Suppression of thermo-acoustic instability using air injection in horizontal Rijke tube

Accepted Manuscript Suppression of thermo-acoustic instability using air injection in horizontal Rijke tube Nilaj N. Deshmukh, S.D. Sharma PII:

S1743-9671(16)30044-7

DOI:

10.1016/j.joei.2016.03.001

Reference:

JOEI 208

To appear in:

Journal of the Energy Institute

Received Date: 29 January 2016 Revised Date:

27 February 2016

Accepted Date: 1 March 2016

Please cite this article as: N.N. Deshmukh, S.D. Sharma, Suppression of thermo-acoustic instability using air injection in horizontal Rijke tube, Journal of the Energy Institute (2016), doi: 10.1016/ j.joei.2016.03.001. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Suppression of thermo-acoustic instability using air injection in horizontal Rijke tube

RI PT

Nilaj N. Deshmukh* and S. D. Sharma Department of Aerospace Engineering,

Indian Institute of Technology Bombay

Abstract

SC

Powai, Mumbai 400076, India

M AN U

The thermo-acoustic instability arising out of coupling between the pressure fluctuation and the unsteady heat release, when grows sufficiently, is known to cause serious structural damage thereby reducing the life span of systems having combustor for example, jet engines, gas turbines and industrial burners. The present work involves experimental study of thermo-acoustic instabilities occurring in a Rijke tube and their suppression by means of diverting a very small

TE D

fraction of the incoming air flow in the form of radial injection from the wall of the tube through micro jets. A horizontal quartz tube with pre-mixed burner is used as the test model of Rijke

EP

tube. The coefficient Rayleigh Index (RI) is estimated from simultaneous measurement of chemiluminescence using a PMT and pressure in the plane of burner head. Experiments were

AC C

carried out at the equivalence ratio of Φ = 1 for a range of the burner position inside the tube, velocity of micro jets, and total mass flow rate through the tube. The control technique of micro jets was found to completely suppress the thermo-acoustics in a range of jet velocity depending on the burner position and the total air mass flow rate. For complete suppression of the thermos*Corresponding author Department of Aerospace Engineering, Indian Institute of Technology Bombay, Powai, Mumbai, 400076, India. Mobile No.: +91-9867167754, Fax No.: (+91-22) 2572 602. E-mail address: [email protected]

ACCEPTED MANUSCRIPT

acoustics, the phase difference between the pressure and chemiluminescence waves was found to significantly increase and the RI was found to reduce to nearly zero. In such case, the flame luminescence was observed to significantly reduce. The proposed technique of air injection into

RI PT

the flame through radial micro jets is simple yet very effective in controlling the thermosacoustic instability.

ratio.

M AN U

Nomenclature

SC

Keywords: Thermo-acoustics, combustion instability, Rijke tube, premixed flame, equivalence

Axial position of air injection

f

frequency

L

Length of tube

M

Total air mass flow rate

Mi x

EP

Heat input

AC C

φ

Inlet air mass flow rate

Axial position of burner

q Φ

TE D

a

Equivalence ratio Phase difference

Vj

Velocity of micro jet

LPG

Liquefied Petroleum Gas

RI

Rayleigh Index

SPL

Sound Pressure Level

2

ACCEPTED MANUSCRIPT

1. Introduction Combustion instabilities are observed in gas turbine combustors, ramjets, rocket engines, industrial burners and other combustion systems [1–3]. When pressure fluctuations are coupled

RI PT

with unsteady heat release from a combustion process, acoustic oscillations are known to grow substantially to large amplitudes. The sign of the heat energy balance in combustion chamber can either encourage or dampen the acoustic oscillations. For example, when the heat supplied

SC

exceeds the heat loss, it amplifies the thermo-acoustic instability [4]. The growth of these oscillations beyond a certain limit can lead to a serious structural damage and loss of control

M AN U

system hardware [5,6]. Therefore, control of the combustion instabilities in the combustion systems assumes significance. On laboratory-scale in the most simplistic way, thermo-acoustic instabilities can be studied using Rijke tube, which is a vertical open-ended tube with a heat source placed at the lower quarter inside the tube [7]. This heat source can be a wire gauge which

TE D

is heated either electrically [8,9] or by a flame [10,11]. Various numerical and experimental investigations on thermo-acoustics using Rijke tube have been reported in the literature [4,12– 17]. Carvalho et al. [7] and Matveev and Culick [8] have discussed how the position of the heat

EP

source in the Rijke tube dictates the mode of the thermo-acoustic instability and derived the exact locations, on the basis of energy input, where heat source can produce the fundamental and

AC C

different higher modes of the thermo-acoustic instability. Swaminathan et al. [18] carried out a detailed analysis of RANS results and concluded that the noise level is low from turbulent premixed flames having an extensive and uniform spatial distribution of heat release rate.

Active and passive controls are two strategies adopted to control instabilities in combustion systems. A certain change in the design of a combustion system associated with passive control involves modifications to the combustor geometry [5,19], whereas in active control, the system 3

ACCEPTED MANUSCRIPT

dynamics is modified by means of actuation using external energy[20]. Passive techniques are optimized to suit some specific operating conditions [21] and hence the applied concept shows a performance improvement only for a set of specific design conditions [22]. However, the lack of

RI PT

proper understanding of the mechanism of combustion instabilities, dynamic response and occurrence of multiple modes at different operating conditions pose some of the major challenges in the design of passive control techniques [19,23,24]. Various passive methods used

SC

to improve the stability of low-emission combustors in stationary power gas turbines have been reviewed by Richards et al. [24]. Relocation of the fuel injection ports was found to be an

M AN U

effective method to attenuate the pressure oscillations [25]. Numerical experiments suggested that passive devices such as baffles, Helmholtz resonators and quarter-waves in an axisymmetric combustion chamber could control the combustion instability with varying effectiveness, depending on the operating condition [26]. The perforated liners, used in a model

TE D

combustor test with various equivalence ratios, were also found effective in damping the instability [27]. In active control techniques, combustion instabilities are controlled by means of external source of energy for actuation either in an open-loop or close-loop (feedback) system.

EP

The open-loop active device is pre-calibrated for application whereas actuation of the close-loop active device is self-adjusting depending on feedback from the operating conditions. Several

AC C

active control strategies have used techniques of phase-shifting [28–30], fuel flow modulation [30–32] and high momentum air flow modulation [33,34] to decouple the pressure and heat release cycles.

The present study focuses on an open-loop control technique for suppression of combustion instabilities using radial air injection for different operating conditions. A small fraction of the inlet air mass flow is diverted into the radial injection for control purpose. The technique is 4

ACCEPTED MANUSCRIPT

simple yet effective and hence it avoids the more complex route of controller design and requirement of expensive sensors for controlling the thermo-acoustic instabilities. A set of 8 micro air jets are radially introduced from the wall of a horizontal Rijke tube just downstream of

RI PT

a pre-mixed gas burner whose location inside the tube could be changed by means of a traverse. The study was carried out for different burner positions from the inlet of tube for equivalence ratio of Φ = 1. Suppression of thermo-acoustic instabilities with varying degree was achieved

SC

depending on the burner position relative to the plane of radial micro air jets, velocity of jets and the total mass flow rate through the tube. The designed control system is more reliable in any

M AN U

working environment as it avoids complex control system requirements. The organization of the present paper is as follows. The details of the experimental setup, measurement systems and experimental approach are explained in sections 2 and 3. In section 4, experimental results and

2. Experimental Setup

TE D

their discussion are presented followed by a concluding summary given in section 5.

A Rijke tube constructed out of toughened quartz glass for clear visualization of flame is used in

EP

the present investigation. A tube with an internal diameter of 75 mm and length of 750 mm is placed horizontally for the ease of experiments as depicted in the schematic diagram of the test

AC C

setup shown in Fig. 1. The air mass flow through the system is maintained with the help of a centrifugal suction pump, driven by a variable speed 375 W electric motor, installed at the exit end of the system. A plenum chamber having the length of about 1000 mm and diameter of 400 mm is placed between the Rijke tube and the suction pump to minimize the acoustical interaction between them. Inside the plenum chamber, a coarse wire mesh and a pair of baffles, all made of aluminium, are fitted as flow correcting devices to minimise the non-uniformities and the

5

ACCEPTED MANUSCRIPT

acoustical interactions. A 31 mm diameter cylindrical brass burner with rounded ceremic head having 61 holes is fitted at the end of a stainless steel tube which is held horizontally in a adjustable moving post of a traverse which can position the burner coaxially inside the Rijke

RI PT

tube at any desired longitudinal location. The supply of gaseous fuel and air to the burner is made through a pre-mixing chamber which works on vortex flow mixer principle. The fuel used in the present experiments is commercially available Liquid Petroleum Gas (LPG) which a

SC

mixture of 41.1% butane, 57.3% propane and 1.6% odorant. The mass flow rate of LPG and the air is controlled and measured using thermal mass flow controllers of Aalborg make, Models

M AN U

GFC17 and GFC37, respectively, having accuracy of ±1.5% of the full scale. The total mass flow rate through the Rijke tube (air from the atmosphere, the burner product and the radial injection when used for control) is metered by Aalborg thermal mass flow meter, Model GFM57, located between the suction pump and the plenum chamber. The burner flame luminance is

TE D

measured by a photomultiplier tube (PMT, DANTEC 9057X0081) powered by BSA (Burst Spectrum Analysis), DANTEC 57N10 to estimate the combustion heat release oscillations. PMT are sensitive from visible to infrared range (300 to 1200 nm). In order to minimize the signal noise

EP

due to ambient light, the open end of the PMT was fitted with a long tube whose other end was bevelled and attached to the Rijke tube just downstream of the burner so as to focus on the entire

AC C

combustion zone. The wall pressure in the plane of the burner head was registered through a stainless steel hypodermic tube of 3 mm diameter with its right angled tip touching the Rijke tube wall and the other end connected to an ultra-low differential pressure transducer (RS 395257).

6

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 1. Schematic diagram of experimental setup.

Figure 2 depicts details of radial air injection from the Rijke tube wall as the control device and arrangement for PMT. A 25 mm wide aluminum connector ring having internal diameter equal

TE D

to that of the quartz tube and 8 radial injection ports with 1 mm diameter, drilled at 45º circumferentially apart, is used as an intermediate part of the quartz tube. Figure 2(a) shows details of arrangement for radial injection with manifold and pre-calibrated rotameter. Teflon

EP

tubes of 4 mm diameter are used to connect the eight ports with manifold through which constant

AC C

pressure air supply is maintained. The Teflon tubes are used to avoid damage due to excessive heating which was observed for PVC tubes at the time of preliminary studies. Figure 2(b) shows position of radial air injection and the relative position of burner slightly upstream. In the present setup, the aluminum ring is placed such that the injection ports are located at a distance of 150 mm from the inlet of the tube, a/L = 0.2 (the choice of this position is explained later). Figure 2(c) shows a photograph of the assembly of the air injection manifold, the PMT and the Rijke tube. In addition to the wall pressures inside the Rijke tube, acoustic pressures are also measured 7

ACCEPTED MANUSCRIPT

using microphone outside the Rijke tube. As schematically shown in Fig. 1, a Bruel & Kjaer condenser type microphone (model 4939 having dynamic pressure range of 28 dB to 164 dB with sensitivity of 4 mV/Pa and a flat frequency range up to 100 kHz) is placed at a distance of

RI PT

600 mm looking into the Rijke tube from an oblique angle of 30º for measurement of the Sound Pressure Levels (SPL) generated by the thermo-acoustic instability. The National Instruments data acquisition system, USB-6212, was used to acquire the acoustic data and the National

SC

Instruments 4 channel data acquisition system, PCI-4462, was used to simultaneously acquire data of unsteady heat release rate and pressure fluctuations. The Lab View 7.1 was used as

AC C

EP

TE D

M AN U

interface between all the sensing devices and the computer.

(a)

8

TE D

M AN U

SC

(b)

RI PT

ACCEPTED MANUSCRIPT

AC C

3. Methodology

EP

(c) Fig. 3. Arrangement for control technique. (a) Manifold for radial air injection, (b) Position of burner and air injection ports and (c) Photograph showing test setup and PMT with high voltage supply.

Some preliminary experiments were carried out with different burner positions inside the Rijke tube, different fuel and air flow rates to explore the boundary of thermo-acoustic instability. The flame was visibly rich and the thermo-acoustics ceased to occur for the equivalence ratio exceeding 1.0, whereas it was difficult for the lean flame with thermo-acoustics to sustain when the equivalence ratio was reduced below 0.6. From the preliminary experiments, three burner positions of x/L = 0.16, 0.173 and 0.186 (leaving a gap of 30, 20 and 10 mm, respectively from 9

ACCEPTED MANUSCRIPT

the plane of radial air injection) and equivalence ratio of Ф = 1 are chosen as the basic test conditions for investigation of the effect of control technique with various radial air injection flow rates on suppression of the thermo-acoustic instability at 7 different total flow rates ranging

RI PT

from 0.4 to1.6 g/s through the Rijke tube. For a fixed total air mass flow rate through the Rijke tube, any variation in the radial air injection flow rate (for the purpose of control) is coupled with a proportionate change in the air inlet mass flow rate from atmosphere. The microphone was

SC

calibrated on regular basis before the acoustic measurements using a B & K Multifunction Acoustic Calibrator, Type 4226, at a frequency of 1 kHz for sound levels of 94 dB, 104 dB and

M AN U

114 dB. The acoustic data were acquired in blocks of 8192 samples at a speed of 200 kHz and the average of 100 such blocks were obtained for a single data point. The required supply of 8 volt DC to the pressure transducer and high voltage of 2040 volt DC to the PMT was provided from the external power source and the BSA, respectively. The data from PMT and pressure

TE D

transducer were simultaneously acquired at the sampling rate of 100 kHz and for each set a total of 50,000 samples were stored.

EP

To ensure the uniformity of jets from all the injection ports before fitting the ring in the Rijke tube, the jet velocity was measured using a miniature Pitot tube (constructed out of stainless steel

AC C

hypodermic tube of 1 mm diameter) at the exit of each port. Figure 3 (a) shows the velocity of jet from each port obtained for maximum mass flow rate of 0.053 g/s through manifold. The average velocity of 7.07 m/s is represented by a dotted line. The RMS of the jet velocity variation is calculated to be 0.072 m/s which is about 1.02% of the average velocity indicating that the flow uniformity of the jets from all the ports is satisfactory. Effect of close proximity of the ports flanking the supply tube to the manifold is reflected by a slight increase in the jet velocity. The trajectory and growth pattern of a jet in cross flow is known to depend on the ratio 10

ACCEPTED MANUSCRIPT

of their densities and velocities [35]. In the present study, such mass flow ratio is defined as the mean radial air injection mass flow normalized by the total air mass flow in the system (which is the total flow rate registered by mass flow meter located at 12 in Fig. 1 minus the constant fuel

RI PT

flow rate). Figure 3(b) shows variation of the mass flow ratio obtained for different total air mass flow rates, M.

SC

7.5

M AN U

Jet velocity (m/s)

8

7

6.5

6

3

4

Port (a)

5

6

7

8

AC C

EP

10 % Radial Mass Injection

2

TE D

1

1

Vj = 1.01 m/s Vj = 5.08 m/s

Vj = 3.04 m/s Vj = 7.07 m/s

0.1 0.2

0.4

0.6 0.8 1 1.2 1.4 Total air mass flow rate (g/s) (b)

1.6

1.8

11

ACCEPTED MANUSCRIPT

Fig. 3. (a) Variation in velocity of jets for each injection port and (b) Radial air jet mass to total

RI PT

air mass for different total air mass flow rates, M.

In the present experiments, total mass flow rate through the Rijke tube is varied from 0.4 to 1.6 g/s and that through the manifold for the control air is varied from 7.2 × 10-3 to 51.96 × 10-3 g/s

SC

resulting in radial micro jets velocity ranging from 1.01 to 7.07 m/s. Further increase in the micro jet velocity was observed to blow off the flame when the burner was in the plane of

M AN U

injection ports. The air mass flow and the fuel mass flow supply rates into the premixed burner are set at 0.12 and 0.00767 g/s, respectively to supply the heat with a constant power of 355 W at a mixture equivalence ratio of Φ = 1.

TE D

4. Result and Discussion

It is seen from the earlier studies on the Rijke tube [8] that combination of electrical heating power and air mass flow rate determines the boundaries of thermo-acoustic instability which is

EP

altered in size along with change in mode when the location of the heat source is changed. In the present investigation, instead of electrical heater, a gas burner with flame is used and hence

AC C

various parameters, such as, the heat supplied in terms of quantity of fuel and equivalence ratio, air mass flow rate into the tube and the burner position are expected to influence the instability boundaries and also trigger different modes of thermo-acoustic instability. Figure 4 demonstrates occurrence of three different modes of thermo-acoustic instability when the burner with mixture equivalence ratio of Φ = 1 is positioned at x/L = 0.16, and the heat supply (fuel quantity) is varied keeping the total mass flow rate constant at 1.4 g/s. The Sound Pressure Levels over the

12

ACCEPTED MANUSCRIPT

entire frequency range of spectra are much above the background noise acquired for two different conditions – ambient noise without any flow through the Rijke tube (suction pump being switched off) and with the full total air mass flow rate of 1.6 g/s at maximum speed of the

RI PT

suction pump in the presence of the burner inside the tube but without flame. The first mode is seen to occur at the fundamental frequency of 268 Hz for the heat supply of 2318 W. However, the subsequent second and third modes are seen for the reduced heat supplies of 887 W and 355

SC

W at the frequencies of 512 Hz and 780 Hz, respectively. Since all the three modes have been obtained for the same total air mass flow rate of 1.4 g/s, the inlet air mass flow rates vary as 0.99

M AN U

g/s for the mode 1, 1.09 g/s for the mode 2 and 1.27 g/s for the mode 3, respectively. The peak SPL clearly appears to depend on the heat supply. It is worth noting that despite a significant reduction in the heat supply from 2318 W to 355 W, the peak SPL for the third mode along with two harmonics is appreciably high with amplitude of about 88 dB. The thermo-acoustic

TE D

instability with the third mode continued to persist from burner is at x/L = 0.06 until the burner was pushed to x/L = 0.24 as shown in Fig. 5 for different inlet air mass flow rates. The figure also shows positions of burner and radial air injection ports used in the present investigation

AC C

EP

where all the experiments have been carried out with conditions pertaining to the third mode.

13

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

(a)

(b)

14

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

(c)

(d) Fig. 4. Occurrence of three modes of thermo-acoustic instability at different heat inputs with the equivalence ratio of Φ = 1: (a) Mode 1 at 2318 W, (b) Mode 2 at 887 W, (c) Mode 3 at 355 W frequency spectra and (d) Heat input versus inlet air mass flow rate for the burner position at x/L = 0.16.

15

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 5. Mode 3 acoustic instability zone for different inlet air mass flow rates, Mi, different color lines indicates the burner locations and plane of radial air injection. The frequency spectra for three different burner positions are shown in Fig. 6. There appears no noticeable change in the overall pattern of the spectra for different positions of the burner - the

TE D

third mode instability peak occurs at 780 Hz with a SPL amplitude of 88 dB along with the harmonics at 1562 Hz and 2343 Hz with SPL amplitudes of 73 dB and 66 dB, respectively. However, for the burner position at x/L=0.186 the harmonics appear to become weaker with

AC C

EP

reduction in SPL by about 5 dB.

16

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 6. The frequency spectra for three different burner positions at total air mass flow rate M = 1.4 g/s. The thermo-acoustic instability is known to grow in accordance with Rayleigh Index which is

TE D

indicative of strong coupling between unsteady heat release and pressure fluctuation, when both are in phase. For a simple one-dimensional model of Rijke tube, the Rayleigh Index is

EP

represented by the following inequality [7,36–38],

1 T

T

∫ p '( x, t ) q '( x, t )dt > 0.

(1)

0

AC C

RI ( x , t ) =

Here, RI (x, t) is the local Rayleigh Index determined per cycle, T is the period of oscillation, p’(x, t) is local pressure fluctuation at position x, and q’(x, t) is local unsteady heat release at position x.

In the combustion process the chemiluminescence is indicative of the rate of energy released from chemical reactions and hence it is considered to be a measure of heat release from the 17

ACCEPTED MANUSCRIPT

combustion [37, 40 – 44]. In the present investigation, temporal variation of the chemiluminescence and the pressure oscillation are simultaneously measured at the plane of burner head by the PMT and the pressure transducer, respectively for determining the local

RI PT

Rayleigh Index using Eq. (1). The dynamic response of the PMT used in the present experiments is extremely high compared to the frequency of thermo-acoustic instability; therefore, the PMT can accurately capture the temporal variations of chemiluminescence of the burner flame. It is to

SC

be pointed out here that since the absolute measure of unsteady heat release is not known from the chemiluminescence, the estimate of the Rayleigh Index is only arbitrary (outputs from the

M AN U

pressure transducer and the PMT are both in volts). Nonetheless, the measured quantity serves the definite purpose of relating the variation in RI with either the suppression or growth of the thermo-acoustic instability.

Figure 7 shows plots of frequency spectra of signals of wall pressures and chemiluminescence

TE D

simultaneously acquired from the pressure transducer and the PMT for burner position at x/L = 1.73 and total air mass flow rate of M = 1.4 g/s. The two data seem to correlate exceptionally

EP

well including the dominant peak frequency of mode 3 and its harmonics. Since coupling between the unsteady heat release and the pressure fluctuation is expected to occur at the peak

AC C

frequency, the time series of these data are further processed using a band-pass filter of 650-850 Hz and shown in Fig. 8. Both the signals, representing different quantities, are obtained in volts and vary in amplitudes as shown for 20 ms time length in Fig. 8(a). The PMT signal is believed to be representative of the unsteady heat release but no calibration exists to quantify q’(x,t). Therefore, to circumvent this anomaly we introduce a coefficient of Rayleigh Index which is non-dimensional and can be used to compare results from different sets of experiments. Figure 8(b) is the plot of a shorter time sequence (4.5 ms) extracted from Fig. 8(a) wherein the data has 18

ACCEPTED MANUSCRIPT

been normalized by the peak amplitude of the respective signals. Determination of the Rayleigh Index from such a plot will be non-dimensional (so it is referred as coefficient of RI) and its maximum value will be 0.5 for the phase lag of zero degree between the two signals. Thus, the

RI PT

coefficient of RI will now depend only on the phase lag and will progressively reduce with increasing phase lag, eventually becoming zero for the phase lag of 90° and negative beyond 90°. In the present case, the phase lag is estimated to be about 5.58° which results in the coefficient of

AC C

EP

TE D

M AN U

SC

RI about 0.484.

Fig. 7. Frequency spectra of simultaneously acquired signals from Pressure Transducer and PMT for burner position at x/L = 0.173 and total air mass flow rate, M = 1.4 g/s.

19

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

EP

TE D

(a)

AC C

(b) Fig. 8. Simultaneous pressure transducer and PMT signals corresponding to Fig. 7 with burner position at x/L = 0.173, (a) Raw data and (b) Normalized data.

The control technique of the air injection through radial micro jets into the flame was employed over a wide range of the total air mass flow rate through the Rijke tube in the present investigation to assess its efficacy in suppressing the thermo-acoustic instability. For all the three burner positions, a complete suppression of the thermo-acoustic instabilities could be obtained 20

ACCEPTED MANUSCRIPT

with varying mass flow ratio (Radial air jet mass to total air mass flow rate). Figure 9 is a depiction of the efficacy of the control technique at a typical total air mass flow rate of M = 1.4 g/s. The effect of jet velocity and the relative position of the burner are clearly displayed – the

RI PT

control technique becomes apparently more effective, even with reduced mass flow ratio, for narrowing gap between the burner and the plane of the injection ports. It is interesting to note that for complete suppression of the thermo-acoustics, not only the spectral peaks disappear

SC

thereby truncating the SPL by a substantial margin of about 30 dB, but the spectra also show commendable match with the background noise spectrum despite the burner and the control jets

M AN U

being the additional sources of noise. For a stabilizing combustion system, the Rayleigh Index is expected to reduce and tend to zero for complete suppression of the thermos-acoustics. Figure 10 shows pressure fluctuation and chemiluminescence fluctuation (heat release) plots obtained when the control technique is employed for the burner position at x/L = 0.173 and the total air mass

TE D

flow rate of M = 1.4 g/s with the mass flow ratio of 2.63% (Vj = 5.08 m/s) resulting in a complete suppression of the thermos-acoustic instability as indicated by the frequency spectrum in Fig. 9(b). The normalization of data is carried out using respective peak values obtained for

EP

existence of instability (Fig. 8(a)). Attention is drawn at the scale on y-axis which is magnified by 75 times as both the curves were seen to lie on the x-axis when the original scale of Fig. 8(b)

AC C

was used. The phase difference is found to significantly increase to φ = 56.25° and the coefficient of RI is almost non-existent (3.8×10-4).

21

(b)

AC C

EP

TE D

M AN U

(a)

SC

RI PT

ACCEPTED MANUSCRIPT

(c) 22

ACCEPTED MANUSCRIPT

M AN U

SC

RI PT

Fig. 9. Suppression of thermo-acoustics with control device at total air mass flow rate, M = 1.4 g/s for three different burner positions: (a) x/L = 0.16, (b) x/L = 0.173 and (c) x/L = 0.186.

AC C

EP

TE D

(a)

(b) Fig. 10. Simultaneous pressure transducer and PMT signals for burner position at x/L = 0.173 and total air mass flow rate, M = 1.4 g/s when the control is switched on with 2.63% of radial mass of air (a) Raw data and (b) Normalized data. The peak SPL data obtained at various total air mass flow rates from spectra similar to those shown in Fig. 9 are compiled and shown in Fig. 11. The SPL value for the case of complete suppression of the thermos-acoustics is determined from the spectrum at the frequency

23

ACCEPTED MANUSCRIPT

corresponding to the peak without any control. It is seen that the jet velocity plays key role in attenuating the peak SPL at higher total air flow rate through the Rijke tube. Since a very small fraction of the incoming air flow is diverted in the form of the radial injection from the wall of

RI PT

the tube, no extra air is need for complete suppression of thermo-acoustic instability. A modulated axial air jet was used as an open loop control technique by Uhm and Acharya [34] for suppression of thermo-acoustic instability. They used fixed burner position and control air jet

SC

with less than 1% of air of total air supplied (velocity ratio of 2) and obtained reductions in pressure oscillation by 20 dB. In the present study radial air jets completely suppresses the

M AN U

pressure oscillations without any extra air. It is worth noting that the burner position relative to the plane of injection ports is crucial. As the burner approaches closer to the plane of injection ports (a/L = 0.2), even the lower velocity jets become effective in suppressing the thermoacoustic instabilities over a wider range of the total air mass flow rate through the tube as can be

TE D

seen in Fig.11(c). The dynamics of the interaction of micro jets with the flame at higher mass flow ratio seems to be effective in decoupling the fluctuating pressure and the unsteady heat release as substantiated by the increase in the phase difference seen in Fig. 10 thereby resulting

EP

in cessation of thermo-acoustics. In the present investigation, the range of effective mass flow ratio is found to vary depending on the burner position relative to the radial plane of micro jets.

AC C

For example, referring to Fig. 3(b), the range of percentage of mass flow required over the total air mass flow rate for complete suppression in Fig. 9 for x/L = 0.16 is from 3.22% to 12.78%; for x/L = 0.173 it is from 2.3% to 12.78% and for x/L = 0.186 it is from 1.38% to 12.78%.

24

ACCEPTED MANUSCRIPT

100

Peak SPL (dB)

90

Control off Vj = 1.01 m/s Vj = 3.04 m/s Vj = 5.08 m/s Vj = 7.07 m/s

60 50 0.4

0.6

(a) (a)

0.8 1 1.2 Total air mass flow rate (g/s)

Peak SPL (dB)

90 80 Control off Vj = 1.01 m/s Vj = 3.04 m/s Vj = 5.08 m/s Vj = 7.07 m/s

50 0.4

0.6

90 80

AC C

Peak SPL (dB)

0.8 1 1.2 Total air mass flow rate (g/s)

1.4

1.6

EP

100

(b)

TE D

60

1.6

M AN U

100

70

1.4

SC

70

RI PT

80

Control off Vj = 1.01 m/s Vj = 3.04 m/s Vj = 5.08 m/s Vj = 7.07 m/s

70 60

(c)

50 0.4

0.6

0.8 1 1.2 Total air mass flow rate (g/s)

1.4

1.6

Fig. 11. Peak SPL variation for different inlet flow rates and jet velocity for three different burner positions: (a) x/L = 0.16, (b) x/L = 0.173 and (c) x/L = 0.186. 25

ACCEPTED MANUSCRIPT

Figure 12 shows photographs of the flame front obtained for burner position atx/L = 0.173 and

RI PT

the total air mass flow rate of M = 1.4 g/s. Figure 12(a) exhibits the flame front when no control is employed and the combustion instabilities result in loud noise of about 90 dB and relatively more luminous flame. The flame in the present case is essentially a cluster of small bright flames

SC

produced by each hole of the burner whose confluence appears to be elongated. For the complete suppression of the thermos-acoustics with the onset of control having the mass flow ratio of

M AN U

2.6%, both brightness and the length of the flame are apparently reduced as can be seen in Fig. 12(b) which visually complements the chemiluminescence results from the PMT shown in Fig. 12. As radial air jets penetrates into the flame it changes the local equivalence ratio and hence the heat release dynamics. The radial air jet changes the phase angle between the two waves which help to suppress the thermo-acoustic instability. A luminous flame results in higher radiation heat

TE D

loss, whereas in a non-luminous flame combustion is spread out uniformly resulting in a lower heat loss [45]. Also, the luminous flame contains higher carbon particles than the non-luminous

EP

flame. Thus, the present control technique promises not only a better and quiet combustion with

AC C

less heat loss but also a lower exhaust emission.

26

RI PT

ACCEPTED MANUSCRIPT

SC

(a) (b) Fig. 12. Photographs of flame front for burner position at x/L= 0.173. (a) Without control of thermo-acoustic instabilities and (b) with complete suppression of the thermo-acoustic instabilities at total air mass flow rate, M = 1.4

M AN U

5. Conclusions

The present investigation has experimentally demonstrated occurrence of the first three modes of thermo-acoustic instabilities in a Rijke tube for a pre-mixed burner position at x/L = 0.16 and the equivalence ratio of Φ = 1 but with different heat inputs. Higher heat input is required for lower

TE D

mode which sustains for a wide range of inlet mass flow rate through the tube. The thermoacoustic instabilities are shown to be entirely suppressed by using a simple technique of air injection into the flame through micro jets issued radially from the tube wall. The effectiveness

EP

of the control technique is found to increase with the burner approaching the radial plane of the micro jets. The micro jets alters the phase angle between pressure fluctuation and unsteady heat

AC C

releases as well as their amplitudes, which help to suppress the thermo-acoustic instabilities. The present investigation proposes a coefficient of the Rayleigh Index as more appropriate indicator of coupling between the pressure fluctuation and unsteady heat release as it can circumvent the anomaly involved in measurement of absolute value of the Rayleigh Index. The chemiluminescence of the flame was recorded with considerable reduction for complete

27

ACCEPTED MANUSCRIPT

suppression of the thermos-acoustic instability and the flame was seen to be shrunk in size with less glow indicating decrease in radiation heat loss.

RI PT

References

B. Kosztin, M. Heckl, R. Müller, J. Hermann, Thermo-acoustic properties of a burner with axial temperature gradient: Theory and experiment, Int. J. Spray Combust. Dyn. 5 (2013) 67–84.

[2]

T. Beke, Modelling of thermoacoustic phenomena in an electrically heated Rijke tube, Eur. J. Phys. 31 (2010) 1331–1344.

[3]

J. K. T. Feldman, Review of the Liturature on Rijke Thermoacoustic Phenomena, J. Sound Vib. 7 (1968) 83–89.

[4]

K.I. Matveev, F.E.C. Culick, Limit-Cycle Properties of a Rijke Tube, Electron. J. Tech. Acoust. (2003) 1–13.

[5]

K.R. McManus, T. Poinsot, S.M. Candel, A review of active control of combustion instabilities, Prog. Energy Combust. Sci. 19 (1993) 1–29.

[6]

D. Zhao, X.Y. Li, A review of acoustic dampers applied to combustion chambers in aerospace industry, Prog. Aerosp. Sci. 74 (2015) 114–130.

[7]

J.A. Carvalho, M.A. Ferreira, C. Bressan, J.L.G. Ferreira, Definition of heater location to drive maximum amplitude acoustic oscillations in a Rijke tube, Combust. Flame. 76 (1989) 17–27.

[8]

K.I. Matveev, F.E.C. Culick, A study of the transition to instability in a Rijke tube with axial temperature gradient, J. Sound Vib. 264 (2003) 689–706.

[9]

W.S. Song, S. Lee, D.S. Shin, Y. Na, Thermo-acoustic instability in the horizontal rijke tube, J. Mech. Sci. Technol. 20 (2006) 905–913.

[11]

M AN U

TE D

EP

AC C

[10]

SC

[1]

J. Majdalani, B. Entezam, W. Van Moorhem, The Rijke tube revisited via laboratory and numerical experiments, in: 35th AIAA Thermophys. Conf., American Institute of Aeronautics and Astronautics, Reston, Virigina, 2001.

P. Chatterjee, U. Vandsburger, W.R. Saunders, V.K. Khanna, W.T. Baumann, On the spectral characteristics of a self-excited Rijke tube combustor—numerical simulation and experimental measurements, J. Sound Vib. 283 (2005) 573–588.

28

ACCEPTED MANUSCRIPT

M.P. Juniper, Triggering in the horizontal Rijke tube: non-normality, transient growth and bypass transition, J. Fluid Mech. 667 (2010) 272–308.

[13]

K.I. Matveev, Energy consideration of the nonlinear effects in a Rijke tube, J. Fluids Struct. 18 (2003) 783–794.

[14]

T. Beke, Heat transfer in a thermoacoustic process, Eur. J. Phys. 33 (2012) 1487–1503.

[15]

J.P. Moeck, A two-way coupling for modeling thermoacoustic instabilities in a flat flame Rijke tube, Elsevier. 32 (2009) (2009) 1199–1207.

RI PT

[12]

SC

[16] D. Zhao, Transient growth of flow disturbances in triggering a Rijke tube combustion instability, Combust. Flame. 159 (2012) 2126–2137. K. Balasubramanian, R.I. Sujith, Non-normality and nonlinearity in combustion–acoustic interaction in diffusion flames, J. Fluid Mech. 594 (2008) 29–57.

[18]

N. Swaminathan, G. Xu, a. P. Dowling, R. Balachandran, Heat release rate correlation and combustion noise in premixed flames, 2011.

M AN U

[17]

[19] S. Candel, Combustion dynamics and control: Progress and challenges, Proc. Combust. Inst. 29 (2002) 1–28. doi:10.1016/S1540-7489(02)80007-4. S.J. Illingworth, A.S. Morgans, C.W. Rowley, Feedback control of flow resonances using balanced reduced-order models, J. Sound Vib. 330 (2011) 1567–1581.

[21]

B. Zinn, Y. Neumeier, An overview of active control of combustion instabilities, in: 35th Aerosp. Sci. Meet. Exhib., American Institute of Aeronautics and Astronautics, Reston, Virigina, 1997.

[22]

R. Kulkarni, K. Balasubramanian, R.I. Sujith, Non-normality and its consequences in active control of thermoacoustic instabilities, J. Fluid Mech. 670 (2011) 130–149.

[23]

B. Eisenhower, G. Hagen, I. Mezi, Passive Control of Limit Cycle Oscillations in a Thermoacoustic System using Asymmetry, (2006).

AC C

EP

TE D

[20]

[24] G.A. Richards, D.L. Straub, E.H. Robey, Passive Control of Combustion Dynamics in Stationary Gas Turbines, J. Propuls. Power. 19 (2003) 795–810. [25]

[26]

R.C. Steele, L.H. Cowell, S.M. Cannon, C.E. Smith, Passive Control of Combustion Instability in Lean Premixed Combustors, J. Eng. Gas Turbines Power. 122 (2000) 412. D. Frinief, Origin and Control of Thermoacoustic Instabilities in Lean Premixed Gas Turbine Combustion, Swiss Federal Institute of Technology, Zurich, 2005.

29

ACCEPTED MANUSCRIPT

L. Lei, G. Zhihui, Z. Chengyu, S. Xiaofeng, A Passive Method to Control Combustion Instabilities with Perforated Liner, Chinese J. Aeronaut. 23 (2010) 623–630.

[28]

M. a. Heckl, Active control of the noise from a rijke tube, J. Sound Vib. 124 (1988) 117– 133.

[29]

K.B. Ariyur, A. Banaszuk, M. Krstic, Identification of averaged dynamics of a controlled combustion instability, Proc. 39th IEEE Conf. Decis. Control (Cat. No.00CH37187). 3 (2000) 2017–2022.

[30]

D.U. Campos-Delgado, K. Zhou, D. Allgood, S. Acharya, Active control of combustion instabilities using model-based controllers, Combust. Sci. Technol. 175 (2003) 27–53.

[31]

D. Guyot, M. Rößler, M.R. Bothien, C.O. Paschereit, Cairns • Australia 9-12 July, 2007, (2007).

[32]

F. Hathout, J.P; Annaswamy, A. M.;Ghoniem A., Modeling and Control of Combustion Instability Using Fuel Injection, in: Act. Control Technol. Enhanc. Perform. Oper. Capab. Qf Militan’ Aircraft, L. Veh. Sea Veh., 2000: pp. 23/1–23/12.

[33]

J.H. Uhm, S. Acharya, Control of combustion instability with a high-momentum air-jet, Combust. Flame. 139 (2004) 106–125.

[34]

J. Uhm, S. Acharya, Role of low-bandwidth open-loop control of combustion instability using a high-momentum air jet—mechanistic details, Combust. Flame. 147 (2006) 22–31.

[35]

S. Muppidi, K. Mahesh, Study of trajectories of jets in crossflow using direct numerical simulations, J. Fluid Mech. 530 (2005) 81–100.

[36]

M.A. Ferreira, J.A. Carvalho Jr., A Simple Derivation of the Rayleigh Criterion in Integral Form, J. Sound Vib. 203 (1997) 889–893.

[37]

K.M. Kopp-Vaughan, S.G. Tuttle, M.W. Renfro, G.B. King, Heat release and flame structure measurements of self-excited acoustically-driven premixed methane flames, Combust. Flame. 156 (2009) 1971–1982.

[39]

SC

M AN U

TE D

EP

AC C

[38]

RI PT

[27]

C.A. Martins, Pedro Tira Lacava, Marco Aurélio Ferreira, J.A. dCarvalho Jr., Energy Balance in a RijkeTube Pulsating Combustor, in: 17th Int. Conf. Mech. Eng., São Paulo, Brazil, 2003. T. Lieuwen, V. Yang, Combustion Instabilities in Gas Turbine Engines: Operational Experience, Fundamental Mechanisms, and Modeling, American Institute of Aeronautics and Astronautics, 2005.

30

ACCEPTED MANUSCRIPT

T.M. Muruganandam, S. Nair, D. Scarborough, Y. Neumeier, J. Jagoda, T. Lieuwen, et al., Active Control of Lean Blowout for Turbine Engine Combustors, J. Propuls. Power. 21 (2005) 807–814.

[41]

S. Balusamy, L.K.B. Li, Z. Han, M.P. Juniper, S. Hochgreb, Nonlinear dynamics of a selfexcited thermoacoustic system subjected to acoustic forcing, Proc. Combust. Inst. 35 (2014) 3229–3236.

[42]

C.O. Paschereit, E. Gutmark, W. Weisenstein, Structure and Control of Thermoacoustic Instabilities in a Gas-turbine Combustor, Combust. Sci. Technol. 138 (1998) 213–232.

[43]

A.F. Ghoniem, A. Annaswamy, S. Park, Z.C. Sobhani, Stability and emissions control using air injection and H2 addition in premixed combustion, Proc. Combust. Inst. 30 II (2005) 1765–1773.

[44]

B.O. Ayoola, R. Balachandran, J.H. Frank, E. Mastorakos, C.F. Kaminski, Spatially resolved heat release rate measurements in turbulent premixed flames, Combust. Flame. 144 (2006) 1–16.

[45]

R.A. Sherman, O. Columbus, Radiation From Luminous and Non-Lumi- nous N atural-G as Flames, Am. Soc. Mech. Eng. (1933).

AC C

EP

TE D

M AN U

SC

RI PT

[40]

31

ACCEPTED MANUSCRIPT

The highlights of manuscript are as follows

RI PT

SC M AN U TE D

•

EP

•

The present investigation has experimentally demonstrated occurrence of the first three modes of thermo-acoustic instabilities in a Rijke tube. The thermo-acoustic instabilities are shown to be entirely suppressed by using a simple technique of air injection into the flame through micro jets issued radially from the tube wall. The simultaneous pressure fluctuation and unsteady heat release measurement in the plane of burner to estimate a coefficients of Rayleigh Index.

AC C

•