Elsevier

Combustion and Flame

Volume 159, Issue 3, March 2012, Pages 1109-1126
Combustion and Flame

Cinematographic OH-PLIF measurements of two interacting turbulent premixed flames with and without acoustic forcing

https://doi.org/10.1016/j.combustflame.2011.09.006Get rights and content

Abstract

This paper describes an experimental investigation into the interactions that occur between two lean turbulent premixed flames stabilised on conical bluff-bodies when they are moved closer together. Cinematographic OH-PLIF measurements were acquired to investigate adjacent flame front interactions as a function of flame separation distance (S). Flame surface density (FSD) and curvature were determined to characterise the unforced flames. Acoustic forcing was then applied to explore the amplitude dependent thermo-acoustic response. Phase-averaged FSD and global heat release measurements in the form of OH chemiluminescence were obtained for a range of forcing frequencies (f) and amplitudes (A) as a function of S. As the flames were brought closer together the adjacent annular jets were found to merge into a single jet structure. This caused adjacent flame fronts to merge above the wake region between the two flames at a location determined by the jet efflux (flame angle) and S. This region of flame–flame interaction we refer to as ‘interacting region’. In the unforced flames, a trend of increasingly negative curvature for decreasing S produced a small net increase in flame surface area via cusp formation. When subjected to acoustic forcing, S-dependent regimes were found in the global heat release response as a function A. The overall trend showed that the occurrence of jet/flame merging reduces the value of A at which non-linear response occurs. In support of previous findings for flames stabilised along shear layers, the phase-averaged FSD showed that the flame dynamics that drive the thermo-acoustic response result from the roll-up of vortices which generate large-scale vortex–flame interactions. Compared with axisymmetric flames, the occurrence of jet merging alters the vortex–flame interactions resulting in an asymmetric contribution to the heat release between the wall and interacting regions. The majority of the heat release was found to occur in the interacting region through the rapid production and destruction of flame surface area. The occurrence of jet merging and large-scale interactions between adjacent flames result in different physical mechanisms that drive the thermo-acoustic response compared with single axisymmetric flames.

Introduction

It is well known that premixed combustion offers significant performance benefits in terms of emissions and as a result has increasing influence on gas turbine combustor design. It is also well known that premixed flames are inherently more susceptible to self-excited thermo-acoustic oscillations. These instabilities arise when velocity fluctuations imposed by acoustic waves cause fluctuations in the heat release rate which, under the right conditions, feedback causing the amplitude of the instability to grow until a limit cycle is established. The non-linearity that leads to finite-amplitude limit cycle oscillations is usually considered to originate from the flame response via a number of different mechanisms, namely changes in flame surface area [1], [2], [3], [4], [5] and/or fluctuations in local equivalence ratio which alters the flame speed [6], [7], [8], [9]. In self-excited or forced flames stabilised along shear layers the presence of an unsteady base flow results in the formation of vortex structures that interact with the flame front [4], [10], [11], [12], [5], [13]. The formation of these structures play an important role in modulating the heat release response. In turbulent premixed flames Balachandran et al. [4] showed that the non-linear response was driven by vortex–flame interactions which both generates flame surface area via large scale distortion of the flame front and annihilates flame surface by bringing flame fronts close together which then propagate into each other. This saturation mechanism is also present in swirling flames [14], [15].

Much of our understanding of combustion instability is based on careful experiments and theoretical modelling of the thermo-acoustic response of single axisymmetric flames undergoing longitudinal oscillations. From a theoretical standpoint, the main challenge is to develop nonlinear models which can account for saturation phenomena to predict limit-cycle amplitudes [1], [16], [17], [18]. In general, this involves incorporating a simplified nonlinear flame model within a stability analysis to account for the amplitude dependent coupling between the heat release and velocity disturbances. Accurately defining this acoustic-flame description is non-trivial and such models typically depend on flame transfer functions derived from global chemiluminescence measurements, numerical simulations or phenomenological considerations.

However, in practical applications such as gas turbine combustion, multiple turbulent flames are typically confined within an annular geometry and are free to interact with their adjacent neighbours. In annular configurations the fundamental instability mode is circumferential and appear as standing or rotating modes. Since these instability modes act normal to the flame they can promote interactions between neighbouring flames altering the mean and local flame structure, blow-off limit, and thermo-acoustic response. Consequently, there is great interest in understanding whether or not the flame response in the presence of circumferential modes are fundamentally different to the flame response driven by longitudinal instabilities. A pressing issue is whether or not existing modelling approaches and experiments based on single axisymmetric flames can be applied to transverse instabilities across multiple flames or whether flame models need to be developed to account for interactions between adjacent flames. An overview of the importance and complexity of combustion instabilities relevant to gas turbine combustion can be found in Ref. [19].

Recent numerical and experimental studies have begun to investigate this issue by studying transverse excitation in single flames and multiple flame configurations. Staffelbach et al. [20] performed LES in a full annular geometry with realistic burners, concluding that for their configuration a single burner may offer a good transfer function approximation. However, in their simulation consecutive burners are spaced several diameters apart mitigating any interactions with their neighbours. In a series of experimental studies [21], [22], [23] comparisons were made between an industrial burner in a single sector and annular geometry showing that differences in the flame transfer function occur due to aerodynamic changes to the flame shape. From global heat release and particle image velocimetry (PIV) measurements Hauser et al. [24] observed that transversal excitation resulted in a tilted asymmetric flame structure due to the occurrence of a rotational instability at the forcing frequency. Similar results were found by O’ Connor et al. [25], [26] showing that in a swirl stabilised flames the instability mode switches from axisymmetric under longitudinal forcing to a helical mode under transverse forcing. An earlier study of multiple jets in a dump combustor by Poinsot et al. [11] showed that interactions between multiple jets can lead to instability.

Most studies of multiple flames have been carried in swirling flames and with a fixed spacing between adjacent flames. In this paper we present an experimental investigation into the effect flame separation distance S has on the mean and local flame structure and the thermo-acoustic response of two bluff-body stabilised turbulent premixed flames. The conjecture is that regimes exist where interactions between adjacent flames are important. Emphasis is placed on the nonlinear flame dynamics that occur between adjacent flames as a function of S so that comparisons with the existing literature on single axisymmetric flames can be made.

In the next section of the paper we describe the experimental methods including: details of the measurement techniques, and characterisation of the boundary conditions. The effect of separation distance on the instantaneous and mean flame structure of the unforced flames is then presented and the importance of jet merging on the flame structure in the interacting region discussed. Results from the forced flame experiments are then presented in terms of their global heat release response followed by a discussion of the phase-averaged flame dynamics derived from the cinematographic OH-PLIF measurements. The final section discusses how the flame dynamics are affected when simulating circumferential modes by varying the relative phase of forcing between the two flames.

Section snippets

Apparatus

The apparatus consisted of two identical bluff-body stabilized turbulent premixed flames that were placed side by side. A photograph of the apparatus is shown in Fig. 1a and the general arrangement was based on the experimental set-up found in Ref. [4]. Methane–air mixtures were premixed upstream and flowed into two cylindrical plenum chambers of length 200 mm and inner diameter 100 mm containing flow straighteners. Each flame holder consisted of a 400 mm long circular tube of inner diameter D = 35 

Effect of separation distance on flame dynamics

The effect of three separation distances on the time-averaged FSD is shown in Fig. 4. For consistency, the same coordinate system is used throughout the paper for all the imaging results with the origin (x, y) = (0, 0) positioned to the left. This is consistent with the experiments as adjustments to the separation distance required proportional adjustments to the position of the wall on the right which is represented by the blank space on the right hand side of Fig. 4b and c. The weak OH signal at

Conclusion

In this paper a systematic study of the effect of flame spacing on the mutual flame interactions between two turbulent premixed flames has been presented. Cinematographic OH-PLIF measurements have enabled the instantaneous, mean, phase-averaged FSD and curvature to be determined. Both unforced and forced flames were studied and in the latter case global OH∗ chemiluminescence measurements were obtained to compare the global response with the planar flame dynamics.

A major result is the occurrence

Acknowledgements

This work was funded by the EPSRC and is part of the SAMULET Project 2 Combustion Systems for Low Environmental Impact. Dr. J.R. Dawson is funded by the EPSRC with an Advanced Research Fellowship. The authors would also like to thank Dr. R. Balachandran of University College London and Professors Mastorakos and Cant of the University of Cambridge for helpful discussions as well as Dr. Clemens Kaminski for the loan of one of the bluff-body burners.

References (41)

  • T. Schuller et al.

    Combustion and Flame

    (2003)
  • R. Balachandran et al.

    Combustion and Flame

    (2005)
  • A. Birbaud et al.

    Proceedings of the Combustion Institute

    (2007)
  • J.H. Cho et al.

    Combustion and Flame

    (2005)
  • A. Birbaud et al.

    Combustion and Flame

    (2008)
  • K. Schadow et al.

    Progress in Energy and Combustion Science

    (1992)
  • C. Külsheimer et al.

    Combustion and Flame

    (2002)
  • P. Palies et al.

    Combustion and Flame

    (2010)
  • G. Yu et al.

    Combustion and Flame

    (1986)
  • K. Kumar et al.

    Combustion and Flame

    (2008)
  • S.A. Filatyev et al.

    Combustion and Flame

    (2005)
  • Y. Xiong et al.

    Combustion and Flame

    (2001)
  • A.P. Dowling

    Journal of Fluid Mechanics

    (1997)
  • S.R. Stow, A.P. Dowling, ASME Conference Proceedings 2004 (2004)...
  • J.G. Lee, K. Kim, D. Santavicca, Proceedings of the Combustion Institute 28 (2000)...
  • L. Zimmer, S. Tachibana, Proceedings of the Combustion Institute 31 (2007)...
  • T.J. Poinsot et al.

    Journal of Fluid Mechanics

    (1987)
  • B.D. Bellows, M.K. Bobba, A. Forte, J.M. Seitzman, T. Lieuwen, Proceedings of the Combustion Institute 31 (2007)...
  • S.K. Thumuluru, T. Lieuwen, Proceedings of the Combustion Institute 32 (2009)...
  • A.P. Dowling

    Journal of Fluid Mechanics

    (1999)
  • Cited by (0)

    View full text