This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).
3D numerical combustion simulation in a can burner fed with methane was carried out in order to evaluate pollutant emissions and the temperature field. As a case study, the General Electric Frame 6001B system was considered. The numerical investigation has been performed using the CFD code named ACE+ Multiphysics (by EsiGroup). The model was validated against the experimental data provided by Cofely GDF SUEZ and related to a real power plant. To completely investigate the stability of the model, several operating conditions were taken into account, at both nominal and partial load. In particular, the influence on emissions of some important parameters, such as air temperature at compressor intake and steam to fuel mass ratio, have been evaluated. The flamelet model and Zeldovich’s mechanism were employed for combustion modeling and NO_{x} emissions, respectively. With regard to CO estimation, an innovative approach was used to compute the Rizk and Mongia relationship through a userdefined function. Numerical results showed good agreement with experimental data in most of the cases: the best results were obtained in the NO_{x} prediction, while unburned fuel was slightly overestimated.
In recent decades, stationary gas turbines have become firmly established as prime movers in the gas and oil industry, acquiring new ranges of application in many areas of utility power generation and, in particular, in combined cycle plants [
In the present article, the numerical study was set with reference to operating conditions of one of the ten burners of the FRAME 6B General Electric gas turbine machine, having the nominal electric power of 40 MW_{e}. In particular, the simulations referring to both nominal and partial load conditions were investigated, focusing attention on the thermal field and pollutant emissions in order to firstly validate the numerical model through experimental data. Secondly, the model was used to obtain regression laws to predict pollutant emissions out of the simulated ranges or to improve the employed correlations. This information is important, since it represents a starting point of a more complex study for the evaluation of the power plant conversion to bioliquid. The first step is thus the construction of a numerical model to reproduce the actual configuration and its operating conditions. The experimental data were provided by
The gas turbine we studied is installed within a combined cycle power plant whose steam turbine has a nominal electric power of 12 MW_{e} (
The power plant can be fed with both methane and offgas, the latter produced by one of the thermal users. Since the offgas contribution is minimal, our studies have focused on methane. The most important parameters regarding the fuel injection system are listed in
Power plant schematization.
Gas turbine specifications at the design condition.
Engine Type  Frame 6B (General Electric) 

Compression ratio (β)  11.8:1 
Temperature at the end of compression  670 K 
Mass flow rate (kg/s)  139 kg/s 
Engine speed (at nominal conditions)  5,115 rpm 
Injection pressure  Variable (up to 40 bar) 
Temperature  620 K 
Mass flow rate  Minimum: 0.208/Maximum: 0.445 kg/(s·can) 
Nozzle hole diameter  2 mm 
Fuel injector: system specifications at nominal conditions.
Injector Type  Pressure swirl Atomizer 

Injection pressure  40:1 
Temperature  350 K 
Mass flow rate  0.28 kg/s 
Number of nozzles  4 
Fuel to air ratio (FAR)  0.019 
Fuel  Methane/offgas 
The steam can be produced by both the heat recovery generator and the post combustion unit: when the steam demand is at the maximum level, these two sections operate in parallel, with the gas turbine working at partial load.
Typical gas turbine load diagram.
Considering the most frequent operating conditions of the plant, simulations for 75, 80, 90 and 100% of the nominal thermal power have been taken into account. Obviously, changes of the air conditions due to compressor regulation have been considered, too. The information about the pollutant emissions will be reported in the following paragraphs, where they will be compared with the numerical results.
The FRAME 6B can is a reverseflow combustor, i.e., the air coming from the compressor is passed inside a preheating space heading from the base towards the dome. This path allows the compressed air to be heated and to enter the combustion chamber with a higher temperature than that at the end of compression, thus increasing the combustion efficiency.
When the air arrives close to the dome, it is swirled and pushed into the primary combustion zone. The injectors of the fuel gas and of the steam are placed in the terminal part of the dome. The central part of the liner, where the combustion chemical reactions develop, is followed by a dilution zone: the holes on the walls of the liner have the dual function to dilute the exhaust gas flow and cool the walls of the combustion chamber. The total length of the burner and its maximum diameter are about 1.50 m and 0.56 m, respectively (
The computational mesh was created using the ACE+ Multiphysics tool. Because of the symmetry of the system and its components, the CFD calculations are performed on 90° sector meshes. Two different domains have been defined: the first is related to fluid volume, the second to the metal walls of the liner. This choice allows the preheating of the compressed air that occurs in the interspace to be described, solving concurrently fluiddynamic and heat transfer equations. In both cases, unstructured meshes have been chosen. With regard to the fluid mesh, more dense discretizations have been located near the outlet sections of the fuel and steam injectors, where the speed of the jets takes on values on the order of hundreds of meters per second and within the primary and secondary zones.
(
Fluid mesh properties.
Region  Characteristics 

Preheater region  Cell type: hexahedral 
Averaged cell resolution: 

Injectors region  Axial length: 5 cm 
Cell type: hexahedral  
Averaged cell resolution: 

Combustion chamber: primary and secondary zones  Cell type: hexahedral 
Averaged cell resolution: 

Combustion chamber: dilution zone  Cell type: hexahedral 
Averaged cell resolution: 

Wall region  Two layer of tetrahedral cells 
Total number of cells  1.5 million cells 
Boundary conditions.
Boundary  Type  Value (at Nominal Condition)  Turbulence 

Air inlet  Velocity inlet  28 m/s (normal to boundary)  Turbulence intensity: 0.08 
Dissipation rate: 0.055 m (hydraulic diameter)  
Gas outlet  Pressure outlet  11.8 bar  Back flow kinetic energy: 0.01 m^{2}/s^{2} 
Dissipation rate: 0.135 m (hydraulic diameter)  
Fuel inlet  Mass flow rate inlet  0.07 kg/s (for the single nozzle)  Turbulence intensity: 0.12 
Dissipation rate: 0.001 m (hydraulic diameter)  
Steam inlet  Mass flow rate inlet  0.032 kg/s (for the single nozzle)  Turbulence intensity: 0.12 
Dissipation rate: 0.002 m (hydraulic diameter) 
As to the turbulent closure, the kε approach with standard wall function has been used to model the turbulent effects. With regard to the combustion model, a chemistry with 35 species and 177 reactions was employed. This is a homemade chemistry derived from the detailed mechanism, GRIMech 3.0 [
The flamelet model, used in the nonpremixed combustion approach to account for chemical nonequilibrium, regards the turbulent flame as an ensemble of thin, locally onedimensional flamelet structures. In this work, three flamelets with strain rate χ equal to 0.01, 0.1, one and 57 flamelets with initial strain rate χ_{0} = 2 and Δχ = 1 have been generated to characterize combustion. Using flamelets, species mass fraction, temperature, as well as the chemical reactions can be computed as a function of the mixture fraction (
Assuming that Z and χ_{st} have independent distributions, the previous equation can be written as:
Equations (1) and (2) for species mass fraction and energy are solved simultaneously with the flow: the flamelet equations are advanced for a fractional timestep using properties computed from the fluid flow, and then, the latter is advanced for the same fractional timestep using properties from the flamelets (
Flamelet model computing. PDF, probability density function.
For the estimation of CO emission, an innovative approach has been used. The following semianalytical correlation proposed by Rizk and Mongia has been implemented into a userdefined function (UDF) [
Values of the constants for CO prediction [
a_{1}  a_{2}  a_{3}  a_{4} 

66.48 ×10^{6}  28.16  2.69  0.5 
The extended Zeldovich’s mechanism has been taken into account to model reactions for thermal NO_{x} formation:
The reaction rates are assumed as follows [
Prompt NO is caused by the presence of radicals in the flame zone; the following equation was adopted to compute its contribution [
Heat transfer modeling involves the wall of the liner. The gasside (
A comparison of the results with experimental data can be carried out focusing attention only on the discharge temperature of the combustion chamber. The values reported in
Comparison between experimental and numerical result at 100% load and ISO conditions.
Firing Temperature  NO_{x} Pollutant Emission  CO Pollutant Emission  

Experimental measurement  Numerical result and percentage error  Experimental measurement (ppm)  Numerical result and percentage error (ppm) (at 15% O2)  Experimental measurement (ppm)  Numerical result and percentage error (ppm) 
1,373 K  1,420 K  83.7  84.7  11.8  11.9 
3.42%  +1.25%  +0.82% 
These simulations are also able to examine the response of the gas turbine model to variations in operating condition. In particular, six further simulations have been performed, using inlet temperatures ranging from 268 K to 298 K. In
The effect of the air temperature at the compressor inlet on pollutant emissions.
Air Temperature Inlet (K)  NOx (at 15% O_{2})  CO  

Experimental Value in Comparison with ISO Condition (%)  Numerical Value in Comparison with ISO Condition (%)  Experimental Value in Comparison with ISO Condition (%)  Numerical Value in Comparison with ISO Condition (%)  
268  90.05  87.15  102.33  102.91 
273  94.16  92.22  101.87  102.66 
278  96.54  95.41  101.29  102.27 
283  98.19  98.63  100.50  101.15 
288  100  101.25  100  100.82 
293  103.21  104.50  99.76  100.33 
298  104.93  106.77  99.55  99.77 
(
It is evident from the previous figures that the predicted trends are qualitatively similar to the experimental values. In particular, they capture the trend of reduced CO and increased NO_{x} with air temperature rising. As for CO emissions, although the model tends to overestimate the real data, it is evident how the simulations can completely follow the small variations measured on the real machine. In fact, as a typical characteristic of all heavyduty machine series, the carbon monoxide emissions increase quickly only if the firing temperature is drastically reduced. Therefore, the stability of the numerical model is based on the ability to appreciate these very small variations. However, the methodical overestimation is due to a systematic error in the evaluation of the main parameters of Equation 6. Two different sources of error are the residence time estimation, since UDF elaborates an averaged value, and the numerical temperature at the exit of the burner. Similarly, the supposed pressure loss may deviate from the real value, because there is no experimental data available for its estimation.
To completely evaluate the pollutant emissions, the effect of different mass flow rates of steam injected within the combustion chamber has been studied at full load and ISO condition, also. In a STIG (STeam Injected Gas Turbine) cycle, the amount of injected steam represents a sensible parameter to take into account, since there are practical limits to the amount of steam, as well as water that can be injected into the combustor before a serious flame stability problems occur [
NO_{x} emission
It is evident how the computational model has a good agreement with experimental results, primarily for intermediate values of
The model analysis has been extended for higher values of
The effect of steam injection on CO emission is almost negligible. Steam temperature is very similar to the air temperature, so that experimental data at nominal conditions show no appreciable variations with increasing
Since the previous study demonstrated the validity of the present CFD model, it is interesting to analyze some physical characteristics of the internal fields of the combustion chamber.
(
Secondly, the flame develops into a thin interface with the air flux and produces a compact body flame: this structure is typical of the diffusive reaction mechanism. The cutting plane on the dilution zone allows one to verify the influence of the cold air flux in preserving walls and decreasing the temperature of exhaust gases before the first stage of the gas turbine.
The strong connection between the temperature field and the NO_{x} production rate is evident in
Keeping attention on fluid dynamics, one of the most important parameters that can be verified is the Mach number of the methane jet (
NO_{x} production rate on the symmetry plane and on two different normal
Fuel Mach number: contour plot and chart.
Recirculation zone.
From this point of view, the attention has been focused on the Mach number and recirculation flow to characterize the primary zone, since the employed correlation for CO contains the assessment for
As mentioned earlier, the analysis has been extended to 75, 80 and 90% of the nominal thermal power.
Discharge temperature at different loads.
Comparison between numerical and experimental temperatures at different load values.
However, for the studied cases, the decreasing of the temperature peak reached within the combustion chamber determines a reduction in terms of NO_{x} pollutant emissions. This trend is well described also by the numerical model (
NO_{x} prediction at different load values.
The evaluation of the numerical model performance in terms of CO assessment at partial load has been summarized in
CO prediction at different load values. UDF, userdefined function.
To improve the UDF and minimize the error, a hybrid strategy can be adopted (
UDF flow chart for the CO minimization.
In this paper, a numerical model for the prediction of pollutant emissions within a Frame 6B combustion chamber was presented. The analysis involved both nominal and partial load conditions in order to completely evaluate the setting of the models that have been implemented. Furthermore, a sensitive analysis was developed varying the intake temperature of the air and the steam to the fuel mass ratio at full load.
The real discharge temperature was used for a direct comparison with the numerical result: in any case, the differences provided negligible percentage errors. As to the internal temperature field, there is no experimental data available; its numerical analysis confirmed some peculiarities of nonpremixed diffusion combustion, such as the body flame structure and the mixing phenomena between reactants. However, an indirect control is possible analyzing the NO_{x} emissions at the outlet section of the combustion chamber. In fact, excluding the results obtained for extremely low values of the steamtofuel mass ratio, NO_{x} concentrations have a good agreement with experimental results. Therefore, it can be concluded that also the thermalfluiddynamic field within the burner reflects the real phenomenon. A regression numerical curve was also obtained to compute NO_{x} for a high steam mass flow rate, producing an excellent fitting with real data.
Ritz and Mongia correlation was implemented through an innovative UDF for CO computing. Even if the discrepancies with the experimental data are always acceptable, a systematic overestimation was verified at both nominal and partial load. In this latter case, the influence of discharge temperature on calculation was studied: instead of the numerical temperature, the experimental value was used to improve the code and limit the percentage error.
The author wishes to acknowledge
The author declares no conflict of interest.