Noise in an entrance plenum

Experiments were carried out on the large-dimensional (external diameter 1,2m) low-speed (frequency of rotation of 2000rot/min) compressor, contain­ing system of rows IGV-R-S and in detail described in [4-6]. Numbers of sta­tors vanes are identical (Zigv=Zs=36), and 2 variants of the investigated axial gaps were selected so that in one of them clocking effect were insignificant (as­sembly 1), and in other on the contrary clocking effect was clearly expressed (assembly 2). As shown in [4], to these conditions axial gaps Д12 (between IGV and R) and Д23 (between R and S) on the hub section are corresponded: assembly 1 – Д12=60mm, Д23 =5mm; assembly 2 – Д12=15mm, Д23=15mm. Comparison of results of noise measurement at two specified assembly allows to allocate the effects, connected with IGV and S clocking.

The compressor flawing path with an entrance site and an entrance plenum is presented on Fig.1. The cascades of airfoils, appropriate to cylindrical sec­tion of a flawing path on mean radius, are presented on Fig.2.

Noise pressure in an entrance plenum was estimated on a reverberation chamber method. For certification of reverberation properties of a plenum the cycle of measurements of acoustic pressure was preliminary executed at radiation of broadband noise by an electromechanical standard source. Such measurements, executed in 78 points, have allowed to establish a point of a microphone position (the central point) in which vicinity rather uniform noise field is formed with the RMS deviation d within the limits of 0,75-2,5dB on the basic 1/3 octava frequencies exceeding 250Hz. It is necessary to notice that for the basic 1/3 octava frequencies f0, near to frequencies, multiple to R blade passing frequency (1266,7Hz) the value d appeared within the limits of 0,75- 1,25dB. It allows to accept that changes of noise pressure in the plenum, re­ceived in the central point and exceeding specified limits, are caused by change of acoustic properties of researched system of rows.

Other important acoustic characteristic of the plenum is time of reverber­ation t, which was determined experimentally for everyone 1/3 octaves fre­quency in a range 500-10000Hz. The range of t change has made 0,6-2,4 second. Comparison to the data for drown chambers (t ~0,2 sec.) allows to count reverberation properties of the plenum satisfactory for relative estima­tions of noise pressure change.

Measurements in the central point were carried out by a microphone of Firm Larson Davis (diameter 0,5 inches, type 2559). As a result of the received data processing levels of noise intensity in dB in 24 of 1/3 octaves frequency bands from 25Hz up to 20kHz with linear averaging during 30 sec. were determined.

All measurements were carried out at independent circumferential shifts of IGV and S to N positions in regular intervals distributed on one S vanes pitch
(all N2 points of measurements). The matrix is formed from the received noise levels for the given 1/3 octava band

D=(Dlk); l, k=1, 2, …, N,

where k-th column corresponds to shift of S relative IGV on value ■ hs (h„ is the IGV and S pitch) in a direction of R rotation, and the data in l-th line correspond to shift of IGV concerning its starting position on value • hs in the same direction.

According to the reverberation chamber theory elements of matrix D are connected to noise intensity Llk by a ratio

Dik = 101g^, (1)

Lo

where L0 is some normalized level of the noise intensity (~10-13kgm/s. m.). According to (1) relative level of the noise intensity, received for various l and k, it was estimated by value size

N N

where L = – p – J2 (10’°"’ )1/10.

l=1 k=1

As have shown experiments noise level in a plenum for all 1/3 octava bands at given mutual circumferential position of the IGV and S (the parameter v) essentially depends on coordinate Y, determining their common circumferen­tial position. In the given researches circumferential position of the struts, lo­cated between IGV and an entrance aperture in a working part, did not change. Therefore it is possible to expect that inflience of coordinate Y is connected to shift of pair rows IGV and S concerning struts, vortical wakes for which are intensive enough.

The received results are presented on Fig.3 (assembly 1) and Fig.4 (assem­bly 2). On the figures values є = є (v) for 1/3 octava bands are given with the basic frequencies: 1000 Hz (Fig.3a, 4a), 1250Hz (Fig.3b, 4b), 2500Hz (Fig.3c, 4c) and 4000Hz (Fig.3d, 4d). Besides on Fig.3e and 4e the results, received for total noise in a range of frequencies 25-20000Hz are presented.

The choice of the specified 1/3 octava bands was determined by what in them dependences є = є(^) at fixed value Y were the most essential, whereas in other bands such dependences were absent. The data are represented as de­pendences

Єо (v )

It allows to estimate both an infbence of the IGV and S shift relative each other (parameter v), and total infbence of their joint shift concerning struts. On each of Figs.3 and 4 the value is given

N N 1=1 k=1

determining average value of the noise, appropriate to the given band of fre­quencies.

Apparently from the presented results, the most essential dependence of radiated noise intensity from stator clocking is observed in 1/3-octava bands of the frequencies, containing the R blade passing frequency (1266,7 Hz) and multiple to it frequencies. It testifies that optimization of the stators mutual po­sitioning allows to lower first of all tone noise, which basic source are pressure pulsations on the R blades. It is typical that in the specified bands of frequen­cies the most intensive noise is observed, and dependence from v of total noise (Figs.3e and 4e) is similar to dependence for 1/3 octava band, containing the R blade passing frequency (Fig.3b and 4b).

As one would expect, all dependences є = є (v) are more essential to as­sembly 2, where all clocking effects are stronger [6]. The attention appreciable enough (especially for assembly 2) dependence of relative intensity of a noise from stators clocking v for 1/3 octava band with the basic frequency 1000Hz pays to itself. In this connection it is necessary to notice that, besides pulsa­tions of pressure on the R blades, stators clocking in system of rows stator – rotor-stator essentially changes intensity of periodic, sweeped by a fbw the vortexes descending with the R blades [5].

If to assume, that periodic vortexes, sweeped by a ft>w in a turbulent vorti­cal wake, generate sound waves, that taking into account velocity of vortexes movement downwards on a ft>w concerning the motionless observer, Doppler frequency of generated sound waves appears equal:

where vx is axial ft>w velocity behind the R and a is a sound velocity. This frequency is in 1/3 octava band with the basic frequency 1000Hz.

Thus, results of measurements of noise intensity, radiated by system of rows IGV-R-S to an entrance plenum, shows that it depends on mutual circumfer­ential position of the IGV and S. Thus infhence of the stators clocking on intensity of radiated noise is caused by change of the tone noise, generated by
pressure pulsations on the R blades and periodic vortexes, descending with the R blades.