Thermo-acoustic Instabilities of Annular Combustion Chambers

I. Background and motivation

The reliability and flexibility of the modern cycle power gas turbine has imposed this technology as standard energy supply source in the present and in the foreseeable future.

The need for high thermal efficiency and low levels of pollutant emissions requires continuous research and development. To comply with today’s stringent regulations, the concept of lean premixed combustion has been adopted as it offers a certain number of advantages in controlling the emissions level.

Nevertheless, industrial practice has proven this technology to be sensitive to the the development of self sustained thermo-acoustic combustion instabilites resulting in high levels of pressure pulsations as it can be seen in Figure 1. These disturb the normal operation and can produce extensive hardware damage to the system. The driving mechanism for these undesired phenomenons is mainly the feedback loop with a positive growth rate linking the unsteady heat release of the turbulent premixed flame and the acoustic field of the combustion system.

Fig 1. Time domain of the measured acoustic pressure in the combustor during unstable operation

II. Theory and research

Research is performed to learn more on these undesired phenomenon and to develop the necessary know-how needed to predict eventual unstable behaviour in the early design stages of the combustor. Consequently, design rules could be undertaken and future instability problems followed by costly infrastructural interventions during the nominal operation avoided. The preferred approach, due to robustness and reliability, is the stability analysis of the system made with the acoustic network modelling approach. For this, the dynamic behaviour of the burner and flame, in terms of the transfer matrices/transfer functions, represents the key input element.

The goal of this project is the experimental determination of the burner and flame transfer matrices for a swirled stabilised premixed burner in a single and an annular combustion chamber. A special attention is dedicated to the comparison between the two burner configuration for a qualitative and quantitative assessment of the differences and similarities respectively. This latter aspect has a great importance from the design point of view in general and concerns the gas turbine producers in particular. Nowadays, the standard procedure is to measure the dynamic characteristics of the burner and flame in a single burner configuration, in test rig conditions, and then use these results for the analysis and development of a full scale annular combustion chamber with assuming the similarity between the two cases. The research undertaken in the project is conducted to look more into this aspect and to address the assumption.

III.Hardware and infrastructure

The experimental infrastructure consists of a single and an annular combustion chamber. The latter is a scaled down model of a real gas turbine combustor and has a maximum thermal power of 1500 kW. It can be operated in perfectly premixed and preheated mode. An important characteristic of the test rig, essential for undergoing work, is the reproduction of the acoustic boundary conditions found in a real engine.

Fig.2 The annular combustor test rig

Furthermore, both test rigs can be operated in externally pulsated mode for the measurements of the burner and flame transfer matrices. This is achieved with air sirens mounted on the up- or downstream side of the burner. As a particularity for the annular combustor, multidimensional acoustic modes including axial and azimuthal can be excited here.

The two combustion chambers are provided with glass windows which allow the optical access inside for flame characterisations. The combustor wall is fitted with a large number of slots for dynamic pressure transducers. In the annular configuration these are distributed in both axial and azimuthal directions, arrangement which is necessary to capture the complex 3D character of the acoustic field.

IV.Measurement techniques

A wide spectrum of measurement techniques are employed for the actual project. Dynamic pressure transducers coupled with high speed data acquisition cards are used to measure the acoustic pressure in the combustor. Afterwards, post processing algorithms followed by a modal analysis method, developed in this project, are applied to the raw data to gain important information on the configuration and spatial distribution of the acoustic field. Additionally to the acoustic pressure, the acoustic velocity can be measured in the burner using the constant temperature anemometer. This, together with high speed optical measurement techniques like UV photomultipliers or diodes are employed for the experimental determination of the flame transfer functions.

A high speed camera system provided with UV filters is used to optically characterise the reaction zone in the combustor. Thus, important informations on the flame position and form are obtained. The quality of the combustion process is additionally assessed by analysing the composition of the exhaust gas. In this sense, a probe is mounted at an axial position further downstream of the reaction zone close to the exit of the combustor.

Last but not least, a Particle Image Velocimetry System (PIV) is used for velocity field measurements in the single and annular combustion chambers. The purpose is to investigate the aerodynamic characteristics of the flow which determine the static and dynamic behaviour of the flame observed. In the case of the annular combustor an important issue to be investigated is the burner to burner interaction, effect which is not present in the single burner configuration.

V. Acknowledgment

The project is funded by Alstom Power Generation AG, Bayerische Forschungstiftung (BFS), Staatsministerium für Wissenschaft, Forschung und Kunst (StMWFK) und Staatsministerium für Wirtschaft, Infrastruktur, Verkehr und Technologie (StMWIVT) whose support is gratefully acknowledged.