Large Eddy Simulations (LES) are used to study a lean swirl-stabilized gas turbine burner where the flow exhibits two stable states. In the first one, the flame is attached to the central bluff body upstream of the central recirculation zone which contains burnt gases. In the second one the flame is detached from the central bluff body downecirculation zone which is filled by cold unburnt gases and dominated by a strong Precessing Vortex Core (PVC). The existence of these two states has an important effect on the dynamic response of the flame (FTF): both gain and phase of the FTF change significantly in the detached case compared to the attached one, suggesting that the stability of the machine to thermoacoustic oscillations will differ, depending on the flame state. Bifurcation diagrams show that the detached flame cannot be brought back to an attached position with an increased fuel flow rate, but it can be re-attached by forcing it at high amplitudes. The attached flame however, behaves inversely: it can be brought back to the detached position by both decreasing or increasing the pilot mass flow rate, but it remains attached at all forcing amplitudes. IntroductionSwirling flows are commonly used to help flame stabilization in gas turbine combustion chambers. They feature several types of vortex breakdown and can exhibit bifurcation phenomena where different states can co-exist and the flow can jump spontaneously from one to another [1,2]. Bifurcations of flames in configurations which are close to real gas turbine chambers have not been investigated so far even though engineers report that they observe these mechanisms and that there is a link between flame states and thermoacoustic instabilities: when the flame changes from one state to another, its acoustic stability characteristics also change.Two dynamic phenomena are usually observed in swirled combustion chambers: (1) a helical flow instability, the so-called precessing vortex core (PVC) and (2) thermo-acoustic instabilites.The PVC is an hydrodynamic instability in swirling flows [3].I t is a large scale structure characterized by a regular rotation of a spiral structure around the geometrical axis of the combustion chamber. It can occur at high Reynolds and swirl number flows [4][5][6][7][8][9][10] and its precession frequency is controlled by the rotation rate of the swirled flow [3]. Several studies show that combustion can suppress the PVC [6,7,11], but other cases also show PVCs which are present in reacting flows [12][13][14][15]. The interaction of PVC with flames has been analyzed for example by Stöhr et al.[16]: they found the PVC to enhance mixing and to increase the flame surface. This was associated to structures in the inner shear layer, whereas Moeck et al. [15] observed the outer shear layer to create most of the flame perturbations. Both researchers as well as Staffelbach [13] using LES evidenced a ''finger-like'' rotating structure at the flame foot around which the PVC is turning. Furthermore, asymmetric fluctuations of the he...

The necessity for a combustion system to work with premixed flames and its capability to cope with rapid load variations avoiding the occurrence of thermo-acoustic instabilities, has led to investigate the complex dynamic phenomena that occur during combustion. Thanks to numerical simulations it is possible to examine these complex mechanisms getting useful information to optimize the combustion system. The aim of this work is to describe a numerical procedure developed in Ansaldo Energia for the investigation of combustion dynamics occurring in Ansaldo Energia gas turbines. In this paper, firstly the experimental apparatus of a full scale atmospheric test rig equipped with Ansaldo Energia burner is described. Secondly, the flame behavior is studied by means of a Large Eddy Simulation (LES). Once the LES has reached a statistically stationary state, a forcing is added to the system to compute the Flame Transfer Function (FTF), in terms of amplitude n and delay time τ, related to initial phases of humming. Thirdly, the forced flame simulations are used as the input of an Helmholtz solver to analyze the acoustic behavior of the system, which is then compared to experimental data. Finally, to evaluate the feasibility of a less computationally intensive approach, a RANS simulation of the same configuration is described and the results are transferred to FEM (Finite Element Method) Helmholtz solver: a comparison between the LES approach and the RANS approach is carried out with reference to the experimental data.

In the framework of the non-standard fuel combustion research in micro-small turbomachinery, a newly designed micro gas turbine combustor for a 100-kWe power plant in CHP configuration is under development at the Ansaldo Ricerche facilities. Combustor design starts from a single silo chamber shape with two fuel lines, and is associated with a radial swirler flame stabiliser. Lean premix technique is adopted to control both flame temperature and NO x production. Combustor design process envisages two major steps, i.e. diagnostics-focussed design for methane only and experimentally validated design optimisation with suitable burner adaptation to non-standard fuels. The former step is over, as the first prototype design is ready for experimental testing. Step two is now beginning with a preliminary analysis of the burner adaptation to non-standard fuels. The present paper focuses on the first step of the combustor development. In particular, main design criteria for both burner and liner cooling system development are presented. Besides, design process control invoked both 2D and 3D CFD analysis. Two turbulence models, FLUENT standard k-ε model and Reynolds Stress Model (RSM), are refereed and the results compared. Here both a detailed analysis of CFD results and a preliminary analysis of main chemical kinetic phenomena are discussed

Chemical Reactor Modelling approach has been applied to evaluate exhaust emissions of the newly designed ARI100 (Patent Pending) recuperated micro gas turbine combustor developed by Ansaldo Ricerche SpA. The development of the chemical reactor network has been performed based on CFD reacting flow analysis, obtained with a global 2-step reaction mechanism, applying boundary conditions concerning the combustion chamber at atmospheric pressure, with 100% of thermal load and fuelled with natural gas. The network consists of 11 ideal reactors: 6 perfectly stirred reactors, and 5 plug flow reactors, including also 13 mixers and 12 splitters. Simulations have been conducted using two detailed reaction mechanisms: GRI Mech 3.0 and Miller & Bowman reaction mechanisms. Exhaust emissions have been evaluated at several operating conditions, obtained at different pressure, and considering different fuel gases, as natural gas and a high H2 content SYNGAS fuel. Furthermore, emissions at different thermal loads have been investigated when natural gas at atmospheric pressure is fuelled. Simulation results have been compared with those obtained from combustion experimental campaign. CO and NOx emissions predicted with CRM approach closely match experimental results at representative operating conditions. Ongoing efforts to improve the proposed reactors network should allow extending the range of applicability to those operating conditions whose simulation results are not completely satisfying. Given the small computational effort required, and the accuracy in predicting combustor experimental exhaust emissions, both CO and NOx, the CRM approach turnout to be an efficient way to reasonably evaluate exhaust emissions of a micro gas turbine combustor.

No abstract