Discrete tones in subsonic jet engine test cells

Progress in Aerospace Sciences 101 (2018) 49–60

Contents lists available at ScienceDirect

Progress in Aerospace Sciences journal homepage: www.elsevier.com/locate/paerosci

Discrete tones in subsonic jet engine test cells G. Desmarais, J. Rocha

T

∗

Department of Mechanical and Aerospace Engineering, Carleton University, 1125 Colonel By Dr, Ottawa, ON, K1S 5B6, Canada

A R T I C LE I N FO

A B S T R A C T

Keywords: Jet engine test cell Noise Discrete tones Super resonance

Jet Engine Test Cells used in testing large engines are often plagued by the generation of discrete frequency tones of up to 170 decibels necessitating emergency stoppage of tests to prevent infrastructure damage. The mechanism(s) behind these tones is not well understood and they present a significant issue to airlines, engine manufacturers' and test cell operators. This work provides an overview of typical u-shaped jet engine test cells before briefly presenting the mechanisms of jet impingement, edge tones, diffuser noise and room acoustics as well as the possibility of each of these mechanisms being a source of discrete tone noise in jet engine test cells. The limitations of noise prediction models applied to jet engine test cells is discussed for each aforementioned mechanism and where possible noise predictions are made. Select literature relating to test cell tone generation and super resonance is presented before examining duct resonance and finding it to be a likely source.

1. Introduction With the commercial aviation industry growing, the need for larger and larger airplanes - thus larger and larger engines - is on the rise. Larger engines output greater thrust and mass flow rates than those of the past and are routinely tested in ground facilities, named jet engine test cells (JETC), to assess their performance and reliability. Fig. 1 adapted from Ref. [1] shows a typical u-shaped JETC which is commonly used for commercial aircraft engine tests. Airflow enters through the inlet silencer before being turned 90° by turning vanes. In addition to these turning vanes, it is typical for test cells to have flow conditioning devices at the inlet. A portion of the inlet air is ingested by the engine and the remainder bypasses the engine. The high velocity engine exhaust entrains the bypassed air into the circular augmentor tube. This entrainment is responsible for the net airflow through the test cell as there are no pumps or fans. The exhaustair mixture travels down the augmentor tube (often referred to as the de-tuner in the UK, the ejector or simply as the duct in certain literature) where it mixes before reaching the diffuser and blast basket. The blast basket is a round perforated duct allowing the exhaust gases to escape while attenuating some noise. Located at the rear inside the blast basket is a forward facing cone. This cone serves to direct the exhaust gasses out through the porous blast basket. Finally, the air-exhaust mixture passes through the exhaust silencer and out to atmosphere. Although the blast basket and exhaust silencers provide some attenuation, engine tests are often done at airports near residential areas where the test noise can be of large concern [2,3]. In addition, testing

∗

larger sized engines has created an environment in which acoustic phenomena (which are non-existent in small engine tests) have emerged. Under certain conditions, a test facility generates high amounts of infrasound (＜20 Hz) as well as a low frequency (20–100 Hz range) broadband and tonal noise of great intensity. There have been reports of tonal noise generation of up to 168 dB inside the Arnold Engineering Development Complex's test facilities [4] leading to large physical loads and emergency stops of testing in order to prevent or minimize damage to the facility. Although this phenomenon can be traced back to the 1980's [5] even today the mechanism of tone generation is seemingly not well understood [1,6,7]. This lack of understanding has led to very little progress in eliminating these tones despite a fair amount of research. It is worth mentioning that the noise source is hydrodynamic in nature and does not originate from structural vibrations. This report theoretically investigates potential sources of discrete tone noise generation in the 20–100 Hz range inside JETC's operating at subsonic conditions (the case for most commercial engines). More specifically, the aft portion of the test cell (the augmentor tube, blast basket, cone and exhaust) is examined. This report also amalgamates recent literature and studies of JETC noise, which is fairly disseminated. 2. Potential noise sources and mechanisms of discrete tone generation 2.1. Jet impingement noise Although a JETC is a complex environment with many potential

Corresponding author. E-mail addresses: [email protected], [email protected] (J. Rocha).

https://doi.org/10.1016/j.paerosci.2018.07.003 Received 4 May 2018; Received in revised form 7 July 2018; Accepted 19 July 2018 Available online 27 July 2018 0376-0421/ © 2018 Elsevier Ltd. All rights reserved.

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

Nomenclature A C c Ca Ci d D Dj f fd fD fH fn fR, C

fR, O h h1 i

j k L l Lc Lp Lx Ly Lz Le M m n n1 n2 n3 St u v Vol β

Cross sectional area of the Helmholtz Resonator neck Constant used in the edge tone Speed of sound Phase velocity of the impingement feedback acoustic wave Phase velocity of the impingement downstream propagating instability wave Diffuser inlet diameter Diameter of the small whistler nozzle pipe Impingement nozzle diameter Edgetone frequency Diffuser natural frequency Diffuser frequency of oscillation Helmholtz frequency Impingement frequency Room acoustics frequency, rectangular and close ended room Room acoustics frequency, rectangular and open ended room Edgetone nozzle to wedge distance Step height of the whistler nozzle Longitudinal duct mode

Azimuthal duct mode Transverse, or radial, duct mode Impingement nozzle to wall distance Room resonant mode in the x direction Length of the whistler nozzle step segment Length of the small whistler nozzle pipe Room dimension in the x direction Room dimension in the y direction Room dimension in the z direction Length of the Helmholtz Resonator neck Mach number Room resonant mode in the y direction Room resonant mode in the z direction Constant used in the edge tone Jet impingement mode of oscillation Diffuser mode of oscillation Strouhal Number Flow velocity at the diffuser inlet Jet velocity Volume of the Helmholtz Resonator cavity Length of the wall of the diffuser

Fig. 1. A typical U-shaped Jet Engine Test Cell, Adapted from Ref. [1].

noise sources, jet impingement noise seems to be one of the most apparent sources. Jet impingement noise is the creation of both broadband and discrete tone noise resulting from a high velocity jet

impacting a surface normal to the jet axis with the resulting noise potentially being greater than that of the free jet [8]. Jet impingement tones have been a subject of research for over half a century and the flow field generated by impingement is surprisingly complex. Even in recent times the details of the relationship between the flow field and noise field are still not well understood [9], [10]. In a JETC the high velocity jet impinges on the cone before passing through the blast basket and exiting the exhaust stack. Given that augmentor tubes tend to be of circular cross section, only axisymmetric impingement is relevant. Reference [11] reasons that a small scale instability, or coherent structure, at the nozzle exit grows as it moves downstream in the shear layer of the jet. When the instability makes contact with the impingement surface, a pressure wave is reflected upstream. This pressure wave moves upstream until reaching the nozzle exit (the most susceptible region of the jet to disturbances), where it causes the creation of another small scale instability thus locking the jet into a feedback loop. Upon impact with the wall the large coherent structures emit sound waves. The frequency of the tone in the subsonic case depends primarily on the jet Mach number and the nozzle diameter to plate distance ratio. Reference [12] presents a good schematic of the feedback situation, shown in Fig. 2. The coherent structures generally take on one of two forms depending on the mode of oscillation; the axisymmetric or helical modes.

Fig. 2. Jet impingement feedback phenomenon, from Ref. [12]. 50

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

before impingement. Reference [23] states that diffusers are often operated near the point of stall in order to have maximum pressure recovery. Such a flow is sensitive to mean velocity distribution, turbulence intensity and boundary layer state at the throat of the diffuser. Operating near stall may induce transitory stall which results in excessive losses and loud audible noise. The unique flow in the augmentor of a JETC may further the complexity related to diffuser noise. The advantage of having a lower impingement velocity may not be worthwhile if the diffuser is problematic and acting as a source of noise. Reference [24] investigates the effect of a 4.6°, 6 foot long diffuser on the wall pressure fluctuations at low frequencies in a wind tunnel. It is found that diffuser-on and diffuser-off testing gives markedly different trends with the diffuser-on giving two definite peaks at 40 and 80 Hz. This is attributed to resonance of the working section. Reference [23] studies transitory stall in 16, 20 and 24° diffusers with Mach numbers at the diffuser inlet varying from 0.05 to 0.95. It is found that for Mach numbers from 0.6 to 0.85, a large increase in velocity fluctuation intensity occurs and is accompanied by the generation of a loud audible tone. It is concluded that the tone is caused by a coupling of the ”preferred” frequency of the jet and the fundamental resonant frequency of the diffuser. The tone is found to be quite sensitive, with the placement of a probe in the flow or day-to-day variations leading to a decrease in tone amplitude. The variation caused by the probe may justify the use of a sponson (also known as a core buster), screens or a harp mixer (also known as koppers harp, coopers harp or ring diffuser) to modify or eliminate tones in a JETC (see Refs. [3] and [25] for more information on these devices). Reference [26] finds that transitory stall in conical air diffusers is significantly affected by the ducting to which the diffuser is connected. The experimental work reports on low frequency pressure fluctuations in diffusers and provides a model to calculate the frequency of said fluctuations. The peak frequencies are found to range from 2 to 3.2 Hz which is below the anticipated frequency for JETCs. In addition, scaling from the 95 mm inlet diameter tested to a typical 5–6 m augmentor diameter gives a frequency well below the anticipated values. While the works of references [23,24,26] make use of different experimental setups, some similarities can be found. The frequencies themselves vary significantly and are hard to compare due to the different experimental geometries, however, the Strouhal number, St, f d (St = Du where fD is the frequency of diffuser oscillations, d is the diffuser inlet diameter and u is the flow velocity at the diffuser inlet) can be used to non-dimensionalize the frequencies and make comparisons. Table 1 shows the frequencies observed in these experiments and the resulting Strouhal numbers. While reference [23] leaves the St = 0.01 frequencies unexplained, these appear to be those detailed by Ref. [26]. These works also mention how frequency coupling can affect the tones. Reference [24] explains that the frequencies observed shift for the “diffuser on” configuration, likely caused by the addition of the diffuser changing the duct length thus shifting the organ pipe modes. Reference [26] finds that longer sections of duct preceding a diffuser cause lower frequency tones to be observed. This may be due to the same organ pipe length mechanism. Reference [23] explains that tones occur when the jet

See Ref. [12] for details on the modes. It should be noted that [13] found that supersonic jets ”Impingement frequencies vary with jet temperature” and the tones ”grow more pronounced at elevated versus ambient temperature”. Although this is found for supersonic jets, it may be of interest for subsonic heated engine plume impingement. It was thought that the helical mode of oscillation was not possible in subsonic impingement, however [14], proves that for small standoff distances (L/Dj<2.3) and high subsonic velocity (M> = 0.9) cases the helical mode is possible. It is also stated that a velocity threshold for helical mode onset varies with the temperature of the jet, with an increased temperature decreasing the velocity threshold for helical oscillation. For the hot jet in test cells, it is likely that the onset of impingement tones could be occurring sooner than predicted, with the added potential of helical modes at lower Mach numbers. Impingement onto a cone as opposed to a flat plate may further the likelihood for helical oscillations. There have been studies into impingement onto cones and cylinders at subsonic, sonic and hypersonic velocities however; these studies are either focused on topics other than noise generation or not for the subsonic case (for example see Refs. [15–18]). As far as the authors are concerned, subsonic jet impingement noise generated by flow onto conical surfaces has not been studied. 2.2. Edgetones The edge tone is very similar to jet impingement, however, it concerns a jet of air impacting a wedge as opposed to a flat surface (a roughly 25° wedge is typical). It should be noted that edge tones can be generated not only by a wedge but by any sharp surface inserted into a flow, such as a flat plate parallel to the flow or a conical probe. It is worth mentioning that most edge tone studies make use of planar jets (rectangular cross section with a high aspect ratio) as opposed to axisymmetric jets [19], however, edgetones have been observed in round jets with the edge being a conical tip, much like the conical cone in a test cell (see Ref. [20] for example). Like jet impingement, edge tone flow can cause discrete tones under suitable conditions and the nozzle to wedge distance plays an important role in the frequency of oscillation. Fig. 3 from Ref. [21] provides an illustration of the phenomenon where the jet oscillates about the apex of the wedge. We can instantly see the resemblance of this experiment to the sharp cone inside the blast basket. For the purpose of this work the feedback mechanism can be assumed to be the same as that of Jet Impingement, however, unlike jet impingement the edge tone jet oscillates in a flapping mode, that is the planar jet flaps from one side of the wedge to the other. This is much like the way a flag flaps in the wind. Fig. 4 from Ref. [21] shows the first mode of oscillation (Fig. 4 a) and the second mode of oscillation (Fig. 4 b) of the edge tone, with each mode oscillating at and emitting different frequencies. For the detailed mechanism of the edge tone see Ref. [21]. This flow mechanism is similar to jet impingement but the resulting tone may differ given the different physics. Most experiments make use of low velocities and low Reynolds numbers although [22] investigates the phenomena up to speeds of Mach 1 and finds tones to be present even at high speed. A JETC appears to be somewhat of a hybrid between jet impingement and edge tone situations thus both need to be considered. With that being said, a JETC is more complex than either individual situation and predictions are difficult to make given the deviation from laboratory or ”classical” studies. This is discussed further in the predictions section. 2.3. Diffuser noise Although a diffuser is a seemingly simple device, the complexity of the flow physics in diffusers should not be underestimated. While not all test cells make use of a diffuser at the rear of the augmentor, their use may be of value given their ability to decrease the jet velocity

Fig. 3. Edge tone Feedback Phenomenon, from Ref. [21]. 51

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

Fig. 4. First (a) and second (b) modes of edgetone oscillation, from Ref. [21].

is that given correct conditions, diffusers can be important sources of noise. Finally an interesting concept is that of the whistler nozzle, as shown in Fig. 5. The whistler nozzle is an abrupt increase in the diameter near the exit of a pipe. In a way this is representative of a 180° diffuser. The length and diameter of the smaller pipe (Lp and D), length of the step segment (Lc ), height of the step (h1) and jet velocity (v) all contribute to tone generation. For brevity the mechanism is not discussed but see Ref. [37] for more information. The multitude of possible parameter configurations (Lp , Lc , D, h1 and v) can generate a wealth of tones of different frequencies and amplitudes. The generated tone exhibits a strong staging behavior with dead zones between stages. This behavior is also found in JETC tones with the tones being produced intermittently. It is a possibility that the complex rear end of a JETC has conditions which act like a whistler nozzle. The possibility of strong jet flow in the augmentor tube or a poorly functioning diffuser may increase the tendency for the JETC to act as a whistler nozzle.

Table 1 Approximate diffuser oscillation frequencies and strouhal numbers measured by refs. [23,24,26]. Reference

Frequency (Hz)

Strouhal Number

[23]

52 625 29 58 87 3

0.01 0.12 0.22 0.44 0.66 0.01

[24]

[26]

preferred frequency of oscillation couples to the diffuser natural modes (corresponding Strouhal number of roughly 0.32) and interestingly, reference [27] (discussed later) points out that the preferred Strouhal number of a jet can actually shift when natural modes of a duct are found at nearby frequencies. Reference [3] reports that ”Frandsen and Chamberlain Architects showed that low frequency noise measured in Cell #8 at Hill AFB was a result of fluctuating flow across a diffuser placed at the aft of the augmenter between maximum air flow and a situation where there is less flow due to flow separation”. The Frandsen and Chamberlain Architects reference is not available to the author thus the details are unknown; however, this supports the argument of diffuser generated noise. There appears to be little other research beyond this point which can apply to the JETC diffuser. Other publications such as [28–31] present interesting findings on diffuser flow oscillation and noise generation, however, the test conditions are far from those in a JETC thus are of limited value. There has also been significant investigation into supersonic diffuser separation and noise generation (see for example [32–36]), however, only the subsonic case is relevant to this work. The common argument of these publications which can be applied to JETCs

2.4. Room acoustics of exhaust JETCs are typically constructed of reinforced concrete [3] which reflects sound waves quite well and leads to sound waves inside a JETC being poorly attenuated. When the frequency of one of these sound waves matches a resonant frequency of one of the portions of a JETC (referred to as a room from here on), the room may experience resonance. A broadband sound may also excite a room given the existence of noise of several frequencies within the broadband spectrum, however, the authors do not believe a broadband source is the cause of tones in JETC. If a broadband source (such as the engine jet noise) were exciting a room, a tone would be present under all engine operating conditions and for all engines (which is not the case). This argument is also presented in Ref. [3].

Fig. 5. The Whistler Nozzle, adapted from Ref. [70]. 52

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

As briefly mentioned in the jet impingement and edge tone sections, the nozzle to plate/wedge distance is one of the driving factors dictating the frequency of oscillation. It is unclear what the effective nozzle to plate/wedge distance is in a JETC. As can be seen in Fig. 6, several regions could be considered the nozzle and several regions could be considered the impingement surface. The effective nozzle could be at the engine itself, at the augmentor inlet upstream, the point at which the divergence begins, the point where the basket begins or at some arbitrary point found experimentally. As for the impingement surface, the base of the cone, the tip and midway are all sensible impingement surface datums. This leads to a variety of logical nozzle to plate/wedge spacing values which could be used in the models. The aforementioned arguments make ”classical” predictions of certain mechanisms very difficult if not impossible to apply. In addition, no measurements available to the author have been made to characterize the flow anywhere downstream of the augmenter inlet thus the true jet characteristics are unknown at the point of interest.

The exhaust stack is highly reflective as it is constructed of concrete and steel. It is also a fairly regularly shaped room thus resonance in the exhaust stack may be a source of discrete tone noise in JETCs. The field of room acoustics deals with sound generation inside rooms and provides tools for estimating the natural frequencies of rooms. The natural modes of a simple rectangular room (the exhaust stack) are discussed in the predictions section. The argument of discrete tone generation in the exhaust stack is supported by an existing patent [38]. This patent outlines a method of dissipating infrasound generated by the exhaust stack by the use of vertical slots made into the exhaust stack. Although the patent claims this is to alleviate noise caused by a Helmholtz resonance, it shows that resonance noise in the exhaust stack is indeed an issue. The authors believe this noise may well stem from room resonance and that the vertical slots may change the reflectivity of the walls thus reducing the tone. A simple solution to resonance in the exhaust stack may be to acoustically treat the interior of the exhaust, however, acoustic insulation tends not to be able to withstand the high flow rates, temperature and turbulence created by modern jet engines in test cells. In addition, low frequency noise is of large wavelength and the larger the wavelength, the less effective classical attenuation methods (acoustic insulation) become [3,38,39]. With that being said [39], makes use of an insulated panel in a JETC exhaust stack to reduce the overall sound power by 5.9 and 6.7 dB in the 31.5 and 63 Hz octave bands, with a 5.5 dB overall reduction. This is for a ground based power turbine application thus it is unknown how the velocity and flow rates compare to JETCs. There is also no discussion of long term performance of the silencer.

3.1. Impingement prediction Reference [40] states that ”in the subsonic impingement case, [] Tam and Ahuja (1990) model is the only available theoretical method for frequency prediction of a self-excited subsonic impingement tone”. That being said, Tam and Ahuja (1990) [41] use the frequency prediction formula presented by Neuwerth in 1974 [42].

fn =

n2 L

(

1 Ci

+

1 Ca

)

;

n2 = 1,2,3... (1)

Where fn is the impingement frequency, n2 is the mode of oscillation (integer), L is the nozzle to wall distance, Ci is phase velocity of the downstream propagating instability wave and Ca is the phase velocity of the feedback acoustic waves. Ci is well defined for ”classical” problems and the wave properties can be found using the hydrodynamic instability wave theory. Unfortunately, the ”classical” edge tone studies do not apply to typical JETCs given that the impingement is onto a conical surface as opposed to a flat plate. In addition, the flow is not a free jet, but rather the flow is confined within the augmentor and blast basket. As previously mentioned, it is also unknown what impingement distance L should be used in the model, let alone if the model is valid for impingement onto a cone. There are numerous assumptions which would need to be made in order to estimate a frequency and consequently the result bears little significance. An attempt at this type of analysis is performed on an open jet wind tunnel [43] where it is necessary to perform manipulations in order to make the model more closely match the experimental results. This may be satisfactory for

3. Model limitations and frequency predictions There are several issues which arise upon attempting to predict the frequencies of some of the aforementioned mechanisms. One cause of concern is the unsteady, chaotic, underdeveloped, compressible, high speed, high temperature and high flow rate air-exhaust mixture in JETCs. The three stream combination of engine core, engine fan and test cell bypass air inside the augmentor also gives rise to some form of mixing and non-uniformity. In addition it is unknown how the shear layer and turbulent structures of this particular flow grow and propagate when being contained in the augmentor and how they interact with the diffuser, blast basket and cone. The effect of the porous blast basket on the wall jet resulting from impingement is also unknown. These issues would require experimentation or high level simulation in order to quantify and this is beyond the scope of this work at this time.

Fig. 6. The Possibilities of Impingement Datums, adapted from Ref. [1]. 53

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

relatively high frequency tone found by Ref. [23]. The 3 Hz tone found by Ref. [26] would roughly scale to 0.05 Hz with a 5.5 m diffuser inlet (based on the scaling laws presented by Ref. [47]) which is far below our frequencies of interest. In addition, the model presented by Ref. [26] assumes incompressible flow which is not the case given highsubsonic velocities in JETCs. Reference [48] provides a formula for predicting the frequency of a whistler nozzle, however, it would be unwise to apply it to a JETC given the closed rear end and the non-discrete step of the diffuser. In addition, the formula is empirical, with the data being found for low speed flows (on the order of 30 m/s) with a small test rig [49]. did find whistler tones to be present in high subsonic flows but unfortunately presents no formula for the prediction of said tones.

estimating tones in a wind tunnel and developing a predicatory tool but the complex acoustic signature of test cells make it hard to match the model results to experimental results. Again, given the added level of complexity in JETCs it is believed an analysis of this type would bear little significance. 3.2. Edgetone prediction Reference [21] provides the following formula for determining the frequency of edge tones:

f=

Cv hn1

(2)

Where f is the frequency, v is the jet velocity, h is the distance between the jet and the edge, C is a constant and n1 is also a constant who's value varies between 1 and 3/2 depending on literature [21]. and [44] state that the value of exponent n1 (k in Ref. [44]) has been the subject of a long debate. Given that this constant is typically found for low speed planar flows, even once the debate is settled, the value found may not apply to JETCs. As can be noted in the formula provided above the exit velocity profile, shape of the nozzle, sharpness of the wedge (tip radius) and transversal position of the wedge are not variables in the formula. While these have an impact, they are deemed ”secondary factors” which have a much smaller effect on the oscillation [21] (from results by Ref. [45]). For example [45], found that the wedge angle played no role under 40° [19]. found that all wedge angles of his study (10, 20, 60 and 180°) produced tones. This formula is not the only one available for predicting edge tone frequencies [44]. provides an overview of the theories and formulae which have been proposed over the years. Many of the formulae contain experimentally determined constants and given that there are no experiments performed which mimic JETC conditions it is difficult to use any of the presented constants with certainty. Most experiments make use of a planar nozzle impacting a wedge at low speed. In a test cell, we find an axisymmetric nozzle impacting a cone at high speed. A JETC appears to be somewhat of a hybrid situation between the edge tone and jet impingement situations. As far as the authors are concerned, there exists no research of edge tones on conical impingement surfaces, let alone formulae for their frequency prediction. Adding the confined jet, flow through the blast basket and highly turbulent flow to the problem once again renders predictions very difficult to make.

3.4. Room acoustics of exhaust predictions Reference [3] calculates the room acoustic and Helmholtz frequencies for the test cell of his study (the dimensions are not provided but it appears to be for large engines) using:

forgan

pipe rectangular closed

fhelmholtz =

c 2π

=

2

c ⎛ l ⎞ m n + ⎜⎛ ⎟⎞ + ⎜⎛ ⎟⎞ 2 ⎝ Lx ⎠ ⎝ Lz ⎠ ⎝ Ly ⎠ ⎜

⎟

2

(4) 2

c ⎛ l ⎞ m n + ⎜⎛ ⎟⎞ + ⎜⎛ ⎟⎞ 4 ⎝ Lx ⎠ L L y ⎝ z⎠ ⎝ ⎠ ⎜

⎟

A Vol∗Le

(5)

(6)

where c is the speed of sound, l, m and n are the resonant modes in the x, y and z direction respectively, L x , L y and Lz are the room dimensions in the x, y and z directions respectively, A is the cross sectional area of the neck, Vol is the volume of the cavity and Le is the length of the neck. It is estimated that the exhaust stack Helmholtz resonance is roughly 6 Hz when the exhaust baffles are acting as a neck. Standing waves (room acoustics) in the exhaust stack are found to have a lowest value of 2.29 Hz with the cutoff frequency being 11.78 Hz. It is also found that there are 1402 different modes of standing waves between the cutoff frequency of 11.78 Hz and 60 Hz. If the natural modes of the inlet and augmentor are added to this, it can be concluded that there are far too many natural modes that fall within the anticipated range to pinpoint a single one as a source. There remains some uncertainty in the values found given that the formulae used do not take into account mean flow and there is no information available on the acoustic reflectivity of the exhaust baffles. The reflective surface of the blast basket inside the exhaust stack is ignored for simplicity, but would also have an effect on the generated noise. In summary, a JETC is a complex environment and there are multiple noise generation mechanisms possible, with many potential discrete frequencies being generated. Table 2 provides a summary of the frequency prediction formulae of the different mechanisms which may be of use to the reader.

Reference [23] provides no formula or model for predicting diffuser tone generation although it is stated that the tone is produced due to coupling of the jet instability and the diffuser fundamental plane wave natural mode. The diffuser natural mode can be estimated using conical frustrum pipe frequency predictions (see Ref. [46]), that is:

n3 c 2β

pipe rectangular open

=

2

3.3. Diffuser noise prediction

fd =

2

2

forgan

(3)

Where fd is the natural frequency of the conical diffuser, n3 is the mode of oscillation and β is the length of the diffuser wall from inlet to outlet. Using an augmentor diameter of 5.5 m, a divergence angle of 16° and a length of 7 m gives frequencies of 24, 49, 73 and 97 Hz for the 1st, 2nd, 3rd and 4th modes respectively. Reference [23] found that diffuser transitory stall is sensitive to mean velocity distribution, turbulence intensity and boundary layer state at the throat of the diffuser. Unfortunately many of these conditions are unknown in JETCs, moreover, the previous prediction is for a no flow condition without end corrections thus the resonance results must be used with caution. The work of [26] provides a frequency prediction formula for diffuser noise, however, given the low frequencies found in their experiments, the mechanism investigated most likely differs from the

4. JETC noise from a systemic approach As seen so far, there are many potential noise generation mechanisms acting in the aft portion of a JETC. No individual mechanism can be singled out and the authors believe it would be imprudent to do so [21]. states that sound production in a flute profits from edge tone oscillations but the oscillations transform under the presence of the resonator. This is similar to the work of [26] where diffuser noise is found to be greatly influenced by the connected ducting. The combination of augmentor, diffuser, cone, blast basket, exhaust stack and silencing baffles makes for a unique environment where potential coupling is hard to predict. 54

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

unacceptable aeroacoustic interaction between the engine plume and the test cell. The F119 engine tested in nearly the same conditions as the F100 engine gives widely different acoustic interaction results which are unacceptable whereas the F100 engine produces acceptable (nominal) results (see Fig. 7). Tests of the F119 engine with the facility in baseline configuration generates a discrete tone at 85 Hz with a level of 168 dB (see Fig. 8). This is roughly 28 db, or 25 times, higher than the maximum level of a typical test. This result arises with the engine at an intermediate power level as opposed to full power. Reference [55] states that the 85 Hz tone is most likely ”the result of an acoustic resonance of the cylindrical duct […] in the first transverse mode” into which the jet exhausts. This vibrational mode could be seen as ”coffee sloshing in a cup” or waves that move wall to wall inside the duct rather than end to end. Several fixes are attempted to alleviate the problem. Changing the gap in the doors at the plume deflector opening, increasing secondary airflow, varying air injection rates inside the doors, and the use of projections into the plume prove unsuccessful. Using tuned Helmholtz resonators with significant trial and error, the 85 hz tone can be eliminated, however, this triggers the appearance of a 140 Hz tone (corresponding to the second transverse mode of the duct). Water injection into the exhaust ducting successfully eliminates the 140 Hz tone. An interesting note from Ref. [39] is the mention of a tuned resonator to attenuate a tone in the exhaust stack of an industrial turbine. Ultimately this option is roughly 5 times more expensive than an acoustic panel thus disregarded, even though it proves successful at the AEDC. Reference [55] goes on to mention that a full scale CFD analysis is performed on the nozzle and a self-induced motion is present, producing a large scale flapping motion of the downstream plume which is consistent with the character of the excited 85 Hz diffuser mode [63]. propose a mechanism of tone generation based on supersonic screech. This cannot be the cause of excitation in commercial test cells given the subsonic flow conditions. Ducted jets are then considered as a possible cause of discrete tones. The explanation is that ”A duct can support several families of stationary normal acoustic modes. The insertion of a jet inside the duct allows the instability waves of the jet to couple to these duct modes, producing feedback resonance. […] As the instability wave grows in amplitude as it propagates downstream, it causes the jet to oscillate. The oscillatory motion of the jet pumps energy back into the normal acoustic mode and closes the feedback loop.” This interaction between the jet instability and duct resonance is termed ”super resonance”. Scale model testing is conducted in order to attempt to replicate the super resonance. 1/2, 1/12 and 1/48 scale models are used but even after extensive testing the super resonance is not replicated. The information presented in the two reports for the USAF [53,54] is quite similar to that summarized by Ref. [55]. Reference [53] predicts resonant frequencies for the duct and compares those with the experimental tones. The lack of super resonance in the small scale facilities prevents research on the coupling mechanism between the jet and duct. The remainder of the report is spent examining supersonic screech. Several recommendations for future work are laid out including the need to identify modes inside the duct. Reference [54] follows up the work from Ref. [53] by providing extensive experimental results and some baseline theoretical results. Of interest, the frequency of oscillation seemingly increases linearly with temperature. Roughly 15 years after the initial work at the AEDC [56], reinvestigates the tones generated in the AEDC. Computational tools are used in hopes of better understanding the super resonance mechanism and determining the factors necessary to predict super resonance. Reference [56] uses 2-D Direct Numerical Simulation with idealized geometries to model the system. The result is a 2 dimensional flow which resonates at 160 dB within 3% frequency of the expected mode. The resulting mechanism appears similar to that predicted by Ref. [5]. The simulated flow velocity is sonic thus of little use in subsonic test cells, however, it is found that ” [the] mechanism does not depend upon any shock-cell structure in the jet, and since strong resonances have also

Table 2 Summary of frequency prediction formulae. Noise Source

Frequency Prediction Formula

Jet Impingement Tones

n2 1 1 ⎞ L ⎛⎜ + ⎟ ⎝ Ci Ca ⎠ Cv f = n1 h n c fd = 3 2β

Edgetones Diffuser Noise

fn =

Room Acoustics (Rectangular, Open)

Helmholtz Resonance

2

2

l 2 Lx

m n +⎛ ⎞ +⎛ ⎞ ⎝ Lz ⎠ ⎝ Ly ⎠

fR, C =

c 4

( ) ( )

c 2π

A Vol*Le

fH =

2

m n +⎛ ⎞ +⎛ ⎞ ⎝ Lz ⎠ ⎝ Ly ⎠

c 2

Room Acoustics (Rectangular, Closed)

2

l 2 Lx

fR, O =

4.1. Literature on JETC tones and super resonance Many JETC manufactures or operators have pointed out noise issues in JETCs. General Electric [50–52], Arnold Engineering Development Center (AEDC) [53–56], Wright-Patterson Air Force Base [6], Aero Systems Engineering [1,38], Ceneco [3], MDS1, Nasa Glen [7], Honeywell [57] and Seimens [58] have all mentioned (directly or indirectly) noise issues in JETCs. In addition, there have been many academic investigations into test cell noise. This section is an overview and summary of the findings from select JETC noise literature. Ejector shroud noise in test facilities is seemingly first identified in the 1960's [59] and is named ”howling”. Reference [5] from 1987 provides a description of low frequency tone generation in test cells. It describes the situation as: ” [altitude test cells] often exhibit large amplitude self-sustained acoustic oscillations and structural vibrations. Under suitable conditions intense, discrete frequency sound is produced which can be damaging to both cell and engine structures …” Tone generation is not limited to altitude test cells, but has also been reported in sea level test cells such as SL-2 at the AEDC [55] [5]. mentions that the operational conditions producing tones falls within a fairly narrow window and seems to depend on the individual test cell. It is also mentioned that the mechanisms which generate the disturbances are not understood and several possibly competing mechanisms could be responsible for the tonal observations. In the early 90's, an effort by Ahuja, Tam, Massey, Fleming and Jones was undertaken in order to solve a discrete tone generation problem at an AEDC altitude test cell. There appears to have been two reports generated for the United States Air Force (USAF), the first in 1992 and the second in 1994 [53,54]. In Addition, 5 publications ([4,60–63]) date from this period, with varying involvement from each author and with additional authors. The topics of these 5 papers seemingly stem from the research conducted at the AEDC and are well summarized in Ref. [55]. Although the focus of this research is on the AEDC test cells, as previously mentioned this problem has also been reported in other test cells such as the Propulsion Systems Laboratory at the NASA Glenn Research Center and in ”a few other facilities” [4,7]. It should be noted that the AEDC test cells do not have a silencer assembly (diffuser, basket and cone) as do modern commercial test cells. This could be evidence that the tone generation in modern cells may not originate from any of these components, but rather from the augmentor. Reference [55] describes the problem in the AEDC test cell as an

1 MDS Aero Support Corporation, ”Cell Structure – Very Large Engine Test Cells regularly suffer from infrasound disturbance, but the latest research demonstrates that the science exists to alleviate this concern,” Dec. 2012, https:// mdsaero.com/2012/12/cell-structure-very-large-engine-test-cells-regularlysuffer-from-infrasound-disturbance-but-the-latest-research-demonstrates-thatthe-science-exists-to-alleviate-this-concern/.

55

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

Table 3 Blast basket end types tested by Ref..[51].

Fig. 7. Typical altitude test cell acoustic signature, from Ref. [55] (reproduced from Ref. [4]).

commercial JETCs since they have no sharp trailing edge. Again building on this work, reference [65] proposes that a pressure pulse from the supersonic jet acts as the noise source exciting the duct mode which excites the shear layer at the nozzle of the jet causing a feedback phenomenon. It is found that resonance is present for a M1.2 jet, but not for a M1.5 jet. Again, this approach does not explain resonance experienced for subsonic jets with non-sharp trailing edge ducts, although, these three works all indicate that the frequency of oscillation is close to that of an acoustic normal mode of the duct. There are no follow up publications from these authors relating to implementation of a solution to super resonance based on the findings from this work. A 2007 report produced by Dr. Tam [66] presents a consolidation of the present knowledge of the aeroacoustics of ducted jets in a 115 page report. This report comes roughly 15 years after the initial investigation at the AEDC and attests to the complexity of noise in JETCs. Part of the report proposes that the jet instability waves grow quickly given the large amount of energy in the jet. As these waves move downstream, they excite a duct mode. This excited duct mode excites instability waves in the jet thus closing the feedback loop. When the frequency of the most excited instability wave of the jet corresponds to that of the duct, super resonance is generated. This is essentially the same mechanism as that proposed by the investigation in the early 90s as summarized in Ref. [55]. Potential methods for super resonance elimination are proposed including the use of Helmholtz resonators, water injection and longitudinal fins in the augmentor. The fins, shown in Fig. 9, are meant to prevent the flapping or helical mode of the jet from forming (the flapping or helical modes are the most amplified instability waves of high temperature circular jets). Reference [27] from 2013 points out that the understanding of the mechanism of unwanted noise in JETC is still far from complete. They perform tests on a simplified model scale experiment (essentially discharging a jet through a pipe) and test various techniques for noise suppression. Parametric trials are performed to determine the conditions required for resonance in this simple test case. A 0.58 inch nozzle discharges into a duct with a geometry range chosen such that the NASA Glen Research Center and AEDC test cells are mimicked. The 2 inch diameter duct produced frequency spikes at 1650, 1100, 820 and 710 Hz for lengths of 3, 5, 6 and 9 inches respectively. This is found to be near resonant frequencies of the duct. It is found that a ”howl stick” is an effective method of damping super resonance. Screens are found to damp velocity fluctuations in the pipe (which are greatest at the center of the pipe) and also decrease noise generation. Placing screens downstream prevents the generation of excess impingement noise and leads to better noise reduction than upstream screens (although low

Fig. 8. Acoustic signature of problematic engine, from Ref. [55] (reproduced from Ref. [4]).

Fig. 9. The Fins Proposed by Tam [66] to avoid helical or flapping oscillation (as viewed from the end of the duct), from Ref. [66].

been observed in subsonic ducted jets, it has been argued that shocks are not a necessary component of the mechanism”. Aeroacoustic simulations are also undertaken using the Rocflu solver which proves unsuccessful given that the solver is too dissipative for reliable simulation of the flow. Reference [64] continues the work of [56] through a highly theoretical and simulative approach. Reference [64] consider that the scattering of vertical disturbances at the sharp trailing edge of the duct as the noise source instigating the duct mode. This applies to both subsonic and supersonic jets. This work is of limited application to 56

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

Fig. 10. Scale model used by Ref. [1].

of the howl stick [52] and the exhaust sound suppressor [58]. Patent [52] outlines the use of a flow distorter placed near the outlet of the engine exhaust to suppress cell howl. In essence the device is a rod placed inside the augmentor, perpendicular to the axis of the augmentor and in the jet flow from the engine. This is described as being effective in eliminating howl and having minimal impact on the engine. The patent is for supersonic flow, however, as previously mentioned howling is also an issue in the subsonic case [27]. found the howl stick to be effective in damping subsonic super resonance when placed in upstream duct locations.

Table 4 Modes of the augmentor. i

j

k

Modal Frequency (Hz.)

1 0 0 1 1 0 1

0 1 0 1 0 1 1

0 0 1 0 1 1 1

7 37 76 37 77 106 106

Fig. 11. Schematic of Acoustic Longitudinal, Azimuthal and Transverse (or radial) modes in a cylindrical resonator, adopted from Ref. [71].

The exhaust sound suppressor [58] outlines a sound suppression apparatus (external to the engine) and method to reduce low frequency exhaust flow noise while avoiding the creation of additional acoustical standing waves (duct modes). The method employed is to insert flow obstruction members (ex: pipes or airfoils) of different widths extended within the exhaust passage. Each member is positioned at an angle between 10 and 30° to the normal of the flow direction. The flow members can be placed at different locations along the length of the duct. The purpose of the varying widths of the airfoils is so that adjacent wake shed vortex sheets are unlikely to be established. The width of the members should be chosen such that low frequency eddies of the

frequency noise is better reduced by upstream screens). The screens are also found to suppress super resonance. The fins proposed by Tam [66] are found to be relatively ineffective, although provide some benefit at higher speeds. Although limited to the simplest case of a jet exhausting into a straight pipe, this study helps further the understanding of tone generation and super resonance. Patent [67], ”System and method for suppression of unwanted noise in ground test facilities” is filed by the same authors as [27]. This patent seems to be the description of the successful resonance cancelling tools from Ref. [27]. There are other patents which have the goal of suppressing resonance in Jet engine ground facilities. Two notable examples are those 57

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

4.2. Duct acoustics

incoming gas stream are disrupted and also so that the shed Eddie frequency has a low probability of coupling with other resonant frequencies. Again, this method uses physical means to disrupt and/or prevent coupling to the duct modes. Reference [51] investigates pressure fluctuations of roughly 2 Hz having caused structural damage in a test cell. The publication focuses on presenting test results more so than investigating the mechanism. Particular emphasis is placed on the basket termination (cone). A 1:12 scale model is used to verify the primary cause and identify practical solutions for the pressure fluctuations. Blast basket end configurations of: concave dish (baseline), flat end, convex dome, 50° cone with hemispherical nose and 55° truncated cone (see Table 3) are tested providing 59%, 59%, 68% and 74% reductions (respectively) from the baseline peak-to-peak pressure fluctuations and reductions of 7.7 dB, 7.7 dB, 9.9 dB and 11.7 dB (respectively) from the baseline 128.3 dB. The 50° cone is then implemented on the full scale test cell and a matching 68% reduction in peak-to-peak pressure fluctuations is observed. This result suggests that small scale testing can be quite a good estimator of full scale results and that the cone can have significant effects on the acoustics of a JETC. [1] performs acoustic testing on a 1/43rd scale model, shown in Fig. 10, in order to better understand the aeroacoustics of the augmentor, basket and exhaust stack. Again, this study focuses mainly on presenting results as opposed to developing an understanding of the phenomena. The 2.5–160 Hz frequency range (full scale values) is of interest in this study. It is found that the dominant peak shifts up in frequency with increasing engine thrust (i.e. engine velocity). As the thrust is increased from low (45k lbf) to medium (77k lbf) to high (102k lbf), the scaled tone changes from 21 to 26–35 Hz. A second peak near 6 Hz, which is highest in the aft portion of the augmentor, is also present and unexpectedly decreases in frequency with increasing thrust. Further testing with an increased open area of the basket decreases the flow resistance and the mass flow rate required for the given thrusts but actually increases the sound pressure level. The secondary tone also increases in amplitude by 3–5 dB. In addition, a tone at 12.4 hz (which is barely noticeable previously) emerges. [57] uses CFD with the DES turbulence model to simulate a complete transient, compressible turbulent test cell. The goal is to understand the origin of severe infrasound encountered when testing a problematic engine which caused structural damage to the exhaust stack. The simulated test cell varies from that presented in the introduction. Rather than a diffuser at the rear of the augmentor, this test cell has a step diameter increase with the jet impinging upon a convex dome. The augmentor length also varies as it is adjustable and has a secondary air inlet. The engine velocities reach values of up to M1.6. It is found that the jet mixes very little before impacting the dome. It is also found that the flow turns and moves upstream in the augmentor almost all the way to the engine nozzle. The flow in the blast basket is found to be particularly volatile with the exhaust plume buffeted by the air that is reflected off the blast plate. In addition, in the upstream portion of the blast basket there is reversed flow through the perforations, little to no flow in the intermediate perforations and high flow through the downstream perforations. Adding rear facing inserts to the augmentor tube outlet is found to prevent reversed flow and helps alleviate the aforementioned issues. This insert causes an overall reduction of 32 dB in the 0–100 hz range at the impingement location. The usefulness of such a device in subsonic test cells is unknown at this time. For brevity the works of [3,6,68,69] are omitted, however, the findings are quite interesting. All deal with ground based operation of jet engines and deal either with noise [3,6,69] or alterations to aft structures of a JETC to improve aerodynamics [68]. Again, duct resonance appears to be a re-occurring theme as a source of noise in these papers.

In the literature presented there is little agreement on the noise source and varying solutions are implemented almost on a case by case basis. This may be attributed to the complexity of JETC aerodynamics/ aeroacoustics and unfortunately provides no clear mechanism or solution to tones in JETCs. Many of the literature sources do point towards duct modes of the augmentor as a source of noise. The resonant modes for ducts are fairly easy to estimate (see Ref. [53] for example). In a JETC it becomes difficult to estimate the effective length and diameter of the duct due to the diffuser basket and cone. Using an approximation of a 5.5 m diameter augmentor of 25 m length with no mean flow would give rise to the fundamental modes shown in Table 4. Fig. 11 provides a visual representation of the different modes of oscillation possible in the augmentor. As can be seen, many of the fundamental duct modes fall within the 20–100 Hz range. The first transverse mode is reported as a potential problematic mode in many publications. With our assumed dimensions, the first transverse resonance is estimated to be 76 Hz which agrees with the first transverse mode hypothesis. These results should however be taken with caution given the no flow approximation and complex flow in JETCs. There are also multiple fundamental duct modes which together cover a fairly broad range of frequencies in our range of interest. This makes it difficult to pinpoint duct modes as the source of tones but it appears to be a likely source. 5. Conclusion Although this work is not a complete literature summary it introduces the reader to the phenomena of tonal noise in JETC's. The discussion of all potential noise sources is not exhaustive and as a result there could be (and likely are) undiscussed mechanisms acting in test cells (such as cavity tones, perforate noise or impingement on the augmentor inlet). Of the many noise generation mechanisms which could be present in JETC's, impingement tones, edge tones, diffuser noise and room acoustics seem likely. Unfortunately the complex environment of a JETC makes it very difficult to pinpoint the source(s) of tonal noise. The respective model of each aforementioned mechanism is of limited use and the results bear little significance for varying reasons. The presented literature on tonal noise in JETC's shows that duct resonance is a likely noise source and preliminary calculations seemingly support this hypothesis. Experimental testing will hopefully allow for the confirmation or negation of duct resonance as a noise source and lead to progress on new designs for tonal elimination. Acknowledgements The authors wish to thank the Consortium of Aerospace Research in Canada (CARIC/CRIAQ ENV-715) as well as the Natural Sciences and Engineering Research Council of Canada for their financial support. References [1] K. Hatlestad, Scale model testing of infrasound produced by various jet engine test cell configurations, 12th AIAA/CEAS Aeroacoustics Conference (27th AIAA Aeroacoustics Conference), 2006, https://doi.org/10.2514/6.2006-2540. [2] C.A. Kodres, Jet Engine Test Cell Noise Reduction, Tech. Rep Naval Facilities Engineering Command, Port Hueneme, CA, August 2000 technical Rept. TR-2118ENV. [3] H. Wei Hua, Investigation into the Vortex Formation Threshold and Infrasound Generation in a Jet Engine Test Cell, Ph.D. thesis University of Canterbury. Department of Mechanical Engineering, 2009. [4] R.I. Jones, G. Lazalier, The acoustic response of altitude test facility exhaust systems to axisymmetric and two-dimensional turbine engine exhaust plumes, DGLR/AIAA 14th Aeroacoustics Conference, Proceedings, vol. II, Aachen, Germany, 1992, pp. 785–794. [5] M. Howe, Self-excited oscillations in jet engine test cells, J. Fluid Struct. 1 (1) (1986) 121–148, https://doi.org/10.1016/S0889-9746(87)90230-1. [6] V. Topalian, Acoustic Resonances in Ducted Jet Systems, Ph.D. thesis Dept. of

58

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

[37] A. Hussain, M. Hasan, The ’whistler-nozzle’ phenomenon, J. Fluid Mech. 134 (1983) 431–458, https://doi.org/10.1017/S0022112083003420. [38] D. Long, Aero Systems Engineering Inc., Us. Patent Application for a ”jet Engine Test Cell Structure,”, Minnetonka, MN, US Patent No. 5,837,890 (Filed 12 Dec. 1994). [39] C. Lucas, G. Hessler, The reduction of low frequency gas turbine exhaust noise - a case study, ASME 1996 International Gas Turbine and Aeroengine Congress and Exhibition, Birmingham, UK, 1996, pp. 1–6. [40] B. Vinoth, E. Rathakrishnan, Effect of impinging plate geometry on the self-excitation of subsonic impinging jets, J. Fluid Struct. 27 (8) (2011) 1238–1251, https://doi.org/10.1016/j.jfluidstructs.2011.05.006. [41] C. Tam, K. Ahuja, Theoretical model of discrete tone generation by impinging jets, J. Fluid Mech. 214 (1990) 67–87, https://doi.org/10.1017/S0022112090000052. [42] G. Neuwerth, Acoustic Feedback of a Subsonic and Supersonic Free Jet Which Impinges on an Obstacle, Tech. Rep NASA, 1974 technical Rept. TT F-15719. [43] M. Rennie, Effect of Jet Length on Pressure Fluctuations in 3/4 Open-jet Wind Tunnels, Published on Research Gate (2000), pp. 1–14. [44] I. Vaik, R. Varga, G. Paal, Frequency and phase characteristics of the edge tone, part i, Period. Polytech. - Mech. Eng. 58 (1) (2014) 55–67, https://doi.org/10.3311/ PPme.7028. [45] C. Segoufin, B. Fabre, L. De-lacombe, Experimental investigation of the flue channel geometry influence on edge-tone oscillations, Acustica/Acta Acustica 90 (5) (2004) 966–975. [46] G. Scavone, An Acoustic Analysis of Single-reed Woodwind Instruments with an Emphasis on Design and Performance Issues and Digital Waveguide Modeling Techniques, Ph.D. thesis Department of Music, Stanford University, 1997. [47] K. Kutruff, Room Acoustics, Spon Press, Milton, U.K., 2009 (chapter 9). [48] M. Hasan, A. Hussain, A formula for resonance frequencies of a whistler nozzle, J. Acoust. Soc. Am. 65 (5) (1979) 1140–1142, https://doi.org/10.1121/1.382779. [49] A. Szumowski, W. Selerowicz, G. Meier, (ed) G. Meier, F. Obermeier, Oscillations of a subsonic flow in an abruptly expanding circular duct, Flow of Real Fluids, Lect. Notes Phys. 235, doi: 10.1007/3-540-15989-4_68. [50] R. Dickman, H. Hehmann, W. Hoelmert, R. Freuler, A solution for aero-acoustic induced vibrations originating in a turbofan engine test cell, 13th Aerodynamic Testing Conference, San Diego, Ca, Usa, 1984, pp. 99–108, , https://doi.org/10. 2514/6.1984-594. [51] R. Freuler, K. Montgomery, Reducing large pressure fluctuations in an engine test cell by modifying the exhaust blast basket end configuration, First Joint CEAS/AIAA Aeroacoustics Conference (16th AIAA Aeroacoustics Conference), Munich, Germany, 1995, pp. 903–909. [52] K. Kinzie, General Electric Company, Us. patent application for a ”methods and apparatus for suppressing engine test cell howl”, Cincinnati, OH, US Patent No. 6, 092,621 (Filed 29 June 1999). [53] K. Ahuja, K. Massey, A. Fleming, C. Tam, R.I. Jones, Acoustic Interactions between an Altitude Test Facility and Jet Engine Plumes - Theory and Experiments, Tech. Rep, Georgia Tech. Research Institute, Atlanta, GA, USA, January 1992 technical Rept. NASA-CR-190423, NAS 1.26:190423, AD-A245463, GTRI-A-8612-001, AEDCTR-91-20. [54] K. Ahuja, K. Massey, C. Tam, R.I. Jones, Temperature Effects on Accoustic Interactions between Altitude Test Facilities and Jet Engine Plumes, Tech. Rep, Georgia Tech. Research Institute, Atlanta, GA, USA, January 1992 technical Rept. AEDC-TR-94-10. [55] C. Sebourne, F. Shope, Research summary on the aedc astf c-2 aeroacoustic resonance phenomenon, 11th AIAA/CEAS Aeroacoustics Conference, Monterey, CA, USA, 2005, pp. 1–12, , https://doi.org/10.2514/6.2005-2932. [56] J. Freund, V. Topalian, A. Samanta, J. Kim, Super-resonances in aedc altitude test cells, 2007 U.S. Air Force T&E Days, Destin, FL, USA, 2007, pp. 131–137. [57] J. Pearson, Y. Sheoran, B. Schuster, Cfd analysis of a gas turbine engine test cell to understand and alleviate infrasound, ASME 2015 International Mechanical Engineering Congress and Exposition, Houston, NV, USA, 2015, https://doi.org/10. 1115/IMECE2015-50453 V013T16A001–1 – V013T16A001–8. [58] R. Putnam, Siemens Power Generation, Us. Patent Application for a ”gas Turbine Exhaust Sound Suppressor and Associated Methods”, Orlando, FL, USA, US Patent No.2009/0277714 (Filed 9 May 2008). [59] P. Bradshaw, J. Flintoff, D. Middleton, Unexplained scale effects in ejector shroud ”howling”, J. Sound Vib. 7 (2) (1968) 183–190, https://doi.org/10.1016/0022460X(68)90266-6. [60] J. Caruthers, A. Vkili, G. Raviprakash, R.I. Jones, Aeroacoustic interaction of jets/ test cells, 14th DGLR/AIAA Aeroacoustics Conference, Aachen, Germany, 1992, pp. 795–799. [61] R.I. Jones, K. Ahuja, C. Tam, M. Abdelwahab, Measured acoustic characteristics of ducted supersonic jets and different model scales, 31st Aerospace Sciences Meeting, Reno, NV, USA, 1993, pp. 1–12, , https://doi.org/10.2514/6.1993-731. [62] K. Massey, K. Ahuja, R.I. Jones, C. Tam, Screech tones of supersonic heated free jets, 32nd Aerospace Sciences Meeting and Exhibit, Reno, NV, USA, 1994, https://doi. org/10.2514/6.1994-141. [63] C. Tam, K. Ahuja, R.I. Jones, Screech tones from free and ducted supersonic jets, AIAA J. 31 (5) (1994) 917–922, https://doi.org/10.2514/3.12074. [64] A. Samanta, J. Freund, Finite-wavelength scattering of incident vorticity and acoustic waves at a shrouded-jet exit, J. Fluid Mech. 612 (2008) 407–438, https:// doi.org/10.1017/S0022112008003212. [65] V. Topalian, J. Freund, Acoustic resonance in a model ducted-jet system, AIAA J. 48 (7) (2010) 1348–1360, https://doi.org/10.2514/1.45191. [66] C. Tam, Engine Test Cell Aeroacoustics and Recommendations, Tech. Rep, Florida State University, Department of Mathematics, Tallahassee, FL, USA, October 2007 technical Rept. AEDC-TR-06-3.

Theoretical and Applied Mechanics, Univ. of Illinois at Urbana-Champaign, 2009. [7] K. Zaman, C. M.M, F. A.F, Noise from a Jet Discharging into a Duct and its Suppression, Tech. Rep NASA, January 2012 technical Rept. E-18163. [8] A. Marsh, Noise measurements around a subsonic air jet impinging on a plane, rigid surface, J. Acoust. Soc. Am. 33 (8) (1961) 1065–1066, https://doi.org/10.1121/1. 1908894. [9] B. Henderson, The connection between sound production and jet structure of the supersonic impinging jet, J. Acoust. Soc. Am. 111 (2) (2002) 735–747, https://doi. org/10.1121/1.1436069. [10] F. Alvi, H. Lou, C. Shih, R. Kumar, Experimental study of physical mechanisms in the control of supersonic impinging jets using microjets, J. Fluid Mech. 613 (2008) 55–83, https://doi.org/10.1017/S0022112008003091. [11] C.-M. Ho, N. N.S, Dynamics of an impinging jet. part 1. the feedback phenomenon, J. Fluid Mech. 105 (1980) 119–142, https://doi.org/10.1017/ S0022112081003133. [12] T. Davis, A. Edstrand, F. Alvi, L. Cattafesta, D. Yorita, K. Asai, Investigation of impinging jet resonant modes using unsteady pressure-sensitive paint measurements, Exp. Fluid 56 (5) (2015) 1–13, https://doi.org/10.1007/s00348-0151976-9. [13] J. Gustavson, P. Ragaller, R. Kumar, A. F.S, Temperature effect on acoustics of supersonic impinging jet, 16th AIAA/CEAS Aeroacoustics Conference, Stockholm, Sweden, 2010, pp. 1298–1315, , https://doi.org/10.2514/6.2010-3785. [14] P. Panickar, G. Raman, Criteria for the existence of helical instabilities in subsonic impinging jets, Phys. Fluids 19 (10) (2007), https://doi.org/10.1063/1.2798804 106103–1–106103–17. [15] R. Mehta, P. J.K, Flow structure of a supersonic jet impininging on an axisymmetric deflector, Indian J. Eng. Mater. Sci. 4 (1997) 178–188. [16] R. Mehta, Numerical simulation of supersonic turbulent jets impinging on an axisymmetric deflector, Indian J. Eng. Mater. Sci. 9 (2002) 227–236. [17] J.-H. Park, T.-S. Lee, Y.-G. Kim, A study of supersonic jets impinging on an axisymmetric cone (article in Korean), Journal of the Korean Society for Aeronautical & Space Sciences 31 (7) (2003) 26–31, https://doi.org/10.5139/JKSAS.2003.31.7. 026. [18] R. Seryu, J. Iwamoto, Impingement of underexpanded jet on cone, 10th International Conference on Fluid Control, Measurements, and Visualization, Moscow, Russia, 2009. [19] M. Ibrahim, Experimental and theoretical investigations of edge tones in high speed jets, J. Fluid Sci. Technol. 8 (1) (2013) 1–19, https://doi.org/10.1299/jfst.8.1. [20] J. Lepicovsky, K. Ahuja, Experimental results on edge-tone oscillations in highspeed subsonic jets, AIAA J. 23 (10) (1985) 1463–1468, https://doi.org/10.2514/3. 9111. [21] G. Paal, I. Vaik, Unsteady phenomena in the edge tone, Int. J. Heat Fluid Flow 28 (4) (2007) 575–586, https://doi.org/10.1016/j.ijheatfluidflow.2007.04.011. [22] A. Krothapalli, K. Karamcheti, Y. Hsia, D. Baganoff, Edge tones in high-speed flows and their application to multiple-jet mixing, AIAA J. 21 (7) (1983) 937–938, https://doi.org/10.2514/3.8180. [23] K. Zaman, M. Dahl, Some observations on transitory stall in conical diffusers, 28th Aerospace Sciences Meeting, Reno, NV, USA, 1990, pp. 1–11, , https://doi.org/10. 2514/6.1990-48. [24] J. Willis, Spurious pressure fluctuations in wind tunnels, J. Acoust. Soc. Am. 43 (5) (1968) 1049–1054, https://doi.org/10.1121/1.1910938. [25] G. Ramos, Study of a Test Cell for Commercial Jet Engines, Master’s thesis Dept. of Mechanical Engineering, Tecnico Lisboa, Lisbon, Portugal, 2015. [26] A. Kwong, A. Dowling, Unsteady flow in diffusers, J. Fluid Eng. 116 (4) (1994) 842–847, https://doi.org/10.1115/1.2911859. [27] K. Zaman, M. Clem, A. Fagan, Noise from a jet discharged into a duct and its suppression, Int. J. Aeroacoustics 12 (3) (2013) 189–214, https://doi.org/10.1260/ 1475-472X.12.3.189. [28] K. Fehse, W. Neise, Low frequency sound generated by flow separation in a conical diffuser, Eur. J. Mech. B Fluid 19 (5) (2000) 637–652, https://doi.org/10.1016/ S0997-7546(00)00114-X. [29] L. Van Lier, S. Dequand, A. Hirschberg, Aeroacoustics of diffusers: an experimental study of typical industrial diffusers at Reynolds numbers of O (105), J. Acoust. Soc. Am. 109 (1) (2001) 108–115, https://doi.org/10.1121/1.1329618. [30] A. Martin, F. Thiele, Investigation of flow separation and transition to turbulence in an axisymmetric diffuser, , in: 42nd AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, USA, 2004, pp. 1–12, , https://doi.org/10.2514/6.2004-1102. [31] L. Hoefener, W. Nitsche, A. Carnarius, F. Thiele, Experimental and numerical investigations of flow separation and transition to turbulence in an axisymmetric diffuser, in: H. Rath, C. Holze, H. Heinemann, R. Henke, H. Honlinger (Eds.), New Results in Numerical and Experimental Fluid Mechanics V. Notes on Numerical Fluid Mechanics and Multidisciplinary Design (NNFM), vol. 92, Springer, 2004. [32] T. Bogar, M. Sajben, J. Kroutil, Characteristic frequencies of transonic diffuser flow oscillations, AIAA J. 21 (9) (1983) 1232–1240, https://doi.org/10.2514/3.8234. [33] K. Zaman, M. Dahl, T. Bencic, C. Loh, Investigation of a ’transonic resonance’ with convergent-divergent nozzles, J. Fluid Mech. 463 (1) (2002) 313–343, https://doi. org/10.1017/S0022112002008819. [34] D. Papamoschou, A. Zill, Fundamental investigation of supersonic nozzle flow separation, 42nd AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, USA, 2004, pp. 1–17, , https://doi.org/10.2514/6.2004-1111. [35] D. Papamoschou, A. Johnson, Unsteady phenomena in supersonic nozzle flow separation, 36th AIAA Fluid Dynamics Conference and Exhibit, San Francisco, CA, USA, 2006, pp. 1–11. [36] B. Donald, W. Baars, C. Tinney, Sound produced by large area-ratio nozzles during fixed and transient operations, AIAA J. 52 (7) (2014) 1474–1485, https://doi.org/ 10.2514/1.J052588.

59

Progress in Aerospace Sciences 101 (2018) 49–60

G. Desmarais, J. Rocha

[69] G. Kroeff, Low Frequency Noise Generated by Industrial Gas Turbines, Ph.D. thesis Loughborough University, 2004. [70] H. Husain, F. Hussain, The elliptic whistler jet, J. Fluid Mech. 397 (1999) 23–44, https://doi.org/10.1017/S0022112099005923. [71] A. Miklos, P. Hess, Z. Bozoki, Application of acoustic resonators in photoacoustic trace gas analysis and metrology, Rev. Sci. Instrum. 72 (4) (2001) 1937–1955, https://doi.org/10.1063/1.1353198.

[67] K. Zaman, M. Clem, A. Fagan, The United States of America as Represented by the Administrator of the National Aeronautics and Space Administration, Us. Patent Application for a ”system and Method for Suppression of Unwanted Noise in Ground Test Facilities”, Washington, DC, USA, US Patent No. vol. 8,939,253 B1 (Filed 26 April 2013). [68] J. Gilmore, Computational Fluid Dynamics Analysis of Jet Engine Test Facilities, Ph.D. thesis Mechanical Engineering Dept., Univ. of Canterbury, 2011.

60