Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility

ARTICLE IN PRESS

JID: CNF

[m5G;February 26, 2016;21:12]

Combustion and Flame 0 0 0 (2016) 1–13

Contents lists available at ScienceDirect

Combustion and Flame journal homepage: www.elsevier.com/locate/combustflame

Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility Max Weidmann a,∗, David Honoré b,∗∗, Vincent Verbaere c, Guillaume Boutin b, Simon Grathwohl a, Gilles Godard b, Carole Gobin b, Reinhold Kneer c, Günter Scheffknecht a a b c

Institute of Combustion and Powerplant Technology, University of Stuttgart, Pfaffenwaldring 23, 70569 Stuttgart, Germany CORIA – CNRS UMR 6614, Normandie Université, INSA de Rouen, Université de Rouen, Avenue de l’Université, 76801 Saint-Etienne-du-Rouvray, France Institute of Heat and Mass Transfer, RWTH Aachen University, Eilfschornsteinstrasse 18, 52062 Aachen, Germany

a r t i c l e

i n f o

Article history: Received 14 October 2015 Revised 25 January 2016 Accepted 26 January 2016 Available online xxx Keywords: Flameless combustion MILD combustion Low NOx burner Pulverized coal LDV OH∗ chemiluminescence imaging

a b s t r a c t This paper presents a series of measurements on pulverized coal MILD flameless combustion in a pilotscale facility for nitrogen oxides (NOx ) emissions reduction. Local measurements of gaseous species concentrations, gas temperature and velocity associated to reaction zone imaging by OH∗ chemiluminescence highlight specific features of this MILD flameless combustion regime. The flameless burner used during the investigation could abate significantly NOx emission levels. Specific aerodynamics of the flow in the furnace induced by the burner geometry are discussed. High momentum turbulent air jets favor a large recirculation of hot flue gas in the combustion chamber. Such a recirculation induces a large diluted combustion regime from the entrainment in the air jets of devolatilized species from the pulverized coal jet and of recirculating hot inert combustion products. The main reaction zone is lifted from the burner exit as it starts in the mixing layers of separated pulverized coal and air jets. As is typical of a flameless combustion regime, such a large dilution of reactants induces low local heat release and temperature increase in the reaction zone. Nitrogen oxides are generated from the fuel NO route at quite a limited rate due to the low temperature environment. The effect of the carrier gas of the pulverized coal jet is also analyzed. The change from carbon dioxide to air as carrier gas generates a first reaction zone attached to the burner exit. This reduces the dilution and increases the heat release in the main lifted reaction zone, leading to higher NOx emissions. © 2016 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction Measures aimed at reducing NOx emissions from combustion processes have been developed and implemented for several decades. Primary measures, such as low NOx burners or air staging are widely used. In order to comply with environmental directives, secondary – post-combustion – measures are, however, necessary. For reason of costs and ever-increasing environment restrictions, research on primary abatement measures remains a topical issue. Since the 1990s, combustion of gaseous fuels with high recirculation ratio of burnt gases without visible flame, referred to as flameless oxidation or FLOXۚ , has demonstrated its potential, especially for high air preheating, in reducing NOx emissions resulting from thermal N conversion [1]. For pulverized coal (PC) ∗

Corresponding author. Fax: +49 711 685 69403. Corresponding author. Fax: +33 232 95 97 80. E-mail addresses: [email protected] [email protected] (D. Honoré). ∗∗

(M.

Weidmann),

combustion, the IFRF has initiated experiments on flameless combustion by preheating combustion air to about 1300 °C, on one hand, and separately injecting combustion air and coal into the furnace, on the other hand. A remarkable NOx abatement could be achieved if sufficient separation between coal and air jets is provided and if the oxygen availability is controlled in the primary combustion zone [2,3]. Cavaliere and de Joannon [4] defined as moderate or intensive low-oxygen dilution (MILD) combustion a diluted regime characterized by air and fuel preheating over selfignition temperature, as well as a low adiabatic temperature. In this context, the high exergy of exhaust gases presents a significant advantage. From the pollutant point of view, flameless combustion is able to abate the NOx formation usually found in highly preheated air combustion. However, a high level of preheat remains difficult to implement in solid-fuel fired utility boilers. But, flameless combustion can be operated with moderate preheating temperatures (those encountered in utility boilers) as explained by Wünning [5] and even without any preheat [6–10]. In fact, MILD flameless combustion remains independent of the extent of

http://dx.doi.org/10.1016/j.combustflame.2016.01.029 0010-2180/© 2016 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF 2

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

preheat: the mixing alone of fuel and combustion air with recirculated exhaust gas gives rise to a diluted MILD combustion regime. Wünning stated that air preheating along with a complete separation between incoming fuel and air are by no means always necessary [4]. A common way to generate a strong mixture within the furnace is to provide the incoming combustion air with high velocity. This ensures a reduction of thermal NO due to the achievement of a rather homogeneous temperature field. Furthermore, it reduces the conversion of volatile nitrogen, which here represents a large share (up to 50% [11]) of the total nitrogen contained in coal. According to Chen and Niksa [12], this can be explained by the shift of the devolatilization into regions heavily diluted by exhaust gas; this delays the mixing between coal and oxidizer. Pershing and Wendt [11] already obtained in 1979 a reduction of NOx emission by 40% by using a low mixing fuel injector. In their pioneering work, Weber et al. [3] showed the potential of MILD flameless combustion applied to PC by comparison with gaseous and liquid fuels with uniform heat transfers and low pollutant emissions for all operating conditions. An upscaling of PC MILD flameless combustion in a 12 MW industrial-scale test facility has been also reported by Zhang et al. [13]. Other experiments performed at semi-industrial scale (500 kW) by Smart and Riley [14] demonstrated the feasibility of MILD flameless combustion of a high-volatile pulverized coal under oxy-fuel conditions. In this context, their burner was provided with a large separation between coal and the oxidants. They recommended improvements on their burner and operation conditions to ensure stable and safe combustion features. More recently, an experimental study of several PC MILD flameless burner configurations has been done in a 300 kW pilot facility [15]. By reference to a conventional swirl jet burner, all burner configurations provided with separated injections achieved a MILD flameless regime. The result of this was good heat homogeneity, low NOx and CO emissions, but a low char burnout. It has to be noted that volatile combustion seems to be invisible as it is for MILD flameless combustion of gaseous fuels, whereas visible sparks of burning particles are still present. Other experiments have also been performed in a 15 kW lab-scale facility with different kinds of coal and carrier gas [16]: incomplete combustion of low-volatile anthracite has been observed as the result of an insufficient residence time. Reynolds-Averaged Navier–Stokes (RANS) CFD simulations have been also carried out. These showed that an increase of air jet momentum facilitates the combustion of volatiles, leading to better MILD flameless performance in terms of homogeneity and pollutant emissions. A similar simulation was presented by Mei et al. [17]. Their post-processing approach unveiled NOx formation: the fuel NO path has been found to prevail over both thermal NO and prompt NO. They showed the existence of NO reburning, as it has been previously envisaged in MILD flameless of gaseous fuels [18,19]. This study also confirmed the predominant role of the aerodynamics of jets to achieve local diluted conditions. The potential of NOx emission reduction with pulverized coal combustion has also been investigated during a previous European research project in a 20 kWth and a 100 kWth furnace, both electrically heated [20,21]. Within this framework, the burner included a central coal nozzle and three combustion air nozzles evenly allocated on a circle centered on the coal nozzle. The results can be summarized as followed: (i) the potential of NOx reduction depends strongly on the stoichiometric air ratio at the burner and to a lesser extent on the coal type; (ii) beyond 100 m/s the combustion air velocity has little effect on NOx emissions; (iii) flameless combustion ensures a strong cut in thermal NO, but slightly increases fuel NO, resulting primarily from enhanced char N conversion, with regard to ’standard’ flame combustion. In view of these results, the burner geometry has been adapted for the present

Fig. 1. Schematic diagram of the furnace and flue gas treatment.

Fig. 2. Configuration geometry of the FLOX burner.

project and scaled up to a thermal load of 300 kWth [22]. During these experiments presented here, detailed in-flame profile measurements have been performed: temperature and gas concentrations by probe sampling, local velocity by Laser Doppler Velocimetry and reaction zone visualization by OH∗ chemiluminescence imaging. The present paper gives a complete overview of the results and provides an analysis to point out the specific combustion features of MILD flameless regime applied to pulverized coal.

2. Experimental setup 2.1. Pilot-scale test rig and burner The 500 kWth pilot-scale test facility is based on a conventional down-fired pulverized fuel combustion reactor. The furnace is cylindrical in shape and its axis is vertical to minimize asymmetry due to natural convection and ash deposition. The furnace is made up of six water-cooled segments with a total length of 70 0 0 mm and an inner diameter of 750 mm. Each of the three upper segments is equipped with five series of four measurement ports distributed every 90° along the segment periphery. They enable measurement of gas composition, gas temperature and ash sampling in vertical and horizontal directions by means of specially designed probes. The upper segments of the furnace, as well as, the burner plate are protected from heat by a refractory. Cooling water flowing in upwind direction provides additional protection. An overall sketch of the furnace is shown in Fig. 1. The design of the FLOX burner (see Fig. 2) used in this work derives from that of previous studies conducted at bench-scale [23]. Coal along with the carrier gas is injected through a central annular nozzle. Two combustion air nozzles are disposed eccentrically

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

ARTICLE IN PRESS

JID: CNF

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13 Table 1 Proximate and ultimate data of the Calentur coal. Calentur Ash Volatiles Fixed C C H O N S

3

Table 3 Operational data for FLOXCO2 , FLOXAIR and FLAME cases.

Water-free basis (wt%) 11.03 37.62 51.35 68.81 4.53 13.32 1.34 0.96

a

VFRcombustion air , m /h (stp) Tc ombustion air , °C VFRc oal carrier gas , m3 /h (stp) Tc oal carrier gas , °C Gas type (coal carrier gas) MFRc oal b , kg/h Air ratio a

Table 2 Coal particle size distribution.

3

b

FLOXCO2

FLOXAIR

FLAME

250.0 143.9 26.0 45.0 79 vol% CO2 21 vol% Air 33.7 1.15

215.7 149.2 34.5 45.0 Air

205.2 147.8 36.2 40.0 Air

32.4 1.16

32.4 1.15

VFR: volumetric flow rate. MFR: mass flow rate.

Diameter (μm) d10 d50 d90

44.53 73.10 100.26

from the center to delay the mixing between reactants and then to enhance their dilution with recirculated flue gas. 2.2. Fuel The fuel used in the experiments is a high-volatile bituminous Columbian coal referred to as ‘Calentur’. The proximate and ultimate analyses are given in Table 1. In general terms, Calentur coal exhibits very similar properties to ‘El Cerrejón’ coal, also of Columbian origin and widely used in the German energy industry. The Calentur coal is characterized by a medium nitrogen content which prevents an excessive fuel NO formation. Its high volatile content is beneficial to reduction reactions. Its low heating value is 26.8 MJ/kg. The moisture of the coal is 4.10 wt% in the state ‘as fired’. The coal particle size distribution (see Table 2) which affects the reduction mechanisms and the burnout has been chosen similar to that commonly used in utility boilers. 2.3. Operating conditions The necessary conditions for a flameless combustion are a fast coal devolatilization, on the one hand, and an oxidation within a diluted atmosphere, on the other hand. This may be carried out with a large internal recirculation of hot flue gases used to heat up the coal and the combustion air and dilute them before reactions with the devolatilization products take place. The recirculation is induced by the high-momentum combustion air jets. However, the applied velocity shall not exceed 100 m/s, since future applications of flameless combustion technology in utility boilers should not give rise to heavy pumping power requirements. Nowadays, it is acknowledged that flameless combustion can be achieved without air or fuel preheating [5–7,10]. Nevertheless, air preheating helps establishing flameless conditions since specifically more heat is made available for fuel heat-up. Air preheating temperatures of 300 °C is the state-of-the-art in utility boilers. In the present tests, it was not possible to exceed 150 °C. It has to be noted that these severe conditions can only have a positive impact on future up-scale scenarios. As a result of restrictions imposed by the combustion air velocity and the burner geometry, the thermal load was set to be approximately 250 kWth . Two MILD flameless configurations differentiated by their coal carrier gas stream have been examined throughout the present investigation (see Table 3). The first configuration, referred to as FLOXCO2 , made use of a 79 vol% CO2 stream (21 vol% air due to air

Fig. 3. Trends of furnace temperature, NO concentration and O2 concentration for FLOXCO2 operated at two different periods.

ingress) as fuel carrier. The intention behind such a large amount of CO2 was to shift the oxidation reaction zone away from the burner. The second configuration, referred to as FLOXAIR , used air as carrier gas as is usual in utility boilers. To keep the overall air ratio constant at 1.15, the combustion air volumetric flow rate was adjusted when air was used as carrier gas. The primary air velocity was set to 15 m/s and 20 m/s for FLOXCO2 and FLOXAIR respectively. In each case, the flame supervision devices were cooled with 5 m3 /h (stp) CO2 . Both configurations have been compared to a reference case of a ’standard’ PC flame generated by a low-NOx swirl burner, referred to as FLAME. In the FLAME configuration, flame supervision devices were cooled with 7 m3 /h (stp) air. This is considered in the calculation of total air ratio in Table 3. The volumetric flow rates were recorded every 10 s (over the whole measurement period). Based on this, average and standard deviation values have been determined. Standard deviations of 5.3 m3 /h (stp) for VFRcombustion air and 0.2 m3 /h (stp) for VFRcoal carrier gas have been found. The coal mass flow rate varied ±1 kg/h around the set point. The FLOXCO2 configuration has been investigated twice: a first time for probe measurements and OH∗ chemiluminescence imaging and, a second time, for measurement of the velocity field. The monitoring of furnace characteristics, such as near-wall temperature at the end of the reaction zone, NO and O2 concentrations at the furnace exit (see Fig. 3) demonstrated a very good reproducibility of the experiment. 3. Measurement techniques 3.1. Reaction zones characterization by OH∗ chemiluminescence imaging The chemiluminescence imaging technique has been implemented on the furnace to characterize the reactions zones in terms of structure and intensity. OH∗ chemiluminescence centered on

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

ARTICLE IN PRESS

JID: CNF 4

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

Table 4 List of species concentrations measured by probe sampling analysis. Analyzer

Group

Species

Meas. range

Analyzer

Group

Species

Meas. range

FTIR

Water vapor Simple carbonaceous compounds

H2 O CO2 CO COS NO NO2 N2 O NH3 HCN SO2 HCl O2 CO2 CO

0–26 0–95 0–80,0 0 0 ppm 0–50 ppm 0–10,0 0 0 ppm 0–10 0 0 ppm 0–200 ppm

FTIR

Alkanes

CH4 C2 H 6 C3 H8 C4 H10 C5 H12 C8 H18 C2 H4 C3 H 6 C2 H2 CH2 O CH2 O2 C2 H 4 O2 C3 H6 O C6 H6 C7 H8 C8 H 8 C9 H12 C10 H8

0–50 0 0 0–20 0 0

Nitrogenous compounds

Paramagnetic NDIR

Sulfur oxides Hydrogen chloride Oxygen Simple carbonaceous compounds

UV Chemilumi-nescence

Sulfur oxides Nitrogenous compounds

SO2 NO NOx

0–10 0 0 ppm 0–50 ppm 0–21 vol% 0–20 vol% 0–50 0 0 ppm 0–5 vol% 0–20 0 0 ppm

308 nm has been selected to ensure, on the one hand, the filtering of visible soot and thermal radiation of hot walls and, on the other hand, the collection of radical spontaneous emission coming from reaction zones [7,10,24]. The optical setup consisted of an ICCD camera Roper Princeton Instruments IMAX (512 × 512 pixels – 16 bits) equipped with a Sodern 45 mm UV lens f/1.8 and three Schott UG11 filters. Squared UV silica quartz windows (100 × 100 mm²) were mounted on the side apertures of the facility by means of a specific holder provided with four nitrogen jets to clear the window from coal particles. The measurement ports limited the effective field of view to a circle of about 188 mm diameter. Image acquisitions have been done for each measurement port (between y = 80 mm and y = 1400 mm from the burner exit). The optical axis of the imaging system was set perpendicularly to the vertical plane of the three jets. The exposure gate time was fixed at 1 ms because of the low OH∗ emission level under MILD flameless conditions. Raw images have been corrected by subtraction of the mean background image, acquired once the facility was shut down. High frequency variations caused by electronic noise have been filtered in a post-processing step by using an adaptive Wiener filter type. For all operating conditions, a series of 20 0 0 images has been recorded from each port to ensure the convergence of the statistical calculations. 3.2. Species concentrations measurements by probe sampling analysis Local concentrations of several major and minor species have been measured by a set of analyzers. In order to avoid any possible fluctuation between them, both lines of analyzers were supplied with sample gas, simultaneously. The sample gas was entrained through the suction probe by a circulation pump before being cleaned from particles in a filter. The complete sample gas line was heated to prevent condensation of water and sulfur. The FTIR spectroscopy analysis has been carried out with a Gasmet DX20 0 0 FTIR analyzer. Based on the absorption spectra, the concentration of several species (see Table 4) has been determined. FTIR can measure all species except O2 which is taken from the paramagnetic analyzer. The last seven groups – from CH4 to C10 H8 – are merged under the designation ‘volatiles’. It was soon realized that NO2 specie could not be reliably detected by FTIR. This is due to the fact that its absorption spectrum superimposes that of an unknown organic compound. This leads to a high degree of uncertainty in the measurement of NO2 and, thus, to no useable

Alkenes Alkynes Aldehydes Carboxylic acids Carbonyls Aromatic hydrocarbons

results. Table 4 presents the major and minor species simultaneously measured with the default measurement line including NDIR, UV, chemiluminescence and paramagnetic analyzers. Gas concentration profile measurements were performed once for each measurement position over a steady-state period of 1 min with data logging every 10 s. These single measurement values were used to calculate the time mean and standard deviation. The time means are used below in the contour plots, and the standard deviations are available in the full data set upon request. Uncertainty in measurement has been evaluated. For paramagnetic, chemiluminescence, UV or NDIR analyzers, the accuracy depends mainly on both calibration and linearity. The related uncertainty is assumed to be 1% of the maximum measurement range. With FTIR spectroscopy, the accuracy depends essentially on the reference spectra, which by their very nature only cover a narrow concentration range. An uncertainty of 2% has been recommended by the manufacturer. However, the possible overlaps of CO2 and H2 O spectra on other species emissions lines may induce additional uncertainty when measuring concentrations in flames [25,26]. It can reach dozens ppm for minor species such as HCN, NO and CO [26,27]. Despite this drawback, FTIR measurements allow a detailed characterization of a large range of major and minor species evolutions. As CO2 , CO, NO and SO2 were measured by both measurement lines simultaneously, a direct comparison could be done to assess the measurement quality. The same radial concentrations profiles have been obtained for the four species in the whole of the measurement domain. The difference between absolute values of CO concentrations remains within the uncertainty range of 2%. For CO2 , SO2 and NO, slight discrepancies are punctually observed in the central coal jet from the burner exit up to 530 mm. Although uncertainties depend firstly on the individual measurement techniques, they could also be attributed to the disruptive effect of the immersion of the probe in the coal jet and to the difference of measurement inertia between analyzers. The standard deviations have been determined: They are in the range of 2% to 6% with regard to the mean value for CO2 and of 4% to 9% for O2 , CO, NO and SO2 . In very reactive regions, standard deviations up to 30% of the mean value were calculated. 3.3. Local velocity measurements by Laser Doppler Velocimetry In-flame velocity measurements were performed by Laser Doppler Velocimetry (LDV). The optical setup consisted of a

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

5

standard 2D LDV back-scattering probe (Dantec 9060 × 0831) equipped with a 10 0 0 mm focal length and a beam expander × 1.98. The relatively large optical window dimensions (100 × 100 mm²) provided by the facility enabled us to measure local velocity farther than the central axial axis. This setup gave a measurement volume of 0.15 × 0.15 × 4.11 mm3 without inducing any experimental bias as it would be with a water-cooled probe. For each measurement port, the probe was horizontally moved along the radial axis by means of a motorized bench. In this configuration, the axial U and tangential W components of the velocity were measured. The grid of local velocity measurements was set to correspond to other local measurements (temperature and species concentrations) carried out by probe sampling. Moreover, the grid attempted to embrace local velocity gradients especially in the turbulent jets. LDV measurements required the seeding of the flow by fine particles. In the present case, this was naturally achieved by coal particles in the central jet and the large recirculation streams in the furnace. The combustion air jet on the probe side was, however, seeded with ZrO2 particles by using a fluidized bed seeding system. The dimensions of the measurement volume in axial and tangential directions are believed to be small enough to consider the measurements punctual without significant volumetric averaging. Assessment of fluctuations induced by turbulence or the radial dimension of the measurement volume was carried out using the rms-velocity (this is available upon request as the complete measurement data set).

3.4. Thermometry by suction pyrometer and heat flux measurements The gas temperatures in the furnace were measured by using a suction pyrometer, which consists of a thermocouple shielded from radiation by a ceramic tube. This tube features an opening which is located close to its front end. The opening directs downstream in the assembled state. Gas is sucked into the pyrometer and heats up the thermocouple via convective heat transfer. The gas flow rate needs to be maintained steady. The application of a suction pyrometer in a coal-fired furnace is challenging in two ways: areas with a high particle density i.e. close to the burner can cause clogging in the probe and the amount of sucked gases must remain low. The latter stems from the averaging character of this measurement technique: The greater the amount of sucked gas is, the wider the temperature field is perturbed. This averaging aspect has to be considered carefully if a comparison between measurements and simulation results e.g. from CFD is foreseen. From the authors experience the averaging can be well captured applying a sphere with a radius of 25 mm as proposed by Parente et al. [28]. An uncertainty of 12 °C for temperatures around 1200 °C has been estimated [28]. This leaves gas temperature fluctuations aside, given the slow response time of the suction pyrometer. Heat flux measurements were made using (i) a heat flux meter for the total heat flux and (ii) an ellipsoidal radiometer for the radiative heat flux to the walls. Probes were set flush with the wall. Both probes allow the hemispherical measurement of total and radiative heat flux, respectively, and are characterized by a long response time. Their design is based on that proposed by the IFRF. The heat flux probe measures the total heat transfer (including conduction, convection and radiation) on its front face. Its principle is based on the measurement of a temperature gradient through a steel plug of known thermal conductivity mounted at the tip of the probe [29]. The ellipsoidal radiometer measures radiative flux through an orifice at its tip and reflects it onto a thermopile mounted at its backend. As done for the heat flux probe, two thermocouples are located at the collection point and further

Fig. 4. Grid of temperature, species concentrations and velocity local measurements.

inside the probe [30]. The ellipsoid is kept clean by steadily purging nitrogen. 3.5. Grid of local measurements A presentation of the measurement grid is given in Fig. 4. This starts 80 mm under the burner and extends 2500 mm longitudinally and 350 mm radially. The angular orientation of the grid is such that it contains the combustion air nozzles. The strong intensity of reaction at the level of the first segment prompts an increase in the density of measurements in this region. With some exceptions located in the burner vicinity, the measurement time for temperature and species concentrations was one minute per position in steady conditions. For LDV measurements, a series of 20 0 0 instantaneous measurements were carried out for a maximum of 90 s at each location. Because of the strong velocity gradients in the present configuration, the grid was refined in the vicinity of the jets and the range of velocity measurements was adapted on both velocity components for each radial profile recording. 4. Global features of pulverized coal MILD flameless combustion This section presents a global characterization of pulverized coal MILD flameless combustion regime achieved with two different carrier gases (FLOXCO2 and FLOXAIR ), compared with the standard FLAME conditions. CO, CO2 , O2 , NOx and SO2 averaged concentrations at the furnace exit are reported in Table 5 for a reference O2 -level of 6 vol% and standard pressure and temperature (stp) conditions. It can be seen that the three configurations are similar in terms of excess O2 concentration. The CO2 concentration depends significantly on the use of CO2 as carrier and cooling gas. As regards the

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

ARTICLE IN PRESS

JID: CNF 6

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13 Table 5 Flue gas analysis results at the furnace exit. Coal: Calentur

FLOXCO2

FLOXAIR

FLAME

O2 , vol%, dry CO2 , vol%, dry CO @ 6 vol% O2 , mg/m3 (stp) NOx @ 6 vol% O2 , mg/m3 (stp) SO2 @ 6 vol% O2 , mg/m3 (stp)

2.8 22.3 188 391 1206

2.9 17.6 50 611 1323

2.9 16.1 172 473 1607

pollutant emissions, FLOXCO2 leads to better results than reference FLAME: NOx and SO2 concentrations decrease by 17% and 25%, respectively, whereas CO emissions increase moderately (+9%). On the other hand, CO emissions decrease in FLOXAIR by 73% when compared with FLOXCO2 . This can be correlated with the variation of the carbon content from ash samplings taken at the furnace exit which is 15.3 wt% and 4.0 wt% for FLOXCO2 and FLOXAIR , respectively. The inhibition in carbon burnout might be caused by the low overall temperature in FLOXCO2 and, additionally, by increased CO2 concentration in the surrounding atmosphere, which reduces the oxygen diffusivity [31–33]. This could also be observed by comparing FLAME and FLOXCO2 : carbon in ash raised from 10.6 wt% to 15.3 wt%, respectively. The application of air as coal carrier gas greatly influences NOx emissions: these increase by more than 50% when compared with FLOXCO2 . This agrees well with the findings of Mei et al. [34] showing the inhibition of fuel N conversion caused by a lack of O2 in the reaction zone. It can also be noticed that NOx emissions in FLAME are less than in FLOXAIR . These unexpected results come from the fact that the FLAME burner was operated in an unstaged low NOx mode. Carbon burnout is inhibited as indicated by the carbon in ash measured at 10.6 wt% and 4.0 wt% for FLAME and FLOXAIR , respectively. Moreover, NO initially formed in the flame might be eventually reduced by char and CO [35,36].The detailed characterization of the effect of the carrier gas composition discussed in Section 6 allows to point out the differences of overall performance of the two MILD flameless combustion modes. The decrease in SO2 emissions in MILD flameless conditions is quite significant and has already been reported by Ristic [37] for experiments in the same facility, but with different coal and burner configurations. So far, no mechanism could have been identified to explain such a decrease. Only assumptions can be made here. The flame temperature is lower under MILD conditions than it is in conventional flame conditions. This allows the desulfurization product CaSO4 , exhibiting thermal instability above 1200 °C, to live on [38]. The mean residence time of the gas and particles is also increased under MILD conditions by the high recirculation, enabling unconverted particles to further desulfurize the gas [38]. Furthermore, the strong mixing of recirculated flue gas with the impinging streams lowers O2 and enhances CO2 partial pressures, respectively. This phenomenon is also known to promote desulfurization reactions [38], and may explain why SO2 emissions are lower in FLOXCO2 . Ash analysis from different positions in the flue gas duct showed an increasing sulfur content in the ash from FLAME to FLOXCO2 conditions. At the position of emission measurements, FLAME and FLOXAIR were found to have similar sulfur content in fly ash, which is however in contradiction with the difference in emissions. Under FLOXCO2 conditions, however, an increase of sulfur content by 23% in fly ash could be identified. Figure 5 shows flames at a vertical distance of 80 mm from the burner under FLOXCO2 , FLOXAIR and FLAME conditions. The difference between the two FLOX regimes is remarkable. The use of CO2 as fuel carrier (FLOXCO2 , Fig. 5a) postpones the combustion, here evidenced by a discernible coal particle stream. When air is used as carrier gas (FLOXAIR , Fig. 5b), an intermittent flame structure

Fig. 5. Photographs at the exit of the burner, for the FLOXCO2 , FLOXAIR and FLAME conditions (from left to right).

is observable. A stable attached visible flame is not possible, contrary to the FLAME reference case. It has been reported elsewhere [3,14,15] that the visual appearance of MILD coal combustion incorporates visible sparks of burning particles, but invisible combustion of volatiles, typically of flameless combustion. These observations are confirmed in both FLOXCO2 and FLOXAIR conditions. 5. Detailed characterization of PC flameless combustion using diluted coal jet (FLOXCO2 ) This section provides a comprehensive analysis of the results obtained under nominal operating conditions of PC MILD flameless combustion with carbon dioxide as carrier gas (FLOXCO2 ). 5.1. The aerodynamic features in the combustion chamber Figure 6 presents the maps of mean and rms of the axial velocity U measured by LDV in a plane containing the 3 jets. The actual measurement locations are given by the small dots. Their interpolation was first done on the right side of the map. The left side was then reconstituted by symmetry. The zero axial velocity iso-line in white - represents the boundary of the recirculation zones characterized by a negative axial component. It is apparent that the aerodynamics field is strongly marked by the high velocity air jets. These are characterized by a length of about 1.8 m, a maximal axial velocity of 115 m/s and turbulent rate of 26.4% at the upper port. The annular coal jet, along the centerline, has a maximum axial velocity of 12.5 m/s and a turbulent intensity of 84%. On account of high momentum, small recirculation zones appear in the near field around air jets: Their maximum axial velocity is equal to −1.5 m/s. At the second measurement location (x = 230 mm), the mean axial velocity between the air and coal jets still remains positive. Local recirculation does not exist anymore between the air and coal jets, indicating the onset of an interaction. As is typical for turbulent round jets, the fluctuations are large along their boundary. One can also notice a slight increase of turbulent rate in the air jets at x = 380 mm, as the result of the interaction between the coal and air jets. The central coal jet is then rapidly entrained by the air jets: Its axial velocity decreases up to x = 530 mm, whereupon, it cannot be distinguished anymore from the air jets. Downstream (x = 1060 mm), the air jets continue their expansion in the combustion chamber. The axial velocity in the centerline region increases and a thin recirculation zone along the wall can be observed. The latter turns out to be stronger in the plane perpendicular to the 3 jets (Fig. 7). Along this plane, the recirculation is only limited by the coal jet expansion (close to y = −100 mm at x = 1060 mm) and large negative axial velocities can be achieved: down to −10 m/s, until the large recirculation zone ends before x = 1890 mm. All these results show that the high momentum flux from the air jets totally controls aerodynamics within the furnace and therefore the achievement of MILD flameless combustion regime. As a matter of fact, the related entrainment process in a confined combustion chamber induces large recirculation zones beside the jets.

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

7

Fig. 6. Maps of axial velocity in the jets plane (left: mean, right: rms).

5.2. Reaction zone identification by OH∗ chemiluminescence imaging Figure 8 presents mean images of OH∗ chemiluminescence obtained from optical accesses normal to the jets plane for the FLOXCO2 and FLAME operating conditions. The difference in location and intensity of the reaction zone can be clearly seen: for a ‘standard’ flame it is attached to the burner, whereas for the flameless combustion regime it is shifted away in the furnace. In the latter case, no OH∗ spontaneous emission, i.e. no reaction zone, can be observed at the exit of the burner. The reaction intensity then increases between x = 400 mm and x = 1200 mm with a maximum at around x = 900 mm. The onset of the reaction zone corresponds to the region where air jets start interacting with the central coal jet (see Fig. 6). The maximum of OH∗ chemiluminescence level here is less than 7% of that obtained from the reference flame. This highlights the very small heat release in this reaction zone, owing to the dilution of the reactants. 5.3. Gas concentrations, temperatures and heat fluxes from probe sampling measurements

Fig. 7. Radial profiles of mean axial velocity in the jets plane and perpendicular to the jets plane.

Throughout their expansion, air jets entrain progressively a part of the central coal jet and a large amount of recirculated hot burnt gases. The main reaction zone takes place in a diluted environment curbing the local heat release density and so the temperature gradient.

Species concentrations and temperature fields are presented in Figs. 9–14. As indicated earlier, the maps presented here were obtained by interpolations of local measurements. The typical structure of the high velocity air jets is found, once again, on the O2 concentration map in the vicinity of the burner exit (Fig. 9). Further downstream, the oxygen concentration decreases as the result of the dilution induced by the entrainment process and the progressive consumption of oxygen in the reaction zone. Oxygen concentration decrease ends at x ≈ 1200 mm and attains the stack value. It has to be noted, that a homogeneous atmosphere is established well upstream the homogeneous flow pattern (x ≈ 1800 mm). The map of volatile concentration presented in Fig. 9 shows that the devolatilization process, resulting from coal pyrolysis,

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF 8

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

Fig. 8. Mean images of OH∗ chemiluminescence (left: FLAME, right: FLOXCO2 ).

Fig. 9. Maps of concentrations of O2 (dry vol%) and volatiles (mol/(n)m3 ).

Fig. 10. Maps of concentrations of CO (dry ppmv) and CO2 (dry vol%).

starts at the burner exit. The volatile concentration decreases from the beginning of the lifted reaction zone. As shown in Fig. 10, the formation of carbon monoxide starts once coal enters the combustion chamber. CO concentration reaches a peak just before the onset of the mixing between air and coal jets. CO is further entrained by the high momentum air jets and generated within these regions as shown by its radial expansion. CO is also progressively oxidized to form carbon dioxide (Fig. 10), down to the end of the reaction zone at x ≈ 1200 mm, in concordance with OH∗ chemiluminescence images (Fig. 8). The persistence of CO compared to volatiles is due

to the fact that CO is produced, amongst others, during volatile decomposition, but also during char burnout reactions. The maps of CO2 and H2 O concentrations in Figs. 10 and 11, respectively, are almost complementary to that given for O2 : the availability of carbon dioxide and water vapor in the high momentum jets progressively increases owing to their formation in the oxidizing reaction zone and their entrainment in the turbulent jets. Downstream the reaction zone and beside air jets, where flue gas is massively recirculated, CO2 and H2 O concentrations are quite homogeneous. As carbon dioxide is chosen as carrier gas, CO2

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

Fig. 11. Maps of concentrations of H2 O (wet vol%) and SO2 (dry ppmv).

concentration is larger at the burner exit until the coal jet is entrained and diluted in the air jet. Some comparisons can be done between major and minor species formed by coal devolatilization in the central jet. SO2 concentration presented in Fig. 11 is similar to that of the main combustion products. Downstream of the main reaction zone, ongoing char burnout reactions can be identified by the slowly increasing SO2 concentration. The map of NH3 concentration (Fig. 12) is similar to that of the volatile species: NH3 is formed in the pulverized coal jet from the injector exit and decreases rapidly in the reaction zone from the beginning of the mixing layer between air and coal jets. The distributions of HCN and CO in the combustion chamber are almost identical. The formation of HCN starts also from the burner exit and its concentration increases up to the beginning of the mixing layer where a maximum is reached (Fig. 12). Then, as for carbon monoxide, HCN is entrained by the air jet and reacts progressively until it has totally vanished at the end of the reaction zone. The map of temperature presented in Fig. 13 shows typical features of MILD flameless combustion regime in a furnace. As the result of the massive recirculation of hot combustion products induced by the high momentum air jets, the thermal distribution is rather homogenous in the combustion chamber aside from the reactant jets. This creates the impression that ’cold’ high momentum turbulent air jets develop in a confined homogenous environment consisting of hot combustion products, and then progressively heat by entrainment process. This is just a conceptual point of view. In reality, a slight progressive increase in temperature in the lifted reaction zone can be observed. The region of maximum of temperature (Tmax ≈ 1100 °C) corresponds to the end of the reaction zone and remains low compared to a ’standard’ – i.e. non-diluted – combustion regime. The diluted MILD flameless regime induced by the large flue gas recirculation is responsible for this low maximum value and, hence, a minimization of NO formation from the thermal route. Downstream in the combustion chamber, heat transfer to the wall induces a progressive decrease

9

Fig. 12. Maps of NH3 and HCN concentrations (dry ppmv).

Fig. 13. Maps of NO concentrations (dry ppmv) and temperature (°C).

in temperature. The homogeneity in heat release is shown by heat flux measurements (Fig. 14). The increase and the peak are less pronounced than for a conventional flame. This was already observed by Orsino et al. [2] and Plessing et al. [39]. In contrast to systems working with highly preheated combustion air like those tested at IFRF [2,40] in Germany [39] or in Japan [41], the overall

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF 10

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

Fig. 14. Profiles of total and radiative heat fluxes in FLOXCO2 and conventional FLAME condition).

temperature level does not increase, which does not alter the radiative heat transfer. The map of NO concentration is presented in Fig. 13. The absence of any local peak of high temperature in the combustion chamber keeps down the formation of nitrogen oxides. In the upper part of the combustion chamber, NO recirculates along with other combustion products making its concentration relatively constant around the air jets. The upstream recirculation of NO in the hot zone around the central coal jet can induce some reduction of nitrogen oxides as an ’auto-reburning’ process and, then, contributes to the low level of NOx emissions [17–19]. A slight formation of NO is observed along the lifted reaction zone. Its maximum concentration is reached close to the end of the reaction zone where the temperature is maximal and the presence of oxygen and HCN are still significant. This tends to show that, as expected in pulverized coal combustion, the main route of nitrogen oxides formation remains the fuel NO route, corresponding to oxidation of nitrogen species generated during coal devolatilization (mainly NH3 and HCN). This however occurs at a limited rate as the result of the low temperature environment. The formation of thermal NO remains marginal as temperatures are well below 1500 °C [42]. As Mei et al. concluded from their computational investigation made with a similar coal, prompt NO and intermediate N2 O route make a very small contribution [34]. The aerodynamical control of fuel NO formation is an ongoing challenge when designing a burner. An idea would be to devolatilize as much fuel N as possible and subsequently reduce it to molecular N2 . During char oxidation regime, it is believed that the release of the remaining fuel N to NO and HCN can hardly be controlled.

6. The effect of carrier gas composition on PC flameless combustion (FLOXCO2 vs. FLOXAIR ) This section assesses the impact of the carrier gas composition on flameless combustion features, by comparing FLOXCO2 (carrier gas CO2 ) and FLOXAIR (carrier gas air) configurations. Zooms on OH∗ chemiluminescence images made at the burner exit for both configurations are given in Fig. 15. The change from CO2 to air modifies the localization and intensity of the reaction zone. For air, a first reaction zone attached to the burner exit can be found. Further downstream, a lifted reaction zone similar to that observed before for FLOXCO2 appears from the beginning of the mixing layers. Higher OH∗ chemiluminescence levels collected for FLOXAIR with regard to FLOXCO2 are the sign of a higher local heat release in the main reaction zone. OH∗ chemiluminescence images obtained from lower ports show that the main reaction

Fig. 15. Mean images of OH∗ chemiluminescence close to the burner exit (left: FLOXCO2 , right: FLOXAIR ). (Note the same scale of false colors for ease of comparison.) (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 16. Comparison of O2 concentrations (dry vol%) maps: FLOXCO2 (left) vs. FLOXAIR (right).

zone terminates at around x ≈ 1200 mm for both operating conditions. Figures 16–22 present direct comparisons of species concentrations and temperature measured for FLOXCO2 and FLOXAIR . For each configuration, the results are obtained from interpolations of local probe measurements. Despite the change in jet flow rate to compensate the use of air as carrier gas, both O2 concentration fields presented in Fig. 16 show external turbulent air jets with very similar shapes. The only noticeable difference is the presence of oxygen in the coal jet for FLOXAIR . Oxygen consumption in the attached reaction zone causes

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

Fig. 17. Comparison of CO concentrations (dry ppmv) maps: FLOXCO2 (left) vs. FLOXAIR (right).

Fig. 18. Comparison of CO2 concentrations (dry vol%) maps: FLOXCO2 (left) vs. FLOXAIR (right).

a first decrease in its concentration at the burner exit. Then, an increase due to the expansion of the turbulent jets, as for FLOXCO2 , can be observed. The first attached reaction zone for FLOXAIR case is accountable for a large amount of carbon monoxide ([CO] > 2%) at the burner exit. In continuity with the first attached reaction zone, high levels of CO have been also measured in the entire main reaction zone (up to x = 1200 mm). Peaks in CO concentration and OH∗ chemiluminescence (Fig. 15) occur concurrently. These appear

11

Fig. 19. Comparison of H2 O concentrations (wet vol%) maps: FLOXCO2 (left) vs. FLOXAIR (right).

Fig. 20. Comparison of SO2 concentrations (dry ppmv) maps: FLOXCO2 (left) vs. FLOXAIR (right).

further downstream and with double intensity than for FLOXCO2 . This highlights an intensification of heat release in the reaction zone, which is also evidenced by the rapid formations of H2 O and SO2 from the coal injector exit (Figs. 19 and 20). As shown in Fig. 18, the formation of carbon dioxide along the reaction zones is more gradual owing to the fact that (i) there is no more CO2 in the carrier gas and (ii) the oxidation of CO is more progressive.

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF 12

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

the coal jet. Most significant NO is observed in the lifted reaction zone. Further downstream, NO concentration is constant and 37% higher for FLOXAIR ([NO] ≈ 330 ppmv) case than for FLOXCO2 ([NO] ≈ 240 ppmv). Despite the large flue gas recirculation and the dilution of reactant jets that arises from it, FLOXAIR exhibits higher local temperature in the mixing layers caused by the attached reaction. This gives rise to larger NO formation in the main lifted reaction zone.

7. Conclusions

Fig. 21. Comparison of NO concentrations (dry ppmv) maps: FLOXCO2 (left) vs. FLOXAIR (right).

This paper presents a detailed analysis of the specific features of flameless combustion applied to pulverized coal. For this purpose, several measurements techniques were implemented on a pilot-scale facility: concentrations and temperature of gaseous major and minor species were obtained from in-furnace probe sampling, local velocity was measured by Laser Doppler Velocimetry and OH∗ chemiluminescence imaging was used to obtain topology and intensity of reaction zones. The flameless burner consisted of two high momentum combustion air jets set symmetrically away on either side of a central pulverized coal annular jet. The study was also focused on the effect of the carrier gas composition on the flameless combustion regime achieved in such configuration. It is shown that the aerodynamics is totally controlled by the developments of the high momentum turbulent combustion air jets and their entrainment of the recirculating flue gas and the central pulverized coal jet. The main reaction zone appears as lifted from the burner exit in the mixing layers of the coal and air jets in diluted conditions as typical of MILD flameless combustion regime. These conditions ensure a stable combustion and a low heat release. Nitrogen oxides are mainly formed in the lifted reaction zone by oxidation of devolatilized nitrogenous species (NH3 and HCN) via the fuel NO route. But the NOx emissions remain limited as the result of the low temperature in the diluted reaction zone. When carbon dioxide is replaced by air as coal carrier gas, a significant change is observed around the coal jet by the presence of a first reaction zone attached to the burner exit. This induces higher CO formation, heat release and temperature level in the main lifted reaction zone, and explains the higher NOx emissions measured. These results also illustrate the interest of experimental study on large-scale combustion facility close to industrial operating conditions with several complementary measurements techniques for detailed analysis of novel combustion concepts. All data is available upon request and can be used as reference test-cases for the development and validation of CFD simulations adapted to pulverized coal flameless combustion.

Full data availability

Fig. 22. Comparison of temperature (°C) maps: FLOXCO2 (left) vs. FLOXAIR (right).

The two temperature fields presented in Fig. 22 confirm the higher local heat release under FLOXAIR conditions. For FLOXCO2 , the thermal distribution remains quite homogeneous, albeit with a slight peak in the vicinity of the coal jet. Below a slight increase in temperature up to 1224 °C at x = 890 mm can be observed. In the end, the temperature decreases due to heat transfer to the walls. Such increases of local heat release and temperature have a direct effect on NO formation in the reaction zone (Fig. 21). NO formation remains low in the attached reaction zone around

The authors would provide to the scientific community the full data set of FLOXAIR and FLOXCO2 cases including burner dimensions, in-flame measurements of gas concentrations and gas temperature, velocity data and wall temperatures. The authors would like to give possibility to the validation of simulation models for MILD flameless conditions. In case of interest please contact the corresponding authors.

Acknowledgment The authors gratefully acknowledge the financial support from the European Commission which supported this project within the Research Fund for Coal and Steel (Project number RFCRCT-20110 0 0 05).

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029

JID: CNF

ARTICLE IN PRESS

[m5G;February 26, 2016;21:12]

M. Weidmann et al. / Combustion and Flame 000 (2016) 1–13

References [1] J. Wünning, J. Wünning, Flameless oxidation to reduce thermal no-formation, Prog. Energy Combust. Sci. 23 (1997) 81–94. [2] S. Orsino, M. Tamura, P. Stabat, S. Constantini, O. Prado, R. Weber, Excess enthalpy combustion of coal, Results of high temperature air combution trials, IFRF Doc. F46/y/3, IFRF, IJmuiden, 20 0 0. [3] R. Weber, J.P. Smart, W.V Kamp, On the (MILD) combustion of gaseous, liquid, and solid fuels in high temperature preheated air, Proc. Combust. Inst. 30 (2005) 2623–2629. [4] A. Cavaliere, M. de Joannon, Mild combustion, Prog. Energy Combust. Sci. 30 (2004) 329–366. [5] J. Wünning, Flameless oxidation, in: 6th International HiTACG Symposium , Essen, 2005. [6] G. Szegö, B. Dally, G. Nathan, Operational characteristics of a parallel jet MILD combustion burner system, Combust. Flame 156 (2009) 429–438. [7] M. Ayoub, C. Rottier, S. Carpentier, C. Villermaux, A.M. Boukhalfa, D. Honoré, An experimental study of mild flameless combustion of methane/hydrogen mixtures, Int. J. Hydrog. Energy 37 (2012) 6912–6921. [8] J. Mi, P. Li, C. Zheng, Impact of injection conditions on flame characteristics from a parallel multi-jet burner, Energy 36 (2011) 6583–6595. [9] P. Li, B.B. Dally, J. Mi, F. Wang, MILD oxy-combustion of gaseous fuels in a laboratory-scale furnace, Combust. Flame 160 (2013) 933–946. [10] C. Rottier, C. Lacour, G. Godard, B. Taupin, L. Porcheron, R. Hauguel, A.M. Boukhalfa, D. Honoré, On the effect of air temperature on mild flameless combustion regime of high temperature furnace, in: 4th European Combustion Meeting, Vienna, 2009. [11] D.W. Pershing, J.O.L. Wendt, Relative contributions of volatile nitrogen and char nitrogen to NOx emissions from pulverized coal flames, Ind. Eng. Chem. Proc. Des. Dev 18 (1979) 60–67. [12] J.C. Chen, S. Niksa, Coal devolatilization during rapid transient heating. 1. Primary devolatilization, Energy Fuels 6 (1992) 254–264. [13] H. Zhang, G. Yue, J. Lu, Z. Jia, J. Mao, T. Fujimori, T. Suko, T. Kiga, Development of high temperature air combustion technology in pulverized fossil fuel fired boilers, Proc. Combust. Inst. 31 (2007) 2779–2785. [14] J.P. Smart, G.S. Riley, Combustion of coal in a flameless oxidation environment under oxyfuel firing conditions, J. Energy Inst. 85 (2012) 131–134. [15] P. Li, F. Wang, Y. Tu, Z. Mei, J. Zhang, Y. Zheng, H. Liu, Z. Liu, J. Mi, C. Zheng, Moderate or intense low-oxygen dilution oxy-combustion characteristics of light oil and pulverized coal in a pilot-scale furnace, Energy Fuels 28 (2014) 1524–1535. [16] M. Saha, B.B. Dally, P.R. Medwell, E.M. Cleary, Moderate or intense low oxygen dilution (MILD) combustion characteristics of pulverized coal in a selfrecuperative furnace, Energy Fuels 28 (2014) 6046–6057. [17] Z. Mei, P. Li, J. Mi, F. Wang, J. Zhang, Diffusion MILD combustion of firing pulverized-coal at a pilot furnace, Flow Turbul. Combust. 95 (2015) 803– 829. [18] E. Masson, S. Maurel, L. Porcheron, F. Aguilé, P. Meunier, A. Quinqueneau, D. Honoré, A.M. Boukhalfa, An experimental characterization of flameless combustion at semi-industrial scale, in: 14th IFRF Members Conference, Noordwijkerhoot, 2004. [19] A. Nicolle, P. Dagaut, Occurrence of NO-reburning in MILD combustion evidenced via chemical kinetic modeling, Fuel 85 (2006) 2469–2478. [20] D. Ristic, Development of a pilot-scale flameless oxidation burner for ultralow NOx combustion of pulverised coal, Final report, Publications Office of the European Union, Luxemburg, 2010. [21] D. Ristic, R. Berger, G. Scheffknecht, C. Lacour, D. Honoré, J.G. Wünning, Experimental study on flameless combustion of pulverised coal under air staging conditions, in: 15th IFRF Members Conference, Pisa, 2007.

13

[22] M. Weidmann, V. Verbaere, G. Boutin, D. Honoré, S. Grathwohl, G. Goddard, C. Gobin, H. Dieter, R. Kneer, G. Scheffknecht, Detailed investigation of flameless oxidation of pulverized coal at pilot-scale (230 kWth ), Appl. Therm. Eng. 74 (2015) 96–101. [23] D. Ristic, M. Zieba, G. Scheffknecht, Effect of near burner aerodynamics on NOx reduction in pulverised coal flameless oxidation, in: 25. Deutscher Flammentag, Karlsruhe, 2011. [24] D. Honoré, Advanced measurements in industrial combustion systems, in: L. Vervisch, D. Veynante, J. van Beeck (Eds.), Turbulent Combustion, VKI Lecture Series 2007-08, von Karman Institute for Fluid Dynamics, Sint-GenesiusRode, 2007. [25] D. Fleig, S. Hjärtstam, D. Kühnemuth, FTIR measurement of NH3 , HCN, SO2 , ˚ H2 S and COS in pulverized lignite oxy-fuel flames, Abo Akademi University, ˚ Abo, Finland, 2009. [26] L.-E. Amand, C.J. Tullin, The theory behind ftir analysis: application examples from measurement at the 12 MW circulating fluidized bed boiler at chalmers, Chalmers University of Technology, Gothenborg, Sweden, 1996. [27] A.S. Bonanno, C.M. Nelson, M.A. Wojtowicz, K.S. Knight, M.A. Serio, P.R. Solomon, In-situ FT-IR measurements at industrial combustion facilities, Optical Remote Sensing for Environmental and Process Monitoring (September 25–27, 1995), San Francisco, CA, pp. 611–616. [28] A. Parente, G. Coraggio, C. Galetti, L. Tognotti, Verification, validation and uncertainty quantification in industrial combustion modelling: some practical tools, IFRF Doc. G25/y/01, IFRF, Livorno, 2010. [29] N. Fricker, How do I measure total heat transfer with a heat flux meter? Combustion File No. 138, IFRF online combustion handbook C76/y/1/6, IFRF. [30] N. Fricker, How do I measure thermal radiation with an ellipsoidal radiometer? Combustion File No. 136, IFRF online combustion handbook C76/y/1/3, IFRF. [31] D. Kim, S. Choi, C.R. Shaddix, M. Geier, Effect of CO2 gasification reaction on char particle combustion in oxy-fuel conditions, Fuel 120 (2014) 130–140. [32] J. Brix, P.A. Jensen, A.D. Jensen, Coal devolatilization and char conversion under suspension fired conditions in O2 /N2 and O2 /CO2 atmospheres, Fuel 89 (2010) 3373–3380. [33] S. Leiser, Numerical simulation of oxy-fuel combustion (Ph.D. thesis), University of Stuttgart, Shaker, Aachen, 2011. [34] Z. Mei, P. Li, F. Wang, J. Zhang, J. Mi, Influences of reactant injection velocities on moderate or intense low-oxygen dilution coal combustion, Energy Fuels 28 (2014) 369–384. [35] A. Molina, E.G. Eddings, D.W. Pershing, A.F. Sarofim, Nitric oxide destruction during coal and char oxidation under pulverized-coal combustion conditions, Combust. Flame 136 (2004) 303–312. [36] L.K. Chan, A.F. Sarofim, J.M. Beér, Kinetics of the NO-carbon reaction at fluidized bed combustor conditions, Combust. Flame 52 (1983) 37–45. [37] D. Ristic, Feasibility and NOx reduction potential of flameless oxidation in pulverised coal combustion (Ph.D. thesis), University of Stuttgart, Cuvillier, Göttingen, 2012. [38] J. Cheng, J. Zhou, J. Liu, Z. Zhou, Z. Huang, X. Cao, X. Zhao, K. Cen, Sulfur removal at high temperature during coal combustion in furnaces, Prog. Energy Combust. Sci. 29 (2003) 381–405. [39] T. Plessing, N. Peters, J.G. Wünning, Laseroptical investigation of highly preheated combustion with strong exhaust gas recirculation, Symp. (Int.) Combust. 27 (1998) 3197–3204. [40] R. Weber, S. Orsino, N. Lallemant, A. Verlaan, Combustion of natural gas with high-temperature air and large quantities of flue gas, Proc. Combust. Inst. 28 (20 0 0) 1315–1321. [41] G.-M. Choi, M. Katsuki, Advanced low NOx combustion using highly preheated air, Energy Convers. Manag. 42 (2001) 639–652. [42] H. Stadler, D. Ristic, M. Förster, A. Schuster, R. Kneer, G. Scheffknecht, NOx emissions from flameless coal combustion in air, Ar/O2 and CO2 /O2 , Proc. Combust. Inst. 32 (2009) 3131–3138.

Please cite this article as: M. Weidmann et al., Experimental characterization of pulverized coal MILD flameless combustion from detailed measurements in a pilot-scale facility, Combustion and Flame (2016), http://dx.doi.org/10.1016/j.combustflame.2016.01.029