Studies on the pollutant emission characteristics of a number of oil-fired industrial process burners and the relationship to firing conditions

R P Q P A R C H IJmuiden, October 1980 REPORT STUDIES ON THE POLLUTANT EMISSION CHARACTERISTICS OF A NUMBER OF OIL-FIRED INDUSTRIAL PROCESS BURNERS AND THE RELATIONSHIP TO FIRING CONDITIONS by R. PAYNE, T. AKIYAMA and J. WITKAMP International Flame Research Foundation c/o Hoogovens IJmuiden B.V. Building 3G - 25 P.O.Box 10.000 1970 C A IJmuiden The Netherlands Tel. [phone] Flames Telex 3 5 2 1 1 INTRODUCTION In recent years much effort has been expended in the investigation of techniques for the control of nitrogen oxides (N0X)-emission from stationary combustion systems. Control techniques investigated and applied to date include for example, flue-gas recirculation, off-stoichiometric firing, staged combustion, and the use of specially developed "low N0 X " burners. For fuels containing chemically bound nitrogen, experience has shown that staged combustion is perhaps the single most effective NOx-control technique. Staged combustion is achieved either by separating the combustion chamber itself into two distinct zones and dividing the total combustion air flow, or by appropriate burner design, in both cases to promote the formation of localized fuel rich conditions under which fuel NO-formation is minimized. Whilst generalized NOx-control strategies and burner designs have been developed for the larger utility boilers firing coal, the industrial and process furnaces include a wide range of operating conditions and flame shape requirements which often precludes a general approach. Indeed, in such furnaces the many differences in configuration and firing conditions can influence NOx-emissions independently of burner design, and since requirements for flame quality and turn-down and heat transfer performance can often be more critical there is often greater restriction in the scope for the application of combustion modifications. In the industrial sector the fuels used are in general drawn from a wide range of liquid and gaseous fuels. In the future it is believed that this situation will continue, with perhaps a greater tendency towards an increased consumption of heavy fuel oils (both petroleum derived and synthetic) and a greater emphasis in some cases on flexibility in handling fuels of different quality and type. A considerable emphasis in the near future will however continue to be on the utilization of heavy fuel oils, and this should be seen against the background of increasingly stringent environmental and energy conservation requirements. The experience in pilot_scale development of staged combustion and advanced burner design [l] has indicated that it is more difficult to achieve high percentage reductions of N0 X for heavy liquid fuels than for coals. Evidence exists that the difficulties may arise due to the way in which the nitrogen is bound in the fuel, and therefore the reaction history of intermediate nitrogen species f"2J. More refractory nitrogen in the petroleum derived residual oils may be released later in the combustion process when more oxygen is available for the conversion to N0 X . Additionally, a major limitation on NOx-control from residual fuel oil flames has been observed to be the onset of smoke formation, particularly as the primary combustion zone becomes more fuel rich as the application of staged combustion j 2, 3J. BACKGROUND The work presented herein concerns the reduction of NOx-emission from commercial and experimental burners firing heavy fuel oil and forms part of a larger programme dealing with combustion modifications in coal and oil fired burners for boilers, and oil and gas fired burners for industrial furnaces. Since the work is sponsored by the Government of The Netherlands*, the choice of equipment and fuels represents to some extent the areas of interest and current practice within Dutch industry. A strong emphasis has been placed on aspects related to the potential for the practical application of NOx-controls, where due to the wide variety of equipment employed, it is often necessary to consider each case on an individual basis. Also, for financial reasons, it is usually more desirable to modify existing equipment rather than to carry out extensive furnace modifications or to install new burners. For these reasons it is, in the first instance, important to know what range of emission performance can be achieved through the adjustment of available parameters, and to what extent such changes effect other aspects (e.g. the heat transfer) of the total combustion system performance. In this study a primary objective has been the characterization of a number of commercial burners with special attention to pollutant emissions, flame stability and heat transfer, and to the determination of the influence of simple modifications to burner parameters. In a second step major modifications to these burners have been carried out, in order to determine the possible minimum acceptable emissions levels which might be achieved. Such modifications have included major changes to the burner hardware and also the application of staged combustion. This work has been greatly facilitated by testing the burners in an experimental environment, but with due regard to industrial requirements, where optimization work can be carried out without the need to consider process constraints. The intention of the above approach has been to provide information, in the first instance for the trimming and modification of burners for low NOx-operation in industrial practice. To this end the experimental study is linked with a practical on-site measurement and burner modifications programme, through which it is hoped eventually to provide information on the effects of scale on operating parameters and NOx-emissions. Jointly by the Ministry of Public Health and Environment (VOMIL) and the Ministry of Economic Affairs (EZ) -3- A further major area of interest concerns the translation of data to a wider range of industrial operating conditions and scales, which requires a knowledge of the interaction of the flame with its surroundings. To provide information in this area, a second major objective of this study has been the investigation of the effects of confinement and heat extraction rate, as independent variables, on the emissions of N0 X and solid particulates. This has been achieved by firing different burners in two differently sized furnaces, both operating with two levels of heat extraction. EXPERIMENTAL SYSTEMS Burners Schematics of the burners used during this programme are shown in fig. 1. Burners A to E represent a selection of commercial burners produced in The Netherlands, and commonly employed in industrial water-tube and fire-tube boilers and in a range of process furnaces Burner E Experimental bumer z~ Pcrcllel flow j j)urrer aCfj^iaom twrl Fig. 1 : Schematics of the test burners The remaining two burners were constructed at the IFRF with the intention of providing for a wider range of operating conditions for the interpretation of test results. All burners were capable also of firing either natural gas and/or heavy fuel oil and extensive testing was carried out on both fuels, although results relating only to heavy oil are presented here. For each burner a range of variable parameters was available, the range varying from burner to burner depending upon the flexibility incorporated into the design. Major variables were: swirl level and distribution of the combustion air; fuel atomizer design and the flow rate of atomizing fluid; fuel injector position; operational variables such as excess air. For some burners the combustion air swirl was fixed whilst for others it was possible to vary not only the overall swirl level but also to investigate for example the influence of swirled or axial flow primary air. A wide range of different atomizers was also available, including two types of sonic atomizer, various designs of internal and external mix twin fluid atomizers, and pressure jets. Additionally some atomizers were produced with modified drillings to permit so-called bias firing, and spray angles in the range 50-120° were investigated. Not all atomizers were tested in all burners but generally a sufficiently wide and appropriate choice was made for each burner to allow the investigation of this important parameter. The above mentioned variables fall into the class of simple burner modifications, being parameters that can be easily changed on operating burners. Major modifications carried out included the use of inserts for the modification of combustion air velocity at the burner throat and at the exit, and in some cases complete changes to the air register. For all burners however the possibility to apply staged combustion was incorporated, and this was achieved by the provision of four tertiary air ports positioned around the burner exit. At this point tertiary air was supplied and metered separately and no attempt was made to adapt burner registers for this purpose. The range of nominal burner inputs varied from 3 to 4.8 MW, mostly without preheat of the combustion air, although a selected number of tests was carried out with preheat levels up to 300°C. Furnaces The experimental work was carried out in two different furnace facilities. The majority of the burner charaterization work was carried out in the larger of the two facilities, this being a horizontally fired refractory tunnel furnace having dimensions 1.9 m x 1.9 m square and 6.25 m long. This furnace was equipped with a cooling load comprising a number of axially symmetric watercooled loops, and leading to flue gas temperatures in the range 850-1000°C (corresponding to a heat extraction of 60-54%). At a later stage the cooling load was modified to provide for two separate levels of heat extraction, referred to later in the text as hot wall and cold wall conditions respectively. The smaller of the two furnace facilities had the dimensions 1.2 x 1.2 m square and 4.5 m long. This furnace is also horizontally fired and is made up of separate water-cooled elements. Again two levels of heat extraction were investigated, the first with a completely water-cooled furnace (cold walls), and the second with a lower rate of heat extraction achieved by lining the furnace with a thin layer of refractory. The thickness of this refractory layer was chosen such that for equal thermal inputs the percentage rate of heat extraction was the same as that of the bigger furnace operating under "cold wall" conditions. For both furnaces the flow rates and temperature differences of the cooling water at individual cooling elements was carefully monitored to permit the determination of axial distributions of heat flux to the furnace walls. Both furnaces were also equipped with a comprehensive range of flue gas sampling and analysis systems, which allowed for the continuous monitoring of NO, NOx, CO, CO2, °2' s 0 2 ' s o 3 ' temperature, soot and total solids. Flue gases were withdrawn with sampling probes and passed through a sample conditioning train appropriate to each type of measurement. Fuel The work reported herein was carried out mainly with one type of heavy fuel oil. This oil was taken as delivered from normal fuel suppliers and had the following mean composition: 0.874 kg/kg 0.110 H 0 0.003 0.0030 N S 0.0090 0.0021 Ash LCV 9720 kcal/kg Viscosity . 3500 Redwood sec at 100°F c The oil was preheated to a nominal 110°C prior to delivery at the burner. A limited number of tests have also been carried out with oils containing up to 0.45% bound nitrogen. -7- COMMERCIAL BURNER INVESTIGATIONS Experimental work concerning the characterization and optimization of emissions from the test burners was carried out in the larger of the two furnace facilities. Under nominal firing conditions the furnace thermal loading was in the range 0.15 - 0.2 MW/m^ with flue gas temperatures, at the exit of the combustion chamber, of 870 - 1000°C. For each burner a test procedure was followed in which a baseline operating condition was first established, where appropriate burner settings and input conditions were defined by individual burner manufacturers. From this starting point a series of simple burner modifications were then applied, followed by more severe modifications to the burner hardware and the application of staged combustion, and changes in emissions and other flame characteristics were monitored. In this way it was possible to draw up a range of characteristics for the different burners Q.4J. To summarize the results of this exercise in terms of the presentation of a series of general guidelines for burner parameter effects on emission levels is very difficult. It was found that parameters that had a significant influence in one burner, resulted in little or no change when applied to other burners; trends which generally confirm that each burner type must be considered on an individual basis. Certain general trends do however emerge which are applicable to all the burners tested in this study. Under normal (non-staged) operating conditions excess air level was found to have a significant effect on the NOx-emissions, and to allow comparisons to be made all burners were trimmed for operation at 5% excess air and subsequent measurements made at this level. The total range of NOx-emissions measured for all burners and all parameters without staging was 195 - 540 ppm (corrected to 0% O2), and ranges for individual burners are indicated in fig. 2. When changing burner parameters (e.g. atomizer type) it is of course to be expected that the flame performance changes, and it was often necessary to trim other parameters (e.g. swirl level) in order to produce stable flames without smoke formation. In general any parameter changes which tended to reduce the fuel/ air mixing in the early flame region, tended also to reduce NO x emissions. For this reason parameters such as fuel spray angle and atomizer hole arrangement tend to be more effective than for example atomizer type. (Narrower spray angles tend to produce longer flames with delayed mixing, whilst reducing the number or the repositioning of atomizing holes can lead to the production of separate flame fingers, with the same overall effect.) Poor mixing in a flame leads however to smoke formation or to excessive particulates emission, and in the limit one is often forced to make a compromise between low NOx~emission and high smoke emission. Important here is the careful matching of atomizer performance and fuel distribution with the air distribution. One observation common to all burners, which confirms also the general trends associated with NO x -emissions, is that long or lazy flames tend to produce low N 0 X , whilst short intense flames tend to be high NO x -emitters. Burner type o A A B o C « 0 * Parallel • Experimental (300 *C preheat Firing density 0.15-0.2 MW/m* T furnace exit * 870-1000 *C - 500 o o §450 iOO 350 300 250- 200 X ^4-L| * .* 150 0.L Fig. 2 06 0.8 1.0 prsmary zone stoichiometry Burner optimization for low NOx-emissi en with staged combustion By introducing tertiary air through separate air ports, staged combustion was applied to each of the burners in a similar manner. Resulting NO x -emissions are presented in fig. 2 as a function of the primary zone stoichiometry, defined as the fraction of theoretical air required for complete combustion. Fig. 2 shows again the trends typical of staged combustion where increased staging leads to a reduction in N 0 X and minimum levels are achieved in a range of primary zone stoichiometries frcm 0.5 to 0.7. For the different burners it can be seen that minimum emissions are achie- ved at slightly different stoichiometries, according to individual burner mixing patterns. Two other factors emerge as significant in the results of fig. 2. Firstly the minimum emissions achieved by individual burners is very similar, falling generally into the range of 150 - 180 ppm NO x . When account is taken of slightly different firing densities due to differences in nominal load of the different burners, then this agreement is even better. Secondly it can be seen that as air staging is increased and minimum emission levels are approached, the spread in results decreases, indicating a reduced sensivity to changes in burner parameters. This was generally true for all the parameters investigated, although factors such as excess air level, burner velocity and atomizer performance and the careful matching of input conditions still remain important. It should also be pointed out that the results of fig. 2 already include some optimization of burner parameters, since low NOx-emissions were sometimes associated with heavy smoke or particulates emissions and flame instability. For these reasons it was necessary to reject some conditions, the flames being unacceptable for industrial application. The onset of smoke formation as NOx-emissions are reduced is a familiar problem, and fig. 3 presents selected results for a comparison of these emission levels. The different points represent optimized conditions where, for individual burners, the NO x emission has been minimized and burner parameters trimmed to reduce or control smoke emission levels. The figures include also values for staged and non-staged combustion, and values taken from recent IFRF burner trials (some with specially developed low NOx-burners) under similar firing conditions. Fig. 3 shows that there is a .strong correlation between NO x and smoke emissions, and that in the limit this is virtually independent of burner design. If a smoke number in the range 3 - 4 is considered to be the maximum acceptable, then fig. 3 would indicate (for the range of conditions tested) a minimum achievable NOx-emission of between 150 and 200 ppm. It was of course possible to produce flames with lower NOx-levels, but such flames were usually associated with high smoke emission and/or instability. The minimum level of NOx-emission of 150 ppm, for a 0.3% N oil, appears to agree however with results from other works (fig. 4, ref. [2J) obtained under more idealized staged combustion conditions. The results shown in fig. 4 might thus be used to extend the findings of this present study to oils of different composition. Fig. 3 shows also that there is a similar correlation between NO x and total particulates emissions, although in this case it is believed that total particulates will depend more on the oil composition (e.g. ash and asphaltenes content), whilst smoke formation results from gas phase reactions. The need to compromise between NOx and smoke emissions was apparent throughout this study, and simultaneous minimization of both emissions leads to conflicting requirements from the point of view of burner design. For the minimization of NOx-emissions an initial rapid vaporization of the fuel under oxygen deficient conditions is -10- required [ 2 ] , conditions w h i c h are g e n e r a l l y ideal for smoke formation. T e s t results indicate that it is the o p t i m i z a t i o n of r e s i dence time and m i x i n g in the rich p r i m a r y c o m b u s t i o n region w h i c h is important. A strong correlation was found for example b e t w e e n burner velocity and NO - e m i s s i o n s , w i t h lower v e l o c i t i e s r e s u l t i n g in lower N O x independently of the application of staged c o m b u s t i o n or changes in burner p a r a m e t e r s . H; H: B • x i O o MMhultf cr _3C0 9 Sumer A Burner B Burner C Burner 0 Experimental burner 300°C preheat N,KB Commercial burners H 0 ,D from recent trials • air preheat to LSQ'C Firing density 0.1-0.2MvV/m* 7 6, Srroce nr 0J2 pcrrjculctes . J iv0wt cf Kje . 3 : Optimized burner performance N O x , particulates and smoke emission The use of very low velocities however results in poor initial fuel air m i x i n g , and residence times under fuel rich conditions which become too long, and smoke formation r e s u l t s . For m o s t burners an optimum burner throat velocity in the range 15-20 m / s was found, b e l o w which smoke formation became e x c e s s i v e , and above which it was not possible to optimize for low N O x . The use of low burner velocities has also some practical implications for burner d e s i g n , in that the use of low initial velocities may lead to an inability to control the flame during turn-down o p e r a t i o n . This m a y in turn lead to the need for more complicated burner d e s i g n s in w h i c h for example turn-down is achieved by acting independently on the tertiary (staging) air supply. In this respect the use of staged combustion provides for an extra degree of flexibility in burner operation. T T 300 Boiler Simjlttor^' Tunnel furnace liquid Fuel Pttroleua Shalt 0 0.2 J. 0.4 -L 0.6 O.S Boiler Simulator • Tunnel Furnace O ± J 1.0 1.2 Coal • A L 1.4 1.6 1.8 2.0 Fuel Nitrogen {Wt. t) Fig. 4 : Minimum NOx achievable under staged conditions |_ref. 2J One further observation which was made, and which was again common to all conditions, was that a reduction in NOx-emissions was accompanied by a simultaneous lengthening of the flames. This was confirmed by detailed in-flame measurements, which showed also a reduction in peak flame temperatures. Such changes are important for industrial process applications, where the interaction of the flame with its surroundings can be critical. In many process heating applications for example flame impingement on heat transfer surfaces, or high peak fluxes, cannot be tolerated. In an attempt to characterize this influence, measurements of the heat flux to the furnace cooling elements were made, and fig. 5 compares the axial distributions of heat flux for four burners operating under baseline and optimized low NOx-conditions. It should be noted that in this figure the heat flux scales for the different burners are not co-incident since the intention was not to compare individual burner performance. It is apparent that the total heat flux or efficiency (obtained by integration along the furnace length) is not significantly effected by the reduction in NOx-emissions. The distribution of heat flux is however effected, and the general trend is one of a slight reduction in peak heat fluxes and a -12- shifting of this peak in the downstream direction. These trends are consistent with the observations of flame lengthening and the reductions in peak flame temperatures. Despite a similar low N 0 X emission level it is clear that there is a strong dependency of the distribution of heat flux on burner design, which, independently of NO x -considerations, leaves open the possibility to select or design burners for their suitability in different process applications. Burner A NOx 280ppm NO x 170 ppm NOx 270 pern' NOx ' 80 ppm' t 4 5 axid distance!ml Fig. 5 : Influence of NO x -control on axial distribution of heat flux -13- THE EFFECTS OF FIRING CONDITIONS ON NC^-EMISSIONS The studies described above have indicated that burners can be adapted to produce low NOx-emissions, and that in the limit the minimum NOx-levels achievable were limited by the onset of smoke formation and were virtually independent of burner design. This work was carried out with burners of a similar size, and in one experimental furnace only where the boundary conditions were fixed. It is clear however from the experimental results and from the literature that other non burner related parameters, such as firing density (MW/m3) , confinement and heat extraction rate, can have a significant effect on both NO x - and smoke-emissions. Information is therefore required concerning the relative effects of these parameters if test results are to have an application to a wider range of conditions on different furnace and burner scales. In order to provide some of the relevant information a series of experiments have been carried out, based largely on the experimental swirl burner design shown in fig. 1. Two versions of this burner have been constructed, having nominal operating loads of respectively 4 MW and 2 MW. In order to provide further data on the influence of burner parameters (particularly velocity and scale), both burners have been fired at thermal inputs of 1, 2 and 4 MW and their characteristics established for a range of staged combustion conditions. To provide data on the independent influence of firing density, confinement, and heat extraction rate, both burners have been fired over the same operating range in the two differently sized furnaces described earlier. By changing the available cooling surface it was possible to operate both furnaces with two different levels of heat extraction, which for ease of identification are referred to as the hot-wall and cold-wall furnace conditions. In this way it was possible to investigate a comprehensive range of furnace conditions, and these are defined in table 1 below. The range of flue gas temperatures quoted corresponds a heat extraction from the flame in the range 44-75% of the total thermal input. FURNACE Nr. 1 hot walls Thermal J input jMW/m3 MW | 4 |0.16 FURNACE Nr. 2 | cold walls MW/m3 T-flue 1220 0.16 1030 880 T-flue ?C hot walls cold walls MW/m3 T-flue\', ? oc MW/m3 T-flue 1050 0.63 1180 0.63 1039 0.08 870 0.32 950 0.32 820 0.C4 740 0.16 670 0.16 620 9c °C ! 2 JO.08 i 1 JO. 04 Table 1 : Range of furnace conditions investigated -14- For the full range of test conditions fig. 6 shows the resulting NC^-emissions for the 2 MW experimental burner as a function of staged combustion. The degree of staging is again identified by the burner primary zone stoichiometry and all results -relate to an excess air level of 5%. Under non-staged conditions (SRI • 1.05) there are clear differences between the different firing conditions, where both the higher firing density and higher temperature environment lead to increased NO x -emissions. The effect of firing density (MW/m3) is however not so significant as one might expect, since increasing the firing density by a factor 4 (same flame in both furnaces) results in approximately 50 ppm increase in N O x , which is largely accounted for by changes in thermal conditions. Also an input of 4 MW in the large furnace is equivalent to 1 MW fired in the small furnace, and here despite the differences in heat extraction, differences in NO x -emission are significant, implying a greater dependence on burner related parameters. With the application of staged combustion the expected trends in N O x are observed, and minimum emission levels are found at a primary region stoichiometry of 0.5. Under these conditions it can be seen that the N O x emissions are independent of the furnace in which the respective flames are fired, but that there remains a dependence on thermal environment as influenced by the heat extraction rate. | H = hot walls C = cold walls 0,4 0.6 0.8 1'JD 0£ 0.5 0.8 primary region stoichiometric rctio SR1 Fia. 6 : Effect of staged combustion and furnace conditions on NOxemission levels of nominal 2 MW experimental burner Results obtained with the larger 4 MW burner showed trends very similar to those reported above. Under non-staged conditions hewever, NO x -emissions were significantly lower than for the 2 MW burner (fig. 8 ) , corresponding to the influence of burner velocity reported earlier. Under staged combustion conditions the minimum NO x -emission levels were comparable for similar firing conditions. 1.0 -15- The NOx-emission results shown in fig. 6 indicate the influence of firing conditions and heat extraction, but are presented without any consideration of smoke formation or particulates emissions. For these flames, it was again found that as N0X-emissions were reduced, the tendency towards smoke formation increased, but that there was also an additional influence of firing conditions which effected the freedom to compromize between the NO x ~ and smoke-emissions. Lower furnace temperatures for example are favourable for low NOxemissions, but result also in an increased tendency towards smoke formation. The firing density on the other hand was shown to have no significant influence on the lower limit of NOx-emissions, but at the higher firing densities the overall furnace residence time available for the burn-out of smoke and particulates formed early in the flames is reduced, which in turn limits the extent to which NOx-emissions can be controlled before the onset of smoke formation. Despite this strong interaction between firing conditions and flue gas emissions a good correlation between NOx and smoke-emissions was found and results for the complete range of test conditions of table 1 are presented in fig. 7. This figure relates to optimized burner performance under staged combustion conditions and indicates clearly a strong dependency on firing density of the minimum practical NOx-emissions achievable. LOO 0I 0 Fig. 7 T~ I s 5 T~ Bacharcch smcke n'. Effect of firing density on optimized burner performance under staged combustion condition -16- With regard to thermal efficiency under these different firing conditions, fig. 8 presents measured axial heat flux distributions for both furnaces operating with an equal thermal input and similar overall levels of heat extraction. Results are presented for both the 4 MW burner (burner 1) and the 2 MW burner (burner 2) both operating at 2 MW thermal load, and baseline (non-staged) and staged combustion (minimum N 0 X ) conditions are compared. It is clear from this figure that the reduction in NO x -emissions by the application of staged combustion does not lead to significant changes in overall thermal efficiency. The distribution of heat flux under conditions of fixed confinement is however affected, where the application of staged combustion results in a slight reduction in peak heat fluxes and a shifting of the peak downstream. small furnccel035 MW/m*) large furnace (0.C8 MW/m s I 150 100 - - -1150) ppm N0X!C%02! (53)%Keat extraction •x o IX 3 s ft s - < 5 CD SO <L 1 Fig. 8 L 5 cxicl C'stance!m) Axial heat flux distributions - the effect of burner size, confinement and staged combustion for conditions of equal thermal input and similar heat extraction c -17- DISCUSSION OF RESULTS In the presentation of experimental results above it has been shown that emission levels of N0 X are the result of a complex series of interactions between burner related and furnace related parameters. The extent to which NOx-emissions can be reduced for a given burner is limited by the onset of smoke formation, whilst for fixed firing conditions the minimum levels of N0 X achievable appear to be virtually independent of burner type. When furnace boundary conditions are changed the NOx-emissions depend primarily on the heat extraction rate, which affects mean temperature levels, and the firing density related to the furnace volume. From an inspection of the observed trends it seems clear that the emission of N0 X can be explained or correlated in terms of the effects of a number of independent parameters. These studies show for example that for a given burner configuration NOx-reduction is always accompanied by an increase in flame length or an adjustment of the flame volume, and the results of fig. 6 suggest also that in the first instance it is the flame characteristics which are important. As a correlating parameter it is therefore proposed that the actual combustion intensity, as defined by the heat release within the flame volume itself, is primarily responsible for the N0 X levels obtained. It is significant that this parameter is also proportional to residence time and relates also to burner characteristics through flame volume dependence on burner size and velocity. Other parameters which subsequently affect the NOx-levels are the heat extraction in the furnace, which controls mean temperature levels and is related also to firing density, and the physical confinement of the flame which affects the ability to entrain combustion products. In an attempt to correlate the NOx-values found in this study a factor has been defined which incorporates the above three parameters. This factor takes the form m/ 3 *t • 1/He -fe¥& where MW/m _ relates to the combustion intensity based on the actual flame volume (assessed from visual observations). He is the fraction of the total thermal input extracted in the furnace and Mr/Ma is the Thring-Newby entrainment parameter which takes into account both burner and furnace diameters. The relationship between the above parameter and the NOx-emissions measured in this study is shown in fig. 9. The figure includes selected results from all burners and all firing conditions, both with and without staged combustion and with air preheat up to 300°C. Flames with high smoke emission (Bacharach smoke-numbers greater than 4) are excluded. Considering the difficulties associated with the accurate assessment of flame volume there is a good basic agreement which expresses the relationship between NOxemissions and operating parameters for the range of conditions covered in this study. -18- j-jr/rr Ac #n AA r *7 ™L . JL.J m', /ye V Ma-Mr Fig. 9 : Correlation of NOx-emissions with combustion intensity, heat extraction and confinement The transfer of experimental data to systems at different scales is also an important consideration and has been included in this study. One of the commercial burners has also been tested in a 12 MW version under conditions of firing density and wall temperature similar to those found in refinery heating furnaces. Measurement data from these tests are also included in fig. 9 and follow the trends of the data obtained at the smaller scale under different boundary conditions. Also included in fig. 9 are results obtained in a current IFRF trials series concerning the scaling of combustion systems L6J, and the two data points for the parallel flow burner, denoted (S), represent 2.3 MW scaled down versions of a 34 MW datum flame. The data points refer to flames scaled down according the criteria of constant velocity and constant mean flame residence time, where the constant velocity flame gives the higher N0X-emission. It should additionally be noted that the two small scale flames were fired with an oil having a nitrogen content of 0.45%, compared to the 0.3% used in this study. The good general agreement of this data however indicates that the selected correlating parameters are capable of taking into account the effects of scale on NOx-emissions. In the above correlation the key parameter is seen to be the combustion intensity within the flame, which relies on a measurement of the flame volume. If however this type of approach is to be more generally applicable, then it is of interest to relate the term MW/m3* more directly to burner design parameters. It has been C -19- suggested ]"5"j that the combustion intensity within the flame can be related to burner pressure loss (Zip) and burner diameter (D3) by the expression MW/m 3 f a Ap0'1*8 -1. 5 This expression was found to apply for one type of burner fired in a range of sizes in one experimental furnace, but failed to correlate data from the range of conditions included in this study. It is believed that the flame volume is affected not only by burner conditions but also by confinement and heat extraction, and fig. 10 represents a preliminary attempt to include these effects into a more complex correlation parameter. For most of the burners investigated this correlation appears to hold, but it fails badly in the treatment of burners with a precombustor (e.g. burner C) and in the case of the parallel flow burner. MW, m. 4 S fl^.He.Ap.Db; Fig 10 : Correlation of combustion intensity MW/mf, burner design and furnace conditions In conclusion it may be stated that a relatively clear relation has been observed between the NOx-emissions from oil flames and the parameters combustion intensity, heat extraction and confinement. The correlation appears to hold for a wide range of firing conditions and furnaces scales. The studies upon which these conclusions have been based have however been carried out largely with one type of fuel only, and whilst it is known that NOx-emissicns are closely linked to fuel nitrogen volatility f2~j, more work is required to establish the link between this paramiter (via fuel type) and the trends observed here. -20- ACKNOWLEDGEMENT The studies reported above have been supported by the Government of The Netherlands through the Ministry of Health and Environment (VOMIL) and the Ministry of Economic Affairs (EZ) under contract reference DGMH/AOO. This support is gratefully acknowledged. REFERENCES [1] MARTIN, G.B. and BOWEN, J.S.: NOx-control overview American Flame Research Committee, International Symposium on NOx-reduction - 22/23 October 1979 - Houston/Texas [2~1 ENGLAND, G.C., PERSHING, D.W., TOMLINSON, J. and HEAP, M.P.: Emission characteristics of petroleum and alternative liquid fuels American Flame Research Committee, International Symposium on NOx-reduction - 22/23 October 1979 - Houston/Texas [3^ AKIYAMA, T. and PAYNE, R.: Report on the AP6 trials - The reduction of NOx-emissions from burners for coal and oil fired boilers Prepared for the "Ministerie van Volksgezondheid en Milieuhygiene" The Netherlands IFRF Doc.nr. F 37/a/5 - 1980 [4J AKIYAMA, T. and PAYNE, R.: Report on the AP8 trials - The reduction of NOx-emissions from oil and gas fired industrial furnaces Prepared for the "Ministerie van Volksgezondheid en Milieuhygiene" The Netherlands IFRF document in preparation [5] MILLER, V.H.: Industrial burner design In publication [6J SALVI, G. and PAYNE, R.: IFRF investigation into the scaling of combustion systems IFRF Doc.nr. F 31/a/49 - 1980