Unit 6: THERMOACOUSTIC INSTABILITY - Subjective Questions

Thermoacoustic instability is the self-excited growth of pressure and velocity oscillations caused by constructive interaction between unsteady heat release and the acoustic field of a combustor.

The feedback cycle consists of:

• Acoustic pressure and velocity perturbations modify the fuel-air mixing, atomization, evaporation, and flame position.
• These changes produce fluctuations in the heat-release rate .
• The fluctuating heat release generates acoustic energy and alters the pressure perturbation .
• The resulting acoustic disturbance returns to the flame through reflection and propagation.
• If the disturbance returns with a favorable phase, it reinforces the original heat-release fluctuation.

Instability develops when acoustic energy supplied by the flame exceeds losses due to damping, radiation, viscosity, and heat transfer. Nonlinear effects eventually limit the growth and produce a finite-amplitude limit cycle.

For a small acoustic perturbation, the pressure-energy source associated with unsteady heat release is proportional to . Over one oscillation period , the net acoustic-energy contribution is

The Rayleigh criterion states:

• : heat release adds net energy to the acoustic field and promotes instability.
• : heat release removes acoustic energy and produces damping.
• : there is no net thermoacoustic energy transfer.

For harmonic fluctuations,

so that

Therefore, constructive coupling occurs when , particularly when pressure and heat release are approximately in phase. Actual instability requires this positive production to exceed all acoustic losses. Thus, Rayleigh's criterion is a necessary energy-based indicator, while the complete stability condition must also account for damping.

Thermoacoustic instabilities may be classified by acoustic mode shape and frequency:

• Longitudinal modes: Pressure varies mainly along the combustor axis. These modes are often represented adequately by one-dimensional models.
• Transverse modes: Pressure varies across the chamber width or diameter. They can cause strong local flame oscillations and require multidimensional analysis.
• Azimuthal modes: Pressure varies around an annular combustor. They may appear as standing waves or waves rotating clockwise or anticlockwise.
• Radial modes: Pressure varies in the radial direction and is important when radial chamber dimensions are acoustically significant.
• Helmholtz or bulk modes: The system behaves like a lumped acoustic resonator, with chamber compressibility acting as a spring and flow through a neck acting as an oscillating mass.

Instabilities may also be described as low-frequency rumble, intermediate-frequency oscillations, or high-frequency screech. The classification identifies the relevant spatial physics and determines whether a lumped, one-dimensional, or three-dimensional model is needed.

Acoustic disturbances can modulate heat release through several mechanisms:

• Velocity coupling: Fluctuating velocity changes the reactant mass flow and equivalence ratio at the flame.
• Fuel-supply coupling: Pressure oscillations alter fuel injection, atomization, and spray penetration.
• Equivalence-ratio coupling: Convected mixture-fraction disturbances reach the flame after a time delay and change its burning rate.
• Flame-surface modulation: Vortices wrinkle and stretch the flame, changing its surface area and total heat release.
• Flame-position oscillation: Motion of the flame relative to acoustic pressure antinodes changes the coupling strength.
• Entropy and vorticity effects: Disturbances generated in the flame may interact with downstream nozzles and produce additional sound.

The response includes a convective or chemical time delay. If this delay places in a favorable phase relative to , acoustic energy is generated according to Rayleigh's criterion.

Thermoacoustic instability can adversely affect both performance and durability:

• Large pressure oscillations increase mechanical and thermal loading on liners, injectors, seals, and turbine components.
• Oscillating flames can create local hot spots, increasing liner temperature and reducing component life.
• Enhanced heat transfer may damage cooling films and thermal-barrier coatings.
• Flame displacement can produce flashback, blow-off, or partial extinction.
• Combustion efficiency may decrease because of incomplete or periodically interrupted burning.
• Thrust and specific fuel consumption may fluctuate as combustor exit pressure and temperature become unsteady.
• Pattern factor at the turbine inlet can deteriorate, increasing turbine thermal stress.
• Pollutant emissions, particularly , CO, and unburned hydrocarbons, may rise.
• Severe oscillations can cause fatigue cracking or catastrophic hardware failure.

Consequently, stable operating limits often constrain the permissible fuel-air ratio, staging strategy, power setting, and emissions-reduction capability.

A one-dimensional calculation assumes that acoustic variables vary mainly along the axial coordinate and that each duct segment supports plane waves. Mean properties are generally uniform or piecewise uniform across each section.

A typical procedure is:

1. Divide the combustor, plenum, injector, and nozzle into one-dimensional elements.
2. Specify mean pressure, temperature, density, velocity, area, and speed of sound in each element.
3. Linearize the mass, momentum, and energy equations about the mean flow.
4. Represent pressure and velocity as downstream- and upstream-travelling waves.
5. Introduce flame coupling through a flame transfer function or an - model.
6. Impose continuity and jump conditions at area changes and across the flame.
7. Apply inlet and outlet acoustic impedances.
8. Assemble a global transfer or scattering matrix.
9. Search for complex eigenfrequencies satisfying the characteristic equation.

The real part of frequency gives oscillation frequency, while the imaginary part indicates growth or decay according to the adopted time convention.

For small perturbations in a uniform duct, the linearized equations are

with . Combining them gives

For harmonic motion proportional to ,

where . Relating the state at to that at gives

Matrices for individual elements can be multiplied to obtain the response of the complete acoustic network.

A flame transfer function relates normalized heat-release fluctuations to an imposed reference disturbance, commonly inlet velocity:

where is the gain and is the phase.

The - model approximates this response as

or, in the frequency domain,

Here:

• is the interaction index or flame-response gain.
• is the delay associated with convection, mixing, evaporation, and chemical processes.
• The factor determines the phase between forcing and heat release.

The model is simple and useful for eigenvalue calculations, but constant and may not represent nonlinear saturation, multiple time delays, or spatially distributed flame dynamics.

The acoustic impedance at a boundary is defined as

while the reflection coefficient for characteristic impedance is

Important limiting cases are:

• Pressure-release boundary: , so pressure has a node and .
• Rigid boundary: , so velocity has a node and .
• Anechoic boundary: , giving and negligible reflection.
• Partially reflecting boundary: Complex represents both phase change and energy loss.

In a transfer-matrix model, the inlet and outlet impedance relations are substituted into the global state equation. The resulting homogeneous system has a nontrivial solution only when its determinant vanishes. Accurate impedance data are essential because boundary reflection changes modal frequency, mode shape, damping, and flame-acoustic phase.

After the duct, area-change, flame, and boundary-condition matrices have been assembled, the acoustic amplitudes satisfy a homogeneous system

A nontrivial acoustic field exists when

Because flame response, time delays, and impedance are frequency dependent, is generally nonlinear and complex. Numerical root-finding or contour methods are therefore used to determine

For the convention :

• is the angular oscillation frequency.
• indicates exponential growth and instability.
• indicates decay and stability.
• represents neutral stability.

The modal frequency is , and the growth time scale is approximately . The sign interpretation must be reversed if the model uses the convention .

One-dimensional methods:

• Assume plane-wave propagation along an acoustic axis.
• Use transfer, scattering, or network matrices.
• Require relatively low computational effort.
• Are convenient for parameter sweeps and preliminary design.
• Represent longitudinal modes effectively but cannot resolve complex transverse or azimuthal structures.

Three-dimensional methods:

• Solve linearized governing equations or a Helmholtz equation over the full geometry.
• Capture longitudinal, radial, transverse, and azimuthal modes.
• Represent spatially distributed heat-release coupling and complex boundaries.
• Require detailed meshes, mean-flow fields, flame models, and greater computational effort.

The one-dimensional method is appropriate when the wavelength is large compared with transverse dimensions and the geometry is approximately duct-like. Three-dimensional analysis is necessary for annular combustors, non-axisymmetric features, transverse modes, and localized flame-acoustic interactions.

For a quiescent, uniform ideal gas, the linearized continuity, momentum, and energy relations can be combined to obtain an acoustic equation. The pressure disturbance satisfies

where is the fluctuating volumetric heat-release rate and is the ratio of specific heats.

Assuming harmonic disturbances,

produces

The left side describes the passive three-dimensional acoustic field, while the right side represents flame forcing. If depends linearly on the acoustic variables, the equation becomes a coupled, generally nonlinear eigenvalue problem. Mean flow, nonuniform temperature, entropy fluctuations, and damping require additional terms or the full linearized Euler equations.

A three-dimensional linear analysis commonly follows these steps:

1. Obtain a steady mean-flow and mean-temperature field from CFD or measurements.
2. Linearize the governing equations about this mean state.
3. Assume harmonic perturbations with complex frequency .
4. Represent flame response using a local or global flame transfer function, flame describing function, or linearized chemistry model.
5. Impose wall, inlet, outlet, injector, and nozzle impedance conditions.
6. Discretize the equations using finite elements, finite volumes, or spectral methods.
7. Assemble the coupled system

8. Solve for complex eigenfrequencies and corresponding three-dimensional mode shapes.
9. Evaluate modal growth rates and the spatial Rayleigh index.

Frequency-dependent flame and boundary models make the eigenproblem nonlinear. Iterative eigensolvers, fixed-point iteration, rational approximation, or contour-integration techniques may be employed. Grid convergence and sensitivity to uncertain flame parameters should also be assessed.

A three-dimensional flame cannot always be treated as a compact source. Its local heat-release response may be expressed as

where contains local gain and phase. A more general formulation uses a response kernel:

This representation accounts for:

• Spatial variation of flame sensitivity.
• Convective delays from the injector to different flame regions.
• Coupling to velocity, pressure, mixture fraction, or equivalence-ratio fluctuations.
• Interaction among multiple burners in annular chambers.

The heat-release model is inserted into the acoustic equations as a complex source operator. Its accuracy strongly affects predicted eigenfrequencies and growth rates, so it is normally calibrated using experiments, large-eddy simulation, or system-identification data.

Each computed eigenmode provides a complex eigenfrequency and a spatial acoustic field.

• The mode shape identifies pressure nodes, antinodes, phase variation, and whether the mode is longitudinal, transverse, radial, azimuthal, standing, or rotating.
• The growth rate indicates whether the mode amplitude increases or decreases with time.
• The local Rayleigh index is commonly evaluated as

For harmonic fields, it is proportional to

Regions with positive drive the mode, while regions with negative values damp it. Integrating over the combustor gives the net flame contribution. These results identify which burners or flame zones cause instability and guide targeted changes to damping, injector design, or flame phase.

Passive control requires no real-time sensing or actuation. Important methods include:

• Helmholtz resonators: Side cavities absorb acoustic energy near a tuned resonance frequency.
• Quarter-wave tubes: Tubes of approximately one-quarter wavelength create a response that opposes the combustor oscillation.
• Perforated liners: Flow through small holes dissipates acoustic energy through viscous losses and vortex shedding.
• Acoustic damping cavities: Properly positioned cavities reduce pressure amplitude at critical frequencies.
• Injector modification: Changes to swirler geometry, pressure drop, or mixing alter flame gain and time delay.
• Combustor geometry changes: Length, volume, and area changes shift acoustic eigenfrequencies away from strongly coupled flame frequencies.
• Fuel staging: Redistribution of fuel modifies flame position and heat-release response.

Passive devices are simple and robust but are normally effective over a limited frequency and operating range. Their design must consider temperature, mean flow, nonlinear amplitude effects, and available installation volume.

Active control detects an oscillation and generates a controlled disturbance that reduces the unstable feedback.

A typical system contains:

• A pressure transducer or optical flame sensor.
• A real-time controller that estimates amplitude and phase.
• An actuator such as a fast fuel valve, loudspeaker, plasma actuator, or secondary-air injector.
• A control algorithm that commands a disturbance with suitable gain and phase.

The actuator may create destructive acoustic interference or modify heat release so that

Advantages:

• Adaptation to changing engine conditions.
• Potential suppression of several frequencies.
• Less dependence on a large resonator volume.

Limitations:

• Sensor and actuator delay can destabilize the loop.
• Actuator bandwidth, authority, durability, and reliability are restricted.
• The controller may spill energy into other modes.
• Failure-safe operation is essential in an aeroengine.

Robust control design must account for uncertain flame dynamics and variations across the operating envelope.

Thermoacoustic instability can be suppressed by breaking the favorable relationship between acoustics and heat release:

• Changing combustor length or volume shifts acoustic eigenfrequencies.
• Modifying injector spacing or burner arrangement changes coupling to azimuthal and transverse modes.
• Increasing injector pressure drop can reduce upstream acoustic feedback and fuel-flow modulation.
• Changing swirl number alters recirculation, flame position, and convective time delay.
• Pilot-main fuel staging redistributes heat release and can reduce the net Rayleigh index.
• Small changes in equivalence ratio may move the flame away from a high-gain response region.
• Adjusting air splits changes local velocity, mixing, and flame stabilization.
• Altering power, pressure, or temperature changes both acoustic frequencies and chemical time scales.

A modification is effective if it reduces flame gain, changes the phase of relative to , increases damping, or separates an acoustic eigenfrequency from a strong flame-response band. Its impact on emissions, efficiency, relight, blow-off, and turbine temperature must also be evaluated.

A suitable experimental procedure includes:

1. Install dynamic pressure transducers at several axial and circumferential positions.
2. Measure heat-release fluctuations using chemiluminescence or another optical diagnostic when access permits.
3. Record synchronized pressure, heat-release, fuel-flow, temperature, and operating-condition data.
4. Calculate power spectral densities to identify dominant frequencies and harmonics.
5. Use cross-spectra, coherence, and phase to determine pressure-heat-release coupling.
6. Reconstruct mode shapes from the relative amplitude and phase of multiple pressure sensors.
7. Estimate growth or decay rates from transient tests or system-identification methods.
8. Construct stability maps against equivalence ratio, pressure, temperature, power, and fuel staging.

A strong narrow-band pressure peak alone does not prove thermoacoustic driving. High coherence and a positive pressure-heat-release correlation provide stronger evidence. Sensor thermal protection, calibration, sampling rate, aliasing prevention, and uncertainty analysis are essential.

An integrated workflow may be organized as follows:

1. Initial screening: Use network or one-dimensional models to identify likely longitudinal frequencies and sensitivity to boundary impedances.
2. Flame characterization: Determine flame transfer functions from forced experiments or high-fidelity simulations over the operating envelope.
3. Three-dimensional analysis: Compute longitudinal, transverse, radial, and azimuthal eigenmodes using realistic geometry and impedance conditions.
4. Driving-region identification: Evaluate mode shapes and the spatial Rayleigh index to locate strongly coupled burners and flame zones.
5. Uncertainty assessment: Vary flame gain, time delay, damping, impedance, and mean temperature to determine stability margins.
6. Control design: Select geometric changes, fuel staging, liners, resonators, or active control according to the dominant mechanism.
7. Experimental validation: Compare predicted frequencies, mode shapes, and growth trends with rig measurements.
8. Engine-level verification: Confirm durability, emissions, efficiency, operability, and control robustness throughout the flight envelope.

The process should be iterative because a control measure that stabilizes one mode may destabilize another. Final acceptance should therefore be based on multimode stability margins rather than suppression of only one measured frequency.