EFFECT OF CONTROL VALVE NOISE ON MULTI-PATH ULTRASONIC GAS FLOWMETERS GUIDANCE TO METER USERS

Transcription

1 EFFECT OF CONTROL VALVE NOISE ON MULTI-PATH ULTRASONIC GAS FLOWMETERS GUIDANCE TO METER USERS A Report for National Measurement System Directorate Department of Trade & Industry 151 Buckingham Palace Road London, SW1W 9SS Report No: 2003/240 Date: December 2003

2 The work described in this report was carried out under contract to the Department of Trade & Industry ( the Department ) as part of the National Measurement System s Flow Programme. The Department has a free licence to copy, circulate and use the contents of this report within any United Kingdom Government Department, and to issue or copy the contents of the report to a supplier or potential supplier to the United Kingdom Government for a contract for the services of the Crown. For all other use, the prior written consent of TUV NEL Ltd shall be obtained before reproducing all or any part of this report. Applications for permission to publish should be made to: Contracts Manager TUV NEL Ltd Scottish Enterprise Technology Park East Kilbride G75 0QU jduff@nel.uk Tel: +44 (0) TUV NEL Ltd 2003

3 NEL East Kilbride Glasgow, G75 0QU, UK Tel: Fax: EFFECT OF CONTROL VALVE NOISE ON MULTI-PATH ULTRASONIC GAS FLOWMETERS GUIDANCE TO METER USERS A Report for National Measurement System Directorate Department of Trade & Industry 151 Buckingham Palace Road London, SW1W 9SS Prepared by: Mr C Coull Mr R J Whitson Approved by: Mrs J A Sattary Date: December 2003 for Mr M Valente Managing Director Report No: 2003/240 Page 1 of 17

4 SUMMARY This report provides guidance on the effect of control valve ultrasonic noise on ultrasonic meters. The guidance is aimed at the operators of gas ultrasonic meters and the designers of metering systems that include gas ultrasonic meters. The guidance is split into two sections, the first considers the effect at the design stage and how its impact can be reduced or eliminated. The second section considers how the effect can be reduced when a meter already in operation is affected. Report No: 2003/240 Page 2 of 17

5 CONTENTS SUMMARY. 2 1 INTRODUCTION 4 2 EFFECT OF ULTRASONIC NOISE ON AN ULTRASONIC METER 4 3 DESIGN STAGE ISSUES. 4 4 OPERATIONAL ISSUES GUIDANCE SUMMARY 12 6 REFERENCES 14 APPENDIX A - PROPOSED NOISE PREDICTION METHOD. 15 APPENDIX B - PROPRIETARY METER PERFORMANCE PREDICTION METHOD. 17 Page Report No: 2003/240 Page 3 of 17

6 1 INTRODUCTION Multi-path transit time ultrasonic meters are increasing being used in the natural gas production and transmission industries. The inherent benefits of the technology are being recognised by industry. However it is evident that these meters can be sensitive to the noise generated by control valves. Reported cases have shown meters becoming inoperable for periods whilst in service. Responding to this industrial metering difficulty and funded by DTI s Flow Programme, NEL has produced this guidance document for ultrasonic meter users, operators and system designers. This document gives advice on how to reduce the risk of valve meter interference occurring at the design stage and gives possible solutions to in-service meters that are being affected. The guidance is based on the results of a study, which included a review of the published literature and a survey of meter operators, valve and meter manufacturers. 2 EFFECT OF ULTRASONIC NOISE ON AN ULTRASONIC METER Transit time ultrasonic meters operate by transmitting short duration ultrasonic pulses through the flow and measuring the time that it takes the pulse to go from the transmitter to receiver i.e. the transit time. The gas velocity can be determined by measuring the difference in upstream and downstream transit times. Transducer Transducer Figure 2.1 Transit Time Ultrasonic Flowmeter In order to measure the transit time the meters electronics must be able to detect the arrival of the ultrasonic pulse at the receiving transducer. Ultrasonic noise at the meter s transmitting frequency can make detection of the received signal difficult, potentially causing the meter to either misread or fail. 3 DESIGN STAGE ISSUES The issue of potential metering difficulties due to valve noise is best addressed during the design stage of a metering facility project. The issue is best avoided completely by ensuring that any type of pressure reduction, control, throttling valve is well separated from the ultrasonic meters in the system. Report No: 2003/240 Page 4 of 17

7 3.1 Metering Installation Selecting the best locations for the meters and valves is a very important aspect to consider during the design. If these are chosen well then the issue of ultrasonic noise need not be considered again. When deciding on the meter and valve locations the following points should be considered: Control valves, throttling valves or pressure reduction valves should be located well apart from the ultrasonic meter. A large unit of process equipment such as an exchanger or tank located between a control valve and the ultrasonic meter is a very effective way to protect the meter. The process equipment is very effective at damping out ultrasonic noise. Straight pipe is very poor at damping ultrasonic noise, pipefittings such as bends and tees are much more effective. One hundred meters of straight pipe has been shown to be as effective as one 90 degree bend in ultrasonic noise damping [2]. Upstream of the valve is the preferred position for the meter. Firstly more noise usually travels downstream although this is not always the case, and secondly the line pressure will be higher, which results in improved meter signal amplitude. In order to increase confidence in a proposed piping design it may be worth carrying out some noise prediction calculations. Appendix A describes a proposed method of predicting ultrasonic noise within the pipe. The method uses control valve noise prediction method from the British / European process control valve standard [1]. This standard determines the audible noise levels outside the pipe however it may be possible to predict the noise levels inside the pipe at the meter by using part of this standard in combination with other techniques. 3.2 Meter Choice All gas ultrasonic meters are different; this section reviews the various meter features that can affect how well a meter will cope with ultrasonic noise Transducer Operating Frequency Ultrasonic transducers generally operate at a specific resonant frequency (F o ). Figure 3.1A shows a typical frequency-response curve for an ultrasonic transducer. The transducer transmits and receives ultrasound only over those frequencies. Hence, when operating as an ultrasonic receiver, it acts like a bandpass filter only responding to frequencies about the operating frequency. In this way the USM is only susceptible to noise at the transducer s resonant frequency. Control valves generate a broad spectrum of noise; if this spectrum overlaps the transducer frequency as shown in figure 3.1A then, depending on the amplitude of the noise, the meter may be affected. If the signal-to-noise ratio drops by enough then the meter will fail to locate the arrival of the signal and fail to give a reading unless further more sophisticated signal-processing methods are employed. Figure 3.1B shows how, in principle, the control-valve noise can be avoided by increasing the transducer frequency. Report No: 2003/240 Page 5 of 17

8 Transducer Frequency Response Curve Valve Noise Spectrum F o Freq F o Freq Fig 3.1A Meter Likely to be Affected Fig 3.1B Valve Noise Avoided Figure 3.1 Influence of Transducer Operating Frequency The literature reveals that many meters operate at, or about, 100kHz, which is in the range that throttle valves can potentially emit. In many cases raising this operating frequency to 200kHz or above moves the signal away from the main noise frequencies, improving the noise immunity of the meter. It should, however, be noted that higher transducer frequencies inevitably lead to lower signal amplitudes due to poorer transmission between the solid transducer and the flowing gas. This in turn could have consequences for meter operation if the received signal amplitude is too low Signal Processing Techniques Signal averaging or stacking is a very effective way of reducing unwanted random noise. The technique essentially stores successive arrival window traces and then averages these. Since the noise is random, its magnitude is diminished by the square root of the stack size (the number of traces that have been averaged) whereas the signal, which is constant in the traces remains. The disadvantages to this technique are two fold; firstly, the response time of the meter is reduced, as a function of the stack size. Secondly, any rapid change in flow rate could potentially cause the meter to misread or fail due to movement of the signal during successive traces, leading to signal degradation when the averaging is applied. Another technique that might be expected to be effective at locating a transmitted signal in noisy conditions is Digital Correlation. As shown in Figure 3.2 this technique involves applying a digital copy of the transmitted signal and the received signal to a digital correlation algorithm. The time at which these correlate to the highest degree, i.e. the correlation function reaches a maximum, is taken as the signal arrival time. Digital Copy of Transmitted Signal Digital Correlation Algorithm corr Received Digital signal t Figure 3.2 Concept of Digital Correlation Technique Report No: 2003/240 Page 6 of 17

9 3.2.3 Path Length The shorter the path length the greater the signal amplitude at the receiving transducer, and a larger received signal means the path will cope with greater noise amplitude. In effect this means that the longer paths in multi-path meters, such as those that using internal pipe wall reflections, will be more susceptible to ultrasonic noise than the shorter paths Transducer Power As with path length, the greater the received signal amplitude the more noise a meter can tolerate. Transmitting a stronger signal will result in a stronger signal being received Transducer Pocket Design The effect of ultrasonic noise may be reduced by locating the transducers in deep pockets. If it can be assumed that the noise propagates directly down the pipe, then the transducers may be shaded from the noise to some extent by the deep pockets. The benefits of this shading may, however, be outweighed by the disadvantages of the long dead leg within the pocket. These include increased risk of debris or liquid accumulation at the transducer face and possible dead leg pressure and temperature effects on the transit time. 3.3 Valve Choice From the limited amount of experimental data on ultrasonic noise inside the pipe, it can be said that all types of control valve, i.e. globe, ball etc, can potentially generate enough ultrasonic noise to affect ultrasonic meters. The experience of one meter manufacturer suggests that ball valves may be less noisy than other types of valve. However, there is insufficient data available to allow conclusions to be drawn or recommendations given as to which type of valve is quieter in the ultrasonic frequency range. General comments on valve trims are given in Section Users should consult their valve manufacturer for details of the in-pipe acoustic spectrum generated by their valve under normal operating conditions. 3.4 Other Potential Sources of Ultrasonic Noise This guide is primarily interested in the noise generated by control / throttling valves and pressure regulators etc. However it has been noted that other items of in-line equipment, (under certain circumstances) can, generate ultrasonic noise to a level high enough to affect ultrasonic meters. These include flow conditioners and filters such as a Witches Hat used during pipeline commissioning. One meter manufacturer advises chamfering the corners of flow conditioners as a way of reducing the conditioner s potential for generating ultrasonic noise. Report No: 2003/240 Page 7 of 17

10 4 OPERATIONAL ISSUES Problems with ultrasonic noise can and do arise in existing installations: this could be due to several reasons, a lack of consideration at the design stage of a project, a subsequent modification to the process plant or service conditions, or possibly the replacement of an original non-ultrasonic meter with a USM. 4.1 Indicators of Effect on Meter It is important to be able to identify the cause of meter failure or poor performance. The main indicator of a USM being affected by ultrasonic noise is the signal-to-noise ratio diagnostic. Ultrasonic noise will reduce the signal-to-noise ratio diagnostic, if the ratio is low (as specified by the meter manufacturer) on all meter paths then noise is likely to be causing a problem. If the diagnostic software shows a discernible difference between the signal-to-noise ratios of the upstream and downstream transducers, then this indicates almost certainly that ultrasonic noise is affecting the meter. In fact, it is possible to determine from which direction the noise originates, as the transducers facing the source will be affected to a greater extent. Low signal-to-noise ratio on individual transducers is more likely to be caused by a malfunction in the transducer or in the excitation electronics, so these aspects should also be considered. If there is a throttling valve in the vicinity of the meter, with a differential pressure drop ( p/p 1 ) of 0.25 or greater, then this may be generating ultrasonic noise, even if it is quiet in the audible range. Although 0.25 p/p 1 is generally backed up by the literature as a threshold at which valve noise increases markedly, it should be noted that several cases reviewed in the industry survey showed meter failures at much lower differential pressure ratios. It should be noted that noise might affect meter performance without causing it to fail. Attention to meter diagnostic functions is essential in assessing whether a meter is affected. 4.2 Options to Reduce or Eliminate the Effect If a meter is affected by ultrasonic noise, then the following range of options are worth considering, in some cases one option may be enough to resolve the problem, in others a combination of options may be required Modify Installation Modifying the pipework between the meter and the valve can lead to a reduction in ultrasonic noise at the valve. The key point here is that it is not so much the distance between the meter and valve that is important but what fills that distance. Where possible, install the meter upstream of the valve. If a valve does generate ultrasonic noise, the levels are likely to be highest on its downstream side, although this is not always the case. This, coupled with the fact that the meter s ultrasonic pulses will be stronger due to the higher pressure upstream, indicates that upstream of the valve is the better location for an ultrasonic meter. Attenuation of ultrasonic noise travelling in the fluid inside a straight pipe is negligible (or very low) and consequently direct lines of sight (along the flow path), between the meter and valve should be avoided regardless of the length. Many forms of standard circuit component, such as bends, elbows and tee-pieces, can help attenuate the noise, so should be placed between the meter and the valve. Table 4.1 shows the attenuation that can be expected for a variety of component types. The effects of multiple fittings are additive (e.g. 1 tee + 1 elbow will give 15 db, which is a reduction ratio of 5.6 : 1). The pipe-elements silencer mentioned in Table 4.1 was made Report No: 2003/240 Page 8 of 17

11 up of 2 elbows plus 2 blind-tees, which collectively would be expected to give 30 db, which is close to the 33 db obtained. The effect of combining components in out-of-plane configurations seems to provide some added benefit. The out-of-plane bends in Table 4.1 may have been made up from two elbows so the expected effect would have been 10 or 12 db, which is slightly less than the 14 db achieved. Tees should be used in a blind configuration as shown in Figure 4.1. Note that the arrows indicate the direction of noise and not necessarily the flow direction. Filters will also act as attenuators. No figures are available for the attenuation that might be expected, although Vermeulen et.al. [2] propose an assumption that db may be achievable. Good No good Figure 4.1 Blind Tee Orientation Expanders can also be used to reduce the noise level: the reduction ratio achieved is directly proportional to the expander s diameter ratio. E.g. An expansion diameter ratio of 2:1 will give an acoustic pressure reduction of 2:1 (= 6 db). TABLE 4.1 ATTENUATION OF VARIOUS COMPONENTS BY VERMEULEN ET.AL.[2] Component Attenuation (db) Reduction ratio 100m straight pipe : 1 Elbow to 2 : 1 Blind Tee : 1 Two out-of-plane bends 14 5 : 1 Tube bundle (heat exchanger) : 1 Perforate plate silencer : 1 Pipe elements silencer : 1 Standard turbine meter to 17.8 : 1 High-speed turbine meter 5 1 : 0.56 Conversely, reducers should be avoided since they will tend to concentrate or amplify the noise again, by a ratio directly proportional to the reducer s diameter ratio. E.g. a reduction in diameter of 2:1 will give an acoustic pressure increase of 2:1 (= 6 db). Some caution should be exercised when applying the attenuation figures for a given component since each of the figures given in Table A.1 is an overall level. All devices will have some form of frequency-dependent attenuation characteristic and, consequently, simply adding figures together may be an over-simplification of the beneficial effect obtained at the USM. When considering adding noise-attenuating components such as bends and tees it is important to ensure that adequate straight length is installed directly upstream of the meter, to ensure the flow profile entering the meter is fully developed. Report No: 2003/240 Page 9 of 17

12 4.2.2 Change Meter Transducers As stated previously, the basic problem is being able to distinguish the meter s pulse or ultrasonic signal from other ultrasonic noise in the system. There are two possible options that involve changing the meter s transducers. either shift the meter s transducer frequency upwards away from the valve noise or, increase the power of the transmitted ultrasound. Shifting the frequency to 500 khz has been shown to overcome problems with valve noise. Ultrasonic noise from control valves has a frequency spectrum that has a haystack shape that typically peaks around 100 khz and drops-off either side. Levels can still be significant at frequencies as high as 250 khz. Shifting the signal from 250 khz to 500 khz may gain 10dB or more in signal-to-noise ratio. Simply boosting the transmitted power may be an option but may not be feasible with specific transducers or with the meter s excitation electronics. Control systems on some meters already have built-in algorithms to adjust beam power automatically when required to improve signal-to-noise ratio. Depending on the meter configuration, the manufacturer may be able to provide alternative transducers with a suitable power output. It should be noted that increasing the transducer power might affect the EEx (explosion protection) rating of the meter. For example, if the transducer power is increased, the energy limitations for a meter that employs Exi (intrinsic safety) protection may be exceeded Alter Meter Signal Processing Options (if available) US valve noise is generally broadband in nature and ultrasonic meter frequencies are specifically tonal in nature, hence the two are uncorrelated. Averaging, or stacking (as some authors refer to it) of such a mixture of tonal and random signals is a common approach used in the acoustics field to improve signal-to-noise ratio and differentiate tones. The meter may already perform averaging but there may be some scope within the meter s control software to increase the averaging period. One disadvantage of increasing the averaging is that it may slow down the rate at which readings are output and make the meter response slower. It is recommended that the meter manufacturer be consulted before making any changes of this kind Alter Process Conditions or Operation It is generally realised that higher differential pressure across a valve leads to increased noise generated from that valve. Several references from the literature review, highlight the fact that noise from control valves increases markedly as the valve pressure ratio ( p/p 1 ) exceeds Therefore in may be possible to silence or considerably reduce the ultrasonic noise emitted from a suspect valve by altering process conditions or operations to reduce the pressure drop at the valve. One aspect worth considering is how much flexibility exists in being able to manage the plant operation and whether a change in schedule or routine could be beneficial. For example, valve noise may be intermittent in that a specific valve may only be used during fixed times or perhaps a number of plant operations are being carried out simultaneously that could be phased so that flow or pressure demand is flattened out. Report No: 2003/240 Page 10 of 17

13 It may be possible to share the pressure load between valves. The one nearest the meter taking the smallest share. This may mean having to control two or more existing valves, but these may already be installed in the system. These options are likely to be used as temporary measures while permanent solutions are considered Silencing Devices As an alternative solution to carrying out what may be major modifications to plant pipe layouts, an in-line silencing device can be installed between the meter and valve. Depending on the design, the silencer may only need to be sandwiched between flanges or may be in the form of a spool piece. A number of companies make such devices, specifically designed to reduce ultrasonic noise to improve USM performance. Claims of db attenuation and more are typical and independent laboratory trials of one manufacturer's devices at NEL justified their claims of between db attenuation. The performance of a given device will depend on its design geometry. The manufacturer should be able to provide performance figures and the recommended installation distances from the meter and noise source. Some designs of silencer will in themselves generate high-frequency noise at high flow velocities, so it is best to keep velocities inside the silencer below a certain design value as specified by the manufacturer. Ideally, noise performance should be given as an attenuation-versus-frequency curve. This is a more precise way of specifying the attenuation characteristics and provides a ready means of checking how much attenuation might be achieved at the particular frequency used by the ultrasonic flowmeter. It is, however, common for only an overall value of noise reduction to be quoted. This is less helpful than a frequency characteristic, but gives some indication of the noise reduction to be expected. However, care should be taken not to rely too heavily on a single-figure value and to allow for perhaps, 10dB or so more noise damping than is actually required. Whenever possible, users should seek to obtain an attenuation-versus-frequency characteristic for the silencer rather than simply an overall noise-reduction value. It should also be noted that instances of silencers blocking up and failing mechanically have been reported Valve Modification This means either changing or modifying the valve for a low-noise design. In general, the problem of high ultrasonic noise levels from valves appear to be associated with standard technology valve designs, where no specific action has been taken in the valve design to reduce noise. Some manufacturers produce special low-noise designs or trims that can be retro-fitted to current valves. Mostly these either split the pressure drop into stages, fragment the flow stream into smaller jets or do combinations of both. These low-noise valves have generally been designed to reduce audible noise from the valve (i.e. noise below ~18kHz) but there is evidence to show that, with the correct design, valve noise in the ultrasonic region is also reduced The published literature seems to suggest that valves with single-stage multi-hole trim designs shift the audible noise into the ultrasonic range. However this frequency shifting Report No: 2003/240 Page 11 of 17

14 effect is not common to all trim designs, the frequency spectrum generated by a trim depends much on the specific design of the trim. More complex trims for example ones which incorporate both staged pressure drops and multi-hole concepts may provide better noise reduction at all frequencies, but there is little experimental evidence to support this view at present. No further comment can be made on valve trims since further work to evaluate the ultrasonic noise characteristics of different designs of valve trims is required. 5 GUIDANCE SUMMARY Gas ultrasonic meter performance can be seriously affected when installed in proximity to control valves. This can lead to total failure of the flowmeter. The presence of ultrasonic noise can be identified by a low signal-to-noise ratio on all meter paths. The transducers facing the source of the noise will be affected to a greater extent than those facing away from the noise. In this way the direction of the noise can be determined. USMs should be positioned well apart from control valves, preferably separated by large process equipment such as exchangers, tanks, vessels etc. If an ultrasonic noise problem is identified in an existing system, then the following options should be considered. The options are not listed in any order of preference. It may also be necessary to use one or more options in order to resolve the issue. Increase meter transducer frequency Increase meter transducer power Install noise-damping equipment in the noise path blind tees, out-of-plane bends etc Install an in-line silencing device Modify process conditions reduce differential pressure across control valve Use signal-processing techniques Stacking, Digital Correlation Reposition valve or meter separated by process equipment etc A popular option is to increase the transducer frequency. This will require changing transducers, for smaller high-frequency transducers, which may require a line shut down. However, the effectiveness of this solution can not be determined without measuring the noise frequency spectrum at the meter. Ultrasonic noise is attenuated to a far greater degree by pipe fittings such as bends and blind tees than straight pipe. Bends and blind tees can be retrofitted into a system to act as a silencer. However, it is important that the fittings do not impinge on the meters required upstream straight length. Since these fittings will distort the flow profile, and affect the meters calibration. It is recommended that the noise-damping figures detailed in Table 4.1 should only be used as a general guide since no details of the frequency dependence of these items is given. In-line silencing devices can be highly attenuative for their size, making them worth consideration where space is limited. However, it should be recognised that their use negates one of the major advantages of USMs, namely their non-intrusive nature. The silencers intrusive nature may, in some designs, make them prone to blockage or mechanical failure, hence it is important that careful consideration be given to the specific silencer design before proceeding. It should also be noted that exceeding the maximum design velocity of the silencer might actually cause the silencer to generate ultrasonic noise. Adding an appropriately designed trim to the noisy valve should reduce the level of noise at all frequencies. However, more research is required in this area, as there are many different types of trim and almost none of them have been tested for ultrasonic noise generation within the pipe. Report No: 2003/240 Page 12 of 17

15 Reducing the differential pressure across a suspect valve will generally reduce the noise emitted from the valve at all frequencies. In general reducing the valve pressure ratio ( p/p 1 ) below 0.25, significantly reduces valve noise. Report No: 2003/240 Page 13 of 17

16 6 REFERENCES 1. BS EN Ed.2:2000, (Harmonised from IEC ) Industrial process control valves - Part 8-3: Noise considerations - Control valve aerodynamic noise prediction method. October Vermeulen, M.J.M., DeBoer, G., Bowen, J., A model for estimation of the ultrasonic acoustic noise level emitted by pressure regulating valves and its influence on ultrasonic flowmeters. Instromet Bruggeman, J.C., Hopmans, L.J.M., Blijie, C.J.M., Van-Veghal, H., A comparison of prediction methods for valve noise with experimental results. Proceedings of Internoise 1996, Liverpool. 4. van Bloemendaal, K., Sloet, G.H., Hosten, B., Brassier, P., Vulovic, F., Ultrasonic meters and noise. Task 4 of the Second GERG Project on Ultrasonic Meters, GERG, Stoll, P., Slawig, H., Muller, C, de Boer, G., Vermeulen, M., Ultrasonic noise characteristics of valves with respect to ultrasonic gas flowmeters. 16 th North Sea Flow Measurement Workshop, paper 7, Scotland, October 1998 Report No: 2003/240 Page 14 of 17

17 Appendix A Proposed Noise Prediction Method The prediction of valve noise levels is complex and BS EN :2000 [1] is the latest European Standard giving methods of predicting overall audible noise levels on the outside of a pipeline. The prediction methodology firstly calculates the acoustic pressure inside the pipe and then accounts for wall transmission loss to determine the audible noise outside the pipe. Therefore, by using the initial part of the standard, the method can predict overall sound pressure levels inside the pipe. Despite being targeted at audible noise, published work by Bruggeman et.al. [3] shows that the 1995 version of the standard is also useful in predicting ultrasonic noise on the outside of pipes. It can therefore be speculated that it would be useful for predicting ultrasonic noise inside the pipe. The prediction process is only taken as far as obtaining the peak frequency and the internal sound-pressure level, L pi, in the downstream pipe (as given by equation 36 of the standard). Once the internal sound-pressure level has been calculated the effects of pipe fittings between the meter and valve can be taken into account using the values detailed in Vermeulen et al. [2]. These values are shown in Table A.1. TABLE A.1 ATTENUATION OF VARIOUS COMPONENTS BY VERMEULEN ET.AL.[2] Component Attenuation (db) Reduction ratio 100m straight pipe : 1 Elbow to 2 : 1 Blind Tee : 1 Two out-of-plane bends 14 5 : 1 Tube bundle (heat exchanger) : 1 Perforated plate silencer : 1 Pipe-elements silencer : 1 Standard turbine meter to 17.8 : 1 High-speed turbine meter 5 1 : 0.56 One significant shortcoming of BS EN [1] is that it contains no guidance on the spectrum shape expected from the valve once the internal sound-pressure level, L pi, and the peak frequency, f p, have been calculated. Observation of the ultrasonic spectral shapes obtained in experimental work by various authors, [2], [3], [4] and [5], all show spectra that fall off more rapidly above the peak than below. The fall-off below the peak frequency is an average of 6 db/octave and above the peak is an average of 12dB/octave. More research is needed before a firm conclusion can be drawn from this but these results would seem to indicate that the expression given by (1) may provide a noise spectra prediction suitable for the ultrasonic range. This expression determines the sound-pressure-level correction factor. L i = 10 x log10 {[1+(f i /(2f p )) 4 ]x[1+(f p /(2f i )) 2 ]} (1) Where L i = Sound Pressure Level Correction Factor f i = Frequency of calculated sound pressure level f p = Spectrum s peak frequency. The sound pressure level at the required frequency is then calculated: L pi, f = L pi - L i (2) Where L pi = Overall sound pressure level, L pi,f = level at octave band centre frequency, f, Report No: 2003/240 Page 15 of 17

18 Figure A.1 shows the result of (1) in graphical form Correction Factor Li (db) Frequency Ratio fi/fp Figure A.1 Proposed Frequency Spectra Once the above corrections have been added for spectral shape, the calculation of valve noise at the meter is complete. The level in the octave band closest to the meter s operating frequency can be used to evaluate the meter s signal-to-noise ratio. It should be noted that the results of this prediction method have not been verified and therefore the results cannot be guaranteed. However, based on existing knowledge, this is the best prediction method available without using proprietary data. It is recommended that the noise-level prediction obtained from this method be discussed with the meter manufacturer prior to purchasing a meter. Report No: 2003/240 Page 16 of 17

19 Appendix B Proprietary Meter Performance Prediction Method Vermeulen et.al. [2] have developed a technique to estimate the noise-to-signal ratio of their flowmeter and hence predict its performance. It is based on estimating the valve noise at the flowmeter, N UFM, using a valve specific weighting factor, N v, in combination with a system attenuation factor, N d, by: N UFM = N d. N valve = N d. N v. P Q (3) This is then compared to the Signal strength, S UFM, of a path in the flowmeter: Where P = Pressure at meter T = Transit Time L = Path Length The noise-to-signal ratio, δ, is then simply: S UFM = [P T]/L (4) δ = N UFM /S UFM (5) A series of tests have been performed to determine the specific weighting factor (N v ) for specific valve types under various valve conditions. Further tests using broadband microphones have given rise to the system attenuation factors for various pipe fittings. These results are shown in Table A.1. A further series of tests have been performed on the proprietary meter to determine the δ critical value. δ critical is defined as the maximum noise-to-signal ratio that the meter can be expected to tolerate. The meter is expected to operate acceptably with noise-to-signal ratios below δ critical, above δ critical the meter is likely to fail, although it is stated that it may operate due to the safety margins applied to the model. Hence the software that has been developed from this work is used to predict whether specific valve and meter locations are feasible. Although this technique could be applied generally, all the experimental work has been done by the meter manufacturer using their meter and the valve weighting factors are not publicly available. Hence this is proprietary software, and cannot be applied to other meters. Vermeulen et al state in [2] that the software has been proven to be adequate ; however, it should be noted that the technique has not been subjected to independent verification. Report No: 2003/240 Page 17 of 17