1 Introduction

In addition to their negative impact on environment, increasing prices of crude oil and natural gas, plus depletion of fossil fuel resources motivated many researchers to investigate alternatives and renewable energy sources, and to improve combustion and emission characteristics of conventional fuels with additives. Most of these researchers used inert components such as steam, N₂ or CO₂ as an additive to evaluate impact of these components on fundamental combustion properties [1, 2, 3, 4, 5]. Aside from adding inert or active components [6, 7, 8, 9], relative effect of gas components on flame behaviour has also been investigated [10, 11]. Some of the related studies existing in literature are summarized in the following paragraphs.

Ilbas and Karyeyen numerically investigated the effect of H₂ addition on combustion behavior of low heating value coal gases and concluded that H₂ addition improves fundamental flame characteristics; but, this situation causes increased NO_x emissions in turn [12]. Park and Kim numerically investigated flame structure and emission characteristics of H₂/CO/CH₄ blending synthetic gases by examining the effect of gas blending and the amount of N₂ dilution. They used GRI 3.0 mechanism for NO_x prediction and validated their results with published experimental data. It was concluded that dominant NO_x formation and destruction mechanisms are: (i) thermal, prompt and reburn routes in non-diluted counterflow non-premixed synthetic gas flames; (ii) an increment in H₂ mole fraction in gas composition increases the peak NO_x amount; (iii) when fuel is diluted with N₂, NO formation rate increases with rising H₂ and CH₄, and decreasing CO amount; (iv) CH₄ decreases positive effect of N₂ dilution on NO decrement [13]. Micro mixing injection combustion is one way of decreasing NO_x emissions of gas turbine combustors. But, it increases temperature levels near burner outlet. Motivated by this, Zhang et al. built an experimental setup to solve this problem with N₂ dilution. Results of their study showed that N₂ dilution decreases temperature levels and protects burner from high thermal loads; as the dilution amount increases, temperature and NO_x emission levels decrease [14]. Lee et al. conducted an experimental study to examine N₂, CO₂ and steam dilution effect on combustion and emission behavior of H₂/CO mixtures by measuring temperature, NO_x and CO levels and observing flame shape, with particular emphasis on flame instabilities. It was concluded that increased heat capacity of the mixture with diluent addition decreases NO_x emissions and an increment in diluent amount increases CO emissions [15]. Ilbas and Yilmaz performed experiments on CH₄/H₂ mixture flames in a natural gas burner. They investigated the effect of gas composition on CO, NO_x and O₂ emissions and temperature distribution throughout the combustor. They reported that compared to 100 % H₂ burning, temperature values and NOx emissions decrease with methane addition; contrarily, temperature values and CO emissions increase with H₂ addition; H₂/CH₄ mixtures emit less CO (compared to natural gas) because of the reduced C/H ratio [16].

Apart from the effect of fuel composition on fundamental flame characteristics, particular interest has been shown to potential effect of this composition on flame stability characteristics. Operating range of premixed and diffusion flames (defined by flashback and blowout limits) have been investigated by many investigators. These researchers have studied how gas components affect flame stability characteristics. Because, different gases have different chemical and transport properties, which will cause flame speed to alter and so will the flame behavior. For example, H₂ and CO generate high-temperature and high-speed flames by allowing them to operate at low equivalence ratios (this reduces emissions). However, such fuels cause flashback, which is a safety hazard. For these reasons, many researchers have investigated flame stability characteristics of gas mixtures [17, 18, 19, 20, 21] and some of these researchers have tried to improve flame stability limits via introducing swirl into the flow [22, 23, 24, 25].

In this study the effect of gas composition, swirl number and inert dilution on combustion and emission behavior of CNG/H₂/CO₂ mixtures was experimentally investigated in a combustion system.

2 Experimental Setup

2.1 Burner

During burner design, characteristics of the fuels to be used and operating conditions of the burner must be considered. In Figure 1, solid model of the manufactured burner can be seen. This is a premixed burner and it can operate at thermal powers up to 10 kW. Burner outlet houses swirl vane and can be easily dismantled to change swirl wane.

Figure 1

Solid model of the burner.

Swirling flows are used in most combustion devices because of their positive impact on fundamental flame characteristics [26]. Swirling flows are characterized by swirl number which is the ratio of the axial flux of the tangential momentum (G_φ) to the axial flux of the axial momentum (G_x).

SwirlNumber=GφGxRGp=∫oR(Wr)ρU2πrdrGx=∫oRUρU2πrdr

In these equations; U and W are the axial and tangential velocity components, R is the radius of the cross section plane, r is the radial coordinate [27].

For practical calculations, approximate swirl number can be expressed as

SwirlNumber=23⋅1−(dhd0)31−(dhd0)2⋅tan(θ)

d_h, swirler hub diameter; d_o, swirler outer diameter; θ, swirl vane angle [28].

2.3 Combustor

In Figure 2, fabricated combustor is illustrated. It has 1650 mm length, 330 mm outer diameter and is made of stainless steel. On combustor wall, there are numerous ports for axial and radial temperature and emission measurements. To make flame optically accessible, demountable tempered glasses are placed at two sides of the combustor.

Figure 2

Combustor.

Temperature and emission measurements are performed with K and B type thermocouples and a flue gas analyser, respectively. Thermocouple (2,4,6,8,9,10) and emission (4,7,9) measurements points are illustrated in Figure 3. Data obtained from these thermocouples are stored in a computer via a data logger.

Figure 3

Thermocouple and emission measurements points.

3 Uncertainty in Experimental Data

During experiments, temperature measurements were performed with thermocouples. Data obtained from thermocouples was transferred and stored in a computer via a data logger which has 100 kS/s sampling rate. Measured values were also corrected taking radiation losses into consideration, and found to be 20–100 K lower than real state.

Emission measurements were conducted with a portable flue gas analyzer. Typical accuracy of the analyzer is: for O₂, ± 0.2%; for CO in the range of 0–4000 ppm, ± 5% (or ±10 ppm, whichever is higher), for CO > 4000 ppm, ± 10%; for NO in the range of 0-1000 ppm, 5% (or ±5 ppm, whichever is higher), for NO >1000 ppm 10%; for NO₂ in the range of 0–200 ppm, 5% (or ±5 ppm, whichever is higher), for NO₂ >200 ppm 10%; for SO₂ in the range of 0–2000 ppm, 5% (or ±10 ppm, whichever is higher), for SO₂ >2000 ppm 10%; and for CO₂ ± 0.3%.

Besides these measurements mentioned above, the amount of each gas in gas mixture was measured with a digital mass flow controller. The accuracy of mass flow controllers was: for gas supply lines, ± 1% of set point for 20 to 100% of full scale, ± 0.2% of set point for 0 to 20% of full scale; for air supply line, ± 1% of full scale.

4 Results and discussion

Different transport and chemical properties of each component of gas mixtures bring many challenges to their usage in practical systems from combustion and emission behavior and flame stability point of view [29]. This study aims to evaluate the relative effect of each gas component on key combustion properties.

In Figure 4, axial temperature distributions of 100% CNG, 90% CNG - 10% H₂ and 80% CNG - 20% H₂ mixtures at 0.4 swirl number and 1.0 equivalence ratio can be seen. Temperature distribution profiles of tested gas mixtures show an agreement in terms of trend and value. Peak temperature values were reached in the flame region and then temperature levels decreased towards the outlet of the combustor. 80%CNG - 20%H₂ mixture formed the highest temperature value. Temperature distributions of 100%CNG and 80%CNG - 20%H₂ mixture were nearly the same near burner outlet. On the other hand, temperature profiles of 90%CNG - 10% H₂ and 80%CNG - 20%H₂ mixtures got closer towards the combustor outlet. Compared to 100%CNG and 80%CNG - 20%H₂ mixture, 90%CNG - 10%H₂ mixture formed the lowest temperature value in the flame region; but, temperature distribution throughout the combustor of 90%CNG - 10%H₂ mixture got a higher value. As the H₂ amount in gas mixture increased, heating value of the gas mixture and heat transfer rate increased. With this and the fact that thermal power of the combustor was kept constant may have caused temperature distribution to be higher. 100% CNG formed the lowest temperature distribution.

Figure 4

Axial temperature distributions (Swirl number: 0.4, Equivalence ratio: 1.0).

To evaluate the effect of inert dilution on combustion and emission behaviour, CNG and mixture of CNG/H₂ was diluted with varying amounts of CO₂ and these mixtures were tested under the same boundary and physical conditions. In Figure 5, axial temperature distributions of 90% CNG - 10% CO₂ and 90% CNG - 20% CO₂ mixtures are illustrated. Combustible mixtures will only burn when concentration of fuel and air are in an experimentally determined range. Upper and lower value of this range is referred as flammability limits [30]. Considering upper flammability limit, we could not attain a stable flame at stoichiometric conditions and thus equivalence ratio was set as 0.9. Temperature profiles of respective mixtures show a similar behavior. Compared to 100% CNG burning, maximum temperature value in the flame region was nearly the same for the 90% CNG -10% CO₂ mixture. In the case of 20% CO₂ dilution, maximum temperature value decreased by a value of 100 K. Unlike N₂, CO₂ is an infrared active molecule and is reported to cause extra heat loss to environment [31].

Figure 5

Axial temperature distributions (Swirl number: 0.4, Equivalence ratio: 0.9).

Temperature decrement with CO₂ addition is attributed to this phenomenon.

Temperature profiles of the gas mixtures that contain varying amounts of CNG, H₂ and CO₂ are demonstrated in Figure 6. Due to the high reactivity of H₂, first H₂ was consumed in the flame region, then other combustible constituents of CNG were consumed. Therefore, H₂ amount in the gas mixture positively correlated temperature value in the flame region. Axial temperature distributions of 80% CNG -10% H₂ -10% CO₂ and 70% CNG - 20% H₂ -10% CO₂ mixtures are nearly the same throughout the combustor. 70% CNG - 10% H₂ - 20% CO₂ had the lowest temperature value in the flame region; but, it formed the highest temperature distribution towards the outlet of the combustor. This is because of reduced burning rate of such mixture. An increment in CO₂ amount decreased burning rate and combustion process was completed further downstream.

Figure 6

Axial temperature distributions (Swirl number: 0.4, Equivalence ratio: 0.9).

60% CNG - 20% H₂ - 20% CO₂ mixture formed the highest temperature distribution (except for the flame region) in all gas mixtures tested (Figure 7). Besides, it was observed that 60 % CNG – 20 H₂ – 20 % CO₂ mixture burns more intensively and shows less axial and radial fluctuations than 60 % CNG – 10 H₂ – 30 % CO₂ and 60 % CNG – 30 H₂ – 10 % CO₂ mixture flames. Hence, to evaluate the effect of swirl number on temperature distribution, 60% CNG - 20% H₂ - 20% CO₂ mixture was used.

Figure 7

Axial temperature distributions (Swirl number: 0.4, Equivalence ratio: 0.9).

In Figure 8, axial temperature distributions of 60% CNG - 20% H₂ - 20% CO₂ mixture at 0.4, 0.6 and 0.8 swirl numbers can be seen. Temperature profiles showed a good agreement in terms of trend for all swirl numbers tested but, they differed by in terms of values. Compared to 0.4 swirl number, temperature profiles were very close at 0.6 and 0.8 swirl numbers. As the swirl number increased, temperature values throughout the combustor reduced. At high swirl numbers; mixing condition of fuel/air mixture improves and heat better diffuses radially . Thus, axial temperature values were reduced.

Figure 8

Axial temperature distributions of 60% CNG - 20% H₂ - 20% CO₂ mixture (Swirl numbers: 0.4, 0.6 and 0.8).

From instantaneous flame images at different swirl numbers (Figure 9), it can be seen that, as the swirl number increases; reaction zone propagates radially out-ward (wider flame), burning intensity increases and flame length reduces.

Figure 9

Flame images at different swirl numbers

Axial CO₂ concentration (in percentage of exhaust gases) profiles of 100% CNG, 90% CNG -10% H₂, 80% CNG - 20% H₂, 90% CNG -10% CO₂ and 90% CNG -10% CO₂ mixtures are illustrated in Figures 10–11. As expected, CO₂ amounts in exhaust gases were very high in the flame region but it reduced towards the outlet of the combustor. As H₂ amount in the gas mixture increased, CO₂ concentration reduced.

Figure 10

Axial CO₂ concentration profiles (Swirl number: 0.4, Equivalence ratio: 0.9)

CO₂ concentration in the flame region decreased with increasing amount of CO₂ in the gas mixture because of the inhibiting effect of CO₂. However, CO₂ concentration increased towards the outlet of the combustor in the case of 20% CO₂ dilution. This indicates that fuel is consumed further downstream.

Based on the temperature decrement in axial direction, CO₂ concentration values decreased in the same direction with increasing swirl number (Figure 12).

Figure 11

Axial CO₂ concentration profiles (Swirl number: 0.4, Equivalence ratio: 0.9).

Figure 12

Axial CO₂ concentration profiles (Swirl number: 0.6, 0.8).

Because CO is an incomplete combustion product, emissions of CO are directly proportional to CO₂ amount. CO₂ presence in a fuel mixture increases heat capacity of the mixture and reduces reaction rate. An increase in CO₂ amount inhibits oxidation of fuel molecules through reducing usability of oxygen for H and leads to increased emissions of unburnt CO [29]. In Figure 13, it can clearly be seen that as the CO₂ amount in gas mixture increases, CO emissions increases.

Figure 13

Axial CO emission profiles (Swirl number: 0.4, Equivalence ratio: 0.9).

Effect of H₂ on CO emissions is positive in terms of environmental effects. As the hydrogen amount increases, CO emissions reduce due to the high reaction kinetics of H₂ (Figure 14).

Figure 14

Axial CO emission profiles (Swirl number: 0.4, Equivalence ratio: 0.9).

Axial NO_x profiles of 100% CNG, 90% CNG -10% H₂, 80% CNG - 20% H₂, 90% CNG - 10% CO₂ and 90% CNG - 10% CO₂ mixtures can be seen in Figure 15 and 16. As the H₂ amount (also swirl number-Figure 17) increases and CO₂ amount decreases, NO_x emissions increase based on the temperature increment.

Figure 15

Axial NO_x profiles (Swirl number: 0.4, Equivalence ratio:

Figure 16

Axial NO_x profiles (Swirl number: 0.4, Equivalence ratio: 0.9).

Figure 17

Axial NO_x profiles (Swirl number: 0.6, 0.8).

5 Conclusions

In this study, the effect of gas composition and swirl number on combustion and emission behavior of premixed CNG/H₂/CO₂ mixtures was experimentally investigated in laboratory scale combustor. Results obtained from this study are:

1. Irrespective of the gas mixture, temperature profiles of tested flames show an agreement in terms of trend. Gas composition is found to be decisive for temperature value.

2. Due to the enhanced heat transfer characteristics in radial direction, axial temperature values decrease as the swirl number increases.

3. Depending on the degree of completion of reaction, CO₂ emissions are high in the flame region. Axial CO₂ concentration reduces with increasing swirl number and H₂ and CO₂ amounts in the gas mixture.

4. Emissions of CO is highly susceptible to gas composition.

5. Thermal Zeldovich mechanism is the dominant NO_x formation mechanism for the tested gas mixtures.