Mengqian Xie,Fangqin Dai,Yaojie Tu

1 The State Key Laboratory of Refractories and Metallurgy,Wuhan University of Science and Technology,Wuhan 430081,China

2 Key Laboratory for Ferrous Metallurgy and Resources Utilization of Ministry of Education,Wuhan University of Science and Technology,Wuhan 430081,China

3 School of Energy and Power Engineering,Huazhong University of Science and Technology,Wuhan 430074,China

Keywords:MILD combustion Off-stoichiometric combustion NO emission Ignition instability Burner design

ABSTRACT Moderate or intense low-oxygen dilution(MILD)combustion has become a promising low-NOX emission technology,while the delayed mixing of reactants and slower oxidation rate could potentially cause ignition instability in some scenarios.This paper proposes a new idea for enhancing the ignition stability for methane MILD combustion by combining with off-stoichiometric combustion (OSC),and its performances have been numerically assessed through a comparison against the original MILD combustion burner.The results reveal although non-premixed pattern has the lowest NO emission,it suffers from a larger liftoff distance,thus less ignition stability.Contrarily,both partially-premixed and fully premixed patterns exhibit excellent ignition stability.Among the considered OSC conditions,the pattern of Inner ultra-rich and Outer lean produces the lowest NO emission while maintains a high ignition stability.Furthermore,the enhancement of the combustion stability by implementing OSC to the original MILD combustion burner is shown by comparing the operational range of furnace wall temperature (Tf),CO and NO emissions,as well as the evolution of chemical flame.The comparison reveals that OSC can extend the lowest operational Tf from 900 K to 800 K.More importantly,OSC can significantly improve the ignition stability in the whole range of Tf as compared to the original MILD combustion burner.

1.Introduction

Combustion will continue serving as the main energy source for human lives for decades to come.Despite the fact that gaseous fuels(such as methane and hydrogen)has less impacts on the environment than solid and liquid fuels,they still can cause serious environmental issues by emitting pollutants to the atmosphere,such as nitrogen oxide (NOX),green-house gas (GHG) and soot.Currently,the worldwide emission regulations for various combustion systems are becoming ever-more stringent.Therefore,new combustion technologies with high thermal efficiency and low environmental impact are still required to be developed.Moderate or intense low-oxygen dilution(MILD)combustion[1],or flameless oxidation [2] is one of the most promising technologies proposed to meet these targets in this century.It can effectively reduce NOXemission while maintaining a high combustion efficiency in a wide variety of applications (e.g.:industrial re-heating furnaces,melting furnaces,steam boilers and gas turbines).The essence of MILD combustion lies in a strong recirculation inside the combustion chamber,which makes the fuel and air simultaneously diluted and preheated by the hot flue gas.As a consequence,the chemical reaction takes place at a slower rate and combustion zone is extended to a larger space.Such processes subsequently result in a more uniform temperature distribution together with a lower peak temperature.Accordingly,thermal NOX,normally accounting for over 90% of total NOXformation during hydrocarbon combustion in air,will be inhibited.Such benefit has been especially favored in scenarios where highly preheated air and oxygenenriched oxidizer are to be utilized for combustion.

At present MILD combustion has been intensively applied in metalogical reheating furnaces,where the combustion chamber is well-insulated,and the wall temperature is normally in the range of 1300-1500 K according to previous experimental measurements in furnaces of different sizes [3–7].However,suchwall condition is totally different from that of boilers,where the internal wall temperature could be as low as 800 K or even lower,and intense heat extraction is also presented.Recently,Tu et al.[4]examined the transition behavior from traditional swirling combustion to MILD combustion in a 1 MW semi-industrial-scale furnace using slightly preheated air (403 K).The switching process was performed at two different furnace wall temperatures,i.e.800 K and 1000 K.The experimental flame photographs near the burner region indicated that the transition process would become unsteady in 800 K case as compared to the 1000 K counterpart evidenced by the frequently oscillating flames.The instability is mainly caused by the local extinction due to less heat release as well as over-cooling effect on the furnace wall.This observation implies that higher preheating temperature of the combustion chamber will be helpful for stable establishment of MILD combustion.In other words,it would be a challenge to realize MILD combustion in boilers due to too low wall temperature

To our best knowledge,most of the designs for MILD combustion burner adopt non-premixed (NP) parallel injection configuration between fuel and air[3,4,6].Such design helps to postpone the mixing of fuel and air,such that the two streams can be preheated and diluted by the recirculated flue gas before combustion happens.However,such design sometimes would result in a longer ignition distance.For example,an obvious liftoff distance was observed under MILD combustion of both natural gas (NG) [3]and pulverized coal (PC) [8] comparing to the traditional swirling flames [9,10] inside the International Flame Research Foundation(IFRF) #1 furnace,even though highly preheated air (>1573 K)was used.In addition,the slower reaction rate due to reduced peak temperature also gives rise to an increased unburned carbon in ash for PC MILD combustion in a 0.3 MW vertical fired furnace[5].Similar observation has been reported by Saha et al.[11] who carried out experimental studies on MILD combustion in a 20 KW labscale furnace with a brown coal and a black coal.Quite high UBC was found in the fly ash for both of the two coal types despite much lower NOXemission was obtained comparing to the traditional firing mode.

Emerging from the above literatures,the combustion instability and the incomplete combustion due to postponed ignition could be potential drawbacks for MILD combustion under low furnace wall temperature conditions,especially without using highly preheated air.This would prevent the further implementation of MILD combustion in systems with intense heat extraction,such as boilers.To this end,MILD combustion burners with both low NOXemission advantage and enhanced ignition stability are needed to be developed or further optimized for utilization in these scenarios.

For a given hydrocarbon fuel,its NOXformation exhibits a high dependence on equivalence ratio (φ).Specifically,NOXformation reaches the peak value at nearly stoichiometric condition(1 <φ <1.1).In fuel-lean condition,NOXformation will decrease in smaller φ due to less heat release and lower combustion temperature;while in fuel-rich condition,it will reduce with larger φ due to higher incomplete combustion as well as stronger reburning effect.Based on this theory,off-stoichiometric combustion(OSC) [12],or fuel rich/lean combustion [13],or bias combustion[14],has been proposed for years to mitigate NOXemission from combustion.Apart from the low NOXemission,OSC has also shown potentials in accelerating ignition stability by reducing the ignition enthalpy in the richer fuel stream.For instance,Bradley et al.[15]investigated the effect of φ on the combustion instability of a leanburnt swirl burner.Quenching and even blow-off would happen when φ was below 0.56 as a result of increased stretch.However,the flame could be stabilized by the hot gas in the inner recirculation zone at φ of 0.6,and by the hot gas in both the inner and outer recirculation zones at a higher φ of 0.75.

Nowadays,OSC has been already adopted in boilers [14,16,17]and gas turbines[18,19].Our previous work has also indicated that combining OSC with MILD combustion is able to further reduce NO emission from coal combustion [20].However,so far there still lacks sufficient understanding of the combination technology,especially on the feasibility of accelerate ignition and combustion stability.The reason is primarily because most MILD combustion burners consider the NP mixing pattern between fuel and oxidizer.Although MILD combustion with fully-premixed (FP) [21] and partially-premixed (PP) [22] mixing patterns have been inspected and indicated to be achievable,OSC,in principle,requires at least two premixed streams at different φ.Unfortunately,relevant studies on FP and PP MILD combustion with multiple jets of different φ are still absent.

This paper intends to propose a new solution to solve the potentially poor combustion stability issue of MILD combustion which would be encountered in strong heat extraction systems,i.e.boilers,by combining with OSC technology.In specific,computational fluidized dynamics (CFD) modeling approach is considered while its validity is firstly examined by comparing against the NP MILD combustion experiment [23].Subsequently,the combustion characteristics under different burner injection configurations (NP,PP and FP) are predicted,and the optimal burner injection configuration is picked up based on ignition liftoff distance and NOXemission.Note that,in this work,the burner injection configuration is changed by exchanging fuel in the center nozzle with the same volume amount of air in the outer annular,thus to keep an almost same momentum ratio between the center and annular streams.At last,the enhancement of the combustion stability attained from the optimal burner injection configuration is presented by comparing with the experimental case in terms of stable operational limit of furnace wall temperature.This paper will provide a new idea for developing novel ultra-low NOXburners with high combustion stability for boilers and gas turbines in the presence of intense heat extraction and high flow stretch.

2.Experimental

The NP MILD combustion experiment mentioned-above was carried out in a lab-scale furnace by firing methane [23].Fig.1 shows the schematic drawing of the MILD combustion furnaceburner system.In detail,the furnace chamber has a cuboid geometry with a height of 550 mm and a cross-section of 250 mm × 250 mm.The burner consists of a central fuel nozzle (ID:6 mm) and an annular air nozzle (ID:8–26 mm),as well as a bluff-body located at the burner end between the two channels which is expected to enhance the combustion stability by generating an inner recirculation zone.There are four exhausted gas exits(ID:26 mm) evenly distributed around the burner at the same plane as the burner inlet.Gas temperature and major species (O2,CO and NO) were measured radially by fine-wire (Pt-Pt-13% Rh,0.5 mm) R type thermocouple and Kane 9106 analyzer,respectively,from the center to the side wall at five furnace heights(x=135,225,315,405 and 495 mm).The measuring accuracies for temperature and gas species are:temperature (±2.5,25–600°C;±0.25%,600–1600 °C),O2(±0.1%,0–25%),CO2(±0.3%,0–99.9%),CO (±20×10-6,0–400×10-6;±5%,401–2000×10-6;±10%,2000–10000×10-6;±0.1%,0–10%),NO (±5×10-6,0–100×10-6;±5%,100–5000×10-6).

Fig.1.Schematic diagram of furnace-burner system of the MILD combustion experiment (unit:mm).

The burner was operated at a thermal load of 9.5 kW during the experiment with ambient temperature of air and fuel at global φ of 0.833,which is typically considered in boilers.Traditional firing mode (see Fig.1d) was firstly maintained for hours to preheat the inner furnace wall temperature to 1000 K by pushing the bluff-body into the furnace.Subsequently,MILD combustion (see Fig.1e)was achieved by pulling back the bluff-body to the location as illustrated in Fig.1c.The furnace chamber turned into ‘‘flameless”under MILD combustion operation,and extremely low NO(<8×10-6) and CO (<5 ×10-6) emissions were detected in the stack.According to these evidences,MILD combustion regime was regarded to be established.

3.Numerical Details

3.1.CFD modeling description

The combustion process was modeled using steady Reynolds Averaged Navier–Stokes algorithm with the aid of commercial software ANSYS Fluent 18.2 [24].Turbulent flow is solved with the standard k-ε model by solving the following transport equations (1) and (2):

where Gkand Gbstand for the turbulence kinetic energy (k) generated by velocity gradient and buoyancy,respectively.YMrepresents the contribution of fluctuating dilatation to overall dissipation rate(ε).σkand σεare turbulent Prandtl numbers for k and ε,respectively.To improve the prediction accuracy of a round jet,the constant C1ε is modified from 1.44 (default value) to 1.6 following previous work [25].

Since the chemical reaction time scale becomes comparable to the mixing time scale under MILD combustion regime,the eddy dissipation concept model is adopted for accounting the turbulence-chemistry interaction,which assumes reactions occur in fine scales [26].The length scale and reaction time scale of the fine scales are calculated by Eqs.(3) and (4),respectively.

To improve the modeling accuracy,the volume fraction constant(Cγ)and time scale constant(Cτ)in the eddy dissipation concept model are adjusted to 3 and 1 according to previous suggestion by Evans et al.[27].Despite of some existing optimized global reaction mechanisms [28–30],the well-known GRI-Mech 3.0 detailed reaction mechanism [31] is considered for describing both methane oxidation and NOXformation,considering the inadequacy of post-processing approach for NOXprediction in Fluent software[32].Radiation of the flue gas is calculated by the discrete ordinate model together with the gray gas assumption.For the sake of saving computing time,a simplified two-dimensional computing domain is considered to assign the major computing cost on solving the 325 steps of chemical reaction.Note that,previous modeling work [33] on the same furnace have already indicated that there is just a minor discrepancy between two-dimensional and three-dimensional computing domains owing to the semisymmetrical geometry of the furnace chamber.According to a grid sensitivity examination performed in our recent work[34],a twodimensional grid with～20,000 fully structured mesh is thus utilized.Convergence of the solutions is considered when all the residuals reach the desired levels,i.e.10-4for continuity and 10-6for others,and the global maximum gas temperature varies within 1 K.

3.2.Validation of the CFD modeling approach

Validation of the CFD modeling approach was carried out for the NP MILD combustion experimental case.Fig.2 shows the comparison of the predicted profiles against experimental measurement for gas temperature,dry O2mole fraction,dry CO concentration and dry NO concentration at five furnace heights.A generally goodagreement between the prediction and measurement has been achieved.More importantly,the detected NO emission in the final dry flue gas (7.9×10-6@ 3% O2) was almost accurately predicted(8.12×10-6@ 3% O2),further giving a confidence to the present CFD modeling.Therefore,the same models were used for the other cases in the subsequent CFD modeling cases.

Fig.2.Comparison between CFD modeling (solid line) and experimental measurement (symbol) for the NP MILD combustion case.

3.3.CFD modeling conditions

As stated above,OSC normally requires two FP streams,i.e.:one is fuel-rich and the other is fuel-lean.In this regard,OSC should be considered as the FP pattern.Although the realization of MILD combustion has been indicated to have nothing to do with the mixing patterns,Li et al.[22]has pointed out that a higher injection velocity,thus a stronger recirculation,should be initiated for the FP MILD combustion comparing to the NP counterpart when using a burner similar as the present one.They reported that NOXemission could achieve the lowest with the FP pattern,while the highest with the PP pattern.The primary reason was indicated to be due to the initial momentum of the streams.Note that,they pulled up the bluff-body in the PP case while premixed the fuel and air in the FP case(see Fig.3a).Consequently,the PP case has the smallest velocity at burner exit due to the largest area,and the FP case has the highest velocity due to the smallest area.However,increasing the injection velocity would bring additional cost to the system.Hence,it would be of practical value to establish FP MILD without significant modification of the burner system.To our best knowledge,the conclusion for NOXemission from the three mixing patterns should be further examined,since the establishment of MILD combustion regime and the NOXemission are strongly associated with the initial momentum,not just the mixing pattern.To assess the feasibility of implementing OSC in MILD combustion for enhanced combustion stability and low NOXemission,it is essential to gain a first understanding of other mixing patterns different from Li et al.[22] in Fig.3a.

To address the above-mentioned issue,the burner is reconfigured for achieving the NP,PP and FP patterns in the present work as shown in Fig.3b.Specifically,NP pattern is the same as in Ref.[12],and is adopted in the present experiment.In PP pattern,the central fuel channel is replaced by air with the same volume amount,and the fuel in shifted to the outer annular channel.In FP pattern,fuel is distributed in both the central and annular channels by 1:1,and the same volume amount of air is supplied into the central fuel channel.Note that,the position of the bluff-body is fixed in all three cases in Fig.3b,thus the exit areas of the central and annular channels are unchanged.Meanwhile,because of the relatively small volume amount in the central channel,the flowing velocity as well as momentum at the exits of the annular and central channels don’t show a significant difference(as can be seen in the Supplementary Material).

In addition to the above three cases (case 1–3 in Table 1),four more cases (case 4–7 in Table 1) are further considered to reveal the effect of switching fuel from the central channel to the outer annular channel on the performance of MILD combustion in combination with OSC.The main operating conditions of the seven cases are specified in Table 1.

Table 1 Main operating conditions of the considered cases

4.Results and Discussion

4.1.Performance comparison among NP,PP and FP mixing patterns

Fig.4 compares the predicted maximum temperature (Tmax)and NO emission among NP (Case 1),PP (Case 2) and FP (Case 3)mixing patterns.Clearly,NP case has the lowest Tmax,thus lowest NO emission,while PP case has the highest Tmaxand NO emission.This observation differs from the conclusion of Li et al.[22]that FP case would produce the lowest NO emission despite the similarburner geometry.The reason behind this discrepancy is caused by the different fuel/air injection configurations in Ref.[22] and the present study.Specifically,in Ref.[22]the blended fuel/air mixture is injected only through the central channel(as in Fig.3a),while in the present study it is injected through both central and annular channels (as in Fig.3b).As a consequence,the injection momentum of the flow at the burner exit in Ref.[22] is larger for FP pattern comparing to both NP and PP patterns.While in the present study the injection momentums of the flow at the burner exit in these three patterns are identical.In another word,injection momentum plays a more important role on the combustion and NO emission than mixing pattern under MILD combustion.On the other hand,Mi et al.[35] has also suggested that there exists a critical momentum rate of the fuel–air mixture below which MILD combustion would not occur for the FP pattern.Note that,in PP pattern Tmaxexceeds 1900 K,while NO emission is merely 6 ppm higher than the FP pattern.This is because the high temperature region (above 1900 K) occupies a relatively small space that the residence time for thermal NO formation is limited.

Fig.3.Schematic illustrations of burner injection configurations under different mixing patterns in:(a) the work of Li et al.[22],and (b) the present work.

Fig.4.Predicted maximum temperature and NO emission in Case 1–3.

For turbulent flames,the flame length and liftoff distance are usually measured through visual observation.However,such method has been reported to be influenced by different observers and observing conditions (e.g.darkened room or lighted room)[36].Moreover,there is no visible flame edge under MILD combustion mode.Therefore,previous studies have been dedicated to propose methods for identifying the frame of the chemical reaction region or chemical flame length based on the information of combustion,like gas temperature [37],species concentrations [38–40]and local equivalence ratio [41,42].Among these,Mei et al.[38]claimed that the use of 1%CO ratio(ratio between the local CO concentration and the global maximum CO concentration)exhibits the flame lift-off behavior and to gain the lift-off height properly,thus could provide the most accurate prediction of the flame shape against experimental observation,in terms of both flame length and liftoff distance.Hence,in this work 1% CO ratio is adopted to indicate the border of the chemical reaction region,or the outline of the ‘‘chemical flame”.Moreover,the flame liftoff behavior will be presented using this approach to compare the combustion stability in various operating conditions.

Fig.5 shows the spatial distributions of gas temperature and CO concentration in Case 1–3 on the left-hand side and right-hand side,respectively.Moreover,a white dashed line is drawn in the CO concentration panel to identify the border of the chemical reaction region.In all three cases the high temperature zone occurs in the central furnace,however it is obviously delayed in Case 1 (NP pattern).According to the outline of the chemical reaction region,Case 2 (PP pattern) and Case 3 (FP pattern) show an immediate ignition at the exit of the annular channel,moreover,the shapes of the chemical reaction region in PP and FP patterns are quite similar.

The ignition behavior can be more clearly understood from the minor species(OH and CH2O)distribution displayed in Fig.6.Note that,OH and CH2O are well-recognized as flame indicator and ignition precursor,respectively.The OH contour further confirms that the NP pattern has a delayed ignition than both PP and FP patterns.In addition,the CH2O contour indicates that methane oxidation by H abstraction (CH4→CH3→CH2O) can be accelerated by mixing air with fuel in the annular channel,which might be the primary reason for the delayed ignition in the NP condition.

Based on the above comparison of combustion and NO emission behaviors among the three different mixing patterns,the NP pattern can attain the lowest NO emission while highest possibility of ignition instability.On the contrary,both PP and FP patterns can achieve much more stable ignition while their NO emissions are beyond expectation.In order to realize simultaneous low NO emission and high ignition stability,OSC concept is attempted in the following section.

4.2.Effect of fuel split ratio in OSC scenario

Fuel split ratio (R) is defined as the percentage of the total fuel designated to the central fuel nozzle.Four cases(Case 4–7)with R being 0.5,0.1,0.05 and 0.01,respectively,are considered to examine its impact on NO emission and combustion characteristics.Specifically,R=0.5 in Case 4 means that the fuel volume is equal between the central fuel nozzle and the outer annular air nozzle.Due to the relatively small air amount in the central nozzle,φ are 10 and 0.5 for the central and annular nozzles,respectively,thus Case 4 is in the pattern of Inner ultra-rich and Outer lean.Accordingly,Case 5 (R=0.1),Case 6 (R=0.05) and Case 7(R=0.01)are in the patterns of Inner rich&Outer lean,Inner lean&Outer lean and Inner ultra-lean &Outer rich,respectively.The predicted Tmaxand NO emission in Case 4–6 are shown in Fig.7.Obviously,Tmaxand NO emission both exhibit an increasing trend with the decrease of R.In other words,transferring fuel from the central nozzle to the outer annular nozzle will break the MILD combustion regime by promoting Tmaxand NO emission.

Fig.5.Spatial distributions of (left) temperature and (right) CO concentration in Case 1–3,the white dashed line denotes the border of the chemical reaction region.

Fig.6.Spatial distributions of (left) OH mole fraction and (right) CH2O mole fraction in Case 1–3.

Fig.7.Predicted maximum temperature and NO emission in Case 4–6.

To gain a further insight into the ignition behavior under different R,Fig.8 shows the spatial distributions of gas temperature on the left-hand side and CO concentration on the right-hand side in Case 4–6,respectively.From the temperature contour,it is seen that the high temperature region is gradually moving downwards as R declines,suggesting the fuel oxidation is to be enhanced by its enrichment in the outer annular nozzle.On the other hand,the outlines of the chemical flame zone in the CO contour reveal excellent ignition stability in all four OSC cases.However,the length of the chemical flame zone is found to become shorter as R is minimized,thus the fuel oxidation will be more concentrated,which explains the reason for the higher Tmaxand NO emission as observed in Fig.7.

The strengthening of the fuel oxidation in smaller R case can be further identified by the OH distribution shown in Fig.9.Specifically,the maximum OH concentration appears in the inner side of the stream in the outer annular nozzle,suggesting that the enhanced fuel oxidation is primarily caused by the increasingly leaner condition in the central jet.Consequently,the peak CH2O concentration together with its size are gradually reduced.

Fig.8.Spatial distributions of (left) temperature and (right) CO concentration in Case 4–6,the white dashed line denotes the border of the chemical reaction region.

Fig.9.Spatial distributions of (left) OH mole fraction and (right) CH2O mole fraction in Case 4–6.

Fig.10.Dependence of Tmax and NO emission on φ.

Fig.10 presents the dependence profiles of Tmaxand NO emission on φ in both the central and annular channels based on the prediction results for the seven considered cases.As φ in the central channel varies in a significantly wide range,the profiles in Fig.10 are only focused in the range of φ from 0.1 to 10,which basically covers ultra-lean and ultra-rich conditions.However,the range of φ in the annular channel is much narrower.As a result,a small change of φ in the annular channel can cause an obvious alteration of Tmaxand NO emission,especially in the rich condition.Fig.10 implies that the predominant reason for the lower Tmaxand NO emission by implementing OSC comes from the reduction of φ in the annular channel in the present study.

4.3.Combustion stability enhancement in Case 4

Based on the comparison of Tmax,NO emission and ignition stability among the four OSC cases in Section 4.2,clearly Case 4 exhibits the best overall performance.Therefore,in this Section,Case 4 will be selected as the relatively optimal OSC case,and its performance on combustion stability at lower Tfwill be further revealed against the original design,namely Case 1.

Fig.11.Profiles of Tmax against Tf in Case 1 and Case 4.

Fig.12.CO and NO emissions at different Tf in Case 1 and Case 4.

In the present experiment(Case 1),the furnace chamber is well insulated and Tfwas measured to be around 1500 K.However,this is not the case for boilers since their Tfwill be much lower.To check whether the combustion can be still sustained by adopting OSC technology at lower Tfconditions,simulations are performed for Case 1 and Case 4 by gradually reducing the Tffrom 1500 K with an interval of 100 K.Fig.11 shows the profiles of Tmaxin the two cases at different Tf.It is seen that Tmaxis always higher in Case 4 than that in Case 1,and the difference is more noticeable at higher Tfconditions.In addition,Tmaxpresents a descending trend with the decrease of Tf,and eventually the extinction occurs at the Tfof 900 K and 800 K for Case 1 and Case 4,respectively.The detailed analysis of the transition process from stable MILD combustion to extinction by heat extraction has been provided in our previous works[43,44].Nevertheless,Fig.11 clearly demonstrates that Case 4 has a wider range of Tffor stable operation with respect to Case 1,indicating the combustion stability of MILD combustion can be indeed further improved by employing OSC technology.

Fig.12 displays the dry CO and NO emission profiles at various Tfconditions for Case 1 and Case 4.It is found the trends of both CO and NO emissions against Tfare basically the same for the two considered cases.Specifically,CO emission gradually increases while NO emission gradually reduces as Tfis minimized.The higher CO emission at lower Tfis caused by the slower oxidation rate of fuel and the delayed reaction zone,and the smaller NO emission results primarily from the inhibition of thermal-NO route due to decreased Tmax.More importantly,in order to achieve nearly zero CO emission,namely complete oxidation of fuel,the lower limit of Tffor Case 1 and Case 4 are 1100 K and 1000 K,respectively.This also suggests that OSC technology is helpful to improve combustion efficiency under strong heat extraction conditions.

In order to further acquaint the performance of OSC technology,Figs.13 and 14 compare the spatial temperature and CO distributions at different Tfconditions for Case 1 and Case 4,respectively.Note that,the evolution of reaction zone at various Tfis also indicated by the white dashed line in each CO counter.Despite the different burner configurations,Case 1 and Case 4 exhibit very similar temperature and CO distribution fields.In specific,the high temperature region gradually shifts away from the burner exit with the reduction of Tf,which can be clearly recognized by the high CO region.At high Tfconditions,the reaction zone is confined in the central furnace and appears with a ‘‘∧”shape,then it changes to a‘‘Π”shape after impacting on the bottom wall as Tffurther goes down.Eventually,extinction occurs due to lifted reaction zone and postponed oxidation,which suppresses the preheating of reactants.

Fig.13.Temperature and CO evolutions with reduced Tf in Case 1.

Fig.14.Temperature and CO evolutions with reduced Tf in Case 4.

Based on the evolution of reaction zone showing in the CO contours in Figs.13 and 14,the variations of average furnace temperature and ignition distance (the axial distance from burner exit to the rim the reaction zone)with the change of Tfin Case 1 and Case 4 are displayed in Fig.15.In Case 1,the reaction zone is found to be detached from the burner exit at all Tfconditions,while in Case 4 it is attached to the burner exit at Tfranging from 1500 K to 1100 K,and slightly lifted at Tfranging from 1000 K to 900 K.Moreover,according to Fig.13,the ignition is so delayed that the reaction zone covers the whole side wall in Case 1 at Tfof 900 K,which almost reaches the lower limit for sustainable operation.However,the reaction zone in Case 4 (see Fig.14) still maintains the ‘‘Π”shape and exhibits quite small liftoff distance at Tfof 900 K.These observations clearly show the ability of enhancing the ignition stability gained by Case 4 via employing the OSC technology.

Fig.15.Average furnace temperature and ignition distance at different Tf in Case 1 and Case 4.

5.Conclusions

This paper numerically assessed the feasibility of combining off-stoichiometric combustion (OSC) technology with MILD combustion for accelerating the ignition and combustion stability of the original MILD combustion burner in a lab-scale furnace fueled by methane.Validation of the employed numerical simulation method was first performed for the experimental case with the original MILD combustion burner,and both in-furnace temperature and species distributions as well as exhausted NO emission have been well-reproduced.By changing the split ratio of fuel amount in the central and outer annular channels,the effects of fuel–air mixing pattern and fuel split ratio on the ignition stability as well as NO emission have been examined.The main conclusions are summarized as follows.

(1) When maintaining the same injection momentum of initial reactants at the burner exit,non-premixed pattern produces the lowest maximum temperature,thus lowest NO emission;while fully-premixed pattern produces the highest maximum temperature and NO emission.As for the ignition stability,non-premixed pattern is found to generate an obvious liftoff distance,however,such instability is eliminated in both partially-premixed and fully-premixed patterns.

(2) In fully-premixed pattern,the maximum temperature and NO emission will increase as fuel is gradually switched from the central channel into the outer annular channel as a result of more concentrated chemical reaction region.In the present study,the optimal OSC burner configuration in terms of both lower NO emission and higher ignition stability is Inner ultra-rich and Outer lean,namely Case 4.

(3) Comparing to the original MILD combustion burner(Case 1),the optimal OSC burner configuration (Case 4) has also shown a wider operational range for furnace wall temperature.The evolution of the chemical reaction region with the drop of furnace wall temperature demonstrates that MILD combustion could become unstable at lower furnace wall temperature,evidenced by gradually postponed ignition point and downstream reaction zone.With the aid of OSC technology,the combustion stability can be accelerated for the original MILD combustion burner with slightly increase NO emission.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

Yaojie Tu gratefully acknowledges the financial support from the National Natural Science Foundation of China (52006077)and Innovation Research Foundation of Huazhong University of Science and Technology (5001120031).

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2020.09.053.