Recent environmental regulations mandate that refinery and petrochemical industries reduce nitrogen oxide (NOx) emissions. Emissions legislations are constantly under review, and they differ from country to country and sometimes even within the same country.
New burner technologies have been developed to meet more stringent emissions requirements. Most of these new technologies use internal flue gas recirculation and fuel staging to reduce NOx. These burners are often referred to as ultra-low-NOx burners (ULNBs). The techniques used in ULNBs for reducing NOx emissions from process heaters affect key aspects of burner and heater performance.
The existing guidelines for burner specifications are generally based on conventional and low-NOx burners (LNBs). ULNBs behave differently from the other burner types, so a new set of guidelines is required to specify new burners in fired heaters.
A better specification is the first and best opportunity to achieve optimum heater performance and to minimize operations and maintenance. The specifications are used as the basis for the burner design. The burner specifications determine the design of the burners and their various critical properties. When specifications are accurate and complete, the burners and heaters can perform optimally.
Several oxides of nitrogen are formed during the combustion process. NOx formation is commonly grouped into three mechanisms or types:
- Thermal NOx
- Fuel NOx
- Prompt NOx.
These three mechanisms are described in the following sections.
Thermal NOx. Thermal NOx refers to the oxidation of molecular N contained in the air or fuel by O2. The rate of thermal NOx formation is very sensitive to local flame temperature. The formation of thermal NOx is described in Eq. 1:
N2 + O → NO + N (1)
N + O2 → NO + O
The first reaction has high activation energy and, therefore, requires high temperature for NOx formation. The formation of an NOx molecule from the first reaction results in a release of an N atom, which rapidly forms another NOx molecule.
The formation rate of thermal NOx increases exponentially with increasing flame temperature and is also directly proportional to the residence time of the reactant in the peak flame zone. The key parameters of thermal NOx formation are temperature, O2 and N concentrations, and residence time in the flame zone.
Fuel NOx. This is the result of reactions between fuel-bound N and O2 in the combustion air. The conversion rates of the fixed N in the fuel can be as high as 50% to 60% of the N present. The conversion rate is a function of the quantity of N compound present in the fuel. Most refinery fuel gas does not have fuel-bound N; therefore, there is virtually no fuel NOx for fuel gas firing.
Prompt NOx. Prompt NOx is a newly recognized mechanism of NOx formation. Although it only represents a small portion of the NOx formed, prompt NOx becomes a significant emissions source when ULNBs are required.
Under the fuel-rich conditions of the flame zone, particularly areas where the stoichiometry is less than 0.6, both hydrogen cyanide (HCN) and ammonia (NH3) can be formed through an extremely rapid reaction of hydrocarbon radicals with atmospheric N. As these components enter areas of the combustion zone where additional O2 is available, HCN and NH3 are oxidized to form CO2, H2O and NOx.
The major source of NOx in a fuel gas-fired heater is thermal NOx emissions created through high-temperature reactions of N and O2 present in the combustion air. Since the reaction is highly temperature dependent, the best direct method of lowering NOx emissions is to lower the peak flame temperature. To accomplish these goals, various burner suppliers have developed a specialized ULNB (Fig. 1).
| Fig. 1. Examples of NOx-reduction methods.|
Types of burners
Burners used in gas firing can be defined as follows:
- In raw gas burners, the fuel gas mixes with the combustion air stream after being released from the gas tip for ignition. These burners have a single combustion zone. There is usually only a single gas tip in the center of the burner throat.
- In pre-mix gas burners, all or part of the combustion air is mixed with the fuel gas before it emerges from the burner. Gas burns more rapidly in a pre-mix burner; therefore, the flame can be much smaller. Also, less draft is required to achieve proper air/fuel mixing. Pre-mix burners are much more sensitive to fuel gas molecular weight (i.e., specific gravity) than are raw gas burners.
- In staged-fuel burners, combustion is carried out in two stages: one fuel-rich and the other fuel-lean. This keeps combustion away from the stoichiometric mixture of fuel and air where flame temperature peaks. There are two ways to create fuel-rich and fuel-lean zones: air staging and fuel staging. Fuel staging is best suited for fuel gas-fired burners. Fuel gas is injected into the combustion zone in two stages, which creates a fuel-lean zone and which delays completion of the combustion process. Fuel supply is divided into primary fuel and secondary fuel. Staged-fuel burners have a longer flame than do raw gas or pre-mix gas burners.
- In external flue gas recirculation burners, part of the cold flue gas (generally 15%25%) from the stack is recirculated along with the combustion air. The flue gas acts as a diluent, reducing flame temperature and suppressing the partial pressure of O2, which reduces NOx formation. External flue gas recirculation is generally used with forced-draft burners.
ULNBs have different features from a conventional burner. A ULNB uses fuel staging and flue gas inspiration (i.e., internal recirculation). Flue gas is internally recirculated using the pressure energy of fuel gas. These burners are designed to recirculate relatively cooler flue gas from the firebox back into the combustion zone. This recirculation of largely inert flue gases into the combustion zone reduces peak flame temperature and average O2 concentration. The internal flue gas recirculation creates a similar effect as the external flue gas recirculation.
The fuel/air mixture is diluted by recirculated flue gas, resulting in lower burning rates. Lower burning rates increase the flame length, which often results in flame-to-flame interaction along with flame impingement on tubes.
Internal flue gas recirculation is a critical parameter in ULNB design. A conventional burner is less affected by poor flue gas flow pattern in the firebox. Many fired heater revamps, from conventional to ULNBs, have experienced problems like flame impingement and flame instability. Flame impingement on the tube will cause high tube metal temperature, which can result in premature coke formation in process fired heaters like crude, vacuum and coker heaters.
Flame-to-flame interaction can result in the merging of flames from individual burners. Merged flame is much longer than standalone flame. The prolonged flames can tilt toward the tubes, and flame tails can impinge on the tubes.
A fired heater with well-designed burners should have several desirable impacts:
- Minimize flame-to-flame interaction
- Prevent leaning of flames
- Prevent flame impingement on heater tubes
- Prevent the burner from recirculating flue gas from its own flame.
Design of fired heaters with ULNBs
For optimum performance of a fired heater, the proper burner selection, number of burners, fired duty per burner and location of burners are very important. Every design step in fired heater and burner design is critical for optimum performance. Oversight at any stage, from fuel gas supply to flue gas disposal, will adversely affect burner performance and, therefore, fired heater capacity, overall thermal efficiency and run length. There are several critical parameters when specifying a fired heater with ULNBs, as discussed below.
Fuel gas delivery system. The ULNB tip orifices are very small compared to conventional burners. Burner ignition ports in most ULNBs are 1⁄16 in. These small ports can and will become plugged if the fuel gas delivery system is not designed to reliably supply clean, dry fuel. The fuel delivery system for ULNB applications must have a filter/coalescer located downstream of the knockout drum. Use of stainless steel fuel piping downstream of the filter/coalescer is also recommended. Also, fuel piping downstream of the filter/coalescer may require insulation and heat tracing, depending on local site conditions.
Burner design parameters. First, the number of burners must be maximized. Increasing the number of burners will reduce the heat liberation per burner. This will result in the shortest possible flame length and a more uniform heat distribution.
Computational fluid dynamics (CFD) modeling was conducted for a fired heater with three burners and also with four burners. A comparison of peak heat flux profiles for the three-burner and four-burner arrangements was plotted for the tubes (Fig. 2). CFD modeling predicts a significant reduction in peak heat flux for the four-burner arrangement. The tube skin temperature was also reduced significantly with the four-burner arrangement, which resulted in significant improvements in overall fired heater run length.
| Fig. 2. Comparison of peak heat flux profile |
with three-burner and four-burner
Second, burner-to-burner spacing must be taken into consideration. ULNBs require maximum clearance between the tiles of neighboring burners. This space is required for internal flue gas recirculation. For example, in a vertical cylindrical heater with burners located too closely together, the inner portion of the burner tips may not have internal flue gas recirculation; this can produce longer flames and higher NOx emissions.
Burner circle diameter. In a competitive market where corners are sometimes cut to minimize overall cost, the API 560 minimum clearance requirement may be barely met. (Note: API 560 defines only the minimum clearance.) Burners installed with minimum API clearance can cause flame impingement for slight changes in fuel pressure or upsets in fired-heater operation.
Traditionally, burner circle diameter in a vertical cylindrical heater is calculated using the formula in Eq. 2:
BCD = Burner circle diameter
Db = Burner controlling diameter (i.e., the front-plate outside diameter)
Cb = Clearance between burner controlling diameter (usually 2 in.)
n = Number of burners.
This formula works well for a conventional burner, although it must be checked for a ULNB. The calculation considers the actual size of the burner to provide the required BCD. A new formula is required in addition to the above formula, which considers the spacing between the tiles of neighboring burners. The formula in Eq. 3 can be used to calculate the BCD in a vertical cylindrical heater with ULNBs:
BCD = Burner circle diameter
Dt = Burner tile outside diameter
Ct = Clearance between burner tiles
n = Number of burners.
Each burner designer has its own set of guidelines for the clearance between the burner tiles. A general clearance guideline is 1 in. per MMBtu/hr of heat release. For example, if the burner heat release is 10 MMBtu/hr, then the clearance between the tiles should be 10 in. Another suggested guideline is to keep the burner center to a center spacing of twice the Dt.
Heat liberation and overdesign
A good engineering practice for specifying the overdesign for a burner, as per API 560, is that burners shall be sized for a maximum heat release at the design excess air based on the following:
- Five or fewer burners: 120%
- Six or seven burners: 115%
- Eight or more burners: 110%.
It has been common practice in the industry to overdesign burners. Additional margin on burner design is sometimes applied on top of the margin on fired heater design. Overdesign shifts the burner design point away from the actual operating point.
The example summarized in Table 1 analyzes a case where a burner is designed with 120% margin on the required firing rate. At normal operating conditions for the heater, the combustion air pressure drop will be 0.41 inchWC instead of the 0.6 inchWC actually available. This overdesign on the burner reduces the draft loss requirements for actual heater operating conditions, which results in a larger flame envelope. It is highly recommended to minimize the overdesign to avoid increasing the overall flame envelope.
Burner test. The burner test is critical for ULNBs. Most test furnaces fire a single burner. With conventional burners and LNBs, the performance seen in the test is similar to multi-burner heaters. With ULNBs, however, this is not always the case. If burners are not placed with sufficient spacing, then flames may merge, resulting in longer flame length. It may also result in flame impingement on furnace tubes. ULNBs are more prone to flame interaction in multi-burner heaters. Therefore, it is always recommended to carry out a burner test. Also, if a burner supplier has a sufficiently large heater, it is recommended to test two or more burners with the same spacing as the actual heater. The test furnace conditions [e.g., bridge wall temperature (BWT), draft, etc.] should replicate those of the actual heater.
A critical parameter to measure during the test is flame dimension. The general practice is to visually establish the flame dimension; however, visual measurement can produce inaccurate or inconsistent results. Therefore, to establish more reliable flame dimension, it is recommended to use carbon monoxide (CO) probing of the flame. CO levels must be low in the firebox outside of the flame envelope. CO levels of more than 2,000 ppm are generally considered to be flame.
Many users do not specify a burner test if they have used a similar burner in the past. Although operating conditions and heat liberation are similar, a performance test is still recommended. A small difference in tip drillingwhich is commoncan have a large impact on NOx emissions and flame stability.
Burner in two circles. It has been a common practice to install burners in two circles for conventional burners (Fig. 3). The two-circle arrangement provides better clearance between tubes and burners. In most cases, an inner circle is added during a fired heater revamp to increase the firing rate. This is not a preferred option with ULNBs, however. The inner-circle burners do not receive cold flue gases for internal flue gas recirculation. This lack of internal flue gas recirculation increases NOx emissions from the inner-circle burner. Overall NOx emissions from the heater will rise and may not meet emissions requirements.
| Fig. 3. Burner layouts.|
Heater design parameters
With respect to burner-to-tube spacing and flue gas recirculation, it is suggested to maximize burner-to-tube clearance, not only to avoid flame impingement on tubes but also for better internal flue gas recirculation. This spacing becomes even more critical for a vertical cylindrical heater.
In a well-designed furnace, the hot flue gases should flow upward through the center of the heater. The cold flue gases should flow downward behind and along the tubes. A smaller clearance between the burner and the tubes may not allow cold flue gases to flow downward. The small space will be filled with hot gases. Fig. 4 shows a typical flue gas flow profile in a heater with a normal firebox and a heater with a tight firebox.
| Fig. 4. Flue gas flow profiles.|
The API recommends a clearance between the tubes and the burners. For natural draft, fuel gas-fired burners, this can be translated into Eq. 4:
CBT = Burner-to-tube clearance, ft
QB = Burner liberation, MMBtu/hr.
For ULNB applications, adding 6 in. to the API-recommended spacing appears to provide the needed clearance for proper flue gas recirculation. The modified equation is shown in Eq. 5:
Table 2 provides suggested clearances for ULNBs, while Table 3 provides a summary of results from a case study comparing the required minimum tube circle diameter (TCD) for a fired heater with both conventional burners and ULNBs.
Firebox or BWT. Conventional burner performance is not very dependent on firebox or BWT. However, ULNB performance is very dependent on firebox temperature. Burner NOx and other emissions are a function of firebox temperature. Higher firebox temperature leads to higher NOx formation.
The firebox temperature calculation methods used most frequently are empirical correlations based on experimental data. The heat transfer in the radiant section is calculated with the widely used Lobo-Evans method. This method assumes complete flue gas mixing in the firebox such that there are no longitudinal or transverse temperature gradients. The equation for radiant heat transfer is as follows:
qr = Total heat absorbed in radiant section, Btu/hr
Tg = Average radiating flue gas temperature in radiant section, °R
Ts = Temperature of cold surface (tube metal temperature), °R
αAcp = Equivalent cold plane area, ft2
F = Overall exchange factor.
These correlations were developed based on operating data from heaters using conventional burners. ULNB flame is cooler than a conventional burner flame. A lower flame temperature will result in lower heat transfer in the radiant section. Therefore, the flue gas temperature leaving the radiant section will be higher than that of a heater with conventional burners. This must be accounted for when designing the fired heater and specifying the burner. A ULNB installed in an existing fired heater (burner retrofit) may result in a loss of capacity or overall thermal efficiency, especially for heaters where the process fluid is heated in the radiant section only.
The overall heat transfer may not be affected in applications where the heater has a convection-section preheat coil followed by a radiant section in which further heating occurs. In this case, the total heat absorption for both the radiant and convection coils will be about the same, regardless of the BWT.
However, it is very important to correctly estimate BWT where there are two or more separate services, one in the radiant section and one or more in the convection section. If the BWT is higher than expected, the radiant-section absorption will be relatively low and the convection-section absorption will be relatively high.
In this case, such as when the radiant section consists of a process coil and the convection section consists of a steam-generating coil, the process coil and steam-generation duties must be guaranteed. Error in the predicted BWT would result in error in one or both of the predicted duties, along with inaccuracy in the guaranteed heater efficiency. An accurate prediction of BWT is required, not only for emissions prediction, but also to correctly predict heat absorption and other heater performance parameters.
An estimated performance of a fired heater with both conventional burners and ULNBs is summarized in Table 4. These parameters are for a vertical cylindrical heater with a process coil in the radiant section only. The convection section is used for waste heat recovery (i.e., a steam coil). As shown in Table 4, there is a substantial increase in BWT. This increase not only affects the firing rate and NOx emissions, but also the mechanical integrity of heater components like tube supports, structural steel, refractory and others.
Heat density per unit area. Heat density is not used for conventional burners. However, it is a very significant parameter for ULNBs. The reason for this is the importance of flue gas internal circulation in ULNBs.
The formula in Eq. 7 can be used to determine area heat density:
Qb = Burner heat release
n = Number of burners
A = Floor area (inside the tube circle or between tube rows).
For vertical cylindrical heaters:
A = πr2
r = Tube circle radius.
For cabin/box heaters:
A = L × W
L = Heater length
W = Heater width (tube center to center).
Flue gas flow patterns in the radiant section of properly designed process fired heaters are usually similar irrespective of heater geometry or type. In most fired heaters, tubes are placed along the walls, and the burners reside in the middle. As heat is transferred from the flue gas, it cools and flows downward close to the tubes. This results in a region of hot upward gas flow in the middle of the heater and a cool downward flow adjacent to the tubes. However, this recirculation decreases as over-firing increases. An over-fired heater may have little or no flue gas recirculation. This extreme case results in no flue gas downward flow, and the flow pattern becomes a plug flow.
This over-firing characteristic can be easily correlated with heat density. It is recommended to limit area heat density to a maximum of 250,000 Btu/hr/ft2. Two heaters with different area heat density but with similar firing rates, BWTs and excess air will have different NOx levels. The heater with the lower heat density will have lower NOx.
Common air plenum design. Common combustion air plenums are sometimes provided to better control the combustion air. These need to be designed to provide equal air to all the burners. If the burner plenum system is designed correctly, it is possible to achieve relatively uniform combustion airflow to each individual burner. It is always recommended to carry out a computational fluid dynamic (CFD) analysis to ensure proper air distribution to each burner. The example in Table 5 demonstrates how air maldistribution can starve some burners of combustion air.
It is common to target 2% or even lower excess O2 (Note: 2% O2 is equivalent to 10% excess air). Most new fired heaters are able to operate at lower excess air because of lower tramp air (i.e., lower air leakage) due to better sealing of the casing.
As can be drawn from Table 5, a heater operating with 10% excess air will have a few burners starving for air at 10% air maldistribution. A lower combustion air supply would result in sub-stoichiometric operation for that burner, which will increase CO emissions.
Reed wall. A reed wall is a refractory wall between the burners and the tubes. Reed walls are installed where flue gas circulation patterns within the firebox are pushing flames into the tubes. These walls safeguard tubes from hot flue gases. The burner flame is contained in the space provided for the burner. However, reed walls must be removed in ULNB applications because they prevent internal flue gas recirculation.
Operating parameters: CO emissions
Most fired heater emissions requirements also limit CO emissions. The factors influencing NOx emissions (firebox temperature, excess air and residence time) also influence CO emissions, although in the opposite direction.
NOx emissions are reduced as firebox temperature decreases, but CO emissions increase. CO emissions are generally not an issue at heater design conditions where firebox temperature is relatively high. However, at low heater throughput, CO emissions will increase.
CO is produced by incomplete combustion of fuel gases. The basic combustion process includes the formation of CO as an intermediate product before it is converted to CO2. CO is simple to convert into CO2 if sufficient O2, temperature and mixing are available. The minimum ignition temperature of CO in air is 1,128°F. A firebox temperature higher than the ignition temperature is required to limit CO emissions. Burner suppliers must be consulted for the minimum firebox temperature below which CO emissions are not guaranteed. Generally, CO emissions from a burner are guaranteed at a firebox temperature above 1,300°F.
Draft is defined as negative pressure at any point inside a fired heater. Draft loss across the burner is the pressure drop of combustion air in the burner. Higher available airside pressure drop to the burner almost invariably improves flame pattern and the ability to effectively use internal flue gas recirculation in burners. It is important to provide the actual calculated draft available to the burner supplier.
Site elevation (altitude above sea level) is one of the most-often overlooked parameters while calculating draft. The stack effect decreases with an increase in altitude. If draft is not calculated based on actual site elevation, the burner as well as the heater will have capacity limited due to insufficient draft.
The correlation in Eq. 8 can be used to calculate the draft (measured in inchWC) generated in a heater.
H = Height, ft
Patm = Atmospheric pressure, psi
Tamb = Ambient air temperature, °R
Tfg = Flue gas temperature, °R.
Atmospheric pressure (Patm) decreases with an increase in site elevation, resulting in a reduction in available draft. Table 6
provides a comparison of draft at approximate site elevations of 0 ft and 5,000 ft.
Fired heater operators tend to operate the heater at a higher draft as a precaution. However, when there is a large change in draft, it can adversely affect the excess air. The example in Table 7 indicates that airflow rate becomes sub-stoichiometric for a draft change from 0.2 inchWC to 0 inchWC. This change may not trigger the low-draft alarm, depending on the instrument setting.
Air leakage. Process fired heaters operate under negative pressure. Even small openings can allow significant quantities of air to leak into the heater. Air leakage into the furnace increases the O2 measured at the arch or stack, independent of the air coming through the burners.
The air coming through the burners will be lower than that measured by an O2 analyzer, which may increase NOx emissions. When furnace air leakage is high, the air coming through the burner may not be enough for complete combustion of the fuel, causing the flames to grow larger and possibly impinge on the furnace tubes. In such a situation, the heater may have to be operated at a higher excess O2, sacrificing heater efficiency to prevent flame impingement.
There are many sources of air leakage in a process heater: tube penetrations, sight ports, joints between heater walls, out-of-service burners, etc. It is important to seal any and all noticeable openings to prevent air leakage and ensure proper burner operation.
With the introduction of next-generation ULNBs, there have been major reductions in NOx emissions. These burners are operating closer to their limits of flame stability and complete combustion. ULNBs are very prone to flame interaction in multi-burner heaters.
ULNBs, along with fired heaters, need to be specified correctly to achieve optimal heater operation. The proper selection of a fuel gas supply system, number of burners, fired duty per burner, location of burners, fired heater design and operating parameters are very important for the optimum performance of a fired heater. Every design step in fired heater and burner design is critical. An oversight at any stage, from fuel gas supply to flue gas disposal, will adversely affect the burner performance as well as the fired heater performance. Many fired heater design and operating parameters may not be of significance with conventional burners, but they become important for ULNBs. HP
||Sultan Ahamad is a fired heater equipment engineer at Bechtel Corp. in Houston, Texas. He has more than 15 years of experience in the design, engineering and troubleshooting of fired heaters and combustion systems for the refining and petrochemical industries. He graduated from the Indian Institute of Technology in Roorkee, India, with a degree in chemical engineering. He is a member of the APIs subcommittee on heat-transfer equipment. |
||Rimon Vallavanatt is the senior principal engineer at Bechtel Corp. in Houston, Texas. He has more than 38 years of experience in the design, engineering and troubleshooting of fired heaters, thermal oxidizers, boilers and flares. He graduated from the University of Kerala in India with a degree in mechanical engineering. He also received a degree in industrial engineering from St. Marys University in San Antonio, Texas. Mr. Vallavanatt is a registered professional engineer in the state of Texas, and he has served on the APIs subcommittee on heat-transfer equipment for the past 30 years. |