Chemical & engineering aspects of low NOx concentration

Two new requirements for low NOx emission from industrial combustion installations are explained. One, which is directed against a previously unsuspected and dangerous form of contamination of certain food-stuffs, calls for a NOx concentration not exceeding 1 ppm at the burner mouth; and the other is a special case of environmental protection in the United States, resulting from greatly accelerated activity in the application of special methods of recovery of indigenous oil. Background mechanisms of NO formation and their reaction rates are reviewed, and the methods of sub-stoichiometric combustion, staged combustion, high excess air combustion, recirculation, interstage cooling and enhanced flame radiation are examined as possible means of producing the recurred low concentrations of NOx. Burner designs capable of meeting both requirements are described.

CHEMICAL & ENGINEERING ASPECTS OF LOW NOx CONCENTRATION W.H. Wheeler ABSTRACT: Two new requirements for low NCbc emission from industrial combustion installations are explained. One, v M c h is directed against a previously unsuspected and dangerous form of contamination of certain food-stuffs, calls for a NOx concentration not exceeding 1 ppm at the burner mouth; and the other is a special case of environmental protection in the United States, resulting from greatly accelerated activity in the application of special methods of recovery of indigenous oil. Background mechanisms of NO formation and their reaction rates are reviewed, and the methods of sub-stoichicmetric combustion, staged combustion, high excess air combustion, recirculation, interstage cooling and enhanced •flame radiation are examined as possible means of producing the recurred low concentrations of NCbc. Burner designs capable of meeting both requirements are described. PROCESS COMBUSTION CORPORA! JUN WHfc/VCP MteftMlSBS 7th July, 1980. CHEMICAL & ENGINEERING ASPECTS OF LOW NOx CONCENTRATION INTRODUCTION: A new aspect of the requirement for minimising the amount of combustion NCbc discharged from burners developed quite suddenly in 1979 from recognition that NOx contained in directly heated air, used for drying food-stuffs, can react with amines associated with the food-stuff proteins, to produce impurities which constitute a ^ serious health hazard when the food is assimilated. A NOx concentration of the order of 0.05 ppn, in gases used for this purpose is seen as the permissible maximum. It means that combustion gases at the turner mouth may average no more than 1 ppn of NOx. As recently as the first six months of 1979, this would have been considered impossible and many are, understandly, sceptical that it is practicable now. But it is now being obtained quite regularly on food-drying kilns in industry. In parallel, for different reasons, a new insistence on minimising the concentration of combustion NOx discharged to stack from steam generators has developed. A major acceleration has occurred in the application of thermal recovery methods for certain types of very viscous crude oil which occur particularly in California, and to a lesser extent in Texas, Oklahoma and elsewhere. These methods demand very large quantities of steam, and the steam must be generated near each well; and for reasonable economy locally available crude or heavy residues will often be used. Consequently, in these areas, big steam generators are being set up adjacent to large numbers of "lazy" wells, with prospects of grossly excessive NOx emission from combustion of fuel with*high nitrogen content. Concern regarding discharge of NCbc, with contamination of local atmospheres and prospective formation of smog, even though tempered by sensitivity to the urgency of increasing oil recovery, has inevitably resulted in the imposition of limits on NOx content in stack gases discharged from the heaters. A concentration of not more than 200 ppn of NOx in flue gases with 13\% CO2 i s being temporarily accepted as tolerable, and at lower CO2 levels the NOx concentration must be correspondingly lower. The alternative presentation is that not more than 0.25 lb. NCbc per million Btu (108 g/gJ) may be discharged to atmosphere. With fuel-nitrogen at 0.8 -1.01 this requirement is by no means easy to meet. The aim of this paper is to examine sane of the methods which may be used to obtain these results, against the very full background of technical and scientific information on N3x formation and suppression which exists already. /3-* v #.^/S!fi±^^ * - 2- B\CKGROUND: No attempt is made here to summarise the wide-ranging published material to which references are made, but certain fundamentals are reiterated. (i) Mechanism of NO Formation: It is accepted chemistry that the formation of NO from air as a result of combustion does not take place by the simple combination of nitrogen and oxygen, but through a set of interrelated reactions in vhich the predominantly active members are atomic oxygen, atomic nitrogen and hydroxyl radicals. The two reversible reactions indicated below as Reactions (1/2) and (3/4) are referred to as the "Zeldovitch Exchange Reactions", and the reaction between atomic N and OH radicals (No. 5), may be taken as an extension of the Zeldovitch mechanism. TABLE I -75,400 1A RT J>\^>\^ (1) N2 + 0-»N0 + N ; ki = 1.36 x 101* exp. -334 (2) NO + N ^ N 2 + 0; k 2 = 3.1 x 10 13 exp. f*$? Rt -6,250 9 RT (3) 0 2 + N-*N0 + 0; k3 = 6.43 x 10 x T exp. -38,640 (4) NO + 0->0 2 + N ; k 4 = 1.55 x 109 x T exp. RT 13 (5) OH + N-»N0 + H; k = 4.101"3 5 f>-7 J>Lo yJ f/> S ^ Units are cm.^ mole.""* sec.""-: with R in calories The reaction-rate constants shown in Table I are extracted from "High Temperature Reaction Rate Data" No. 4, 1969, by the Department of Physical Chemistry at Leeds University. The data on these reactions has been re-evaluated for general publication (Ref. 10) where it is presented slightly differently and with same marginal variations, but the Table I values are well suited to present purposes and very conveniently assembled. To provide something tangible in this rather complicated situation, values of the reaction rate constants have been computed for a particular temperature, which has been taken as 1,600° C. to be fairly high in an industrial context, but not unusual. They are shown in Table II. - 3 It might be helpful at this point to make the reminder that the reason why it is necessary for the reaction to follow this indirect and perhaps seemingly complicated route, rather than formation by direct combination of N and 0 atoms, is that in the case of two directly colliding atoms, no reaction to form a diatomic molecule occurs. If a molecule were formed at the instant of collision, its lifetime would be very much shorter than the time between collisions. This is because the sum of the energy of the reaction and the relative kinetic energy of the colliding atoms is more than enough to dissociate the molecule. The atoms will, therefore, fly apart again unless a third molecule participates in the collision, to remove the excess energy; or as in the present case the collision is between an atom and a diatomic (or poly-atomic) molecule, where the energy can be distributed among more than one degree of freedom. TABLE II Zeldovitch Reaction-Rate Constants at 1,600° C. 5 3 k^ = 2.5 x 10 k 2 = 2.8 x 1 0 1 3 cm. " per gm. mole per second " " " k3 = 2.28 x 1012 " " " " k 4 = 9.62 x 10p k 5 = 4 x 1013 ti it it it it it ti »» it it f| ft ,| ft It It 9 It will be observed that the k value computed for Reaction 1 comes out to a relatively low figure compared with the reverse reaction (k 2 ), but it must be remembered that the mole fractions of N 2 and 0 in the system are, respectively, several orders higher than those of NO and N, which are the components of the reverse reaction. Consequently, although Reaction 2 is known to be rapid, and its activation energy close to zero, the result is that the NO and N produced by Reaction 1 is not all immediately destroyed by Reaction 2. Moreover, what is destroyed regenerates atomic 0, available to react again with the massive concentration of N 2 as Reaction 1, a most important source of N atoms. Looking at Reaction 4 which, again, is intrinsically slow and also takes place between components of low concentration, it is evident that it will be followed by the much faster Reaction 2, and therefore Reaction 4 will not do much to increase the N atom population directly, although Reaction 2 will regenerate atomic 0, able to participate in Reaction 1, which does do so. Since k 3 » k^ it might be imagined that reaction (3) will be the dominant reaction producing NO in high temperature air i.e. combustion with excess air. However the concentrations of the species involved in the reaction need to be considered, particularly the concentrations of the N and 0 atoms. These atoms are produced by the thermal dissociation of N2 and 0 2 at elevated temperatures. Molecular nitrogen is thermally dissociated at a much slower rate than oxygen and an outcome of this is a significant and growing population of 0 early in the reaction even at moderate temperatures, while the N population remains small. The high concentration of 0 relative to N may be sufficient to offset the disparity in rate constants of reactions (3) and (1), and certainly, by the time a steady state concentration of N and 0 atoms is achieved, both reactions (3) and (1) are contributing to the rate of NO production. But if nitrogen-bearing molecules are present in the fuel, a new source of N atoms is available. As each nitrogen-bearing molecule is broken down by pyrolysis, new N atoms are released. They are not in thermal equilibrium with N 2 at the temperature and will progressively become unavailable for direct reaction, but they can only combine to N 2 as they find and collide with suitable molecules , such as NO, and while they continue to exist as free N atoms they participate in Reactions 3 and 5, which result in formation of NO. It is this that accounts for the large increase in NO formation which occurs with nitrogen-bearing fuels. The chemistry is known to be less simple than this, with intermediates such as CN and NH as well as N atoms appearing en route to NO, but the simpler concept is easy to follow and the end result is the same. (Ref. 1.) It is possible to discern a second aspect of this mechanism when note is taken of the variation in NO concentration resulting from combustion of fuels with increasing molecular nitrogen content. (Reference 2.) The amount of NO formed increases, but it does not increase in proportion with the nitrogen content of the fuel. It is shown in Fig. 1. that when the N? content is modest, a high proportion is converted to NO. With a relatively high N 2 content, although the amount of the NO increases, the proportion of nitrogen converted is lower. These results indicate that it is the 0 atom population which is now limiting. - 5 - (ii) Sub-Stoichiometric Combustion: In different circumstances, when a fuel is burnt with less than theoretical oxygen, it could be expected that the absence of free 0 2 would preclude the formation of 0 atoms at all, with consequent disappearance of Reaction 1, which is apparently the most important source of N atoms, and simultaneous disappearance of NO altogether. This does not occur; Reaction 1 is, indeed, reduced in rate proportional to the concentration of free oxygen, but this is not zero because the C0 2 present is partially dissociated-C02 + M ^ C 0 + 0 + M, whereM represents a colliding molecule causing dissociation of the C0 2 by transfer of energy. It is also the case that the population of N atoms directly attributable to thermal dissociation will be hardly less than before, because the temperature will remain high, so NO will continue to form although at a quite rapidly declining rate. None the less, if really low NO levels are being sought, it will not be until the ratio is down to around 0.65X or lower, that this result is achieved. Table III shows equilibrium mole fractions of 02, 0 and NO, determined by Craig & Pritchard, for combusting mixtures of H2 and air, at 251 excess air, stoichiometric air and 0.83 of theoretical air. (Ref. 3.) TABLE III Mole Fractions at Equilibrium 02 0 NO 6 x 10-3 251 excess air -4 3.6 x 10-2 2 x 10 Stoichiometric air -5 3 x 10-4 1.5 x 10-3 7 x 10 -5 6 x 10-6 83% of theoretical air 1.0 x 10 5 x 10-5 Figure 2 shows the variation with time of the mole fractions of 0 NO and N determined by Craig and Pritchard. The high peak in atomic oxygen concentration developed at the instant when primary combustion occurs, declining only slowly to the eouilibrium level, is very well shown in the curves; and associated with the same phenomenon are the exceedingly high - although transient - growth rates in atomic nitrogen and nitric oxide concentrations. - 6 - (iii) Staged Combustion: It has been recognised in many published papers that the formation of NO may be reduced by the use of "Staged Combustion", but it is important to observe that, by itself, staging of the combustion is not an effective means of thermal NO reduction, even though some small low-fire trials create the impression that it is. (Ref. 4.) Imagine that oil is being reacted with air in an autothermic gasifier, with 50% of stoichiometric air. The gas contains CO and H2, and some free carbon, with CO2, H2O and N2 making up the remainder. The temperature of the gas is between 1,250 and 1,300° C. The amount of NO present at this stage is low, for all the obvious reasons. But if this gas is now burnt with excess air, the combustion temperature will be just as high as if it had not been through the gasification stage, and there will be no reduction in thermal NO formation. However, with nitrogen-containing fuel the situation is different. (Para. 6, p.13.) The essential retirement, with staged combustion, is that heat should be extracted from the made gas before it is burnt to completion. In small installations, particularly on low fire, this happens incidentally, but some large boilers have been fired with oil-rich burners at the bottom and excess air burners at a suitable height above, in order to produce the result. The substoichiometric flames radiate heat to the boiler walls and their gases are partially cooled before they meet the excess air from the burners above. The whole combustion process therefore takes place at a reduced temperature and less NO is formed although the same overall fuel/air ratio has been maintained. (Ref. 5.) There are difficulties in the method, especially so when it is adopted as a means of avoiding NO formation during pyrolysis of nitrogen-bearing oils, because to succeed in this the air : fuel ratio must be low and carbon deposition, soot, smoke and particulates, are obvious problems. But there are other less obvious problems: the effects of greatly enhanced emissivity due to flame carbon, and the reducing effects of CO and H2 on the refractory material of the walls. The, high emissivityjcan easily double the ra^e_ofj»adjiat-ivp hggt;transfer in the ^h-stmch^Pt^iig-^TiPj v-fh ^jgg^^g^ g-rftt overheating'of pipes in this region. And at wall temperatures which are easily sustained under normal excess oxygen conditions, silica in the refractory is reduced to the monoxide, SiO, which is volatile at 1,200° C. and disappears from the scene, leaving l the walls porous and friable, and eventually disintegrated. - 7 - Effective precautions against these happenings are quite essential. They include management of radiative effect by control of flame diameter, the use of properly chosen refractory materials and air-washing of the refractory walls. (iv) Rate of Formation of NO as distinct from Equilibrium: It is the rate of formation of NO from the oxygen and nitrogen engaged in the combustion process which governs the amount produced in a given time. The alternative presentation, namely that the amount of NO present in given conditions of temperature and 0 and N atom availability depends essentially on how long the temperature is maintained, is sometimes overlooked. It accounts for certain apparent disparities in published results relating to combustion of similar fuels at the same excess air levels and nominal flame temperatures. For example, when a flame is large the loss of heat by radiation from the centre of the flame is screened by the gases surrounding the centre. The outermost gases radiate successfully to the walls of the furnace, but the central gases can only radiate to the gases immediately surrounding them, which. are radiating back almost as much. Consequently, when the flame is large, gases spend a longer time at a temperature near their maximum, *&ich is also the maximum NO formation condition, and Figure 3 shows how high the concentration can become if the time at temperature is long enough for equilibrium to be reached. - 8 - Conversely, NO concentrations in products from small burners - radiant dish burners are a good example - may be unexpectedly low at a given flame temperature and excess air condition. r These points are illustrated by Figs. (4) and (5). Fig. (4) shows NOx increased by firing rate, partly attributable to flame size; and Fig. (5) shows the two effects (a) increased NOx production due to the increase in flame temperature produced by air pre-heat; and (b) a reduction in NOx production when heat is lost from the flame more qiickly in a larger cross-section furnace. The same result is observed in a cold furnace, with the rate of NOx production increasing as the furnace heats up. Fig. (6). INTERIM NOTE ON NO, N0 2 AND NOx The designation'NOx'has been used in the last paragraph, whereas previously reference has most frequently been made to nitric oxide, NO. Use of 'NOx1 conforms with the designation adopted by the authors responsible for Figs. (4), (5) and (6); and where NOx is used earlier in the paper, to stay in line with the current widely used designation. But the oxide of nitrogen produced in normal industrial flames is almost exclusively NO; the higher oxide, NO2, is decomposed at normal flame temperatures and is barely detectable at the mouth of a combustion chamber. It is formed later, by reaction of NO with atmospheric oxygen, at a rate which increases progressively as temperature falls. The velocity of oxidation is given by - ctPh/o ^ kPfJQ Po, (2N0 + 0 2 = 2N02) at * where P is in mm Hg and t is in sees, and the velocity coefficient at different temperatures is: (Ref. 6.) t° C 0° 25.2° 85.5° 100° 150° 104 x k 10.63 8.73 5.69 4.8 3.35 The significance of this point has recently been enhanced by the newly developed awareness that NOx contained in directlyheated air used for drying foodstuffs, can react with amines associated with the foodstuff proteins, to produce very objectionable impurities. The chemistry of the reaction with amines requires N0 2 . By itself,NO does not appear to react. Therefore a description of burner gases indicating that their N 0 2 content is 1 or 2 ppm could be mistakenly assumed to mean that when diluted these gases would be quite safe for direct drying of foodstuffs. This is dangerously wrong. It is inevitable that combustion products containing 1 or 2 ppm of N0 2 , at the burner mouth, will contain a very much higher concentration of NO which, in transit to and through the foodstuff after air dilution, will be largely oxidised to N 0 2 which will then react with amines in the foodstuff. - 9BURNER DESIGN FOR LOW NOx: The two new industrial requirements for minimisation of combustion NOx mentioned in the Introduction, particularly the first one demanding 1 ppm of NOx in products at the burner mouth, have necessitated some reappraisal of accepted thinking on combustion theory and burner design. The 1 ppm NOx problem will be considered first. DESIGN CONSIDERATIONS FOR A 1 ppn NOx BURNER: Fuel: It is obvious from the technical material which has been reviewed that it is futile to look for 1 ppm NOx combustion products with a fuel containing appreciable combined nitrogen. Although gaseous fuel is not indispensable, it is the most convenient, and is the only fuel being used for this application at the present time. Therefore, natural gas, L.P.G. and some types of cleaned coke-oven gas, may normally be regarded as suitable. Combustion: The combustion requirements, given clean gas, are low burning temperature, low oxygen content and short duration of upper temperatures. These are not mutually consistent in normal excess-air combustion and compromises are necessary. However, staged substoichiometric combustion with inter-cooling is capable of providing the conditions. SUB-STOICHIOMETRIC METH3D: When natural gas is burned with about 50% of stoichiometric air in a suitably designed combustion chamber, the gases contain very little free carbon and their final composition is approximately as shown in Table IV, at a temperature of about 1,350° C. TABLE IV % Vol. N2 H20 H2 CO co 2 AI Enthalpy at 1,350° C. 58.5 43.7 x 10 3 K.cal. 12.3 11.5 x 10 3 " " 15.4 10.8 x 10 3 " " 8.9 5.0 6.8 x 10 3 " " 6.0 x 10 3 " " 100.0 10 3 K.cal. = 4.187 MJ. 78.8 x 10 3 K.cal. Volume percent of products from 0.5 X CH^/air combustion; and enthalpy at 1,350° C, per Kg.Mol. of CHA. fAverage So.h = 0.34.1 - 10 - Rate of Nitric Oxide Formation: A rigorous theoretical estimate of the rate of formation of NO under these conditions is complicated and involves a number of simplifying assumptions. But an expression for the rate of production of NO which considers Reactions 1-4 in Table 1, is derived (Ref. ll)by Sridhar Iya by taking the algebraic sum of the rates of the 4 reactions and assuming that the concentration of N atoms has settled to a steady state. The expression simplifies, for small time intervals, to d [NO] = 2kx [N2][0]. eft Using this equation with an estimated value of 3.2 x 10"8 mole fraction of [0], based on extrapolation of the data produced by Craig and Pritchard, and allowing 100 times this level for a duration of 10- 2 sec.for the transitory much higher than equilibrium concentration of oxygen atoms produced early in combustion, the mole fraction of [NO] produced in 0.5X CH^air combustion in 0.1 sec. is 0.003 ppm and in 0.5 sec. is 0.004 ppm. The omission of Reaction 5 from this calculation, with excess fuel conditions, is justified by the fact that in these circumstances there is heavy competition for OH as well as 0 and 0 2 . The sub-stoichiometric burner assembly will be designed to keep the time the gases spend at high temperature reasonably short. The target is 0.5 sec. which is a reasonable expectation with a turner of sane size. The method is to put the sub-stoichiometric products directly into an air heat-exchanger, diluting efficiently and quickly before entry, with recirculated products at a temperature which has been reduced from 1,3500 ex burner to 500° C. The cooled products finally leaving the heat-exchanger are then to be turned with the minimum excess air needed to ensure completion of combustion. Final Combustion of Cooled Products: The mixing which takes place immediately before entry into the heat-exchanger has the purpose of freezing further NO formation and, also, of avoiding excessive gas temperature at the entry to the heat exchanger. Approximately equal quantities of 1,350 and 500° C. gases will produce 900° C. gases at the entry, quite acceptable with suitable materials of construction. The products (from CH4 + 5 Air) cooled to 500° C. contain 27 Btu/ncf (l.OJ/cm3) of sensible heat and 73 Btu/ncf (2.72J/cm3) of calorific heat: a total of 100 Btu/ncf (3.73J/cm3) when completely burned. Assuming 1.15 volumes of air for combustion (7J-10% excess air), the temperature will be about 1,100° C. The rate at which NO is produced at this temperature is low and the end-gas, before dilution with fresh air for use in drying, will contain no more than 1 ppm of NOx, virtually all NO. - 11 - Combustion Stability: It is important to remember that it is much more difficult to maintain stable combustion with a flame temperature of the order of l,100o c. than in normal gas/air ratio situations, which produce substantially higher flame temperatures. With a low c.v., low temperature flame such as this, it is particularly necessary to retain the heat where the combustion is taking place, for long enough to avoid quenching. On the other hand, the requirement for low NOx calls for as short a residence time as possible in the upper temperature environment. The compromise in this present case is shaded towards stability, because inadvertent lock-out quickly becomes intolerable. Ultimately, it is this which governs the achievable minimum in combustion NOx generation. Several geometrical forms for burners or burner-jets have been described for maintaining flame stability, particularly by the Gas Board's Research Departments in connection with natural gas flames; and proprietary components based on these principles are available. The choice by the designer depends on burner capacity, in the sense of total heat-release; and on individual burner-jet size, depending on whether large numbers of small flames are used in repeating arrays, as in "In Duct" type burners,or a relatively small number of burner-jets is arranged in a chamber; or whether a single gas/air feed in conventional style into a combustion enclosure is used. A repeating characteristic in low-temperature flame stabilisation is the aim to ensure preservation, by gas-velocity control and heat feed-back, of a small proportion of the total flame, for example as a corona, whereby continuous ignition of the main body of mixed gas is sustained and lift-off is avoided. It is difficult to be more specific than this, because the details with low flame-temperature burners of different configurations and size are so varied, but the geometrical forms described in Gas Board papers provide a selection of ideas for application to the purpose. HIGH EXCESS AIR METHOD: The alternative to two-stage combustion with inter-stage Cooling of the gases, as a means of obtaining low combustion temperature, is single-stage burning with very high excess air. It is obviously impossible with this method to have a low free oxygen concentration, so if it is to succeed the combustion temperature must be so low, and the duration of the top temperature pihase so short, that the combined effects of rate of NO formation and time, result in low overall production of NO. The limiting temperature, of course, is set by the lower limit of inflammability of natural gas with air. Unfortunately, at near-limit conditions the speed of burning is not very high. This creates the problem that if the combustion is inhibited too early by dilution with the air which is being heated by direct admixture, the effect is to freeze the intermediate combustion products, among which formaldehyde is prominent. This is objectionable, not only because heat is lost, but also because of the irritant effects of formaldehyde on eyes, nose and throat. - 12 - It is evident, therefore, that some basic combustion difficulties associated with this method are more severe than those of the sub-stoichiometric method. On the other hand, if they can be overcome the equipment may be simpler. In particular, the need for the inter-stage heat-exchanger is avoided. But there is no escape from the practical requirement for stable combustion over a reasonable turn-down range, and this is unquestionably difficult with the combustion over the whole range close to the limit of f lammability where stability can so easily be destroyed by minor fluctuations in mixture strength. The classically accepted lower limit of flammability of methane/air at ambient temperature is 5.3% CH4, (Ref. 7 & Fig. 7) which relates to a mixture of slightly less than 18 :1 air to gas. The "no loss" temperature of combustion at 5.3% CH4/air, if North Sea gas is substituted for CH4, is about 1,375° C , lower in practice, and no designer would normally choose to work at a limit of inflammability. It is not possible, with reaction rate data at present available, to produce a generalised calculation of the amount of NO which will be formed under these conditions, for the two reasons: (i) that the transiently high oxygen atom concentration which is always developed at the instant of burning will decline more slowly at the lower temperature, because the burning is itself drawn cut longer; and (ii) the amount of time needed for completion of combustion (e.g. on account of formaldehyde) is necessarily "time at temperature", during which NO is being formed at a corresponding rate. The former, while it persists, will greatly increase the rate of NO formation, above the rate corresponding to equilibrium of [0] at the temperature: the latter will govern the time for which the temperature must be held before "freezing" of NO formation by sudden chilling of the products. Therefore, the procedure is first to define the size of burner which is to be built: (a) in terms of heat release, and (b) of proportionate physical dimensions, which will be largely governed by the designer's choice of flame stabiliser geometry, referred to earlier. Next is the choice of system and components for controlling the gas/air ratio, so that neither an over-rich mixture, which will produce gross NO, nor an under-limit mixture which will lose the flame, is allowed to occur anywhere over the desired turn-down range. Then the requirements for ignition, flame monitoring, Gas Board regulations and automatic control must be provided for, and brought into even an early experimental design, because they are liable to affect ultimate functioning on account of the extreme sensitivity of combustion near the flammability limit. If these precepts are followed, a burner system capable of stable functioning and any reasonable turn-down range, with NOx concentration never exceeding 1.5 ppm at the burner mouth, and rarely exceeding 1.0 ppm, is entirely practicable. DESIGN CONSIDERATIONS FOR HIGH NITROGEN FUEL BURNERS The second new industrial requirement mentioned in the Introduction relates to NCbc in flue gases discharged from large steam generators burning locally available crude or heavy residues, with high combined nitrogen. The steam is used in very large quantities, in thermal recovery methods for oil of so highly viscous a character and such affinity with the rock or sand strata in which it is distributed, that pre-heating is probably the only way of getting the oil cut of the ground after the first gush. It is of seme interest to recount the essential features of the method. Steam is passed down the well for a period which might be a week, or a month, or more - depending on the plan decided by the Reservoir Engineer - for the purpose of heating up an area around the well, to reduce the viscosity of the oil contained in the porous rock. The condensed hot water penetrates through the pores and, in favourable circumstances, is able to travel along "fingers" surrounding the well, resulting in an extended distribution of the heat in the rock. The steam flow is then discontinued for a period, which may be of the same order as the flow period, to allow the heat to penetrate. The sequence is referred to as Puff and Hiff! Now the normal production procedure at the particular well-water or gas injection, or simple pressure depletion - whatever it may have been, will be resumed, under conditions rendered more favourable by the heating. The method has found particular application in certain areas of California where, in sane cases, residential areas have grown up in the vicinity of these "lazy" but unspent wells. The number of these installations, existing and contemplated, is large, and anxiety regarding serious contamination of the local atmospheres and prospective formation of smog, has resulted in the imposition of limits on the amount of NOx which may be discharged, which are not easy to meet economically when burning high-nitrogen fuel in a large installation. Some factors affecting NO formation in combustion of nitrogenbearing fuels have been mentioned earlier and three points of importance may be recognised here. First that the reaction rate with fuel nitrogen is much faster than with atmospheric nitrogen. De Soete showed (Ref. 8.) with NH3 added to hydrocarbons, burned in nitrogen-free "air" (using Helium), the reaction was complete, and the NOx level higher than the equilibriun value, within 2 milliseconds of the passage of the flame front. Second Thomas & Shaw (Ref. 9.), high NOx levels are produced at low combustion temperatures (c. 850° C.) with organic nitrogen compounds in the fuel. The experiments were conducted with CO as the fuel, and the nitrogen-free simulated air contained 79% argon and 21% oxygen. Third, Turner et al, that reduction of NOx by flue-gas recirculation appears to be due exclusively to reduction of thermally produced NOx. (Ref. 2.) -14 - The conclusion which may be drawn from these three, and earlier points, is that probably the most positive way of effecting a reduction in NOx specifically attributable to nitrogen in the fuel, is by way of two-stage combustion with inter-stage cooling. The sub-stoichiometric phase then creates conditions under which the nitrogen atoms produced by pyrolysis of the fuel very rarely encounter O2 or OH molecules, which would produce NO by Zeldovitch Reactions 3 and 5; and instead the atomic nitrogen reverts to N2 by other routes. However, in the second combustion stage, if the first stage gases cannot be sufficiently cooled to produce a low final combustion temperature, thermal NOx will be produced as usual. Therefore, to minimise this effect the known procedure of recirculating partly or completely expended products of combustion into the secondary burning zone should be arranged. These are the principles upon which the low NOx firing system applied to thermal recovery steam generators has been based, and sane details are given below of the combustion arrangements and engineering construction which have been employed. GENERAL FEATURES OF DESIGN: The firing system developed for the purpose has sometimes been referred to as employing "High Intensity Burners" because of the contrast with conventional natural draught or open-flame burners. The combustion intensity actually used is designed to be just so high as to provide the conditions in the heater enclosure which the system demands. The aim of the system is to put the maximum possible amount of heat into the water without burning the tubes. This requires that each square foot of heat transfer surface, from one end to the other of the radiant section, should be uniformly fluxed up to its stipulated "circumferential average" maximum rating. It entails accurate and comprehensive matching of heater and burner together, to contrive that heater configuration, dimensions and heat-transfer loading are matched by the burner throat diameter, combustion gas entry temperature and velocity, and total heat release. When this matching is correct, low excess air gases leave the radiant section at about 800° C , well suited for transfer to a compact convection section. The heater is then capable of producing substantially more steam - 20-25% more has been claimed at higher thermal efficiency than has been practicable with preceding systems of equivalent dimensions; and NOx levels"of only 100 to 150 ppn (corrected to 3% excess O2) are reported with 0.8% N 2 in the crude or residual oil fuel being burnt. A single burner is used, commonly of 60-65 M Btu/h (17.5 -19.0 MW) heat release, firing axially into a refractory lined cylindrical chamber sane 40 ft. in length, with closely spaced longitudinal tubes near the walls, at 9 ft. centres. The gases leave the radiant section at about 800° C. and leave the convection section at about 200° C. - 15 - COMBUSTION DESIGN DETAILS: The fuel is burned sub-stoichianetrically in the combustion chamber with 65-70% of theoretical air and the partly burned products are discharged through the chamber throat in a relatively narrow forward stream into the heater enclosure. Additional air, sufficient for completion of combustion plus a small excess, is brought in at the throat as a peripheral stream, and the combined forward velocities produce the amount and direction of momentum defined by the system calculations. The momentum is specifically axial. Angular momentum is forbidden: it interferes with the designed flow pattern and creates an unwanted circulation pattern of its own. It means that the required degree of combustion intensity must be developed without reliance on swirl. Completion of combustion, in which the remaining H 2 and CO and any elemental carbon from the first stage are burned,takes place in the fast moving efflux gas column leaving the chamber throat. Certain important mechanisms come into play during this process. The forward momentum of the column enables it to entrain, and give away heat to, the spent and already cooled products drawn back from the heater enclosure. (Fig. 8.) At the same time, the column of gases, consisting of hot, highly emissive products, radiates at a high rate to the waterfilled tubes and is, itself, rapidly cooled. And the result of these two happenings is that combustion of the unburned gases from the first stage is completed near stoichiometrically in the forward-moving column, without ever exceeding the temperature of the gases leaving the chamber. In this way the effect of inter-stage cooling, needed for minimisation of thermal NOx, is achieved automatically; and the substoichiometric conditions of the first stage, inside the chamber, which inhibit direct reaction of the nitrogen atoms liberated during decomposition of the oil, prevent most of the gross NOx production normally associated with combustion of high nitrogen fuel. HEAT FLUX UNIFORMITY: Complete uniformity of heat flux along the whole length of the tubes, which is the second feature the system aims to produce, depends on the compatibility of the flow pattern developed by the forward momentum of the entering column of gas, with the dimensions of the heater enclosure. The momentum (which drives the recirculation) is proportional to the mass flow rate and the axial velocity of the products entering the system through the combustion chamber throat. The mass flow rate is determined by the heat release demanded in the heater, and the velocity by the area of the throat and the momentum added in the annular air stream at the throat. The emissivity and temperature, which govern the radiant flux from the central column of gas, are determined by the degree of completeness of combustion in the chamber. - 16 - r The first phase of the forward travel of the column has a critically important role in securing the desired overall uniformity of heat flux over the full length of the radiant section. The water tubes at the entry end of the heater see the column at its maximum temperature and emissivity. The recirculated gases brought back frcm further along the heater are cooled as they travel, by the water tubes. Those reaching the entry end are cooled the most. Thus, the initially high radiant flux from the hot central gases compensates for this, delivering at a maximum rate near the entry end, and the rate progressively declines as the column is itself cooled by radiation and entrainment, while the temperature of the surrounding body of gases increases. The gas flow pattern as a whole can be understood by reference to Figure (8) which relates to a 60 M Btu/h heater (17.5 MW) designed to be in ideal match with the combustion chamber driving it. As momentum is continuously exchanged during forward travel of the entering jet of gases the velocity of the stream, progressively enlarged by entrainment, declines in proportion to the increase in its total mass (momentum is largely conserved during this phase) and the cross-section of the stream is correspondingly increased. This proceeds until the widening stream is buffered by the slow, plug-flow gases moving out of the heater. The section where this occurs may be referred to as the "neutral plane" because sane gases go forward and out to the convection section and the rest travel back towards the entry, to recirculate again. It is marked X-X on Figure (8)In a well-designed system it should be at about 80% of the full length of the heater enclosure. A concept which may be found useful is to regard the plug-flow zone as a reservoir, continuously fed from the entry end , with N masses of hot gas, in the present case reaching it at about 800° C, and discharging one mass, which is equal to the inflow from ^the burner, forward to the convection section and (N-l) masses rearwards in response to the reduction in pressure generated by the burner jet. This means that the entering jet entrains (N-l) masses of gas, and this amount is specifically related to the heater crosssection and the jet diameter. - 17 - TRIALS WITH A STEAM GENERATOR Results obtained in a compliance test by California Air Resources Board (CARB) in firing trials carried out on a 50 M Btu/hr (15 MW) steam generator using this combustion system, is shown in Fig. 9. The curves relate well to theoretical expectations and provide an excellent illustration of the functioning of the combustion system which has been described. When excess oxygen (i.e. flue-gas 0 2 %) is zero or very low, the system delivers about 70% of theoretical oxygen into the combustion chamber and 30% into the secondary stage. When excess oxygen is 3.5-4.0% the proportionate distribution remains the same, but there is now 85% of theoretical oxygen in the chamber. The effect of this on 0 2 and 0 concentrations inside the chamber is very significant and explains the higher NO levels shown in the curves. If the figures in Table III, from Craig and Pritchard (Ref. 3.) are examined, it will be seen that equilibrium concentrations of 0 2 and 0 decline by times 150 and times 10 respectively between conditions of 100% and 83% of theoretical air, and although corresponding reductions in 0 2 and 0 concentrations at 70% are not quoted they will again be large. It is this big change in the availability of oxygen for reaction with atomic nitrogen which produces the rapid increase in NO concentration shown in the CARB curves between 0.5% and 3.5% flue-gas oxygen levels: a much greater increase than would occur with thermal NO alone, for the same excess oxygen change. The underlying point, which has been referred to earlier, is that if a nitrogen-bearing fuel can be pyrolised under relatively oxygenfree conditions, the nitrogen atoms rarely encounter 0 2 or OH molecules with which they would produce ND, and instead the atomic nitrogen reverts to N? by other routes. This is, indeed, the reason for using a sub-stoichiometric stage for combustion of high-nitrogen fuel. But the second stage of combustion, in which the recirculation of spent products is imposed by the forward movement of the fast-moving column of hot gases leaving the combustion chamber, is also very relevant, in this case to the control of thermal NO which is produced in the second stage. The two effects produced by the system, recirculation of spent products and high thermal emissivity of the forward column as it leaves the combustion chamber, combine to produce a moderate temperature of combustion in the second stage where the 00 and H 2 produced in the first stage are consumed. At the same time, the effect of the recirculation on hot gas radiation to the water tubes is to temper the high radiant flux from the central column during its early travel, and to enhance the flux further along the heater enclosure, by the evening-out of gas temperature which the recirculation achieves. In this way the radiant flux to the tubes is rendered very nearly uniform from end to end of the heater. ACKNOWLEDGMENTS It is a pleasure to acknowledge the generous assistance of Dr. D.R. Baulch of the Department of Physical Chemistry at Leeds University, and of Dr. Paul Eisenklam of the Department of "Chemical Engineering at the Imperial College of Science, London, vho have most kindly provided expert correction of inaccuracies contained in the original text. LOW NOx CONCENTRATION REFERENCES Fenimore, C.P., Combust. Flame 1972. 19. 289. Turner, D.W., Andrews, R.L. and Siegmund. Ch. W. Paper 28E. at A.I. Ch.E. (San Francisco) Nov. 28, 1971. (64th Annual Meeting.) Craig, R.A. and Pritchard, H.O. (NASA). Presented at the 1972 Spring Technical Meeting of the Central States Section of the Combustion Institute, Oklahoma. March 21, 1972. Turner, D.W., and Siegmund. Presented at Am. Flame Res. Ctte. Flame days, Chicago, Sept. 6-7, 1972. Bartok, W., Crawford, A.R. and Piegari, G.J. A. I.Ch.E. Symposium Series 68 No. 126, 66. 1972. Wburtzel: Ccmpt. Rend. 1920. 170, 109, 229. W.A. Bone and D.T.A. Townsend: "Flame & Combustion in Gases". pp 109-112. (Longmans). De Soete, G.G., Formation and decomposition of nitric oxide in combustion products of hydrocarbon flames, presented at "American Flame Days", to An. Flame Res. Ctte., Chicago, Sept. 6-7. (1972). Thomas, J.T. and Shaw, A.C., Conference on Coal Science, Prague, June 10-14 (1968). See: Blokker P.C., Section 2.2 Report No. 4/75. Stichting CONCAWE, The Hague, March 1975. Baulch, Drysdale, Home and Lloyd. "Evaluated Kinetic Data for High Temperature Reactions." (Butterworths, 1973). K. Sridhar Iya: Hydrocarbon Processing. Nov, 1972. Vol. 5i(II) Richard T. Waibel, "Recent Research on Nitrogen Oxide Emissions." Institute of Gas Technology. No. 47. Fall 1979. Chicago, U.S.A. MARK LABORATORIES LTD. O'Z C'3 FUEL 0*9- &G NITRC6EN(%) TURNER. AMt>iZ£h/$ FIGURE I O'S £.£IE6WUMD &EF.2 PAGE OF o>7 O'S icf* lo b/STAHCE. CM X lo loo 1000 16,000 TIME'* LLSBcS • • / " FIGURE £. (/g^ggg£gf P^QM R^^.z) SPCClfS'MOLE FRACTIOUS. EQUILIBRIUM CoNCEtfT«flTlOf/S AT FiHALCONOntoHS tNfilCfiTEb B>* fiRROHS ^ 0*0-8 (fveL LEAH) _ <3 <f)-h2(Fw RicH) (h^£6i<MVfi,LBNC MARK LABORATORIES LTD I 4$oor PAGE i 4-ooo 3Xoo VJ 2440 f6oo or IZOO WO* fS °° EQUILIBRIUM F!6,UR£ 3 <(,C ° No /7 *° f8 °° £6NC£NTMT(ONS i ^°° *°co *fOO C\\+/t\\R* 22oo R*F.2 OF \ i ' i ' l M 1Mi l ' ' 1 i ,1 M ! r-f : ' ' K& ' "i JE** /ffiA - AIR-PK€H£AT TEHPS% r i ;1 i i1 i1 1 • 1 ' i 'li' i ; i '! ! . .. MI 1 ; 1 ; • ' ' : '/ ; 1 / 1 • i i 4*L y 1 *:i I^ U'^ ' >7' r3^4*' LJ IA i i '/ /, Si" 1 / Ta 1 ! T/| 1 1 ' /'A g ,, .f tjscsaB S4*' / i; i J / ! L St^T Bl M •' _1* '1 1 ii c f 1 1!>*T! t A., • ' «?? ^ rfi >T^ £*--] Jll 1 i 1! P" 1! | 1 i i . j f BJO • 1 f • s4 !o ! r |I | 1 ! IXIC, * 1 - . ', TI' ^ ^ (IM. Rtu./6f.=0'293**W) M1 i i i i! 1 _ IEI I4I 1 , •: IX mi • ] 4 *"j ;« . 7 L I'M "' r - p WJ n^Qfljf11 f Hit Tirgg^^*: fill |f, Mi ] 1 j FIGURE 4. # 0 * AS A FUNCTION OF £KC£SS AlR _AT TW<? f/*W$ RAf£S .(LIT* */ft FlUH&i) 3S0mC. 12-6 4? S&TM 22°C. ZZ*C. FIGURE 5\ A/0* N'TH f <?l/fl<- FlRtslC, Kfirei Mb WALL TeHPe*tmme esr biFF£*£MT cko&rSECTION. YAIURTIOWS f/6S 4.k£ l tiCHAlb T. >/A»/a£i.: 7.6.7: A/p. 47. / 3 70. (&*£ -t+W+H °4^4+aOC? ! lit K2C *>^J?sa^-t.c± • •in]-r? r h^ ' / -i-!-(-'- -i I w a i |P» »--•• ; .j -j i i -r I 1 ! ! I • : i / • •i;1 '•! 1 i .-1~ o • 1 i • - i T S1« 1 i :i ! I /' ' ' , fi •JJ-; ' • / i ! !i :W^ "51- 6 • f i yfei I r* 1 | | i <£l j tl i ,_..!„. 1 •wi • i - rH 1 t ,' 1 1 !- i - 1 ! 1?-!-i i 1 T 1 It 4- I / i: M ^ > i -'-H i ! . ..1 ' i - i• +i -Hi 1 ! j ! _ i |, *. 1 ii Ii ! i 1 1I i 1 *|| fr i i T % | M •« 1 ' 1 , T"i£j5^ j ?a«* 3 J 1 A J 1 ins* . uh-f- V i 1 Kr*1] T II 1 1 i n 1 FIGURE 6. NO* vdKitrnoHS tutu NO FVAN#CE HERT-UF. RT.HIUBEL.l.GT Ho47. 1919. X X X II hi ' X X' ! Ml ' 111 M 1 1 ^UkU 1 it it ' <' ¥c- I I ! a T -g>- : XX k Xlt' i XX ^Si M a J ' 1 '1 - 1 -~ 1 , i 1* •g&O f-?-i ! i , M 1 M M - j 1 1 ' 'i 1 Ms -4- " » 9 ,; k >° 4" 3 ; 9~~ -'-r + T l 1 1 !! j ''' k * x : zz ,. .-4 . -. I 1 1 «• i 1 j. ,,, , _fl ftO . 1 1 |j i 1 1 !' lit i1 M ' °i M i' ! i 4, • ' ' > ! 1 i i ' No 1 , 'lit* 1! T I | '! i 'M M 'i . . 7 i 1 | l-l 11M !i i ; | \! j •: 1 ; : '; • ; IMM 11M iM • 9 II IS / f r C*n& Me*Ann*> in Air i f i iff .X-• 7 FIGURE 7. SPEED Of UNIFORM MOVEMENT OF FLAME" IN" £c+n. DIAMETER Tt//3£. (c+n/S.) T/\ <LOH\/ECTtO*f teCTIOH: PIPES. REf-RRCTORY HULL Fl6UH£ %•• ENTRAINMENT L RECIRCULATION &A^ *rL3bOOM./lr. ail***, r TOROIDAL COMbUSTlOhl CHt\M6EP~. . ^ REPfKobUCEO FROM TECH, NOTE A/o.98 URQU H~fttZT ENGINEER IASC Co. 350 300 250 ZOO ISO 100 - t I.O 2.0 3.0 OXYCEN CONTENT IN FLUE GAS (%) - DRY VOLUME