Radiative Heat Flux Measurement in a Semi-Industrial Oxyfuel Combustion Chamber with Biomass and Coal

Abstract

Oxyfuel is a combustion technology where the oxidant consists mainly of oxygen and carbon dioxide instead of oxygen and nitrogen. Since carbon dioxide has strongly absorbing bands in the thermal spectrum, the radiation properties of the flame change in an oxyfuel atmosphere compared to conventional combustion. When retrofitting an existing air-fired combustion system to an oxyfuel process, the oxygen content in the oxidant must be adjusted so that similar values for heat transfer by radiation are achieved. This measure allows the system to be operated with otherwise unchanged parameters. In this work, the thermal radiation of natural gas, pulverised walnut shells and lignite under an air and oxyfuel atmosphere is investigated in a semi-industrial combustion chamber with water-cooled membrane walls, at different oxygen concentrations and combustion parameters. While the radiative heat fluxes for natural gas with an oxygen content of 28 vol% in the oxidant are significantly higher than those for firing with air, the values for lignite are still below the air-firing, even with an oxygen content of 30 vol%. For walnut shells, the oxyfuel results are close to the air case for all oxygen concentrations between 27 and 33 vol%. The walnut shells show higher radiative emissions than the lignite at the same thermal output. For non-swirled flames, the radiative heat flux is lower than for swirled flames.

1. Introduction

Oxyfuel combustion is a promising CO₂-capture technology for the reduction of greenhouse gas emissions in the energy sector. This combustion technology replaces atmospheric nitrogen with carbon dioxide, so the oxidant consists of a mixture of oxygen, carbon dioxide and water vapour. By condensing the water vapour, the carbon dioxide can be separated from the exhaust gas and stored in deep grounds, or to further processing in the chemical industry. While nitrogen is a non-radiating gas, carbon dioxide and water have strong absorbing bands in the thermal spectrum. As a result, the radiative properties in a combustion chamber with an oxyfuel firing system change compared to a conventional firing system with air. Since a large amount of heat is transferred by thermal radiation in industrial systems using diffusion flames, detailed knowledge of the radiative properties in oxyfuel systems is useful when designing a commercial oxyfuel combustion chamber. Especially in the case of an existing air-fired plant that is retrofitted with oxyfuel firing, it is important to adjust the operating parameters for the combustion process so that similar temperatures and heat transfer in the chamber occur.

Due to the high CO₂ content and water vapour in oxyfuel combustion, the non-luminous share of the radiation increases. Because of their high emissivity, the heat transfer by radiation in a combustion chamber can be higher than in conventional furnaces at a comparable gas temperature [1,2]. The luminosity of the flame is primarily affected by the particle and soot concentration, although studies by Wiliams et al. [3] have shown that less soot is formed during oxyfuel combustion than in air combustion [2].

Zabrodiec et al. [4] investigated the total incident thermal radiation of a 40 kW_th flame using an ellipsoidal radiometer in a refractory-lined combustion chamber with a diameter of 400 mm. Torrefied biomass and lignite were used as fuels in an air and oxyfuel atmosphere. It is noted that it is difficult to separate the radiation of the flame from the surrounding radiation, mainly from hot walls. In order to exclude the influence of the walls, the background radiation was measured immediately after the flame was extinguished. It has been found that higher values for the incident radiation at the wall are recorded for air-firing with both fuels than for oxyfuel firing with 27% oxygen content. In addition, the torrefied biomass shows higher values for both air and oxyfuel combustion than lignite. They concluded that the low calorific value of the biomass requires a higher mass flow, and that the higher particle density means the radiation is more intense than for lignite. However, the measurement results for the radiative heat fluxes are only given in relative values. Not a single absolute value is given. This makes it difficult to assess the plausibility of the radiation characteristics in the combustion chamber at a thermal power of 40 kW. Smart et al. [5] investigated the convective and radiative heat transfer under oxyfuel conditions for coal with co-fired biomass at a thermal power of 500 kW. In addition, they investigated the influence of the exhaust gas recirculation rate on the radiative properties. Measurements were taken with a MEDTHERM digital radiation heat flux meter [6] in a refractory-lined combustion chamber with a diameter of 800 mm. Their results show that the total radiative heat transfer measured at the wall decreases as the exhaust gas recirculation rate increases. The residual oxygen of 3 vol% was kept constant during the investigation of all trials. Ramadan et al. [7] investigated the temperature and thermal radiation of a 200 kW_th natural gas flame in a 630 mm wide combustion chamber with refractory-lined walls using different swirl numbers. They used an O₂ to CO₂ ratio of 17/83 mass%. It has been found that the measured incident thermal radiation at the wall increases as the swirl increases. Compared to an air-firing with the same swirl number, the values for the incident radiative heat flux are 25% higher than for the oxyfuel case. The authors explain the higher radiation of the air flame by the fact that nitrogen absorbs less heat than CO₂ in the oxyfuel flames. They state that the oxygen content in oxyfuel combustion must be increased in order to achieve the same values for radiative heat flux at the wall as in combustion with an air atmosphere. However, the O₂/CO₂ ratio is not varied in their work in order to provide a statement about the O₂ content in the oxidant necessary to reach radiation conditions similar to those of an air flame. For their measurements, they used an ellipsoidal radiometer. Corrêa da Silva and Krautz [2] used an ellipsoidal radiometer to investigate the influence of various burner parameters on radiative heat transfer in a square 1 × 1 m boiler with water-cooled membrane walls. Dried lignite was used as fuel at a thermal power of 400 kW. A similar radiative heat transfer was achieved with an oxygen content of 31% in the oxidant stream. A series of other authors investigated the influence of the exhaust gas recirculation rate for oxyfuel combustion [5,8,9].

For semi-industrial scale oxyfuel combustion and measurements of radiative heat flux, studies mainly with refractory-lined combustion chambers can be found in the literature. An exception mentioned in this section is the work of Corrêa da Silva and Krautz [2]. In industrial applications, water-cooled membrane walls are usually used. Water cooled tubes represent a heat sink compared to refractory-lined walls, which emit most of the absorbed heat back into the combustion chamber. For a detailed investigation of the flame in a semi-industrial plant under oxyfuel conditions, the oxyfuel burner used by Zabrodiec et al. [4] was up-scaled from 40 kW_th to 500 kW_th nominal power at the Institute for Energy Systems and Technology (EST) at the Technical University Darmstadt in Germany [10]. In order to investigate the impact on radiative heat transfer due to up-scaling, measurements were carried out at the EST with a Gardon Gauge sensor. The objective of this work is to investigate the thermal radiation of pulverised biomass in a semi-industrial combustion chamber with water-cooled walls. Grounded walnut shells (WS) are used as fuel, since they do not have fibrous properties like most biomass and are most similar to the structure of pulverised coal particles. To show the impact of heterogeneous and pure homogeneous combustion under oxyfuel conditions on thermal radiation, natural gas (NG) is also used as an additional fuel. The influence of oxygen enrichment in the oxidant and different combustion parameters, such as swirl and air distribution between secondary and tertiary channel, are discussed. In addition, Rhenish lignite (RBK) is used as a reference fuel.

2. Materials and Methods

2.1. Oxyfuel Combustion Chamber

Figure 1 shows the top-down fired oxyfuel combustion chamber, in which the experiments were carried out.

The upper section of the combustion chamber consists of a 4160 mm high water-cooled membrane wall comparable to commercial boilers. The upper cross-section is hexagonally shaped. The lower end of the membrane wall is connected to a 2650 mm high refractory-lined wall, which has a circular internal diameter of 750 mm. At the bottom of the refractory-lined wall, the flue gas leaves the combustion chamber through a 680 × 310 m opening and is conducted through a convective heat exchanger. The combustion chamber and the associated cooling system are designed for 1 MW_th. Furthermore, the chamber has numerous measuring ports on the membrane wall. The burner has a natural gas lance in the centre, which is primarily used to heat up the combustion chamber. When a temperature of 600 °C is reached in the chamber, pulverised fuel can be injected through the primary, annular burner channel. A cooling channel is located next to the pulverised fuel channel, which is not used in the current experiments, but serves as a bluff body. A narrow channel for axial flow and a channel with 45° angled vanes, to generate swirl flow, are located at the outer section of the burner. These two channels are used for the secondary air. An additional channel with approx. 40° angled swirl vanes surrounds the gas lance. The burner is mounted in a quarl, which has an opening angle of 21 degrees. The quarl is surrounded by 24 tertiary nozzles arranged in a ring, each with a diameter of 28 mm. The combustion chamber can be operated with air or a mixture of recirculated flue gas with oxygen enrichment. The total radiative heat transfer was measured at the ports with the numbers 1–4. The axial distance (z) is measured starting from the quarl. Measurements at the upper two ports of the combustion chamber are not possible due to insufficient accessibility. The gas temperatures near the wall are measured at position T₁ and T₂.

2.2. Operation Conditions

In the air-fired operation, the combustion air is taken from the environment via a fan and distributed to the different burner channels. The distribution of the air can be adjusted using control valves. The transport air volume flow for the solid fuel is kept constant through a second fan. In the oxyfuel mode, the flue gas is recirculated and enriched with oxygen before distribution to the burner channels. The water vapour in the flue gas is not condensed specifically during recirculation. However, some of the water vapour condenses in the lines, and can be removed after the experiments. The oxidant for the combustion of the fuel is divided between the secondary channels and the tertiary nozzles. A distinction is made between the local and global equivalence ratio Equation (1) and Equation (2), respectively.

$${\lambda }_{loc}=\frac{{\dot{V}}_{O_2,prim}+{\dot{V}}_{O_2,sec}}{{\dot{V}}_{O_2,st}}$$

(1)

$${\lambda }_{glob}=\frac{{\dot{V}}_{O_2,prim}+{\dot{V}}_{O_2,sec}+{\dot{V}}_{O_2,tet}}{{\dot{V}}_{O_2,st}}$$

(2)

The local combustion equivalence ratio ${\lambda }_{loc}$ consists of the primary ${\dot{V}}_{O_2,prim}$ and secondary ${\dot{V}}_{O_2,sec}$ stream. The global combustion equivalence ratio ${\lambda }_{glob}$ additionally includes the tertiary ${\dot{V}}_{O_2,tet}$ nozzles. This allows an air staging for the reduction of NO_x emissions with ${\lambda }_{loc}$ < 1 and full oxidation of the fuel with ${\lambda }_{glob}$ > 1. The necessary oxygen requirement for a stoichiometric ratio ${\dot{V}}_{O_2,st}$ is calculated from the elemental analysis and the mass flow of the fuel Equation (3).

$${\dot{V}}_{O_2,st}=\frac{{\dot{m}}_{fuel}}{{\rho }_{O_2}}\left(Y_C\frac{M_{O_2}}{M_C}+\frac{1}{2}{Y}_H\frac{M_{O_2}}{M_{H_2}}+Y_S\frac{M_{O_2}}{M_S}-Y_O\frac{M_{O_2}}{M_{O_2}}\right)$$

(3)

where ${\dot{m}}_{fuel}$ is the mass flow of the solid fuel, ${\rho }_{O_2}$ the density of oxygen at standard conditions, $Y_i$ the mass fraction of the element i and $M_i$ the molar mass of element i. The transport flow for the solid fuel in oxyfuel operation mode consists of 15 vol% oxygen and 85 vol% carbon dioxide. A transport flow rate of 100 Nm³/h through the primary channel is used for all thermal loads in air operation and 110–120 Nm³/h for oxyfuel. All pulverised fuel-fired operating points are operated with a natural gas support flame of 1–2 Nm³/h in order to ensure the stability of the flame in the case of fluctuations in the dosing of the solid fuel. Since the swirl of the burner can be adjusted via a secondary, axial flow channel and a channel with swirl vanes, the swirl number S is calculated according to Equation (4).

$$S=\frac{L_{swirl}}{r{I}_{axial}}\approx \frac{r_{swirl}{\dot{V}}_{swirl}{u}_{swirl}cos\left(\phi \right)}{r\left[{\dot{V}}_{swirl}{u}_{swirl}sin\left(\phi \right)+{\dot{V}}_{axial}{u}_{axial}\right]}$$

(4)

where $L_{swirl}$ is the angular momentum of the swirl channel, $I_{axial}$ the momentum of the axial channel and the momentum of the axial component of the swirl channel. $r_{swirl}$ is the mean radius of the swirl channel, r the radius of the vortex and $\phi$ the angle of the swirl vanes. Here, r is taken as the mean value between $r_{swirl}$ and $r_{axial}$. The primary swirl channel is not used during pulverised fuel operation.

2.3. Measurement Equipment

A Gardon Gauge Heat Flux Sensor type GG01-250-SW from the company HUKSEFLUX [11] is used to measure the incident total radiative heat flux on the inside of the combustion chamber wall. The sensor is able to measure a heat flux up to 250 kW/m². The sensor is named after its inventor Robert Gardon [12]. The sensor is shown schematically in Figure 2 in copper and dark grey. In addition, the sensor is mounted in an in-house built lance that has purge nozzles to protect the sensor window from slagging.

The sensor consists of a water-cooled copper block, which serves as a heat sink. In the centre of the copper housing, a cable is connected to a blackened foil which absorbs the incident radiation. Another cable is connected to the heat sink. Together, these two cables form a thermocouple. The radiative heat flux is absorbed by the black foil, which is made of constantan, and subsequently dissipates the heat along its radius to the surrounding heat sink. The temperature profile across the radius of the foil is illustrated with a red line in Figure 2. The radiative heat flux $\dot{q}$ can be calculated theoretically by using Equation (5).

$$\begin{array}{c}\hfill \dot{q}=\frac{4\lambda \delta }{R^2}(T_c-T_0)\end{array}$$

(5)

where $\lambda$ is the thermal conductivity, $\delta$ the thickness and R the radius of the foil. $T_c$ is the centre temperature of the black foil and $T_0$ the temperature of the heat sink [12,13,14]. In practice, the heat flux results from the voltage signal U and a constant S ($\dot{q}=U/S$), which is determined during calibration and specified by the manufacturer in the data sheet. In order to exclude convective heat transfer from the measurement, a sapphire window is attached to the front of the sensor, which, according to the manufacturer, has a transmissivity of 86% at a wavelength of 1 µm. This transmissivity is almost constant for the spectral range of 0.2–5.5 µm and is therefore within the required measuring field. As the sensor measures relatively, the error due to the sapphire glass is negligible. Together with the protective window, the sensor has an effective incidence angle of 150°, which means that the entire hemisphere is not covered. According to the manufacturer, the resulting error is also negligible, since the part of the field of view that is cut off has a very small angle of incidence to the sensor surface and has approximately the same temperature as the sensor. The calibration error of the sensor is specified by the manufacturer as approx. 5.8%. Calibration is carried out in accordance with the ISO 14934-3 standard [15]. The uncertainty caused by the transmitter and the control unit is low. In addition to the calibration uncertainties, there are other uncertainties which are difficult to determine, such as spectral properties of the flame. Therefore, the error shown later in the results relates to the time averaging of the measurement data. The measuring points are measured at each port (1–4) at a stable conditions for 5 min.

2.4. Fuel Properties

The fuels used for the experimental investigation of radiative heat transfer in an air and oxyfuel atmosphere are natural gas, Rhenish lignite (RBK) and walnut shells (WS). The solid fuels are milled to dust. Their particle size distribution is shown in

Table 1. Particle size distribution for WS and RBK. Mass fraction of particles larger than the specified mesh size.

	20 µm	63 µm	125 µm	180 µm	250 µm	300 µm	355 µm	425 µm
WS	100	99.8	74.1	13.6	1.4	1.0	0.6	0.3
RBK	99.9	82.6	60.3	43.0	23.9	10.2	6.3	4.2

. The particle size distribution was determined using a sieve analysis with different mesh sizes ranging from 20 µm to 425 µm.

While the WS have an approx. constant diameter, a distinct scattering of particle sizes can be recognised for the RBK. The results of the elementary and proximate analysis are summarised in

Table 2. Elementary and proximate analysis for WS and RBK in mass per cent. The respective calorific value is given in MJ/kg.

	Walnut Shells			Rhenish Lignite
	AR	Dry	DAF	AR	Dry	DAF
C	47.60	51.26	51.65	57.07	63.92	68.12
H	6.04	6.50	6.55	4.99	5.59	5.96
O	38.16	41.09	41.41	19.86	22.24	23.70
N	0.36	0.39	0.39	1.77	1.98	2.11
S	0.00	0.00	0.00	0.09	0.10	0.11
Ash	0.70	0.75	–	5.51	6.17	–
Water	7.14	–	–	10.71	–	–
C (fix)	11.30	12.17	12.26	32.28	36.15	38.53
Volatiles	80.86	87.08	87.74	51.50	57.68	61.47
LHV	16.01	17.24	17.37	18.43	20.64	22.00
HHV	18.88	20.33	20.49	22.15	24.81	26.44

. The quantities are given in mass per cent for as received (AR), dried and dry and ash free (DAF). The lower heating value (LHV) and the upper heating value (HHV) are given in MJ/kg.

It should be noted that the WS have a lower calorific value than the RBK, which means a higher particle mass flow is required for the WS to achieve the same thermal output as for the RBK in the combustion chamber. The RBK has a stoichiometric oxygen requirement of 1.2 Nm³/h, whereas the WS have an oxygen requirement of 0.96 Nm³/h. This means a different oxidant volume flow is required for the same equivalence ratio for both fuels. As the burner was originally designed for the RBK, lower velocities at the burner result for the WS, which leads to an unstable inner recirculation zone. To avoid this, either a higher thermal output can be used or less oxidant can be fed through the tertiary nozzles, which means a sub-stoichiometric zone in the flame core is no longer possible. As the WS have a relatively low nitrogen content, the air staging for NO_x reduction is less relevant than for the RBK.

The natural gas consists of 92.6 mol% methane, 4.2 mol% ethane, 1.4 mol% carbon dioxide, 1 mol% nitrogen and 0.8 mol% other components. The calorific value is 40.7 MJ/m³, and the standard density is 0.779 kg/m³.

3. Results and Discussion

3.1. Results for Natural Gas

The radiative heat flux measurements for natural gas were carried out at a thermal load of 310 kW, a swirl number of 0.96, and a global stoichiometric ratio of 1.4. The oxyfuel cases were performed by enriching the recirculated flue gas to an oxygen content of 28 vol% (Oxy28), 29 vol% (Oxy29) and 31 vol% (Oxy31) at a temperature of 50 °C. Figure 3 shows the heat flux measurement plotted against the axial distance to the quarl.

On the left-hand chart, a maximum heat flux of 60 kW/m² is measured for air-firing, which is reached at a position of 0.65 m. For oxyfuel combustion it can be seen that the heat flux in the upper part of the combustion chamber increases with an increasing oxygen concentration. Compared to the air case, the oxyfuel points already reach their maximum at 0.45 m with up to 75 kW/m² and decrease continuously with further axial distance. The adiabatic flame temperature increases with increasing oxygen content in the oxidant (cf. Corrêa da Silva et al. [2]), therefore higher radiative heat flux is measured at higher oxygen concentrations. From a distance of 0.95 m, all measuring points are close together. The end of the flame is possibly located here, and the background radiation of the combustion chamber is measured. In the chart on the right, the air flame with a swirl number of 0.96 is compared to a non-swirled air flame with the same conditions otherwise. As expected, the short, highly swirled air flame has stronger emissions than the non-swirled jet flame. Due to the high swirl, the reaction zones and, thus, the heat release are concentrated on a smaller local area than with the jet flame, where the reaction zones extend over a longer distance. All three operating points are met at the 1.25 m position, since this position marks the end of the flame. The length of the flame can also be seen through windows installed at the same height as the measuring ports. As described in [7], the emissivity of the flame decreases with a lower oxygen concentration, as the increased proportion of CO₂ during combustion reduces the gas temperature. In addition, a higher CO₂ volume flow results in an unchanged equivalence ratio, due to a lower oxygen concentration in the oxidant. In the case of oxyfuel firing, the radiative heat flux decreases with increasing distance from the burner quarl, whereas the combustion with air first increases and then decreases. One reason for the maximum close to the burner could be the higher O₂ concentration near the burner, which leads to a stronger reaction of the mixture in this zone and, thus, to higher heat release rates, higher temperatures and faster ignition. The near-wall gas temperatures inside the combustion chamber for the various flames are shown in Figure 4 at position T₂.

While the oxyfuel cases have almost the same wall temperature of approx. 800 °C, the value for the air case is below the values of the oxyfuel cases, as with the radiative heat flux measurement. The non-swirled air flame has the lowest temperature here, at just under 500 °C.

3.2. Results for Walnut Shells

The pulverised WS are burned at a thermal load of 500 kW, which corresponds to a mass flow of 110 kg/h. The swirl number is 0.95, and the global stoichiometric ratio is approx. 1.4 for air and 1.6 for oxyfuel combustion. Lower ratios show a significant increase in CO emissions for WS. The local equivalence ratio is changed between 0.8 and 1.0 for air combustion and 1.0 to 1.4 for oxyfuel. The oxygen concentration in the recirculated flue gas is enriched to 27 vol% (Oxy27), 30 vol% (Oxy30) and 33 vol% (Oxy33). The left-hand diagram in Figure 5 shows the results of the radiative wall heat flux measurements plotted against the axial distance to the quarl for air combustion and oxyfuel.

It is immediately noticeable that the shape of the curves does not fall continuously over the axial distance to the quarl, as in the oxyfuel cases for natural gas, but first rises and then falls again as in the air-fired natural gas case. Since particle radiation dominates for pulverised firing, the soot concentration probably has a smaller effect than in natural gas combustion. The highest incident radiation of 112 kW/m² was measured at the 0.65 m position for Oxy33, due to highest O₂ concentration and gas temperatures. In contrast to the investigations with natural gas, deviations of the oxyfuel operations from the air point are significantly lower, whereby Oxy27 and Oxy30 are closest to the values of the air point. At the lowest measuring port, the results for the different flames do not overlap, as is the case of natural gas combustion, as the dust flame is longer than the gas flame. The right-hand diagram in Figure 5 shows the comparison of an air point and Oxy30 with low and high local stoichiometric ratios. The global equivalence ratio is 1.6 for all points. The results from the diagram on the left are also shown for comparison. The sub-stoichiometric condition reduces the NO emissions for the air point from 94 to 61 ppm. CO emissions, on the other hand, increase from 580 to approx. 4900 ppm, as the lower velocities in the flame lead to poor mixing and combustion. For the oxyfuel case, NO is reduced from 87 to 48 ppm, and the CO emissions increase from 4800 to 8800 ppm. In this case, a lower radiative heat flux and temperature is measured than in the comparable stoichiometric case. The low conversion of carbon monoxide, which indicates poor combustion, may leads to lower heat release. Furthermore, a long flame is generated which distributes the heat radiation over a larger area. Both the air and the oxyfuel case show significantly lower values compared to the flames with a higher local equivalence ratio, but otherwise the same boundary conditions. The change in the local equivalence ratio has a stronger effect on the differences in thermal radiation in the air case than in the oxyfuel case. However, the difference between the local equivalence ratio in the air case is 0.2, while in the oxyfuel case, the difference is 0.4.

The difference in heat release can also be seen from the near-wall gas temperatures at T₂ in Figure 6. The air case with an stoichiometric local equivalence ratio reaches the highest temperature, although the lowest radiative heat flux is measured on average over the burner axis under air conditions. As the equivalence ratio for air firing is slightly lower than for oxyfuel firing, the adiabatic flame temperature is higher. Since particle radiation dominates for radiative heat exchange, it is assumed that this does not have a significant effect on heat transfer by radiation. The temperature for the same operation point at ${\lambda }_{loc}$ = 0.8 is approx. 250 K lower. The reduced velocity near the burner for small local stoichiometric ratios may lead to a poor mixing of fuel and oxidiser and, thus, to a lower heat release rate, which is reflected in the temperature.

3.3. Results for Rhenish Lignite

The RBK is burnt with a thermal output of 500 kW, a combustion equivalence ratio of approx. 1.6, and a swirl number of 0.95. The oxygen content in the oxyfuel operation is 26 vol% (Oxy26) and 30 vol% (Oxy30).

Figure 7 shows the radiative heat flux plotted against the axial distance to the quarl in the diagram on the left-hand side for air and oxyfuel combustion. The course of the measuring results along the axial direction are similar for all operating conditions, except for the air-fired case, where there is a significant drop in radiative heat flux from 0.95 m. The reason for this drop is fluctuations in fuel feeding. These are caused by the discontinuous feeding of fuel into the dosing system which, in some cases, leads to fuel blockages. This was also noticeable in the drop of the wall temperatures. The maximum incident radiative heat flux of 92 kW/m² is measured for the lignite at 0.65 m. Both oxyfuel cases (Oxy30 and Oxy26) are below the values of air-firing. A higher oxygen concentration >30 vol% is therefore necessary for the lignite in order to obtain values similar to those of the air case. In comparison with the WS, which are close to the air-fired case for all oxygen concentrations in oxyfuel operation, higher deviations are recognisable here. In Figure 7, on the right-hand diagram, the radiative heat flux of the lignite for air and oxyfuel operation is compared with the measurements of the WS. It can be seen for both the air and the oxyfuel case that the measured incident radiation is significantly higher for the WS than for the lignite. Zabrodiec et al. [4], who investigated RBK and torrefied biomass, came to the same result. Their assumption is that radiation emissions from the flame are larger due to the higher particle load of the biomass, which is necessary to achieve the same thermal output due to the lower calorific value of biomass. However, it is also possible that the higher volatile content of the WS results in faster ignition and, thus, higher local temperatures.

Figure 8 shows the near-wall gas temperature of RBK for air and oxyfuel firing. One would expect the Oxy30 case to be closer to the air case if the radiative heat flux in Figure 7 is considered. However, since fluctuations in the fuel supply occurred during this experiment, as already mentioned, the temperature dropped, so these two cases produce similar temperatures. In addition, the adiabatic flame temperature decreases with decreasing oxygen concentration in the oxidant in an oxyfuel atmosphere. Due to the high heat capacity of CO₂, the temperature is lower than the adiabatic flame temperature of air, which has an oxygen concentration of 21 vol% in the oxidant. This results in the lowest temperature for Oxy26 in the investigation of the RBK.

Despite different boundary conditions, in the study of Corrêa da Silva and Krautz [2] a coal-fired oxyfuel combustion chamber with water-cooled membrane walls also shows the best agreement with a comparable air-fired case at an oxygen concentration of 31 vol%. Overall, the values for the radiative wall heat flux in their work are higher than the values measured here, despite a lower thermal power of less than 400 kW. This may be due to the fact that the coal has a higher calorific value, different properties of the chamber walls or cooling, higher flame temperatures, and different mixing conditions. In general, it is difficult to determine the radiative heat transfer in absolute terms due to reflective walls. Therefore, an absolute comparison with most measurements in the literature, such as [4,7], is not possible, as combustion chambers have refractory-lined walls, which have different absorption and emission properties than water-cooled steel walls. Other authors, such as Zabrodiec et al. [4], therefore only make comparisons in relative terms.

Since the oxyfuel burner used for this study is an up-scaled version of the 40 kW_th burner used by Zabrodiec et al. [4] in their work, the results for the radiative heat flux measurements should be compared to each other. As the combustion chambers have different dimensions, and the measurements were taken at different heights, the distance to the quarl is presented in dimensionless values ($z/d$) and the position of the maximum measured radiative heat flux is indicated. The measurement data are difficult to compare, since Zabrodiec et al. [4] used a combustion chamber with heated, refractory-lined walls, whereas in the current study, the combustion chamber has cooled steel walls. For this purpose, all absolute values are normalised by the maximum measured value. The RBK used in both studies is approx. identical.

Figure 9 shows the results in terms of normalised radiative heat flux (RHF) of Zabrodiec et al. [4], and those of the current study plotted against the dimensionless axial distance to the quarl. The torrefied biomass (TB) used by Zabrodiec et al. [4] has a higher calorific value than the WS, so their results for RBK and TB are closer than RBK and WS in the current work. Major differences in absolute values may caused by the refractory-lined walls of the reference combustion chamber. At position 0.45, the measurements differ for RBK by 0.17 between air and oxyfuel and, at position 0.65, by 0.13 for the 40 kW_th case. For the 500 kW_th case, the first two positions of RBK differ by 0.17 and 0.18 in oxyfuel and air mode. Thus, the differences in thermal radiation in relative values near the maximum for RBK show good agreement for both of the combustion chambers with different thermal loads. The remaining points beyond the maximum value cannot be compared because, as already mentioned, the radiative heat flux measurement for the 500 kW_th case with RBK is distorted by fluctuations in fuel feeding. TB differs by 0.14–0.2 between air and oxyfuel for the 40 kW_th case, while the 500 kW_th cases differ by 0–0.04 between air and oxyfuel for WS. Since the carbon content of TB is higher than that of WS, the endothermic Boudouard reaction at high temperatures may lead to lower local temperatures and radiation intensity of the TB flame. Therefore, the difference in RHF between the air and oxyfuel flame could be greater for TB in the 40 kW_th case than WS at 500 kW_th. Another reason could be the worse combustion of TB under oxyfuel conditions. Both theories are supported by the significantly increased CO concentration measured by Zabrodiec and co-workers during TB combustion with 27 vol% oxygen in the oxidant.

4. Conclusions

In a semi-industrial combustion chamber with water-cooled membrane walls, natural gas, milled walnut shells, and lignite were burnt in an air and oxyfuel atmosphere. NG was investigated to exclude the impact of an oxyfuel atmosphere on pyrolysis and char burnout. The incident radiative heat flux on the wall in the combustion chamber was measured. For oxyfuel combustion, the oxygen concentrations were varied, and the influence of the swirl number and the distribution of the oxidant to different burner channels was investigated.

The investigation has shown that the homogeneous and heterogeneous reactions of solid fuels under oxyfuel conditions probably have a significant influence on the ignition and, thus, the flame temperature and radiative properties. While WS with a volatile content of 80.86% result in higher values for the radiative heat flux, already at 27 vol% O₂, the results of RBK with a volatile content of 51.5% and 30 vol% O₂ are still below the results of the air combustion with RBK. NG shows a considerably higher radiative heat transfer for all oxyfuel conditions than for an air atmosphere. It is assumed that the higher oxygen concentration in the oxidant results in a faster ignition and a stronger local chemical reaction. The radiative heat flux measured for a non-swirled flame is lower than for a swirled flame as the long, non-swirled flame distributes the heat over a larger area. In addition, the mixing of the fuel with the oxidant is poor, which leads to reduced reaction and heat release, which is also reflected in the increase in CO formation. The flames with pulverised fuel all show a slight rise near the burner, and then decrease along the height of the combustion chamber. While similar values for the radiative heat flux are achieved for walnut shells with all oxygen concentrations analysed as for the air-fired case, the radiative heat fluxes for lignite at 26 vol% oxygen are significantly below the values of air-firing. For 30 vol% oxygen in oxyfuel operation, the measured radiative heat fluxes are close to the air condition, but still slightly lower. Comparing the walnut shells and the lignite, the walnut shells show a higher radiative heat flux than the lignite at the same thermal output. Zabrodiec et al. [4] assumed that a higher particle load is necessary for the walnut shells to achieve the same load and, therefore, a higher emission is generated by the particle cloud. The assumption of the authors of this study is that the higher volatile content of the WS leads to faster ignition, higher reaction rate and, therefore, to higher local heat release. A comparison with another combustion chamber and different thermal load, but the same fuel and similar oxygen enrichment shows good agreement in the difference between oxyfuel and air, using relative values for the radiative heat transfer. Overall, it is difficult to compare data from the literature in absolute terms, since combustion chambers with refractory-lined walls are mostly used, which reflect a higher proportion of heat back into the chamber than cooled steel walls.

The study has shown that changing the fuel of an oxyfuel firing system from coal to biomass may change the radiative heat transfer due to the higher volatile content of biomass. Furthermore, additional investigations need to be carried out to determine the O₂/CO₂ ratio at which the radiative heat flux for natural gas is similar to those of air-firing.