You are on page 1of 11

Fuel 124 (2014) 258268

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Design and preliminary results of an atmospheric-pressure model gas turbine combustor utilizing varying CO2 doping concentration in CH4 to emulate biogas combustion
Christopher J. Mordaunt a,, Wade C. Pierce b
a b

Bucknell University, 370 Breakiron Building, United States Bucknell University, Dana Engineering Building, United States

h i g h l i g h t s
 Design details of an atmospheric-pressure model gas turbine combustor are presented.  The primary focus is to study the effect of biogas in land-based gas turbines.  Injection congurations, CO2 doping, and air temperature emulate biogas combustion.  Three different fuel injection congurations were examined.  Stability maps and lean-blowout data are reported.

a r t i c l e

i n f o

a b s t r a c t
Utilization of biogas can provide a source of renewable energy in both heat and power generation. Combustion of biogas in land-based gas turbines for power generation is a promising approach to reducing greenhouse gases and US dependence on foreignsource fossil fuels. Biogas is a byproduct from the decomposition of organic matter and consists primarily of CH4 and large amounts of CO2. The focus of this research was to design a combustion device and investigate the effects of increasing levels of CO2 addition to the combustion of pure CH4 with air. Using an atmospheric-pressure, swirl-stabilized dump combustor, emissions data and ame stability limitations were measured and analyzed. In particular, CO2, CO, and NOx emissions were the main focus of the combustion products. Additionally, the occurrence of lean blowout and combustion pressure oscillations, which impose signicant limitations in operation ranges for actual gas turbines, was observed. Preliminary kinetic and equilibrium modeling was performed using Cantera and CEA for the CH4/CO2/Air combustion systems to analyze the effect of CO2 upon adiabatic ame temperature and emission levels. The numerical and experimental results show similar dependence of emissions on equivalence ratio, CO2 addition, inlet air temperature, and combustor residence time. 2014 Elsevier Ltd. All rights reserved.

Article history: Received 25 August 2013 Received in revised form 29 January 2014 Accepted 29 January 2014 Available online 14 February 2014 Keywords: Gas turbine Biogas combustion Combustion instability CO2 addition

1. Introduction 1.1. Biogas as a fuel source Concern with global warming and fossil fuel consumption has promoted the desire to utilize renewable and sustainable energy sources, such as bio-fuels [1]. The global energy demand is rapidly increasing, with approximately 88% of current demand being met by fossil fuels. Biogas is the byproduct of decomposing organic

Corresponding author. Tel.: +1 570 577 1417; fax: +1 570 577 7281.
E-mail address: cjm039@bucknell.edu (C.J. Mordaunt). http://dx.doi.org/10.1016/j.fuel.2014.01.097 0016-2361/ 2014 Elsevier Ltd. All rights reserved.

matter in the absence of O2, largely produced by landlls and anaerobic digesters. Biogas is often called dirty methane because that produced by a landll is contaminated, principally with CO2. Much of this fuel is produced from biodegradable materials such as municipal waste, sewage, energy crops, green waste, biomass, and manure [2]. These materials would otherwise provide no benet, decomposing in a landll or eld, releasing heat-trapping methane into the atmosphere, and contributing to global warming. Methane, if released directly to the atmosphere, causes approximately 21 times the global warming effects of carbon dioxide because of its ability to trap heat radiating outward from the earth [3]. Therefore, landll gas is generally ared off as it is released,

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

259

wasting the energy potential, and converting it to basic combustion products consisting of H2O, CO2 and N2. Efforts to recover this energy have focused on the technology necessary for collecting and purifying this gas [4]. Recognition of the potential of landll gas as an energy source and revenue generator encourages more efcient collection and reduction of harmful emissions into the atmosphere [1]. Biogas has great potential for use in the generation of electricity, especially if used to fuel a gas turbine or internal combustion engine. However, the use of biogas as a fuel is complicated by its low Lower Heating Value (LHV) level (<25% natural gas) and the variability of its component composition [5]. The use of biogas is becoming more popular among nations trying to reduce carbon dioxide emissions and lower dependence on fossil fuels. Previous research by Lastella et al. [6] investigated the production of biogas in an anaerobic digester. They used the idea of generating biogas from fruits and vegetables, due to the high quantities of waste from markets. The composition of biogas, on average, ranged from 60% to 68% CH4 and 32% to 40% CO2 by volume [6]. Similar amounts were also found by the National Non-Food Crops Centre (NNFCC) for non-woody materials [2]. The NNFCC sets the standards for biogas from anaerobic digesters and most landlls. Several researchers have studied the feasibility of using biogas in spark-ignition and compression-ignition engines, and many comparisons have been made between landll gas, biogas, and CH4. Bade Shrestha and Narayanan [1] compared landll gas to CH4 in a spark ignition engine. Through performance and combustion characteristics, it was shown that the landll-gas-fueled engines performance deteriorated in comparison with CH4 operation. However, by increasing compression ratio and advancing spark timing, the performance was improved for landll gas operation to be equivalent to CH4 operation. The effects due to composition changes in the landll gas were more pronounced at lean and rich mixture operation than at stoichiometric conditions [1]. Bari [7] investigated the effects of CO2 on performance in dual-fuel internal combustion engines. They found that the presence of as high as 40% CO2 in biogas did not deteriorate the engine performance, as compared to the performance of the engine in dual-fuel mode with natural gas (96% methane). With the presence of up to 30% CO2, improved engine performance was found, compared to the same engine running with natural gas [7]. The author concluded that the amount of oxygen created by dissociation of the CO2 in the fuel decreased combustion delay, thereby enhancing the combustion of unburned carbon particles and lowering fuel consumption. Depending on the source of the biogas, diluent gases (N2, CO2, Ar) can range from 4% to 51% in landll or sewage gas [8]. This variation is a signicant problem for high performance low-emission combustion systems that are optimized for extremely narrow fuel specications. Using biogas as a surrogate fuel, Wilson and Lyons [9] studied trends (in terms of burning velocities as a function of equivalence ratio) for a variety of fuels. Flame temperatures and associated burning velocities are reported as a function of diluent composition. Additionally, due to the presence of inert gases, biogas may ow up to 68% more by volume to achieve the same energy content, compared to natural gas [9]. Diluents in the gas can cause what is known as a lifted ame (where the ame is lifted from its anchor point above the injector) which could be a positive asset for atmospheric burners. Non-premixed lifted ame burners reduce the amount of heat transfer back to the burner and are commonly found in commercial steam boilers [10]. 1.2. Lean premixed prevaporized combustion Nitric oxide is formed during the combustion of fuels not containing fuel-bound N2 by three chemical mechanisms that involve

N2 from the air: the thermal (or Zeldovich) mechanism, the Fenimore (or prompt) mechanism, and the N2O-intermediate mechanism [11]. The Zeldovich mechanism tends to dominate NOx (NO + NO2 are collectively referred to as NOx) production in gas turbine engines because of the high temperatures involved. Lean premixed prevaporized (LPP) combustors utilize partially or fully premixed combustion air and fuel. Fuel is introduced into (and partially premixed with) the combustion air upstream of the combustion chamber and burned at equivalence ratios lower than stoichiometric, normally below / 0:65. Combustion temperatures are much lower at the lower equivalence ratios; therefore, lower NOx is produced through the thermal mechanism. Reducing NOx emissions in gas turbines was a major motivation for the development of lean premixed prevaporized combustors [12], since a major source of nitrogen oxides are the combustion of fossil fuels in engines or electrical power plants. The power industry is responsible for about 23% of NOx formed as a pollutant during the combustion process [13]. Operating at lean conditions to reduce ame temperature and NOx, however, often leads to combustion driven pressure oscillations and intense heat release uctuations that can cause wear and damage to combustor components. In extreme cases, this can lead to liberation of combustor components into the hot gas path, damaging downstream turbine components [14,15]. Combustion instabilities occur when small perturbations in the combustion system develop into pressure oscillations that overcome the damping forces of the combustion chamber. If the pressure oscillations are in phase with heat release uctuations, the amplitude of the pressure oscillations will increase. This condition, known as the Rayleigh Criterion, indicates that a dynamic relationship between heat release and pressure oscillations exists. If heat is released by combustion at the same point of the greatest vibration, a pressure increase is encouraged [16]. Eventually, pressure oscillations reach a maximum limit, known as a limit-cycle, where the energy losses due to damping forces becomes equal to the energy gained by combustion [14]. These oscillations are strongly inuenced by the composition of the fuel, the equivalence ratios, as well as the combustion chamber geometry. Exhaust species routinely considered important in energy generation from combustion are: CO2; CO; NOx; unburned hydrocarbons (UHC); and sulfur oxides (SOx) and mercury from burning coal [12]. 1.3. Biogas as a fuel source in LPP combustors Since biogas can be used in a more efcient way, research is being conducted on the utilization of this gas in land-based gas turbines [5,17]. A gas turbine designed to produce low emissions on natural gas may not achieve the same emission goals and performance on a different gaseous fuel. Compressed natural gas (CNG) is typically 7195% CH4 combined with higher hydrocarbons, N2, and CO2 [12]. The presence of contaminates in biogas can have dramatic effects on the combustion characteristics of the fuel, impacting the ame in at least three ways: (1) through changes in mixture specic heat and adiabatic ame temperature; (2) chemical kinetic rates; and, (3) radiant heat loss [5,18]. As seen in diesel/biogas engines, the presence of CO2 in the biogas reduces the burning velocity, which ultimately affects the performance [7]. Combustion parameters such as performance, stability, and emissions must be evaluated prior to establishing the viability of utilizing biogas in gas turbines. The suitability of biogas for use as a fuel depends primarily on its CH4 content. In landll gas, typical CH4 to CO2 concentration ratios range from 60/35% to 35/55%, with the balance being N2; gases from anaerobic digestion of sewage (used commonly in wastewater treatment plants) typically have 6575% CH4, with the balance being N2 [19]. Presently, gas turbines require

260

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

a methane content of at least 55%, but most normally desire greater than 90% [20]. The purpose of this research is to design an experimental apparatus and explore the feasibility of utilizing unrened biogas in the low NOx LPP gas turbines typically utilized in the power generation industry. Of concern during biogas combustion are characteristics such as ame holding, lean ame blow out, and combustor pressure oscillations (limit cycle oscillations). Additionally, the three pollutant species of greatest interest are CO2, CO, and NOx. To accomplish this, an atmospheric-pressure swirl-stabilized dump combustor has been developed. The biogas fuel is emulated by real-time variation of CO2 doping concentration into the CH4 fuel stream. Additionally, several different fuel/diluent injection congurations are explored. Knowledge of the interaction among acoustic disturbances, ame surface area, and heat release rate is imperative to studying the mechanisms that initiate and sustain combustion instability. Chemiluminescence from certain excited radicals has often been used as an indicator of ame heat release regions [21]. One such species is CH* (the excited CH radical), which has been observed to occur only in the primary ame zone and is formed primarily through the following reaction:

can result in signicant uncertainty due to large ow disturbances [2123]. Usually, chemiluminescence imaging requires an Intensied Charge Couple Device (ICCD) camera, especially if phaselocking onto a pressure uctuation or attempting to capture the ame structure by micro-second gating. However, the CH* chemiluminescence signal is strong enough to be collected by a conventional camera if temporal resolution is not required. 2. Experimental setup A brief description of the test apparatus is provided here, a more detailed description can be found in other work [24]. 2.1. Air and fuel delivery systems Dry inlet combustion air is supplied from an air compressor capable of delivering 2800 standard liters per minute (SLPM) at 1.0 MPa (150 psi). The compressed air is regulated, ltered and sent to a Teledyne Hastings (Model HFC303L) ow controller with an equivalent N2 ow range of 0200 SLPM (1% FS). Combustion inlet air temperature is controlled with a 9-kW Watlow Process Heater and a custom-manufactured Power Modules Inc. limit control module. Gaseous fuel or diluent is supplied from bottled gas cylinders and is typically regulated from 13.8 MPa (2000 psi) to 0.83 MPa (120 psi). The two gaseous fuels are metered separately by Teledyne Hastings (Model HFC202) ow controllers. For these

C2 H O 3 P !CHA2 D CO
*

While CH chemiluminescence may be an excellent marker for low-strained, non-curved ames, some studies have indicated that the correlation between emission from CH* and heat release rate

Fig. 1. Cutaway view of combustor detailing injector, swirler, bluff body, and dump plane with mounted PCB transducer.

Fig. 2. (a) Model of fuel injectors; (b) swirler mounted on a different style fuel injector tube.

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

261

experiments, CH4 is capable of owing 20 SLPM (1% FS) of equivalent N2 and the CO2 is capable of 25 SLPM (1% FS) of equivalent N2. The fuel or diluent either enter the combustion chamber through the injector, or are premixed with combustion air
Table 1 Injector designation and ow ranges. Injector B1 B2 B3 Hole D (mm) 0.254 0.305 0.356 Hole D (in) 0.010 0.012 0.014 Flow (SLPM N2) 15.018.5 17.527.5 27.040.5

upstream of the venturi (explained in greater detail below). Each fuel can be independently premixed or injected. 2.2. Combustion chamber A cut-away diagram of the model combustion chamber is shown in Fig. 1 with ow going from bottom to top. The rst section of the combustor contains a 2-mm-thick, 114-mm-long optically accessible quartz window for ame viewing and laser diagnostics; and a stainless steel extension (not shown if gure) is used to achieve the overall combusor length of 37.5 cm. The main air supply ows around the injector tube, encounters the swirler, and entrains the fuel jets in an annular mixing section between the 9.53-mm-diameter bluff centerbody and 20-mmdiameter inlet tube, with a corresponding annular hydraulic diameter, DH, of 15.46 mm. The fuel/air mixture then enters the 45-mm-diameter modular combustion chamber, with an area ratio, AInlet/AChamber, of 0.16. The swirler used in this study (as
Table 2 Emissions equipment measurement range and uncertainty. Species NOx CO CO2 O2 UHC Range 050 ppm 0200 ppm 020% 025% 0300 ppm Uncertainty 0.5 ppm 2.0 ppm 0.2% 0.25% 9.0 ppm Source Horiba PG-250 Horiba PG-250 Horiba PG-250 Horiba PG-250 CA 600 HFID

Fig. 3. The three different fuel injection schemes utilized in the study. Also, note the location of the choked upstream air inlet venturi.

Fig. 4. Sample of limit-cycle combustion instability frequency.

Fig. 5. Stability maps and lean-blowout points for inlet temperatures of 422 K (Top) and 477.6 K (Bottom).

262

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

shown in Fig. 2) consists of a hub and an outer ring which contains eight straight, at vanes; this is designed such that the impinging air ow will always contact a vane surface. With the assumption of uniform axial and swirl velocities the geometrically based swirl number calculated for the 45 swirler is approximately Sn = 0.69 [25]. A separate stream of cooling air ows through the outer passages of the dump plane and impinges directly on the exterior of the quartz window to prevent overheating. Due to the sudden expansion in the combustion chamber at the dump plane, a corner toroidal recirculation zone (CTRZ) is created. Additionally, the bluff body attached to the injector and the effect of the swirler create a strong center recirculation zone (CRZ). Both of these recirculation zones are required for proper ame anchoring [14]. Upstream of the swirler, the air ows through a venturi orice to create a choked ow and to provide an inlet acoustic boundary. Incorporated into the dump plane and downstream of the inlet venturi are PCB piezoelectric pressure transducers (Model 113A21) to measure the dynamic pressure uctuations. These transducers have an average sensitivity of 3.63 mV/kPa, a non-linearity of less than 1% full scale, and response time less than or equal to 1.0 ls. Each transducer outputs 5 VDC@ 137.90 kPa (20 psi), and the data is monitored and recorded at 30,000 Hz. 2.3. Fuel injection and premixing schemes Gaseous fuel is introduced into the combustion chamber through custom manufactured injectors sized specically for the fuel composition. The gaseous injectors are plain-jet air-blast atomizers (named MP injectors after the designers, Mordaunt and Pal) which were originally designed to be used in the highpressure, high-temperature model gas turbine combustor at The

Propulsion Engineering Research Center (PERC) at The Pennsylvania State University [14]. The injectors have eight injection holes created by electric discharge machining (EDM) and sized according to a specic volumetric ow of the gaseous fuel mixture. This is to obtain choked ow conditions in the fuel stream and to stay within acceptable pressure limits. Fig. 2 shows a model of the MP injector and the swirler attached to the injector tube. Fuel injectors are sized to maintain an approximately consistent injection momentum ratio, J, of fuel to air, as calculated by:

J qV 2 j =qV 2 1

where q and V are the substance density and velocity, respectively, and the j and 1subscripts represent the fuel and cross-owing combustion air streams, respectively. It is desired to have choked ow conditions in the injector to ensure a constant mass owrate, even when combustor pressure oscillations occur. However, it is well known that, as a sonic jet encounters a subsonic cross-ow, it expands, forming an inclined barrel shock and a Mach disk [26,27]. The jet velocity at the exit of the injector and the location of the shock point are difcult to interpret; therefore, calculating the momentum ratio for these conditions is not possible. To overcome this, the momentum ratio is calculated at the throat of the injector holes using knowledge of the fuel/diluent owrates and mixture properties as:

_ q V A m

where the superscript denotes properties evaluated at the critical conditions. Due to the large variation in ow rates, three injectors were utilized to maintain relatively consistent injection pressures and momentum ratios. Each injector was designed to operate over a

Fig. 6. CO2 emissions for the three CH4/CO2 injection congurations. Also shown is the CH4/N2 case.

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

263

specic range of total volumetric ow from the combined fuel and diluent(s) as follows in Table 1: The value of A for each injector was determined experimentally for pure N2. Mixing effects were studied using three congurations for injecting fuel and diluent (CO2 or N2) into the combustion air with respect to the combustion chamber and the choked inlet air venturi (see Fig. 3): CONFIG. ANon-premixed: Fuel and diluent are injected through a gaseous injector and mixed with the combustion air in the premixing section upstream of the combustion chamber. CONFIG. BNon-premixed fuel/premixed CO2: Combustion air and CO2 are premixed upstream of the choked inlet venturi; fuel is introduced through the injector and mixed with the combustion air/CO2 in the premixing section upstream of the combustion chamber. CONFIG. CFully premixed: Fuel, CO2, and combustion air are introduced and mixed upstream of the choked inlet venturi. Note that CONFIG. A is representative of the manner in which biogas fuel would be utilized in an actual gas turbine combustor; CONFIG. B is a conguration that would be suitable to EGR studies; and that CONFIG. C is a limiting case that eliminates unmixedness effects entirely. 2.4. Emissions measurement

response time (240 s response time for SO2) with 1% uncertainty at full scale. A California Analytical Instruments Model 600 HFID Analyzer measures the total concentration of hydrocarbons within the gaseous sample. The unit uses Flame Ionization Detection (FID) to determine the total hydrocarbons in the sample, ranging from 03 to 030,000 ppm. Response time is 90% full scale in 1.5 s and repeatability better than 0.5%. Table 2 shows the calibration gas and measurement ranges used in experimentation. Gaseous emission samples from the combustion chamber are collected by a water-cooled collection probe located on the combustor centerline and approximately 25 mm prior to the exhaust tube exit. The quenched emission sample passes through a heated line to prevent condensation of UHC and H2O prior to reaching the emission analyzers. The sample line is split in two, with part of the ow passing through a Unique Heated Products (model FLT-1069) heated particulate lter before entering the HFID analyzer, and the remaining ow passing through an ice-bucket chiller to condense and remove moisture from the sample gas prior to entering the Horiba analyzer. 2.5. Chemiluminescence imaging The CH chemiluminescence images were captured with a Nikon D300 mounted on a tripod at a height level to the combustor dump plane. The camera utilized a Nikkor 50 mm f/1.8D AF lens and an Andover Corporation 430 nm (FWHM 10 nm) spectral lter. 2.6. Experimental procedure

A Horiba Portable Gas Analyzer PG-250 is used to measure NOx/ SO2/CO/CO2/O2 emission components. NOx is measured using chemiluminescence, SO2/CO/CO2 using non-dispersive infrared absorption, and O2 using a galvanic cell. The unit has a 45 s

Each test was conducted in a similar manner following a standardized test matrix, varying combustion air temperature, equivalence ratio, and percentage of diluent by volume. Starting from an

Fig. 7. CO emissions for the three CH4/CO2 injection congurations. Also shown is the CH4/N2 case.

264

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

equivalence ratio of / 0:8 and working down to / 0:6, the percent CO2 diluent was varied from 0% to 50% in increments of 10%. When a 10% increment was not achievable, a 5% increase was utilized. In addition to the three mixing congurations of CO2 described a non-premixed (as in CONFIG. A) set was performed with N2 as diluent to use as a comparison.

indicate a region where lean-blowout conditions prevail and the lower right-hand corner indicates regions of high intensity limitcycle oscillations. The green area marks the range in which all doping scenarios for CO2 were stable. It is observed that an increase in inlet air temperature causes a slight change in operation ranges. These stability maps are useful for determining regions where the combustor can operate safely, as well as indicating regions for further study.

3. Experimental results 3.1. Lean blowout and limit cycle behavior Two of the major factors restricting operation in this experiment were combustion instability (high intensity limit-cycle oscillations) and lean-blowout conditions. Data points were obtained where combustion instabilities were tolerable, meaning that the ame could be safely maintained at its anchor point and ash back or lean blowout did not occur. However, due to combustion instability levels, not all desired equivalence ratios or dilution levels could be achieved. The PCB transducer mounted in the dump plane is set to detect the pressure uctuations in the combustion chamber at a sampling frequency of 30,000 Hz. A Fast Fourier Transform (FFT) of the signal collected at one data point indicates the maximum peak location is at 387 Hz (as shown in Fig. 4) with no major higher harmonics observed. A linear acoustic analysis of an open-ended, unanged pipe predicts the resonance frequency to vary from approximately 450 to 575 Hz; therefore, this can be interpreted as the 1/4 wavelength standing wave [24,28]. Stability maps were constructed for operation with inlet air temperatures of 422 K (300 F) and 477.6 K (400 F) and are presented in Fig. 5. The upper left-hand corner of these images 3.2. Emissions measurements Two things must be noted regarding the emissions measurements: (1) the sampling probe was located on the centerline of the combustor (which tends to be a recirculation zone wake caused by the swirl effecteven at this great of a downstream location); and (2) no UHC measurements above the noise level of the HFID instrument were detected. The former statement has the greatest signicance since changing the probe location (either radially, or azimuthaly at a particular radial location) often led to a change in concentration measurements. While a prole and averaging scheme were attempted, the number of data points to accomplish a successful value was intangible. This will be addressed further in the conclusions section. The latter of the previous statements (i.e., no measurable UHC values) is not surprising given the nature of premixed methane combustion; however, it was important to monitor this species for any signicant results. Carbon dioxide is a major species of combustion of hydrocarbon fuels and, since CO2 is being added into the fuel/air mixture, higher CO2 concentration levels that increase with the addition of excess CO2 are expected in the exhaust stream. Shown in Fig. 6 are the

Fig. 8. NOx emissions for the three CH4/CO2 injection congurations. Also shown is the CH4/N2 case.

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

265

results obtained for the various mixing scenarios. Note that these results are presented on a parts per million volume dry (ppmvd) basis and have not been corrected for O2 concentration, as is typically performed. Such O2 concentration corrections cause the data to grow exponentially at high equivalence ratios (i.e., near stoichiometric). The similarity of trends observed in Fig. 6 (with respect to equivalence ratio and CO2 doping increase) are expected since CO2 is a predominant percentage of the combustion products; furthermore, these similarities are an indicator of good experimental repeatability. Moreover, for the N2 doping case, the observed reduction in CO2 concentration is expected as more diluent is added to the total composition. The CO emissions data for the different injection congurations are presented in Fig. 7. The rst observable trend is that, for xed equivalence ratio and increasing CO2 introduced, the CO emissions

increase drastically. A second, less obvious trend, is most notably observed in CONFIG. A for the CH4/CO2 injection case: starting at the lower equivalence ratio (/ 0:60) the measured CO concentration reduces at / 0:70, and then begins to increase again at the higher equivalence ratios (/ > 0:75). Nitrogen oxides emissions are minor species in the combustion products but are important parameters for lean combustion, especially in land-based gas turbines used in the power generation industry. Shown in Fig. 8 are the NOx emission measurements for the different injection congurations. The most noticeable trend is that NOx concentration decreases with increasing CO2 doping for all equivalence ratios. Emission levels are comparable for both CO2 and N2 diluents as well as premixed or injected fuels. With increased combustion inlet air temperature, NOx emissions increase for all doping scenarios.

Fig. 9. CH* chemiluminescence false-color intensity images at xed equivalence ratio of / 0:70, with ow from bottom to top in each image. Starting from top left to bottom right, CO2 = 0%, 10%, 20%, 30%, 40%, and 50%, respectively, was premixed with combustion air (i.e. CONFIG. B).

266

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

Shown in Fig. 9 are the false-color intensity images collected using a CH* lter mounted on a camera. The images were collected at a xed equivalence ratio of / 0:70 over a range of premixed CO2 (i.e. CONFIG. B) that varied from 0% to 50% by volume of methane. As observed in the images, the emission intensity becomes weaker as more CO2 is premixed in the air, covering a range from unstable combustion (limit-cycle behavior) to approaching leanblowout conditions. Additionally, it is observed that the ame becomes weaker as more CO2 is added and the reactions slow, extending the ame further up into the exhaust tube.

4. Discussion of results The results obtained from the CO2 emissions measurements are not surprising. In some instances, the emissions can be increased by 100% over that of pure natural gas. While this statement may sound unappealing, biogas is made from substances that consume CO2 during their life-cycle; therefore, from a global perspective, there is no net increase in CO2 emissions. To validate the quantity of CO2 emissions measured in the combustor, an analysis was performed using two separate combustion packages: the Nasa Lewis Chemical Equilibrium Analysis (CEA) code; and the Cantera

combustion modeling program. The CEA program provides equilibrium concentrations and adiabatic ame temperatures [29]. The Cantera combustion modeling program [30] has the ability to model different common combustor congurations, uses the GRI-Mech 3.0 rate package [31], and is more applicable to non-equilibrium conditions. Shown in Fig. 10 are results obtained by an equilibrium analysis with CEA, and with the Cantera program running an adiabatic reactor model. Both models utilize an inlet air temperature of 422 K (300 F). While each model shows reasonably similar trends in CO2, the Cantera program results may be more appropriate because the program takes into consideration species reaction rates and combustion chamber residence times. As mentioned earlier, an unusual trend was observed in CO emissions at high CO2 doping conditions, especially in the non-premixed case (CONFIG. A). Initially CO emissions are high at / 0:65, decrease at / 0:70, and increase again at / 0:75. Glarborg and Bentzen [32] observed similar trends in their studies on an atmospheric-pressure ow reactor under highly diluted N2 and CO2 conditions. They observed that high concentrations of CO2 led to a dramatic increase in CO concentration after initiation of reaction; and that high levels of CO2 prevent complete oxidation of the fuel at higher temperatures, causing increased CO

Fig. 10. Predicted CO2 emissions from CEA (Top) and Cantera (Bottom) models. Both models use inlet air temperatures of 422 K.

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

267

concentrations in the near-burner region. The presence of CO2 competes with O2 for atomic hydrogen and leads to formation of CO. Reaction of CO2 with hydrocarbon radicals may also be a contributing factor to CO formation. Amato et al. [33] observed increased CO emission levels, especially at lower adiabatic ame temperatures and shorter residence times, which they attributed directly to higher CO2 levels. An attempt to duplicate the observed phenomena utilizing the Cantera model provided results that, although off by several orders of magnitude, contained trends similar to those observed experimentally [24]. These results were due either to different reaction pathways that opened/closed at different adiabatic ame temperatures, or the direct inuence of high CO2 concentrations. This modeling effort is ongoing and a complete analysis is left for future work. It is well known that unmixedness causes localized equivalence ratio uctuations and can lead to combustion instability and higher pollutant emissions [14,25,34]. Since the fuel is injected so close to the combustor in the mixing section, it often has insufcient time to mix thoroughly with the air. The extent to which fuel is injected into the mixing section depends on the equivalence ratio which, in turn, has an effect on the momentum ratio: smaller momentum ratios will result in lower penetration into the air stream, whereas higher momentum ratios will result in deeper penetration. To determine the extent that premixing and momentum ratio play on the emissions levels observed, the aforementioned injection congurations were devised. Additionally, although it would be impossible to maintain exact values as ow conditions change, the three injectors utilized in this study were sized to maintain relatively constant momentum ratios. Another factor of major importance for CO emissions is the adiabatic ame temperature. To present experimental results with numerous signicant driving parameters in a clearly distinguishable manner, the results are assembled in Table 3. The layout of this table are as follows: the three left-hand columns; the four center columns; and the three right-hand columns. The left-hand columns list the adiabatic ame temperature (as calculated from the
Table 3 CO emission levels, momentum ratio, and adiabatic ame temperatures.

CEA analysis), the equivalence ratio, and the percentage of doping for CH4 doped with CO2. The right-hand columns show the same information for the case of CH4 doped with N2. The central columns (which are color coded for the particular injector utilized) show the CO measurement (in ppmvd) and the square root of the critip cal momentum ratio, J . In theory, if injection conguration, momentum ratio, and unmixedness have no effect, then CO values would remain constant for a particular selection of adiabatic ame temperature, equivalence ratio, and CO2 percentage. For example, for the TAD = 1870.7 K, / 0:70, and CO2 = 30% case (outlined by the dashed box in the table), the CO values are 0.7, 1.4, and 1.5 ppmvd, respectively, for the three different injection congurations. Additionally, it is noted that the square of the momentum ratios are rather close (32.5 and 30.3, respectively). Although it would be tempting to say that the injected conguration had higher CO values, these results are within the experimental uncertainty of the equipment. Continuing rightward on this same row, it is observed that the injected CH4/N2 has an approximately similar CO reading of 2.8 as compared to the 1.5 ppmvd of the injected CH4/ CO2 case; and at approximately the same adiabatic ame temperature of TAD = 1881.0 K. The most signicant results presented in this table are for the cases of higher equivalence ratio and CO2 doping concentration. For instance, for the case of TAD = 1869.4 K, / 0:75, and CO2 = 50% (outlined by the red box in the table), the fully premixed case and the injected CH4/premixed CO2 case have comparable CO emissions of approximately 8.6 ppmvd, whereas the injected CH4/CO2 case has almost twice the CO concentration at 15.8 ppmvd. Comparing these results to the N2 case at approximately the same temperature, it is observed that the CO emissions are 8.9 ppmvd with comparable momentum ratios of 5250.1, respectively. These results would indicate that injection method of CO2 does have an effect and that, compared to N2, the presence of CO2 is involved with the combustion reactions and exhibits more than just a diluent effect. However, further study to verify these trends is required. Two other factors that may contribute to increased CO production with CO2 doping are combustor residence time and adiabatic

268

C.J. Mordaunt, W.C. Pierce / Fuel 124 (2014) 258268

Acknowledgments The authors would like to gratefully acknowledge the signicant contributions of the following Bucknell students: Curtis Saunders10, Devin Weaver10, Michelle Beck11, John Stevenson11, Matthew Tanner M.S.12, and Carleton Knisely14. Additionally, this work would not have been possible without the dedication, expertise, and skill of the following Bucknell staff: Tim Baker, Skip Fegley, Wade Hutchison, and Dan Johnson.

References
[1] Bade Shrestha S, Narayanan G. Fuel 2008. [2] Hopwood L. Fuel and energy-renewable fuels and energy fact sheet; 2009. [3] Forster P, Ramaswamy V, Artaxo P, Berntsen T, Betts R, Fahey D, Haywood J, Lean J, Lowe D, Myhre G, Nganga J, Prinn R, Raga G, Schulz M, Van Dorland R. Changes in atmospheric constituents and in radiative forcing. Technical report, IPCC, Cambridge, United Kingdom, New York (NY); 2007. [4] Tillman D, Sarkanen K, Anderson L. Fuels and energy from renewable resources. New York: Academic Press; 1977. [5] Lafay Y, Taupin B, Martins G. Exp Fluids 2007. [6] Lastella G, Testa C, Cornacchia G. Energy Convers Manage 2002. [7] Bari S. Renew Energy 1996. [8] Zhang Q, Noble D, Lieuwen T. J Eng Gas Turbines Power 2007. [9] Wilson DA, Lyons KM. J Energy Resour Technol 2009. [10] Wilson DA, Lyons KM. Fuel 2008. [11] Turns S. An introduction to combustion. Boston: McGraw Hill; 2000. [12] Richards G, McMillian M, Gemmen R. Prog Energy Combust Sci 2001. [13] Roy S, Hegde M, Madras G. Appl Energy 2009. [14] Mordaunt CJ. Dual fuel issues related to performance, emissions, and combustion instability in lean premixed gas turbine systems. PhD thesis. The Pennsylvania State University; 2005. [15] Lieuwen T, McDonell V, Petersen E. J Eng Gas Turbines Power 2008. [16] Rayleigh J. The theory of sound, vol. II. New York: Dover Publications; 1945. [17] Weiland P. Appl Microbiol Biotechnol 2010. [18] Lee K, Kim H, Park P, Yang S, Ko Y. Fuel 2013. [19] Tillman D, Harding N. Fuels of opportunity: characteristics and uses in combustion systems. New York: Elsevier; 2004. [20] Christensen T, Cossu R, Stegmenn R. Landlling of waste: biogas. New York: E & FN SPON; 1996. [21] Lee J, Santavicca D. J Propul Power 2003. [22] Najm H, Knio OM, Paul P, Wyckoff P. Combust Sci Technol 1998. [23] Najm H, Paul P, Mueller C, Wyckoff P. Combust Flame 1998. [24] Pierce WC. Combustion of biogas in a gas turbine simulation. Masters thesis. Bucknell University; 2011. [25] Seo S. Parametric study of lean premixed combustion instability in pressurized model gas turbine combustor. PhD thesis. The Pennsylvania State University; 1999. [26] Chai X, Mahesh K. In: 40th AIAA uid dynamics conference. Chicago (IL): American Institute of Aeronautics and Astronautics; 2010. [27] Dickmann D, Lu F. In: 38th Fluid dynamics conference and exhibit. Seattle (WA): American Institute of Aeronautics and Astronautics; 2008. [28] Coppens A, Frey A, Kinsler L. Fundamentals of acoustics. New York: John Wiley & Sons; 1982. [29] McBride B, Gordon S. Computer program for calculation of complex chemical equilibrium compositions and applications. Cleveland, Ohio: National Aeronautics and Space Administration, Lewis Research Center; 1996. [30] Goodwin D. Cantera users guide, division of applied science. California Institute of Technology; 2001. Release 1.2. [31] Smith G, Golden D, Frenklach M, Moriarty N, Eiteneer B, Goldenberg M, et al. Available at: http://www.me.berkeley.edu/gri_mech/. [32] Glarborg P, Bentzen L. Energy Fuels 2008. [33] Amato A, Hudak B, DSouza P, DCarlo P, Noble D, Scarborough D, et al. Proc Combust Inst 2011. [34] Shih W, Lee J, Santavicca D. In: Twenty-sixth symposium (international) on combustion. The Combustion Institute; 1996.

Fig. 11. Residence time calculated by the CEA model using an inlet air temperature of 422 K.

ame temperature effects. As shown in Fig. 11, the combustor residence times predicted by the CEA models for the CH4 and CH4/CO2 combinations decreases from approximately 15 ms at lower adiabatic ame temperatures to approximately 10 ms at the higher temperature values. Excess amounts of CO2 introduced to the combustion chamber will lower ame temperatures, allowing exhaust products to reside inside the chamber longer and increase CO formation. Lastly, it has been noted that radiation heat loss increases with higher CO2 concentrations, causing a further reduction of the adiabatic ame temperature and increasing CO concentrations [18]. 5. Conclusions and future work The design details of an optically-accessible atmospheric-pressure model gas turbine combustor have been presented. The primary focus of this work is to study the effect of fuel injection conguration, CO2 doping level, and inlet air temperature to emulate the combustion of biogas in land-based gas turbines. Preliminary results are presented and will be used as points for further study. It was observed through emissions measurements and CH* chemiluminescence images that high concentrations of CO2 cause increased CO emissions and weakens the ame, effecting the lean-blowout point. Additionally, it was observed that injection method and momentum ratio have an effect on measured CO emissions. Emission measurements presented in this work should be considered with caution because the location of the emission sampling probe resulted in observable differences: however, this was not easily quantiable. Future work will include the installation of a convergingdiverging exit nozzle that will provide a choked exit condition and more uniform emission prole. Further study of limit-cycle oscillation and radiant heat loss will be studied with phase-locked chemiluminescence imaging and an ICCD camera.

You might also like