Research Topic

Development of numerical methods for pollution prediction in gas turbines


Nitrogen oxides, such as NO and NO2, often summarized as NOx, emitted into the atmosphere contribute to acid rain, ozone formation and smog problems. They lead to climate change and can cause severe health problems. Therefore, the reduction of NOx is a key issue in today's gas turbine and IC engine development. However, the prediction of nitrogen oxides is difficult because of their slow formation compared to other reaction products. In industrial applications postprocessing tools for Reynolds Averaged Navier Stokes (RANS) computations exist which take the mean temperature profile and include statistic effects in order to compute the amount of NO. In turbulent flames, where unsteady flow phenomena occur, the temperature –among all other properties– strongly fluctuates. Large Eddy Simulation (LES) is conceptually well suited to predict such unsteady flows. However, for the LES context, none of these postprocessing tools exist. With the growing application of LES in industrial applications, the demand for good NO modeling and prediction rises.

The same applies for carbon monoxide (CO). Conventional gas turbine combustors typically feature a rich primary combustion process. Afterwards, the products are quickly mixed with secondary air. The prediction of CO emissions in this regime using Flamelet Generated Manifolds (FGM) built from premixed flamelets is rather difficult, as the burnout behavior and the distribution of species  behind the reaction zone is strongly deviating from pure mixing.

Method and Theory

Comparison of the NO distribution for Transport (left) and FGM (right).

State of the art for chemistry modeling are the different variations of flamelet models. An extension to the basic flamelet approach by Peters using a mixture fraction variable only is the inclusion of finite rate chemistry effects by a reaction progress variable. This so-called Flamelet Generated Manifolds approach (FGM) was developed by van Oijen (2002). The details of the implementation of the FGM in the Fastest-ECL code are given in Ketelheun et al. (2009).

The FGM methodology is based on the direct connection of the species mass fractions, temperature and density to the controlling variables mixture fraction and progress variable. However, minor species like NO have characteristic time scales strongly deviating from those of the major species used as progress variables. Therefore, additional modeling is required. Although the NO mass fraction cannot be described well by the progress variable, its source term is sufficiently well captured. Solving an additional transport equation for the NO mass fraction with taking its source term from the FGM database allows for the slow development of the NO distribution along the flow. To include interaction of turbulence and chemistry, presumed PDF modeling has been applied. More details on the NO modeling can be found in Ketelheun et al. (2010).


Time averaged radial profiles of the NO mass fraction, mean values (left) and fluctuations (right).

The chosen test case is a bluff body configuration of the University of Sydney. The fuel (50 vol% methane & hydrogen) enters the burner through a central nozzle. The recirculation zones establishing in the wake of the bluff body lead to mixing with the air from the coflow. In addition, hot exhaust gases recirculate and preheat the fuel. Albeit its rather simple geometry the burner features flow characteristics typical for technical applications, in particular the recirculation zone and preheating of the fuel.

The new model solving an additional transport equation is compared to results of a standard FGM procedure. Figure 1 shows an instantaneous snapshot of the NO distribution obtained by both methods. One can well observe the different structure and the higher peak values of the standard FGM model. The comparison of time averaged radial profiles of the NO mass fraction to experimental data is shown in Figure 2. The left columns show the mean values, the right columns give the fluctuations.

The standard FGM model does not agree with the shape of the profile in the two lower positions. At the positions further downstream the shape seems to match better with the experimental data, the peak values still do not match. The model including an additional transport equation for the NO mass fraction yields much better results, they are in very good accordance with the experiments, especially in the lower part of the flame. The slopes at the inner and outer shear layers are well captured, although the peak values are not fully reached. The disagreement in the downstream positions is related to a slightly different position of the flame in the computation and the experiment. The fluctuations of the NO mass fraction are much too high in most positions in the flame for the standard model, whereas the additional transport equation yields too low values for the fluctuations. Seen integrally, the amount of NO obtained by the novel model is much closer to the experimental data than the amount obtained by the standard FGM approach. With the transport equation model, an improved prediction of the NO mass fraction could be obtained compared to the standard FGM approach. The higher computational effort for the additional equations is justified by the higher prediction accuracy for NO.

Future Work

Future work will concentrate on the modeling of carbon monoxide. Therefor, the modeling implemented for NO will be extended with an additional model for the post flame zone. An additional time scale describing this burnout behaviour will be introduced. Furthermore, the addition of hydrogen to the fuel is to be considered with respect to its different diffusivity compared to other species and the mixture fraction.


J. A. van Oijen: Flamelet-Generated Manifolds: Development and Application to Premixed Laminar Flames, PhD. Thesis, TU Eindhoven, 2002.

A. Ketelheun, C. Olbricht, F. Hahn and J. Janicka: Premixed Generated Manifolds for the Computation of Technical Combustion Systems. ASME Turbo Expo, Orlando, FL, USA, 2009.

A. Ketelheun, C. Olbricht, F. Hahn and J. Janicka: NO Prediction in Turbulent Flames Using LES/FGM With Additional Transport Equations. Proc. Combust. Inst., 33 (2) pp. 2975-2982.

Key Research Area

Multi-Physics; Large Eddy Simulation, Combustion Modeling, Tabulated Chemistry


Anja Ketelheun


Petersenstrasse 30

D-64287 Darmstadt



+49 6151 16 - 24401 or 24402


+49 6151 16 - 6555




ketelheun (at) ekt.tu...

 Print |  Impressum |  Sitemap |  Search |  Contact |  Privacy Policy
zum Seitenanfangzum Seitenanfang