You are on page 1of 7

On the Modelling of Noise Generation in Corrugated Pipes

Hugh Goyder
Craneld University, Shrivenham, Swindon SN6 8LA, UK e-mail: h.g.d.goyder@craneld.ac.uk

The offshore oil and gas industry uses corrugated pipes because of their exibility. Gas owing within these pipes interacts with the corrugations and generates noise. This noise is of concern because it is of sufcient amplitude to cause pipework vibration with the threat of fatigue and pipe breakages. This paper examines the conditions that give rise to the large noise levels. These conditions, for the occurrence of noise, are investigated using an eigenvalue approach, which involves the effect of damping due to losses from the pipe boundaries and pipe friction. The investigation is conducted in terms of reection conditions and shows why only few of the very many possible natural frequencies are selected. The conditions for maximum noise response are also investigated by means of a nonlinear model of vortex shedding. Here, an approach is developed in which the net power generated by each wavelength is calculated. DOI: 10.1115/1.4001977

Introduction

Gas owing within a pipe, which has internal corrugations, may generate large noise levels. The noise is of concern to the oil and gas industry because it may be of sufcient amplitude to cause pipework vibration that may lead to fatigue after only a few hours of operation. Pipework failure with the consequent loss of gas is generally unacceptable and, thus, a good understanding of the noise generation mechanism is essential. Corrugated pipelines are used because they are exible. They are called risers when they connect between a platform on the sea-surface and equipment on the seabed. Such pipes may be 500 m or 1000 m in length and around 0.2 m in diameter. When these pipes are used on the seabed they are known as jumpers and may be shorter or longer than risers. Short corrugated pipes, for example 20 m in length, are known to produce noise as well as longer lengths. An illustration of a corrugated pipe is shown in Fig. 1, which is based on ISO 13628-11 1. The gas ow within the pipe passes over the carcass, which is designed to keep the pipe open should the pressure within fall to zero. The carcass, Fig. 2, which forms the corrugations, includes pockets or cavities, which contain stagnant gas. A shear layer forms between the moving uid within the pipe and the stagnant gas in the cavity and it is the oscillation of these shear layers that is responsible for the generation of the noise. The production of noise from a corrugated pipe seems to have been rst identied by Petrie and Huntley 2 and then by Ziada 3. More recent publications are by Belfroid et al. 4, by Debut Antunes and Moreira 5, by and Tonon et al. 6. The effect on structural vibration has been described by Goyder et al. 7. Practical experience of noise production shows that there is generally no problem at low owrates but that with increasing ow a threshold is exceeded and for higher owrates noise will be produced. The noise is periodic with a dominant fundamental frequency. As the ow rate is steadily increased the noise frequency remains locked-on to a single frequency but will jump to a new frequency, typically a higher frequency, when the ow rate is sufciently increased. Similar behavior of lock-on occurs while the ow is being decreased. Oil and gas companies must make a decision on how to operate should the noise be produced. Some decide to operate only below the threshold velocity for the onset of noise. Others decide to operate with the noise present and then ensure that their equipContributed by the Pressure Vessel and Piping Division of ASME for publication in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received October 30, 2009; nal manuscript received June 2, 2010; published online July 21, 2010. Assoc. Editor: Njuki W. Mureithi.

ment will withstand the noise and vibration. It is worth noting that subsea jumpers may make a noise but that this may not be apparent to the operators, thus, possibly allowing a potential damage source to go unnoticed. This paper addresses the problem of how a corrugated pipe can be assessed for conditions that lead to unacceptable noise. There are two main difculties. i A corrugated pipe is usually terminated by a very complicated system of pipework at each end. This conguration makes computer modeling very difcult because of the differing length scales involved; a very long length of corrugated pipe with many short lengths of pipe in the terminations. This problem is tackled rst. ii Only limited data is available concerning the source strength and behavior. A typical problem is the need to interpolated and extrapolate from a few measurements of corrugation noise. A basis for such extrapolation is given. Both these problems are tackled in a novel manner in this paper.

The Noise Source

It is well known that ow over a cavity containing stagnant uid causes oscillations in the shear layer, which forms between the moving and stagnant uid 8. In this application, the moving ow is within the pipe and the stagnant ow is within the cavities of the carcass. The oscillations of the shear layer, sometimes called vortex shedding set up an axial acoustic wave within the pipe. The acoustic wave causes all the shear layers on all the cavities to become synchronized to the acoustic wave. The shear layers and acoustic wave form feedback loops in which the shear layer is a source for the acoustic wave and the acoustic wave assists the formation of the shear layer. Only axial waves are considered here. The possibility of high frequency waves with wavelengths of the order of a pipe diameter could be an additional problem but the associated frequencies are beyond those which have been observed within the work considered here. The basic equation that relates the frequency of shear layer oscillation to the ow velocity is due to Rossiter and is described in detail by Howe 8. The equation is St = fW U0 1

where St is the Strouhal number, f is the frequency of oscillation, W is the width of the cavity, and U0 the ow velocity in the pipe. This equation has been checked experimentally by Belfroid 4. He treated the width as a free parameter and determined which dimension best ts the data. Additionally, he determined best values for the Strouhal number although he noted that the cavity width is not a simple quantity and that some account may have to be taken of the rounding of the cavity edges. AUGUST 2010, Vol. 132 / 041304-1

Journal of Pressure Vessel Technology

Copyright 2010 by ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 03/10/2014 Terms of Use: http://asme.org/terms

1 1 2 4 5 6 7 8 7 8 6 2 3 3 3 4 3 5

Fig. 1 Construction of corrugated pipe: 1 antifriction layer, 2 outer layer of tensile armor, 3 antiwear layer, 4 inner layer of tensile armor, 5 back-up pressure armor, 6 interlocked pressure armor, 7 internal pressure sheath, and 8 carcass

The oscillations of the shear layer are considered to be a volume velocity source of the monopole or dipole type. The sources are illustrated diagrammatically in Fig. 3. Both these types of source will generate acoustic wave in the pipe. The recent work of Tonon et al. 6 suggested that the dipole source is more likely to occur.

The Acoustic Boundary Conditions

Acoustic plane waves propagate in both directions within the pipe. A resonant condition is set up when the waves are reected
Cavity

Shear Layer Flow

Fig. 2 Corrugated pipe carcass showing cavity and shear layer

from boundaries at each end. The consideration of boundary conditions forms in important part of this investigation because it is found that the boundary conditions are responsible for determining which acoustic natural frequencies will be selected by the vortex shedding mechanism. Typically, a pipe carrying ow cannot have closed or open ends and consequently it is not immediately obvious how standing waves are formed. As an illustrative, and not untypical, example the conguration shown in Fig. 4 will be investigated. Here a side branch at the left-hand end forms one boundary while the change in cross-section at the right-hand end provides a second boundary. These are simple boundary conditions. In a real conguration, they can be much more complicated with tens or hundreds of side branches and changes in cross-section. A method for handling such boundary conditions is required. The boundary conditions for this conguration or any more complicated conguration are best described by means of reection coefcients. A reection coefcient is the ratio of the pressure in the reected wave to the pressure in the incoming wave. At the right-hand end, in this example, the change in cross-section has a reection coefcient 2 given by

Flow

Fig. 3 Diagrammatic representation of pipe, ow, cavities, and shear layers. The sources from the cavities can be either monopoles shown as straight arrows or dipoles shown as curved arrows.
S1 L1 Side branch S Corrugated pipe L Change in cross section S2

Fig. 4 A corrugated pipe of length L and cross-sectional area S operates between a side branch of length L1 and cross-sectional area S1 and a T-junction where the pipe cross-sectional area changes to S2.

041304-2 / Vol. 132, AUGUST 2010

Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 03/10/2014 Terms of Use: http://asme.org/terms

1.000 0.500 Reflection Coefficient

fn =

nc , 2L

n = 1,2,3. . .

0.100 0.050

0.010 0.005

0.001 0

4 L1 c

10

where f n is the nth natural frequency in hertz. For a long pipe, such as those being considered here, the natural frequencies calculated using Eq. 5 are closely spaced. Thus, with a speed of sound of 400 m/s typical for methane and L = 500 m, the spacing of natural frequencies is 0.4 Hz. Thus, there are apparently very many natural frequencies for the vortex shedding mechanism to select. However, only very few are observed in practice. A more detailed analysis that includes the effects of energy loss from boundaries may be performed by means of an eigenvalue calculation. The starting point is to note that the waves in the pipe may be modeled by p = Aeikx + Beikx 6

Fig. 5 Modulus of reection coefcient for a side branch plotted on logarithmic scale. Here, the area ratio S1 / S = 0.1 and L1 is the length of the side branch. The peaks in the reection coefcient coincide with resonances in the side branch.

2 =

1r 1+r

where r is the ratio of new area to the existing area, in this case r = 2S2 / S. The two arises from the waves being able to exit the tee in two directions. For the left-hand condition the reection coefcient is given by

1 =

r r 2i cot kL1

where p is the pressure at location x along the pipe and A and B are the amplitudes of the waves propagating in the positive and negative directions, respectively. This analysis assumes that the wave motion is harmonic with frequency dependence exp+it. As before k is the wave number and is equal to / c. The same wave number is applicable for both the positive and negative waves for the case where the ow velocity is small compared with the speed of sound small Mach number. This is usually the case for the conguration of a riser or jumper but the equation would have to be modied if large Mach numbers are encountered. Two equations may now be developed using the reection coefcients. At the left-hand end x = 0 the wave of amplitude B is incoming and the wave of amplitude A is reected. Hence, the reection coefcient is related to the wave amplitudes by

where in this case r = S1 / S and the wave number k = / c. The frequency is in radians per second =2 f where f is the frequency in hertz and c is the speed of sound. The length of the side branch is L1. Both reection coefcients have a magnitude that lies between 0 and 1. They may also have a phase. The reection coefcient for the right-hand end is constant with frequency while for the left it is dependant on frequency. This dependence is illustrated in Fig. 5. It can be seen that the reection coefcient is small except at frequencies that correspond to a standing wave in the side branch. These occur when kL1 =

1 =

A B

In contrast at the right-hand end x = L the wave with amplitude A is incoming with the reected wave having amplitude B. Also, at this end, the waves are inuenced by their exponential term, thus,

2 =

BeikL AeikL

L1 = 2n 1 , c 2

n = 1,2,3, . . .

These equations may be organized into a matrix form as follows:

Thus, the left-hand boundary is a good reector only at some frequencies and will allow waves to escape at most other frequencies. In allowing waves to escape from the pipe, the boundaries act in a way that is similar to a damping devicethey allow energy to leave the corrugated pipe. Reection coefcients can be calculated using proprietary computer codes, which can take into account complicated pipework arrangements. Note that the whole pipework system does not have to be modeled. It is sufcient to form two models each of which embrace only that pipework, which is in the vicinity of each boundary of the corrugated pipe. This avoids the need to introduce the corrugated pipe in the acoustic simulation, thus, avoiding the need to have a long length scale, the corrugated pipe, and a short length scale for the side branches in the terminations.

2 e2ikL


A B = 0 0

In a more general analysis the terms on the right-hand side would contain acoustic source terms. As in a standard eigenvalue problem the determinate of the matrix must be zero for values of A and B to exist with no source terms. Thus, e2ikL 12 = 0 10

By moving the second term to the right-hand side, taking the log of both sides and remembering that the log function is multivalued the last equation may be written as kL =

L i = ln 12 + n, c 2

n = 1,2,3. . .

11

Eigenvalue Analysis

What are the natural frequencies of the acoustic wave in the pipe conguration of Fig. 4? Or more generally what are the natural frequencies of a pipe bounded by two discontinuities modeled by reection coefcients? A typical formula for a pipe with closed or open ends gives the natural frequencies as Journal of Pressure Vessel Technology

For the special case of 1 and 2 real and equal to 1 full reections Eq. 5 is recovered. However, for the case where there is only partial reection then complex values must be used for . In particular, this gives rise to complex frequencies, which are the poles of the eigenvalue problem. Hence,

r n=

c 1 arg12 + n L 2

12

AUGUST 2010, Vol. 132 / 041304-3

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 03/10/2014 Terms of Use: http://asme.org/terms

4 i L c

q U0 q p1 U0

Imaginary part of natural frequency

p2
2

a
1

Fig. 7 A cavity with a shear layer. a The shear layer produces a monopole source with uid pumped pistonlike into and out of the cavity. b A dipole source with the shear layer apping in a hingelike manner.
0 200 400 600 r L c 800 1000 Real part of natural frequency

Fig. 6 Complex natural frequencies poles of system in Fig. 4 with L / L1 = 100, S1 / S = 0.1, and S2 / S = 2. The horizontal axis is the real part of the natural frequency and the vertical axis is the imaginary part.

the right half plane. In summary, the fact that only a few poles have small damping explains why only a few resonant frequencies and not all those suggested by Eq. 5 are observed in practice.

Attenuation With Distance

in =

c ln12 2L

13

i where the complex frequency is written as n = r n + in superscript r for real and i for imaginary. Note that 1 and 2, as well as having magnitude of, at most 1, have negative real values and, thus, have restricted phase angles. In Eq. 12, the value of n will, therefore, outweigh the argument of the reection coefcients for all but the rst few n and, thus, Eq. 5 is still approximately true. The damping is described by Eq. 13 and is dependant on the log of the magnitude of the product of the reection coefcients. It is interesting to consider what happens as one of the reection coefcients drops to zero. If both coefcients start as 1.0 then the poles lie on the real axis. As the product of reection coefcients get smaller the poles move into the complex plane according to Eq. 13. As the product drops to zero, Eq. 13 shows that the poles have moved to innity. This limiting case corresponds to an anechoic condition for which there are no acoustic natural frequencies. Thus, Eqs. 12 and 13 remain valid for all boundary conditions. In particular, note that due to damping the poles will lie off the real axis with those poles with least damping being closest to the axis. For the particular case of the conguration of Fig. 4 the location of the poles eigenvalues is given in Fig. 6. Here, for the side branch model, L / L1 is 100 and S / S1 = 10 while for the tee S / S2 = 1. An exact calculation has been made using Eq. 10. Note that the curve is made up of points where each point corresponds to an eigenvalue. This calculation conrms that there are a large number of closely spaced natural frequencies as suggested by Eq. 5. The relationship between Fig. 6 and Fig. 5 should be noted. The poles lie along a curve that is the same as that in Fig. 5 but is inverted. Examination of Eq. 13 conrms that this observation is completely correct and provides a simple method for determining the poles of the conguration. The acoustic damping ratio can be calculated from the pole values by dividing the imaginary part of the pole by the real part. As illustrated in this example and as found in practice, very few poles have small imaginary part and, thus, small damping. It is only at those natural frequencies where the reection coefcient is large and little energy is lost through the boundaries that the damping is small. It is these poles that are most susceptible to the vortex shedding mechanism. The shear layers provide energy to the system in the sense of negative damping. Thus, the addition of negative damping will cause the poles to cross the axis and go to the unstable half of the complex plane. In this Fourier formulation the unstable half of the complex plane is the lower half. This is in contrast to the Laplace formulation where the unstable half is

As well as energy lost at the boundaries a further source of energy dissipation is attenuation along the pipeline. Belfroid et al. 4 consider various types of attenuation and use a standard model in which Eq. 6 is modied with the introduction of an attenuation parameter. The modied equation for the pressure now reads p = Aexeikx + Bexeikx
1

14

where is a small parameter of order 0.01 m . The simplest method for dealing with this additional source of damping is to make k and, hence, complex. This enables the previous eigenvalue calculation to proceed as before but now there is an additional complex term. With this term Eq. 13 is modied to

in = c

c ln12 2L

15

This merely moves the imaginary part of each eigenvalue further upward from the horizontal axis. Thus, the shear layer mechanism must provide negative damping that overcomes attenuation with distance as well as losses from the boundaries to move the poles to the unstable half of the complex plane. For the acoustic system to produce noise it is necessary for it to become unstable and for the poles to move to the unstable half of the complex plane. The vortex shedding mechanism provides negative damping and those poles closest to the real axis can be moved across the axis allowing instability. Poles deep within the stable part of the complex plane with corresponding large damping will not become unstable because the negative damping mechanism provided by vortex shedding cannot overcome the large damping values. The eigenvalue analysis enables these vulnerable poles to be identied. The next step is to determine the strength of the source and how it provides negative damping.

Source Analysis

The mechanism of vortex shedding across a cavity has been investigated in many circumstances. The phenomenon is nonlinear and involves the rolling up of the shear layer to produce vortices. For small amplitudes the shear layer oscillations add energy to the system and, thus, act as a source of negative unstable damping. However, for large amplitudes the shear layer absorbs energy and, thus, acts as positive damping. The noise in the pipe evolves in response to these nonlinear energy sources and sinks. There are at least two ways in which the shear layer can interact with the acoustic wave. These alternative interactions are illustrated in Fig. 7. If the lump of uid adjacent to the shear layer and in the mouth of the cavity oscillates in a pistonlike manner, then each cavity provides a monopole source. Such sources have been frequently observed in deep cavities where resonance within the cavity causes large amplitudes of motion in the mouth. Alternatively, the shear layer and the adjacent uid can rotate, possibly Transactions of the ASME

041304-4 / Vol. 132, AUGUST 2010

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 03/10/2014 Terms of Use: http://asme.org/terms

a
1.0 0.5 0.5 0.9 0.0 0.8 0.4 0.1 0.056 0.7 1.0 0.6 0.0 0.5 1.0

0.5

0.32 0.18 0.5 0.4 0.1 0.056

0.5

0.5

Imaginary

1.0

0.032 0.018 0.01

Imaginary

0.3

1.0

Real
0.0 0.5 1.0

0.032 0.018 0.01

0.3

Real
1.0 0.5 0.0 0.5

1.0

0.5

Fig. 8 The nondimensional acoustic impedance of a shear layer from Graf and Ziada. The real part is shown on the horizontal axis while the imaginary part is shown vertically. The spiral lines correspond to constant acoustic amplitude normalized by ow velocity. The radial lines correspond to constant Strouhal numbers. Part a includes smaller acoustic amplitudes only while part b includes larger acoustic amplitudes.

apping in a hingelike manner from the separation edge of the cavity. In this case, each cavity acts like a dipole source. Possibly a mixture of these two motions can occur. It should be noted that for the long wavelengths being considered there is no resonance within the cavity and the oscillations of the uid in the mouth must simply compress the gas in the cavity, which will behave in a springlike manner with no inertia. Recent experiments by Tonon et al. 6 suggested that the dipole mechanism is more likely in which a grazing ow interacting with the shear layer and coordinated by an acoustic plain wave in the pipe produces a dipole source. Tonon et al. are able to make an estimate of the noise produced in their system based on computer simulations of a shear layer. This work is very useful in identifying the fundamental cause of the noise in corrugated pipes. The approach adopted here starts from the engineering circumstances where some noise data is available and it is necessary to interpolate and extrapolate to other similar circumstances. For this case a simple noise model is required that will facilitate this process. The noise model uses an energy approach and is based on ideas that have been developed for the monopole case. Although a dipole source is being considered, it is assumed that the behavior is analogous to the monopole case and this seems reasonable since both are consequences of the oscillation of a shear layer over a cavity. In particular, the key issue to model is the source strength, which must increase when the acoustic velocity is small and then decreases and become a sink as the acoustic velocity becomes large. A very complete set of data for the monopole source has been generated in experiments by Graf and Ziada 9. This data is shown in Fig. 8. The data model the source strength in terms of both the Strouhal number and the acoustic velocity. This data have received some recent conrmation with numerical simulations by Martinez-Lera and Schram 10. The data in Fig. 8 are expressed as a complex impedance with the real part which is in-phase with the velocity describing the energy supplied or removed by the source and the imaginary part out-of-phase with the velocity describing the extra stiffness loading of the source. It is anticipated that the data for the dipole source will be similar. However, only limited proprietary data from actual occurrences is available. The modeling starts by assuming that resonant conditions are present in the pipe with an acoustic wave that is sinusoidal in both space and time. An appropriate frequency to consider would be one that the eigenvalue analysis has identied as having small damping. In this circumstance, the net energy supplied by all the Journal of Pressure Vessel Technology

sources in each corrugation must balance the energy dissipated by the damping. Furthermore, when the acoustic amplitude is small the shear layers must act as a source providing negative damping and when the amplitude is large the shear layers must act as a sink absorbing energy. The data of Graf and Ziada provide exact modeling of this positive or negative feedback for the monopole source. Figure 9 shows an example of this behavior for the case of a Strouhal number of 0.4. This is an appropriate Strouhal number for corrugated pipes. The data are extracted from Fig. 8 and a quadratic curve tted. The source, thus, has the form s=A u u B U0 U0

16

where s is the source strength, u is the acoustic velocity, and U0 the velocity of the mean ow. The parameters A and B are constants, which may easily be determined for the monopole case from the Graf and Ziada data. For the case being considered here the two parameters A and B must be determined from measured data. The key assumption is that the quadratic form is similar for both the monopole and dipole cases. The source model here is

Acoustic impedance u U0 f St,u U0

0.05

0.00

0.05

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Normalised acoustic velocity u U0

Fig. 9 Real part of acoustic impedance of a shear layer for a Strouhal number of 0.4 as a function of acoustic velocity u divided by mean ow velocity U0. The function fSt, u / U0 is obtained, together with the points from Fig. 8. The smooth curve is a t of Eq. 16 to this data.

AUGUST 2010, Vol. 132 / 041304-5

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 03/10/2014 Terms of Use: http://asme.org/terms

only for the real part of the source impedance the imaginary part is available for the monopole source but not for the dipole case. The imaginary part expresses the change in the speed of sound due to the presence of the corrugations. Fortunately, for an energy analysis only the real part is needed. The effect of all the cavities must be combined to give the amplitude of the acoustic wave. The approach is to assume that the response to the shear layer in each cavity is local with each shear layer responding to the local acoustic velocity. Along one wavelength of the standing wave there are a range of conditions changing from pressure antinodes with zero acoustic velocity to pressure nodes with maximum acoustic velocity. The model is formed by working out the net effect of all the carcass cavities along a single representative wavelength. The acoustic velocity depends on the location along the standing wave. If there are N cavities along one wavelength of the standing wave then the velocity at location n is given by sin 2 n u=U N

1.0 Acoustic Velocity 0.5 0.0 0.5 1.0 0 1.5 Cavity Power 1.0 0.5 0.0 0 2 1 Cavity Power 0 1 2 3 4 0

20 b

40

60

80

100

20 c

40

60

80

100

17

is the maximum acoustic velocity along the wavelength. where U The acoustic power generated by the dipole in each cavity is given by 1 u wn = s 2 U0

20

40

60

80

100

1 U = s 2 U0

Cavity number

sin2 2

n N

18

where the source s is modeled above in Eq. 16. Note that s also depends on u and, thus, on Eq. 17. It can be seen that the power generated by each dipole is a very complicated expression depending on the location within each wavelength and the local value of acoustic velocity. The energy analysis proposed here considers one complete wavelength of a standing wave. For each ow velocity it is necessary to see if any of the poles with small damping could with the addition of negative damping, be drawn into the unstable half of the complex plane. The rst step is, therefore, to determine how much energy is generated by the sources. Figure 10 illustrates a conguration in which a standing wave embraces 100 cavities. Here the parameter B in Eq. 16 equals 0.48. A calculation of the power delivered by each cavity has been determined using Eqs. 1618. In Fig. 10b, all the sources are delivering power but those located in the region of maximum velocity have a reduced output because they are becoming saturated. The power is normalized to the case of maximum power with the power from each source being given as a percentage of the maximum power. The case in Fig. 10c is an extreme limiting case where there is no net power produced with some of the cavities adding power sources and others absorbing power sinks. To determine the operating conditions, the excess acoustic power produced by the combination of sources and sinks must be sufcient to equal the damping provided by the losses at the terminations of the pipe and any attenuation with distance. This condition gives the noise produced by the pipe.

Fig. 10 a The normalized acoustic velocity of a wave with 100 cavities in one wavelength. b The power generated by each cavity for the conditions of maximum power. c Saturated conditions where the positive power from sources is balanced by power absorbed in sinks. The cavity power is given as a percentage of the maximum total power that can be produced in one wavelength.

Discussion

The analysis involving reection coefcients has identied a method for determining both the frequencies and damping of the pipework system. This provides the description of the system as in any stability analysis. The damping is simply related to the loss of energy at the ends of the pipe with small damping values occurring when the reection coefcients are close to 1.0. The fact that reection coefcients are infrequently close to 1.0 explains why so few acoustic frequencies become unstable despite there being very many candidates one every 0.4 Hz in the example illustrated above. The determination of reection coefcients for a complex pipework system is straightforward and commercial software is available. Thus, this novel method can be applied to practical 041304-6 / Vol. 132, AUGUST 2010

systems of considerable complexity. The approach has proved successful when applied to complex pipework systems that have been modeled theoretically. Such calculations have been found to agree with observations. One approach when making theoretical investigations and identifying which acoustic modes have small damping is to suggest changes in those side branches, which generate large reection coefcients. These side branches, because they are resonant, are often vulnerable to vibration and structural modications to prevent vibration may be necessary. However, if such side branches can be shut by valves or made shorter then there is the possibility of removing the resonance frequency completely. The calculation of system damping can be elaborated by including attenuation with distance. The comparison of damping due to the length of the pipe and damping due to losses at the pipe ends is a straightforward calculation and can establish the relative importance of the pipe length and boundaries. For very long pipes it may be possible to neglect the losses at the pipe ends in comparison to the attenuation with distance. Another effect, which has not been included here but which is easy to incorporate is the change in sound speed due to the presence of the corrugations. Belfroid 4 gave a formula. The effect is straightforward to determine and is based on the compressibility of the gas in the cavities in much the same way that pipe distensibility is taken into account in water hammer calculations. The output from the eigenvalue calculation is the damping of the system. This provides one portion of the information needed in an assessment of the energy balance. The second portion is the energy supplied by the sources. When these two portions are combined it is possible to determine at what ow velocity the noise will start and the amplitude of the pressure. The model suggested for determining the acoustic velocity amplitude is a standard nonlinear approach. It is similar to the work of Debut and Antunes 5 who used a simplied form of nonlinear model. Their model has been used to investigate the effects of Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 03/10/2014 Terms of Use: http://asme.org/terms

nonperiodic separation of corrugations. The novel approach in this paper is to develop a method for conditions where there are very many sources. It should be stressed that the shear layer data from Graf and Ziada was collected on a different system a monopole source to that considered here and that consequently it is not directly applicable to the dipole source created by the corrugations. However, the general form of the data are likely to be similar. Examples of equivalent earlier data are that of Coltman 11 and Elder 12, where both show a spiral form for the shear layer impedance. Until equivalent data can be collected for a corrugated pipe this data provides a good basis for understanding and modeling shear layers. However, what is needed is a good set of data for the dipole case that mirrors the very full set of data, which has been obtained by Graf and Ziada for the monopole case.

S cross-sectional area of pipe m2 S1 cross-sectional area of side branch m2 S2 cross-sectional area of each pipe in tee junction m2 St Strouhal number U0 gas ow velocity m s1 maximum acoustic velocity in pipe m s1 U c f fn i k n p r s u wn x z n speed of sound m s1 frequency Hz natural frequency the nth Hz imaginary unit 1 wave number= / c integer, 1, 2, 3 acoustic pressure Pa area ratio for reection coefcient acoustic source for one cavity acoustic velocity in pipe acoustic power from one cavity Nm/s distance along pipe m acoustic impedance kg m4 s1 attenuation coefcient m1 reection coefcient gas density kg m3 frequency radians per second= 2 f natural frequency in radians per second

Conclusions
The following conclusions may be drawn. 1. The acoustic losses at the boundaries of a corrugated pipe provide an effect equivalent to damping. The boundary conditions, including this damping effect may be usefully modeled using reection coefcients. An eigenvalue analysis then shows how poles with small damping are related to reection coefcients that are close to 1.0. 2. Although a long pipe has very many natural acoustic frequencies, only a very few are excited in practice. The selected frequencies are those for which the reection coefcients are large and the damping is small. 3. A full mathematical development for the acoustic poles of a riser has been formulated enabling these to be calculated from a simple equation once the reection coefcients are known. See Eqs. 12, 13, and 15. 4. When use is made of a detailed shear layer model it is seen that within one wavelength there will be locations that add energy to the system and other locations that remove energy. A general model for one wavelength is constructed that gives the net energy production or absorption. 5. An energy approach is suggested as a method for determining noise from corrugated pipes. The terms in the energy equation include losses at the boundaries, absorption due to friction and production and dissipation at cavities. This approach is applicable when some data is available and there is a need to interpolate or extrapolate. It is clear that much more data is needed to enhance and validate the proposed energy method.

References
1 ISO 13628-11, Petroleum And Natural Gas IndustriesDesign and Operation of Subsea Production SystemsPart 11: Flexible Pipe Systems for Subsea and Marine Applications, 2007-09 15. 2 Petrie, A. M., and Huntley, I. D., 1980, The Acoustic Output Produced by a Steady Airow Through a Corrugated Duct, J. Sound Vib., 701, pp. 19. 3 Ziada, S., and Buhlmann, E. T., 1991, Flow Induced Vibration in Long Corrugated Pipes, International Conference on Flow-Induced Vibrations, IMechE, Brighton, UK. 4 Belfroid, S. P. C., Swindell, R., and Tummers, R., 2008. Flow Induced Pulsations Generated in Corrugated Tubes, Ninth International Conference on Flow-Induced Vibration, Prague, Czech Republic, Jun. 3Jul. 3. 5 Debut, V., Antunes J., and Moreira M., 2008, Flow-Acoustic Interaction in Corrugated Pipes: Time Domain Simulation of Experimental Phenomena, Ninth International Conference on Flow-Induced Vibration, Prague, Czech Republic, Jun. 3Jul. 3. 6 Tonon, D., Landry, B. J. T., Belfroid, S. P. C., Willems, J. F. H., Hofmans, G. C. J., and Hirschberg, A., 2010, Whistling of a Pipe System With Multiple Side Branches: Comparison With Corrugated Pipes, J. Sound Vib., 329, pp. 10071024. 7 Goyder, H. G. D., Armstrong, K., Billingham, L., Every, M. J., Jee, T. P., and Swindel, R. J., 2006, A Full Scale Test for Acoustic Fatigue in Pipework, ASME PVP, ICPVT, pp. 119377. 8 Howe, M. S., 2004, Acoustics of Fluid-Structure Interactions, Cambridge University Press, Cambridge. 9 Graf, H. R., and Ziada, S., 1992, Flow Induced Acoustic Resonance in Closed Side Branches: An Experimental Determination of the Excitation Source, ASME Symposium on Flow-Induced Vibration and Noise, ASME PVP, New York, Vol. 247. 10 Martnez-Lera, P., Schram, C., Fller, S., Kaess, R., and Polifke, W., 2009, Identication of the Aeroacoustic Response of a Low Mach Number Flow Through a T-Joint, J. Acoust. Soc. Am., 1262 582586. 11 Coltman J. W., 1968, Sound Mechanism of the Flute and Organ Pipe, J. Acoust. Soc. Am., 444 983992. 12 Elder S. A., 1978, Self-Excited Depth-Mode Resonance for a Wall-Mounted Cavity in Turbulent Flow, J. Acoust. Soc. Am., 643 877890.

Acknowledgment
The author acknowledges the support of Craneld University at the Defence Academy of the United Kingdom.

Nomenclature
A, B A, B L L1 N amplitudes of pressure waves in pipe Pa constants in source equation. length of corrugated pipe m length of side branch m number of corrugations in one wavelength

Journal of Pressure Vessel Technology

AUGUST 2010, Vol. 132 / 041304-7

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/ on 03/10/2014 Terms of Use: http://asme.org/terms

You might also like