen available for side reactions.

Proceeding away from the oxygen-rich region into the dilute phase, the levels of carbon monoxide increase, as do levels of carbon dioxide, to a lesser extent.

For operation in partial CO combustion mode with a combustion promoter, the oxidative nature of the oxygen-rich zone is enhanced. For complete-combustion operation with a promoter, the co-rich region of the regenerator vessel is smaller.

When operating with oxygen-enriched air, the partial pressure of oxygen in the immediate vicinity of the air grid will be higher. And for regenerators operated in partial-combustion mode, the partial pressure of free CO also will be higher in the dilute phase and downstream.

SO_x pickup

The commercial availability of high-quality, SOx-transfer agents over the last decade has made their addition to FCC units (FCCUs) an increasingly popular method of meeting stringent emission regulations.2

During catalyst circulation in the FCCU, these additives work by picking up SOx from the regenerator as SO3 and releasing the sulfur as H2S in the reactor and stripper. The regenerated additive then returns, along with the spent catalyst, to the regenerator.

The H2S leaves the unit with the other reaction products and ultimately is recovered as elemental sulfur in the refinery's sulfur-recovery units.

The reaction mechanisms associated with SOx-transfer additives have been discussed in detail.3 Components of the base FCC catalyst also are known to contribute somewhat to SOx-transfer activity.

Over the years, these additives have evolved from simple compounds to complex, rare-earth-promoted, spinels comprising mixed-solid solutions and incorporating a transition metal. The rare-earth component is added principally to enhance SO2 oxidation activity.

The overall effectiveness of SOx-transfer agents is known to be influenced strongly by regenerator operating parameters.4 Laboratory work has indicated that the rate-limiting step in the SOx-transfer process is the oxidation of SO2 to SO3.5

The kinetics and equilibrium of this reaction are well known. In the absence of a promoter, the reaction proceeds very slowly.6

The formation of SO3 at equilibrium is favored by higher oxygen concentrations and lower temperatures. For example, increasing the oxygen concentration from 0.1% to 1% to 5% increases equilibrium conversion from, respectively, 12% to 21% to 37%.

Commercially, the effectiveness of SOx-transfer additives is known to increase with increasing levels of excess oxygen in the flue gas. Although the additives are effective in regenerators operated in partial CO combustion mode, it is generally accepted that the effectiveness is less than for operation in complete-combustion mode.7

Laboratory experiments were conducted using mixtures of SO2, O2, and N2 in partial-combustion mode, by varying oxygen concentrations from 0% to 1%. The results indicated trends similar to those observed for complete-combustion operation.4

The BOC Group's work was aimed at studying SOx-transfer effectiveness as a function of variations in regenerator oxygen and CO concentrations resulting from the enrichment of combustion air.

Study results

Studies were performed in a microbalance setup with various gas mixtures containing SO2. These mixtures were designed to simulate different regenerator regions.

The experimental apparatus included a microbalance and automated gas-mixing and delivery system designed to operate cyclically. Each cycle comprised an absorption period, simulating conditions encountered during catalyst regeneration, and an additive-regeneration period. The additive-regeneration period simulated the reducing environment and moisture encountered by the additive during circulation in a commercial unit.

The varying fluidization states were not simulated, but the authors do not expect this to alter the trends observed.

Samples were subjected to five or more complete cycles, depending on the phenomenon being investigated. The studies found that, regardless of the CO concentration, the effectiveness of the SOx-transfer additive being tested depended mainly on the free oxygen concentration ([O2] 2 [CO]/2).

For the fourth absorption cycle in each experiment, the relative SOx pickup, as a function of operating conditions (including the free oxygen concentration in the inlet gas), is summarized in Table 1 [19581 bytes].

The experiments performed with shorter cycles (each partial cycle reduced by 50%) were limited by the additive-regeneration capacity. For these tests, the SOx-pickup efficiency was considerably less, even when high concentrations of free oxygen were used.

Although the presence of high CO results in reduced availability of free oxygen, it was found to have no permanent effect on additive performance in multicycle experiments.

Exposure of SOx-transfer additives to high-CO environments did not affect subsequent performance. Such conditions are encountered in certain portions of the regenerator under conditions of oxygen enrichment or partial CO combustion.

This important observation was corroborated by studies performed in the presence and absence of FCC catalysts. Sequential weight-gain plots are presented for the additive in Fig. 1 [63553 bytes], and for the additive/catalyst mixture in Fig. 2 [81061 bytes].

Fig. 1 [63553 bytes] shows the performance of the additive in an environment containing 2% free oxygen (Curve A) after it has been exposed to five cycles involving absorption in a CO-rich, oxygen-poor environment (Curve B). In Fig. 2 [81061 bytes], Curve C illustrates excellent SOx pickup over five cycles for an additive-regenerated equilibrium catalyst (E-Cat) mixture.

In this experiment the mixture had been subjected to five cycles with oxygen-rich (2% free oxygen) absorption steps (Curve A), followed by five cycles with CO-rich absorption steps (not shown, but similar to Curve B in Fig. 1[63553 bytes]). As long as additive regeneration was near-complete, the enriched-CO environments caused little decline in the additive's absorptive pickup activity, even for samples subjected to several cycles.

The mixtures of regenerated E-cat and additive were further tested over several cycles An example of these results is shown in Fig. 3 [85090 bytes].

In this test, the mixture was subjected to 28 cycles in an environment containing 2% oxygen and 0.5% SO2 in nitrogen, with and without 5% CO and 6% CO2.

Performance remained unchanged over the cycles in both cases, except that, in the presence of CO and CO2, the SOx pickup was relatively less. Thus, the presence of enriched-CO environments does not appear to contribute toward additive-poisoning mechanisms.

It is important to note that the results indicate that SOx pickup activity in commercial regenerators is confined mainly to the air grid and dense-bed areas, especially for operation in partial-combustion mode. Thus, SOx-transfer effectiveness can be expected to be much improved for enriched-oxygen operation at constant temperature conditions. This is especially true for partial-combustion operations, in which enrichment affects the gas composition in the vicinity of the air grid.

Even marginal increases in available oxygen in various portions of the regenerator can result in enhanced SOx pickup. Also, the higher partial pressure of oxygen in the air-grid area can be expected to ensure more complete conversion to SO2 of the H2S released from coke sulfur.

The coupling of a SOx-transfer additive with regenerator oxygen enrichment and catalyst-cooler technology can substantially reduce refinery SOx emissions. Pilot studies in a circulating unit are in progress to quantify scale-up of the laboratory observations.

NO_x formation

FCC regenerators are major contributors to refinery NOx emissions. In general, three categories of NOx emissions result from combustion processes: fuel NOx, so-called "prompt" NOx, and thermal NOx.8

Under FCC regenerator conditions, it is known that the fuel NOx mechanism dominates; the nitrogen species from the coke combusted in the regenerator vessel are mainly responsible for FCC-related NOx.

For regenerators operated in complete-combustion mode, the nitrogen species contained in the regenerator flue gas, besides elemental nitrogen, are mostly nitrogen oxides-principally nitric oxide (NO) and some NO2.

In this mode, the amount of NOx produced per unit of coke nitrogen combusted depends on the concentration of combustion promoter in the catalyst inventory and the amount of excess oxygen in the flue gas. The nitrogen content of the coke depends on the amount and type of nitrogen in the feed, as well as on reactor operating conditions, and is the subject of several ongoing investigations.

For regenerators operated in partial-combustion mode, nitrogen in the regenerator effluent, other than elemental nitrogen, is present mainly in reduced forms (mostly ammonia).9 These reduced-nitrogen species subsequently are oxidized to elemental nitrogen and NOx in the downstream boiler, where CO is oxidized to CO2.

Other investigators have reported that the amount of NOx formed in the boiler is strongly affected by boiler operating conditions and, most importantly, by the amount of nitrogen species present in the boiler fuel in reduced forms.

Species like ammonia are less stable than nitrogen. For this reason, the amount of boiler NOx formed by the fuel and prompt-NOx mechanisms can be reduced substantially if it is possible to favor the formation of elemental nitrogen in the regenerator at the expense of reduced-nitrogen species.

Operational options proposed to achieve this preference include the addition of excess promoter to a regenerator operated in partial-combustion mode.10 11

There is little published information regarding the effect of the oxygen concentration in the combustion gas. The authors have attempted to understand this effect by studying, individually and in combination, reactions involving some of the principal nitrogen species.

The free-radical chemistry involved in the release of fuel-bound nitrogen to gaseous species is very complex. A simplified representation, however, involves the release of HCN as a primary product if the fuel nitrogen is in aromatic or cyclic forms, such as that expected for FCC coke.12 13

The HCN is rapidly converted to HOCN by atomic oxygen, OH, and H radicals. The HOCN is subsequently converted to NHx species.

For a regenerator operating in complete-combustion mode, the HCN and ammonia formed from the HCN are oxidized to elemental nitrogen and NOx. In partial-combustion mode, however, only a portion of the reduced-nitrogen species may be oxidized. Associated radicals subsequently are reduced substantially to nitrogen by the reducing atmosphere in the regenerator dilute phase.

The two principal sets of reactions involved in this mechanism are:

Oxidation reactions involving reduced-nitrogen species, especially ammonia and hydrogen cyanide (HCN)
Subsequent reactions involving the reduction of intermediate NOx by the reducing species in the environment (CO, ammonia, and coke).

The hydrolysis reaction also plays a major role in determining the balance between HCN and ammonia. The two reaction types include eight principal reactions, not including the constituent free-radical pathways and hydrolysis reaction (see Reactions).

Equilibrium compositions were calculated for typical conditions in various regenerator regions to identify thermodynamically favorable reactions. Example results are shown in Fig. 4.[81120 bytes]

For ammonia oxidation with air at 700 C., the favored pathway leads to the formation of nitrogen, with very little NOx being formed. The BOC Group's study results, discussed in the following section, show that much more NOx can be formed from the introduced nitrogen species.

This result implies that it is important to use nitrogen as the background gas while studying commercial operation, and that regenerator kinetics play an important role.

Study results

Studies were performed at typical regenerator operating temperatures, residence times, space velocities, and environments. Model compounds were used in tubular reactors made of quartz and Inconel, at process pressures of 5-35 psig.

The authors studied the effect of variation in oxygen concentration on the destruction of ammonia and HCN. The tests were performed in the absence and presence of fluid beds comprising typical metals-laden FCC equilibrium catalyst from a commercial unit.

The concentrations of nitrogen species were quantified using a Fourier Transform infrared analyzer. Similarly, reactions involving NOx reduction were studied by systematically varying concentrations of CO, CO2, and oxygen.

The reactions between NOx and ammonia also were studied briefly.

Some experiments were performed to study the network of sequential and parallel reactions that result from using mixtures of ammonia or HCN and oxygen, carbon oxides, and coke (on catalyst) as feed. In all cases, the researchers attempted to simulate as closely as possible the conditions present in various portions of commercial regenerators. The experiments were performed using nitrogen as the background gas.

For commercial operation with oxygen-enriched air, the partial pressure of oxygen in the vicinity of the gas-distribution system, and in the dense-bed region, is higher. Results of experiments simulating the conditions in these regions showed that higher oxygen concentrations cause higher conversions of ammonia to nitrogen and NOx.

Table 2 [15678 bytes] summarizes results from runs performed at 1,300 F. with similar inlet ammonia concentrations. The tests were performed both with and without fluidized beds of FCC catalyst.

The results show that, even in the absence of catalyst, ammonia conversion increased from 15% at a 3.6 psi partial pressure of oxygen at the inlet, to 36% at 12.9 psi. Parallel trends were observed for runs performed with similar space velocities in the presence of catalyst.

Significantly longer residence time was required to maintain high ammonia conversion as the partial pressure of oxygen was reduced. Additionally, the authors found that increasing the oxygen concentration in the combustion gas increased the ratio of NO2 to NO in the product.

All of these observations underscore the importance of oxygen availability for the destruction of ammonia in the short time that gas spends in the oxidative atmosphere of regenerators operated in partial-combustion mode.

It was found that HCN also was readily oxidized to nitrogen and NOx, even at low partial pressures of free oxygen. In general, the reactivity of HCN in a typical regenerator environment was found to be similar to that of ammonia. Hydrogen cyanide, however, was found to be more reactive, especially at low oxygen partial pressure.

For commercial enriched-oxygen operation, the availability of CO increases in certain portions of the regenerator vessel. Studies were undertaken aimed at understanding the potential for reducing to elemental nitrogen NOx formed previously in other portions of the regenerator vessel. These studies revealed that the rates of these reactions are favored strongly by higher CO concentrations.

The reaction between CO and NO is rapid at 1,300 F., and essentially proceeds to completion in the residence times available in commercial FCCUs, when catalyst fines are present. The reduction of any intermediate NO2 to nitrogen is less rapid in comparison and appears to proceed sequentially through the formation of intermediate NO.

No NO2 was found in the product in any of the studies with CO, indicating that the reduction of NO2 to NO is very rapid.

In experiments performed in the presence of oxygen, no NOx was found in the product, except for the studies in which the CO concentration in the reactor effluent was less than 1%. For these cases, NO was the only nitrogen oxide found in the effluent.

Given sufficient residence times (as short as 4 sec), the reduction of NO2 was found to proceed to completion (100%) to form nitrogen, even at temperatures as low as 1,110 F., in the presence of about 9% excess CO.

With a residence time of about 1 sec, and with excess CO (<3%), a temperature of about 1,400 F. was necessary to completely react NO2. This temperature is well within the operating range of FCC regenerators.

Results for runs performed at different temperatures and residence times, and at excess CO levels, are summarized in Table 3 [16124 bytes].

The authors also found that ammonia and NOx react with each other to varying extents, producing elemental nitrogen in regenerator environments containing FCC catalyst. The reaction between ammonia and NO was found to be enhanced by the presence of oxygen, similar to selective catalytic reduction.

The ability to rapidly oxidize a significant portion of the reduced forms of nitrogen in a commercial unit therefore is key to maximizing subsequent conversion of the unconverted reduced species, such as ammonia and HCN, to elemental nitrogen. This is true as long as the reducing agents are present in excess.

Carbon on the coked catalyst also was found to be an effective reducing agent for NO in the absence of oxygen or CO. Reduction of NO2 with carbon results in the formation of some NO, similar to its reaction with CO.

Studies performed with ammonia or HCN and oxygen in the presence of carbon oxides essentially proceeded to complete conversion, as long as excess oxygen was available. Results from experiments performed under similar conditions (residence time = 1 sec; temperature = 1,300 F.; pressure = 30 psig) are summarized in Table 4 [14513 bytes].

The continuous reduction of the intermediate nitrogen oxides by CO substantially increased the overall reaction rate. Studies performed with ultrashort residence times (<1 sec) showed enhanced reaction rates for mixtures enriched with oxygen and CO.

Additionally, it is clear that increased concentrations of available CO ([CO] 2 2[O2]) clearly favored increased concentrations of HCN, similar to conditions in regenerators operated in deep partial-combustion mode.

Commercial implications

The high reactivities of the various nitrogen species in the FCC regenerator environment indicate that the amount of nitrogen released from coke in regenerators is several-fold the amount deduced from measuring nitrogenic compounds in flue gas. Back-calculating based on the observed reactivities appears to indicate that anywhere from 25% to 65% of FCC feed nitrogen can be converted to coke nitrogen.

This calculation further validates some observations based on overall circulating unit balances, and emphasizes the differences in feed sulfur and nitrogen chemistry.

BOC's studies indicate that, for a regenerator operated in partial-combustion mode with oxygen-enriched air as the combustion medium, the reduced forms of nitrogen are expected to be oxidized to elemental nitrogen and, to a lesser extent, intermediate nitrogen oxides and associated radical species. The nitrogen oxide species are reduced sequentially to nitrogen by reduced nitrogen species, by coke on the catalyst particles, and by CO.

The availability of oxygen in the dense bed is vital for rapid oxidation of the reduced nitrogen species, as well as for immediate, sequential conversion of the formed NOx to elemental nitrogen, which in turn enhances the destruction of more reduced-nitrogen species.

Similar trends can be expected when a CO combustion promoter is used.

Enriched-oxygen operation also emphasizes the reducing nature of the regenerator freeboard. Thus, overall, the flue gas from oxygen-enriched air operation can be expected to have higher concentrations of elemental nitrogen and lower concentrations of ammonia and HCN. Results from laboratory experiments simulating the coupled reactions support this conclusion.

The reduction of ammonia in the regenerator flue gas should reduce downstream NOx production per unit of coke nitrogen burned in the regenerator.

For units operated in complete-combustion mode, all the reduced nitrogen species are converted to nitrogen and NOx in the regenerator. Mechanisms for reduction of the NOx therefore are important.

The authors found that these species are reduced considerably by CO, even in the presence of oxygen, when FCC catalyst particles are present. Thus, for complete-combustion operation with enriched oxygen, two opposing factors are present.

Reduction of formed NOx is enhanced because of the greater availability of CO in the dense bed and portions of the dilute phase. Increased oxygen availability, however, could favor NOx formation because of increased selectivity to intermediate NO2.

The balance between these factors depends on regenerator design, including the catalyst and air-distribution method, hardware conditions, and the presence of excess oxygen in the flue gas.

When a combustion promoter is used for complete-combustion operation, the zone in which CO and other reducing agents are present is smaller and has less reducing strength, compared to partial-combustion operation. This also is the case when regenerators are operated with greater concentrations of excess effluent oxygen.

BOC's data thus explain the increase in NOx observed in commercial units operating in complete-combustion mode with high concentrations of promoter and excess oxygen.

These results refute speculation that increased excess oxygen somehow results in increased dissociation of elemental nitrogen, and that, therefore, FCC NOx formation is thermally based. The results also resolve the opposing effects observed commercially for promoter addition in partial and complete-combustion modes.

In addition, BOC's studies indicate that, for regenerators operated in complete CO combustion mode, the amount of intermediate NO2 present could far exceed that in the effluent gas. Further laboratory and pilot-plant studies are in progress to better quantify the effects of regenerator operation with enriched oxygen.

Overall, the studies suggest possibilities to substantially reduce NOx formation by choosing a combination of partial CO combustion and other appropriate technology options, such as oxygen enrichment, followed by combustion of the flue gas in an optimized low-NOx burner.

References

1. Krishna, A.S., and Parking, E.S., "Modeling the Regenerator in Commercial Fluid Catalytic Cracking Units," Chem. Eng. Prog., April 1985.

2. McDaniel, J.G., and Neumann, D.J., "Reducing FCC SOx Emissions to Very Low Levels Using DESOX," 1992 NPRA Annual Meeting, New Orleans.

3. Yoo, J.S., Bhattacharya, A.A., and Radlowski, C.A., "Advanced De-SOx Catalyst: Mixed Solution Spinels with Cerium Oxide," Applied Catalysis B: Environmental, Vol. 1., 1992, p. 169.

4. Sigan, J.A., et al., "Reducing FCC Emissions with No Capital Cost," AIChE spring national meeting, New Orleans.

5. Powekk, J., et al., "Advanced FCC Flue Gas Desulfurization Technology, 1988 NPRA annual meeting, San Antonio.

6. Leppard, W.R., "Sulfate Control Technology Assessment," prepared by Exxon Research & Engineering Co. for EPA Report 460/3-75-001-a, November 1974.

7. Pane, P.A., and Latimer, J.A., "Controlling FCC SOx Emissions with DESOX," 1991 AIChE annual meeting, Los Angeles, Paper No. 127E.

8. Pourkashanian, M.P., Yap, T.L., and Missaghi, M., , Combustion Modeling: Cofiring and NOx Control, "Reduction of NOx Emissions Using CFD as a Design Tool," FACT Vol. 17ASME, 1993.

9. Johnson, G.L., Samish, N.C., and Altrichter, D.M., U.S. Patent No. 4,744,962.

10. Altrichter, D.M., U.S. Patent No. 5,021,144.

11. Avidan, A., Mathias, M., Menon, R., Sodomin, J., and Teitman, G.J., U.S. Patent No. 5,268,089.

12. Miller, J.A., and Fisk, G.A., Chem. & Eng. News, Vol. 31, 1987.

13. Houser, T.J., Hull, M., Alway, R., and Biftu, T., (Int'l) J. of Chem. Kin., Vol. 12, 1980, p. 579.