Flow Measurement and Instrumentation 13 (2002) 125142 www.elsevier.

com/locate/owmeasinst

Guidelines for the use of ultrasonic non-invasive metering techniques

M.L. Sanderson , H. Yeung
Department of Process and Systems Engineering, School of Engineering, Craneld University, Craneld, Bedford MK43 0AL, UK Received 1 July 2002; received in revised form 7 July 2002; accepted 17 July 2002

Abstract This paper provides a comprehensive set of Guidelines for the application of clamp-on transit time ultrasonic owmeters to a wide range of industrial ows. These Guidelines have been drawn up in conjunction with users and manufacturers and sponsored by the United Kingdoms Department of Trade and Industry. They represent the best practice to be used for the application of this technology to liquid metering. The Guidelines identify the range of possible non-invasive technologies which can be employed for the measurement of pipe ows and installation, pipework, uid and operational effects on clamp-on transit time ultrasonic owmeters, together with effects which may be specic to particular manufacturers. The paper concludes with the identication of further work which needs to be undertaken to strengthen the Guidelines. 2002 Elsevier Science Ltd. All rights reserved.

1. Introduction 1.1. Non-invasive ultrasonic metering technologies One of the signicant advantages in employing ultrasonic techniques is that when they are applied to the metering of liquid, they cannot only be used non-intrusively, in that the sensors may be in contact with the liquid but do not intrude into the ow path, but can also be used non-invasively, that is with the transducers mounted onto the outside of the pipework. Such methods are commonly referred to as clamp-on ultrasonic techniques, since the transducers are clamped or held onto the outside of the pipework. These methods enable measurements to be made without breaking into the pipework and, therefore, measurements can be made where for reasons of safety, hygiene, continuity of supply or cost it is not possible to break into the pipework. The method also provides a basis for checking existing meters. Non-invasive ultrasonic metering techniques can thus be used as a temporary, semi-permanent or permanent method of measurement.

1.2. The origin of the guidelines The Guidelines For The Use Of Ultrasonic Noninvasive Metering Techniques were originally written by Prof. Mike Sanderson and Dr. Hoi Yeung in the Department of Process and Systems Engineering in the School of Engineering at Craneld University under a contract issued by the Department of Trade and Industry for the National Measurement System Policy Units 19992002 Flow Programme. The intention in drawing up these Guidelines was, in the light of increasing use of this technology, to identify and promulgate best practice in its application and also provide material to inform national and international standards. The Guidelines have been drawn up under the guidance of a Steering Group representing users and manufacturers of noninvasive ultrasonic owmetering techniques. They attempt to identify the full range of factors including installation effects, pipework, uid effects, operational effects and manufacturer specic effects which are likely to affect the performance of ultrasonic non-invasive meters. They quantify the likely effect of these factors; identify best operational practice for such meters and provide estimates of the overall uncertainty which can be achieved when employing best practice. The Guidelines are based on our best current understanding of the technology identied from the experience of users and

Corresponding author. Tel.: +44-1234-754696; fax: +44-1234750728. E-mail address: m.l.sanderson@craneld.ac.uk (M.L. Sanderson).

0955-5986/02/$ - see front matter. 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 9 5 5 - 5 9 8 6 ( 0 2 ) 0 0 0 4 3 - 2

126

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

manufacturers and on data currently available in the open literature. The papers identied in the Bibliography have provided the supporting data for the guidelines. 1.3. Scope and content of guidelines This section provides a brief overview of the complete contents of the Guidelines. 1.3.1. Possible non-invasive ultrasonic technologies In employing non-invasive ultrasonic technologies there are three potential technologies which can be employed. These are the Transit Time Ultrasonic Flowmeter (TTUF), the Doppler Ultrasonic Flowmeter (DUF) and the Cross Correlation Ultrasonic Flowmeter (CCUF). The scope of this Guideline is restricted to the use of TTUFs, since such owmeters are likely to give rise to a metering technique with the widest range of application and the highest accuracies and lowest uncertainties. Section 1.4 provides a description of the noninvasive TTUF, typical application areas, the performance measures which should be used in assessing the meter performance and the likely performance in terms of error, repeatability and reproducibility when used under ideal ow, pipe, uid and operational conditions. Because of the need to distinguish between the different ultrasonic technologies and their applications, Sections 1.5 and 1.6 provide a comparative overview of Doppler Ultrasonic Flowmeters and Cross Correlation Ultrasonic Flowmeters. 1.3.2. Installation effects TTUFs clamped on to the outside of the pipework are effectively single diametrical beam TTUFs and as a consequence suffer the velocity prole problems and errors associated with similar spool piece designs with non-intrusive sensors. These problems and errors arise as a consequence of variations in the velocity prole of a fully developed ow as Reynolds Number is varied and also as a consequence of upstream pipework congurations such as bends and valves. The effects of these are considered in Section 2. Section 2.1 provides a guide as to suitable distances downstream of disturbances which should ideally be employed to minimize the level of additional error introduced into the measurement. Estimates of the likely effects on the accuracy of the meter of not achieving these distances are provided in Section 2.2. A method of further reducing the errors of disturbed ows by measuring in more than one diametrical plane is identied in Section 2.3. 1.3.3. Pipework effects In addition to the effects of upstream disturbances on clamp-on ultrasonic owmeters, additional uncertainties arise as a consequence of their use with a wide variety of pipe sizes, pipe materials, wall thickness and lining

materials. Section 3.1 provides guidelines as to the range of pipe sizes and types with which the technology can be applied. Measurements are often made on pipes which have been in service for sometime and in some cases can have suffered corrosion and erosion. Section 3.2 deals with the effect of pipe wall roughness on the measurement performance of the meter. Section 3.3 identies the ultrasonic properties of the pipework which are required to be supplied by the user. 1.3.4. Fluid effects Clamp-on TTUFs can also be used with a wide variety of liquids with varying viscosity and density. The effects of these are considered in Section 4. For clean uids there are two effects. The uid dynamic effects caused by the variation of Reynolds Number with density and viscosity, and the effect of varying speed of sound on the performance of the meter. These are considered in Sections 4.1 and 4.2. Although TTUFs are meant for operation with clean uids, it is possible that such uids may have small amounts of entrained air bubbles or solids associated with the ow. The effects of these are identied in Sections 4.3. Section 4.4 deals with limitations caused by highly attenuative liquids. 1.3.5. Operational effects The overall performance of the meter is likely to depend signicantly on the operational conditions under which the meter is employed. Section 5.1 identies the need for the equipment to be used by trained users if accurate measurements are to be achieved. The transducers can be mounted in such a way so that a single, double or quadruple traverse of the pipe by the ultrasonic beam is used to undertake the measurement. The relative merits and drawbacks of each of these is considered in Section 5.2. The location of the transducers with respect to the pipe methods for obtaining the correct positioning of the transducers and methods of clamping the transducers to maintain their positions are considered in Sections 5.3, 5.4 and 5.5. Operational methods to reduce the effect of ow prole disturbance are identied and discussed in Section 5.6. The ultrasound produced by the transducers has to be well coupled to the pipework. The preparation of the pipework and the use of an appropriate couplant is important to obtaining an accurate measurement. A range of couplants is available depending on the temperature of the application and whether the installation is temporary or semi-permanent. Guidelines for the preparation of the pipework and the application of the couplant are provided in Section 5.7. The process temperature has an impact on the choice of couplant and also transducer selection. These are discussed in Section 5.8. Section 5.9 provides guidelines as to the selection of the frequency of operation of the ultrasonic transducers with differing pipe size. Clamp-on TTUFs require a knowledge of the inner diameter of the pipe. This is

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

127

obtained by requiring the user to input the pipe outer diameter and the pipe wall thickness. Methods available for obtaining the pipe outside diameter and the pipe wall thickness, including the use of ultrasonic thickness gauges are identied in Section 5.10. The likely effect of errors in either of these measurements is included in this Section. 1.3.6. Manufacturer specic effects The overall accuracy achievable in addition to showing a dependency on the factors identied above will also depend on items which are manufacturer specic. These include the method the manufacturer employs to measure the transit time difference, the compensation, if any, provided for Reynolds number and the speed of sound in the liquid. Guidelines as to the best practice in these are identied in Sections 6.16.3. Condence in the measurement can be enhanced by the provision of a series of diagnostic tools. The range of possible diagnostic tools is discussed and compared in Section 6.4. 1.4. Clamp-on transit time ultrasonic owmeters (TTUF) Fig. 1 shows the elements of a typical clamp-on TTUF. The meter consists of three elementsthe transducers, the clamping arrangement and the signal processing and user interface electronic package. The transducers are piezoelectric devices which generate ultrasound which penetrates through the pipewall, and which is then transmitted along the paths shown within the uid. The clamping arrangements enable the transducers to be in good sonic contact with the pipe wall and at the correct distance for the particular pipe and liquid application. The signal processing electronics

measures the difference in time between ultrasonic beams passed upstream and downstream, as a consequence of the ow of the liquid, and converts this to a volumetric owrate which is displayed on the user interface typically either in numeric or graphical format on a user interface, which also provides the means of entry of application specic data by the user. 1.4.1. Operation of TTUFs Fig. 2 shows the principle of operation of the clampon TTUF measuring the owrate of a liquid whose speed of sound is cl in a pipe of internal diameter D. Although different manufacturers undertake the measurement in a variety of ways and employ different algorithms to compute the owrate from the measurements they make, the following gives, in generic form the basis of the measurement. The owrate is measured by using the transit time difference between ultrasound travelling from transducer T1 to transducer T2 and when the ultrasound is travelling from transducer T2 to transducer T1. If the ultrasonic beam in the uid is at an angle q to the pipe axis then the volumetric owrate Qv is related to the time difference T T12 T21 by:Qv k.p.D.c2. T l 16.cotq (1)

where cl is the speed of sound in the liquid, D is the internal diameter of the pipe and where k, the correction factor, is given by:k vpipe vbeam (2)

where vpipe is the actual average velocity in the pipe, and vbeam is the average velocity measured along the beam. If the wedge on which the transducers are mounted is made of a material for which the speed of sound is cw and having an angle a to the pipe axis, then Eq. (1) can be re-written as:k.p.cl.cw.D. Qv cl.sina 2 . T cw 16.sina 1

(3)

Fig. 1. A typical permission.)

clamp-on

TTUF.

(Panametrics

used

with

By measurement of the transit time difference, T, together with knowledge of factors specic to the measurement, namely the internal diameter of the pipe and the speed of sound of the liquid, input by the user (or measured on line) and internally held variables such as the k factor, the wedge angle and the speed of sound in the wedge it is possible to provide a measurement of the volumetric owrate of the liquid. In order to obtain an accurate measurement it is necessary to obtain an accurate measurement of the required separation of the transducer T1 and T2. From simple geometric considerations this separation, s, in the case of a single path meter to be:-

128

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

Fig. 2.

Schematic of a clamp-on transit time ultrasonic owmeter.

2.sw

2.sp

2.sl

(4)

where sw is the separation through the wedge, sp is the separation through the pipe and sl is the separation through the liquid. This can be shown to be given by:cp.sina cw 1 cl.sina cw 1 cl.sina cw
2

2.h.tana

2.t

cpsina cw

(5)

The application areas for clamp-on TTUFs where they show signicant advantages over techniques are those where the pipe cannot be cut for reasons of safety or hygiene or possible process contamination; where the cost of breaking into the pipework is too high; where reasons of continuity of supply installation, maintenance or service need to be carried out without a shut down in the process and on larger pipes where the cost benet over other in-line meters is greatest. 1.4.3. Performance measures for TTUFs TTUFs generally provide outputs indicative of volumetric owrate or totalised volume ow. In assessing the performance of the meter, three measures are of concern. These are the percentage error of the indicated volumetric owrate or totalised volume ow, the repeatability of the measurement and the reproducibility of the measurement. In employing these terms it is important that they should correspond to standard usage. The following provides standard denitions of the three terms. The percentage error of the volumetric owrate measurement is given by:E Vind Vstd 100% Vstd (6)

2.D.

The correct separation of the transducers can be obtained by providing the owmeter with the pipe wall thickness, t, and internal diameter, D, and the speed of sound in the pipe wall, cp, material and the speed of sound in the liquid, cl. This can be used with internally stored values to indicate to the user the required separation. 1.4.2. Application areas Clamp-on TTUFs are designed to be used with clean liquids, although they can operate in the presence of limited amounts of either particulate matter or gas bubbles (Guidance on operating in the presence of particulate matter or gas bubbles is to be found in Section 4.3.). Table 1 drawn from manufacturers application data identies typical application areas to which clamp-on TTUFs are being applied. The largest single sector where these meters are currently being applied is in the water industry, although this does not represent the majority of applications. This is not meant to be an exhaustive list, and for specic areas of application outside those identied in Table 1, potential users should consult suppliers. Over the years, the technology has also found applications in dirty liquids.

where the Vind is the volume of liquid passed as indicated by the meter under test during the duration of the test and Vstd is the recorded reference volume passed during the same time. It is usually expressed as a function of ow velocity through the meter. The repeatability of the measurement is a measure of the ability to provide repeatable measurements under the same set of conditions. It is a measure of the random uncertainty in the measurement, R, which is dened as:n

2.83.m.sE and sE

(Ei E)2
i 1

(7)

n 1

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

129

Table 1 Application areas of clamp-on TTUFs Sector Water industry Application Temporary/semi permanent installation Checking of installed meters; leakage determination; network analysis Permanent installation Applications where hygiene, operation (eg. Supply interruption), maintenance and cost are of major consideration Permanent installations Used for measurement of crude oil after rst stage separator; ofoading of crude oil; water injection; rened oil Permanent installations Blending and batching of food and drink where issues of hygiene and contamination are of major concern Permanent installations Process applications of high purity and aggressive liquids where measurements can be made with a meter which can be serviced without having to shut down the process Temporary installations Employed as a tool for checking, servicing and maintaining existing owmeters Permanent installations Process monitoring and control of liquids where application and maintenance can be achieved without breaking into the pipework and where meter process compatibility can be achieved by using the existing pipework; pump protection Temporary/semi permanent/permanent systems Measurement of cold and hot water ows in large building complexes such as hospitals, schools, factories and ofces to monitor and control usage Heat metering and air conditioning applications

Oil industry

Food industry

Semiconductor industry

Chemical industry

Miscellaneous applications

where Ei is the error of the ith run within the test series; E is the average error of all the runs in the test series and n is the number of runs in the test series. m is a scaling factor which enables comparisons of repeatability to be made between test series which are different in length by normalising all the repeatabilities to a standard length of test run. m Ttest set Tstd (8)

tests are available it may be possible to express the reproducibility as a function of the standard deviation of the errors across the whole range of tests. In this case reproducibility would then be dened as:Reproducibility 2.83.sreproducibility where sreproducibility
n

(10)

(Ei Eall)2
i 1

where Ttest set is the average length of the test run in a particular set of runs and Tstd is the time to which the uncertainty is to be normalised. The reproducibility of the measurement is concerned with the degree to which the measurement can be reproduced if the conditions of measurement are changed. Of particular importance in clamp-on TTUFs is the degree to which the readings of the meter are reproducible under unclamping and re-clamping conditions. The reproducibility of the measurements can then be measured by measuring the difference between the average errors in the two sets of tests, one taken before unclamping and re-clamping and the other after, i.e. Reproducibility E1 E2 (9) where E1 is the average error of the initial tests and E2 is the average error of the test undertaken after unclamping and re-clamping. If more than two sets of

n 1 where Ei is the average error obtained in the ith repro ducibility test series, Eall is the error average overall the reproducibility test series, and n is the number test series undertaken to assess the reproducibility. Reproducibility is generally expressed as a function of ow velocity. 1.4.4. Likely uncertainty, repeatability and reproducibility under ideal ow pipe uid and operational conditions TTUFs are likely to be used over a very wide range of pipe sizes and liquids and because the overall performance achievable in the eld depends on the care with which the operators prepare the pipework and mount the transducers, coupled with an inability to provide traceable calibrations of clamp-on meters in the eld it is difcult to obtain gures for the uncertainty,

130

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

repeatability and reproducibility of such owmeters in the eld. The values manufacturers quote for the accuracy of clamp-on TTUFs range between 1% to 5% of reading with this performance being over a restricted ow range which is generally specied in terms of ow velocity. At the lower ow rates the performance is generally specied as a xed uncertainty expressed as a velocity which may range from 0.01 m/s to 0.03 m/s depending on the pipe size and the particular manufacturer. In addition to these accuracy specications, manufacturers often specify what is termed a calibrated accuracy. By this is meant the accuracy which is achievable in situations where the meter is capable of being calibrated in situ by some other means, or where the clampon transducers are being used as part of a spool piece. Typical values for the accuracy under these conditions range between 0.5% to 2% of reading, dependent on manufacturer and pipe size. Repeatabilities quoted by manufacturers typically lie in the range 0.15% to 0.5% of reading, depending on manufacturer. The evidence of users on performances achieved in the eld indicate that over a range of pipe and uid conditions, the specications provided by manufacturers may not always be achievable, although there is little hard evidence on which to base such judgements. These Guidelines suggest that the calibrations which have been undertaken under laboratory conditions should be taken as the basis for a judgement as to the likely performance that could be achieved in the eld under operating conditions. The results obtainable in the eld are unlikely to be better than the performances achieved in the laboratory under fully developed ow conditions, and with extreme care being taken in the mounting and positioning of the transducers. The results obtainable under non-ideal ow conditions and poor mounting of the transducers and incorrect spacing are likely to increase the error of the measurement. The calibrations on which estimates provided below have been based have been undertaken over a range of pipe materials and sizes, although the majority of publicly available calibrations have been undertaken on pipe sizes greater than 50 mm. These results indicate overall that an accuracy within 5% of reading can be obtained for velocities greater than 1 m/s for a range of pipe sizes and materials. A typical repeatability which can be obtained is 1% of reading and reproducibility under unclamping and re-clamping conditions of 2% of reading. In some applications, such as nuclear power station performance evaluation using the secondary cooling circuit, signicantly better performance has been achieved by providing a full scale model of the pipework and using this to calibrate the clamp-on TTUF on a test stand. This approach can be employed where the economic cost is justied and where the pipework material,

dimension and conditions can be expected to be accurately known. Under these conditions and where the in situ calibration is undertaken with the greatest of care, the claimed accuracies of better than 1% of reading are probably justiable. 1.5. Clamp-on Doppler ultrasonic owmeters It is important that any potential user of non-invasive ultrasonic metering techniques should be aware of the difference between TTUFs and Doppler Ultrasonic Flowmeters (DUFs) which are also commonly commercially available. The technological basis of the two is completely different and the overall accuracy which can be obtained is much poorer using DUFs. Fig. 3 shows the operation of a DUF. DUFs require gas bubbles or solid particles to act as scatterers of ultrasound. Assuming that all the scatterers are moving with a velocity v, then the Doppler shift, fd, of the transmitted ultrasound is given by:fd v 2.ft. .cosq cl (11)

where ft is the transmission frequency, cl is the velocity of sound in the liquid and q is the angle of the ultrasonic beam with respect to the pipe axis. In general the Doppler shift is not a single frequency because broadening of the spectrum occurs as the scatterers have a range of velocities; the beam has a nite width; the angle of the ultrasonic beam to the direction of liquid ow has a range of values and the effects of turbulence. An estimate has therefore to be made of the average Doppler shift from a spectrum which typically looks like that in Fig. 3. The performance of Doppler owmeters is limited, due to the nature, size, spatial of the scatterers which varies the attenuation of the ultrasonic beam. The sensed volume is not well dened. In large pipes it is likely to be close to the wall. The relationship between the sensed velocity and the mean velocity in the pipe is unknown. The velocity being sensed is that of the scatterers, and because of slippage it might not correspond to the liquid velocity. Upstream disturbances, in the form of bends, valves, pipework and probes causing vortices can all cause the meter to read in error. Errors can also be caused by vibration. In order to obtain an accurate volumetric owrate measurement, the internal diameter of the pipework needs to be measured accurately. The uncertainty with such ow meters can be high and typically is likely to be greater than 10% of reading. However, under closely dened ow conditions it is possible for the DUF to provide a repeatability of 2% of reading. Typical application areas for such meters are the measurement of raw sewage, sludge, slurries, paper pulp, tar, sands, and oil-water-gas mixtures.

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

131

Fig. 3.

Basics of Doppler ultrasonic owmeter.

Some manufacturers of clamp-on TTUFs provide dual modes of operation. In the absence of scatterers, the meters operate as conventional TTUFs. When the scatterers in the form of particulate matter or gas bubbles rise above a certain level, the meter changes its mode of operation and it then operates as a DUF meter. 1.6. Cross correlation ultrasonic owmeters Cross Correlation Ultrasonic Flowmeters (CCUFs), although not widely used are commercially available from some manufacturers. Fig. 4 shows the basis of operation of a CCUF. Modulation of the ultrasound by particulate matter, gas bubbles or turbulence is detected

at two stations axially separated by a distance, L, which is typically 0.5 D and D. The ow velocity is estimated by cross correlation of the noise signals generated by modulation phenomenon. The time delay between the arrival of similarly modulated signals provides a peak in the cross correlation function of the noise at each receiver, and provides an estimate of the time for a particular distribution of scatterers to travel along the pipe from one beam to another. In two phase ow, the signal is affected by factors such as the number of scatterers per unit volume, the distribution of scatterer sizes, their spatial arrangement and the velocity prole across the pipe. Consequently a calibration obtained under a particular set of operating

Fig. 4.

Cross correlation ultrasonic owmeter.

132

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

conditions may not apply to another. The overall accuracy of the technique should probably be reckoned to be 5%, although for a particular application where the meter can be calibrated under similar conditions accuracies of 1% of owrate or better may be expected. The application areas to which such owmeters have been employed are in the measurement of single phase liquids, liquid-solid mixtures, liquid-gas mixtures, liquid-liquid mixtures, low pressure gas and high pressure gas.

of the meter and 5D downstream of the meter. Although the downstream conditions are probably satisfactory, the upstream requirement is probably over optimistic. 2.1. Installation conditions for the above accuracies to be achieved In providing advice as to the effect of the disturbances, two approaches have been adopted. The rst is to provide estimates of distances at which the effect of the disturbance is less than 2% reading. If this assumed, together with an inherent uncertainty of the measurement of 5% and the two sets of errors add in a mean square sense, the overall effect of the disturbances at distances greater than these specied distances can be reckoned to be negligible. The data which has been used to provide these distances is based largely on the performance of wetted sensor single beam TTUFs since it is believed that this data is more comprehensive and reliable. They are generally consistent with the limited data which has been obtained from clamp-on TTUFs. Table 2 provides the distances for which a series of disturbances for which the errors of measurement are likely to be less than 2%. 2.2. Effects of upstream pipework If the upstream disturbance is known and it is not possible to obtain the required distances specied in Table 2 then the following data, which is provided in graphical form in Figs. 613, based on the data in reference [16], can be used to provide an estimate of the additional error created by the disturbance. It should be noted that in the majority of cases a single beam diametrical meter under reads the owrate. 2.3. Reducing effects of disturbed ows One method which is often suggested for improving the measurement under disturbed ow conditions is to
Table 2 Distances downstream of a disturbance for less than 2% increased uncertainty Disturbance Number of diameters required to reduce error to less than 2% 4 18 30 22 47 18 15 20

2. Installation effects on clamp-on TTUFs A clamp-on TTUF, when used with only a single beam measurement, is sensitive to velocity prole effects in both fully developed and disturbed ow conditions, because as Eq. (1) shows the device estimates the average velocity across the whole of the pipe cross section by measuring the velocity prole along the beam. Even in fully developed ow the value of k is not unity and depends on Reynolds number and pipe wall roughness. This may or may not be corrected for by manufacturers. Fig. 5 shows the k factor variation with Reynolds number for smooth pipes. A correction based on Reynolds Number may or may not be incorporated by the manufacturer. Fig. 5 shows that the corrections required are greatest in the laminar regime. In the transition regime however the correction factor may not be well determined, and this uncertainty on the correction factor for the meter may add signicantly to the overall meter uncertainty. In the case of disturbed proles, the k factor for fully developed ow is no longer applicable. The ow prole can be disturbed by upstream pipework congurations such as bends, expansion and contractions and the presence of valves and pumps. Manufacuturers often specify a requirement to have 10D of straight pipework upstream

Fig. 5. k factor against Reynolds Number for fully developed ow in smooth pipes. (Taken from reference [3], used with permission of Academic Press.)

Conical contraction Conical expansion Single 90 Bend Two bends 90 in U Two 90O bends in perpendicular planes Buttery valve 2/3 open Globe valve 2/3 open Gate valve 2/3 open

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

133

Fig. 6.

Errors downstream of a conical contraction.

Fig. 10.

Errors from two 90 bend in perpendicular planes.

Fig. 11. Fig. 7. Errors downstream of a conical expansion.

Errors downstream of a buttery valve 2/3 open.

Fig. 8.

Errors downstream of a single 90 bend.

Fig. 12.

Errors from a buttery valve 2/3 open.

Fig. 9.

Errors downstream of two 90 bends in U conguration. Fig. 13. Errors downstream of a gate valve 2/3 open.

134

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

Fig. 14. Two beam measurements.

undertake the measurements in perpendicular planes as shown in Fig. 14 and take the average of the two measurements. Some manufacturers enable this to be undertaken with simultaneous measurements from the two beams being provided using two sets of transducers and one set of processing electronics. If only one set of electronics is available, then it is necessary to provide an independent measure of the mean owrate to account for any changes which may occur in the mean owrate between the two measurements. Most manufacturers recommend that the average of the two measurements should be used as the estimate of the owrate. In the light of the above indications it is probably better to take the maximum reading as the best estimate of the owrate.

3. Pipework effects 3.1. Typical ranges of pipe sizes, pipe materials, lining materials and wall thickness to which the technology can be applied Clamp-on TTUFs are required to operate over a range of pipe sizes, materials, wall thicknesses and lining materials. Particular manufacturers will claim to be able to work with particular pipe sizes, pipe materials, wall thicknesses and lining materials and not all manufacturers will be able to provide a metering system which will operate over the complete range of pipe sizes, materials, wall thicknesses and lining materials shown in Table 3. The claim made by manufacturers is that the pipe has

to be sonically conducting, although the experience of users indicate that whilst some manufacturers may be able to work with particular pipes other cannot. The following commentary should be taken as indicative and not necessarily as denitive and identies pipe materials where the problems most frequently occur. Occlusions or pores in cast iron can attenuate the ultrasound. Concrete and cement pipe can cause problems but often if the pipe is fully soaked, the meters may work better. There is evidence that with stainless steel and GRP some manufacturers experience difculty in achieving adequate signal transmission. For some manufacturers, the pipe material, the pipe size and material may limit the particular mode in which the owmeter can be operated. (see Section 5.2). Although manufacturers do not in general specify the range of pipe wall thicknesses with which their owmeters will operate, at least one manufacturer identies a maximum pipe wall thickness and there is evidence in small diameter pipes having thick walls that there can be signicant errors. Several problem areas have been identied with coatings and pipe linings. Clamp-on TTUFs will only successfully work where the coating/lining material is bonded to the pipe material in such a way that there are no air gaps between the two materials, since air gaps will prevent successful transmission of the ultrasound. If the meter is to be clamped on to the coating material then the thickness of the coating must be accurately known in order that the pipe o.d. can be accurately determined. If the pipe is lined, then the thickness of the lining is required to be known accurately. The correct sound speeds for the wave which is being transmitted through coating/lining materials are required. Methods for undertaking these measurements are provided in Section 5.10. If there is more than one lining material, a lining material and a pipe coating, or a lining material together with a coating which has built up over a period of time, then account of all these in terms of calculation of pipe id from od and separation of trandsucers. Some manufacturers do not enable data from lining materials to be taken in separately, in which case a thickness corre-

Table 3 Ranges of pipe sizes, materials, wall thicknesses and lining materials with which clamp-on TTUFs can be used Pipe sizes over which operation of clamp-on owmeters has been claimed by manufacturers Pipe materials for hich manufacturers have used Clamp-on TTUFs 8 mm id to 6000 mm id Pig iron, cast iron, carbon steel, stainless steel, copper, aluminium, Hastelloy, asbestos, concrete, glass, LDPE, HDPE, PP, PVC,PTFE, PVDF, ABS, FPR, acrylic Most manufacturers do not claim an upper limit, although one manufacturer sets this at 25 mm Tar, epoxy, mortar, rubber

Maximum pipewall thickness for the operation of clamp-on TTUFs Lining materials and external coatings with which manufacturers claim to be able to use clamp-on TTUFs

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

135

sponding to the sum of the wall thickness and the lining thickness has to be input, together with a speed of sound which represents a weighted average of the speed of sound in the two materials, based on their relative thicknesses. For manufacturers who take into account lining materials and coatings, some allow the user to input more than one lining material or coating. Where this is not possible, it is necessary for the user to input an appropriate thickness and speed of sound which accounts for the total thickness and an average speed of sound. 3.2. Effects of pipe wall roughness on measurement performance The roughness of the internal wall of the pipe is likely to cause two sets of problems. The fully developed prole in a pipe depends on the pipe wall roughness. Since the velocity prole is affected, so the k factor required to correct the measured velocity along the beam to the average velocity across the pipe cross section. Fig. 15 shows the correction factor required for fully developed ow for a series of pipe relative roughness, k/d (roughness to diameter ratio) and shows that pipe wall roughness can cause an error of up to 4% in very rough pipes. In addition to affecting the prole, the pipe internal roughness can also cause signicant scattering of the ultrasonic signal. The degree of scattering is related to the absolute level of roughness compared to the wavelength of the ultrasound being employed. Thus it is related to the ratio of / k where is the wavelength of ultrasound in pipe wall material and k is the roughness. Since is inversely proportional to the frequency of excitation of the owmetera typical wavelength in a metal pipe at 1 MHz is 6mm whereas at 500 kHz the wavelength is 12 mm. Therefore, as a general guideline, the rougher the inner surface of the pipework, the lower should be the frequency of operation of the owmeter. Rougher pipe work will also limit the number of wall reection (see Section 5.2) which can be used. If the

owmeter will not work with three reections, then two should be tried and if two will not work then a single pass conguration with no reections should be attempted. Pipe wall roughness can also affect the estimation the pipe id from the pipe od and pipe wall thickness measurements (see Section 5.6). 3.3. Ultrasonic properties of pipework In order to obtain the correct separation of the transducers, both the thickness of the pipe wall and the speed of sound in the material need to be known. Manufacturers have available tables of speeds of sound in different material. The nature of the ultrasonic wave which is transmitted through the pipe may either be a longitudinal wave, a shear wave or a Lamb wave. It is therefore particularly important to input the correct speed of sound in the wall material during the set up phase of the meter. In the case of materials where the speed of sound is uncertain, an ultrasonic thickness gauge applied to a sample of the material of known thickness can be used to provide a measurement of the speed of sound.

4. Fluid effects 4.1. Effects of uid density and viscosity on meter performance As was shown in Section 2.2 in fully developed ow the k factor for the meter depends on the Reynolds number Re given by:Re v.d.r m (12)

Thus, operating a clamp-on TTUF requires a knowledge of the density r and the viscosity m or alternatively the kinematic viscosity n which is given by m / r. It is particularly important to know the viscosity of high viscosity

Fig. 15. Effect of pipe wall roughness on k factor.

136

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

products for which measurements are being undertaken in small pipes since these ows may be in the laminar ow regime. 4.2. Effects of speed of sound on meter performance Clamp-on TTUFs require the user to input the speed of sound into the meter at the initialisation stage as part of the procedure to set the correct distance between the two transducers. Subsequently, depending on the design of the owmeter, this speed of sound will then be used in the dening equation of the owmeter or alternatively a new speed of sound will be estimated from the transit time measurements. The latter enables compensation for the variations which may occur as a consequence of changes in temperature of the uid. The correct separation of the transducers and the correct speed of sound is important to the correct operation of the owmeter. In the case of liquids, whose speed of sound is uncertain then manufacturers generally have facilities for its measurement. 4.3. Effects of air bubbles and solids on meter performance TTUFs are designed for operation with liquids which have only low content of air bubbles or solids. The effects of air bubbles is to scatter the ultrasound but also to affect the speed of sound in the liquid by affecting the compressibility of the liquid. These have the effect of attenuating the signal and degrading the signal to noise ratio in the owmeter. The degree of scattering and attenuation is dependent on the size, number and distribution of the air bubbles. Manufacturers provide typical levels of entrained air with which their meter will continue to operate, typically less than 10%, although generally without any specication of the size of the bubbles. The effect of solid material is to scatter the ultrasound. Low densities of solids having particles diameters less than l / 8, where l is the wavelength of ultrasound in the liquid is likely to cause little effect on the meter. 4.4. Limitations caused by highly attenuative uid media Highly viscous uid generally has associated with them high levels of attenuation of ultrasound. So for example, water has a kinematic viscosity of 1.003 10 6 m2s 1 at 20C and an attenuation at 1MHz of a pressure wave of 0.22dB/m whereas a product such as castor oil has a kinematic viscosity of 1.1 10 2 m2s 1 and an attenuation of 95dB/m at 1 MHz. Thus it is possible that the attenuation in a high viscosity uid may be so high as prevent correct operation of the owmeter. This is usually detected by the

owmeter electronics and agged up. Since viscous attenuation is generally a function of frequency, lowering the frequency of operation of the owmeter may enable measurements to be made.

5. Operational effects 5.1. User training The simple appearance of the meter and the simplied principle often presented belies the fact that it is a very sophisticated owmeter and there are many complex interrelated phenomena involved in producing a measurement. Users who undertake the measurement must be aware of these and take them into account when undertaking the measurement. It must be stressed that clamp-on meters should be installed by trained personnel. 5.2. Transducer conguration There are three congurations to mount the transducers. The two main methods are direct transmission (Z) and single reection (V). Multiple reections (W) are also used. Direct transmission (Z), Fig. 16a, is used

Fig. 16.

a. Z conguration. b V conguration. c W conguration.

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

137

for large pipe sizes as the distance between the path of the ultrasound is shorter than the other congurations and thus less signal loss. They are however much more difcult to install correctly, to measure the distance between the transducers and align. Single reection or reex mode (V), Fig. 16b, is the recommended installation method. The path length is longer giving better time resolution. The set up and alignment of the transducers are much easier as they are on the same side of the pipe. The measurement of the separation is easier and potentially more accurate. The down side is that the axial separation between the transducer could be a problem in conned spaces. Multiple reections (W), Fig. 16c, may be used on small pipes to increase the lapse time, by increasing the ultrasound path between the transducers. Care must be taken to ensure that the correct distance is used, as it is very easy to end up with a V, W or more reections without realizing that the distance is incorrect. 5.3. Location of the transducers with respect to the pipe For pure liquid with no particles nor bubbles, the orientation of the transducers is irrelevant. For horizontal pipe runs, gas has a tendency to oat to the top and particles settle at the bottom. Thus it is preferable to mount the transducers around the side of horizontal pipes (i.e. at 3 and 9 oclock positions). It should be emphasised that if there were gas at the top and sediments on the bottom, the effective ow area will be different from the pipe area and hence the volume ow would be in error. For gas in vertical mains the transducers should mount on the down leg, with ow from top to bottom. For particles in vertical main, mount the transducers on the up leg, with the uid travelling upwards. 5.4. Positioning of the transducers The method of positioning depends on the transducer design and the method of transmission. Most transducers are clamped in position by a clamping mechanism, such as a strap and then located by measurement of the distance apart. Most suppliers also offer a calibrated track to help with the measurement and location. If when setting up the transducers there appears to be no signal, some manufacturers suggest a process called scanning can be carried out. The transducers are moved relative to each other until a signal is found, or the signal becomes stronger. It is better to start too far apart and move the transducers towards each other as they scan better forwards rather than backwards. If a signal is found by scanning, the implications are that the pipe diameter, wall thickness, materials selected and uid could be wrong as there is only one correct position.

A careful examination should be carried out before proceeding with the measurement. As mentioned above, it is difcult to achieve good accuracy for direct transmission. The following methods may be used to set up the meter: A sheet of paper is wrapped around the pipe to mark positions. The sheet is then removed and the nal locations calculated and marked. The sheet is replaced on the pipe and the transducers located in position. In a conned, dirty space this is very difcult. Some suppliers provide locating spigots on the clamping mechanisms. These spigots are located into appropriate holes in a spacer to give the correct transducer spacing. A transducer is xed in its position. The second transducer is moved (scanned) along the opposite side for the best signal. When this is found, the transducer is clamped in place and the distance apart measured. 5.5. Method of clamping There are a variety of methods used for holding the transducers in position. The simplest method is to use straps, usually nylon with quick release and grip buckles. These are easy to locate and tension. They do however stretch with time and should not be used for long term measurement. Chains are often used, these are particularly effective for permanent installation. Steel ropes, with quick release and grip mechanisms are used but tend to be cumbersome. Jubilee clips may be used for smaller pipes and are very effective for permanent installation. With iron based pipes, magnetic clamps can be used. These are very easy to position, but care must be taken not to adjust the clamping mechanism too tightly, as the magnets tend to loosen. Magnetic clamps should not be used when the temperature exceeds 50C. In the event of only an approximate ow rate being required, it is often possible to just use the adhesion of the couplant to hold the transducers in place. 5.6. Getting most information from the installation Velocity prole and swirl have a large inuence on the measurement. The recommendation of Section 2 with regard to mounting downstream and upstream of disturbances should be adhered to. However, the following steps may be used to ensure the best results: By moving the transducers around the pipe circumfer-

138

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

ence and checking the velocity it is possible to determine whether there is a large asymmetry of prole. All the published experimental evidence suggests that because the meter reads low under asymmetric conditions, the highest reading found is likely to be the closest to the actual mean velocity. It is possible to get an indication of swirl by comparing the velocity from the single reection and the single transmission mode. If they are very different, severe cross ow is present. The measurement should be treated with caution. 5.7. Couplant and pipe preparation The purpose of the couplant is to provide a reliable transmission of ultrasound between the transducer and pipe wall. Different couplants are used for long and short term use. The long term types are essentially Araldites, Eurothane resins and Epoxy resins without llers that will diffuse the sound. For permanent installations, regular checking is required. The frequency of checking depends on supplier and couplant type. Flowmeter diagnostics may be used to indicate when couplant replacement is due. Short term types are Silicon grease, axle grease etc. With a short term couplant, it is important to ensure that it does not dry out. It is recommended that a thin bead of couplant, about 5 mm by about 34 mm deep is run along the transducer and then compressed on to the pipe. The following precautions should be taken with couplants: Always clean and degrease the transducer pipe area. lf there is a coating on the outside of the pipe it may be necessary to remove it, particularly a coating containing bres or metal strengthener. Care should be taken to ensure that air is not introduced by excessive spreading or mixing. Excessive amounts of couplants should not be used. For smaller pipes, the transducers could become acoustically connected via the couplant. If there is pitting in the pipe walls, enough couplant should be used to cover the pits and make a full acoustic path. With plastic pipes, it may be necessary to roughen the surface slightly to ensure adhesion of epoxy resin. Care should be taken in dusty/aky environments. Mixing with the couplant can reduce the effectiveness of the couplant. Also it can make the couplants dry out. It must be ensured that users are aware of the Health and Safety regulations as some of the couplants are irritants. The temperature compatibility of the couplant and the process needs to be checked. Pipe surface preparation must be carefully done to

preserve the original curvature of the pipe. It is important that the transducer faces and the pipe axis are parallel as one degree error could lead to approximately one percent change in path length. 5.8. Process temperature Process temperature has several interacting effects. It affects the speed of sound, uid density, viscosity and hence the Reynolds number and the velocity prole. These effects have been covered in Section 2. Changes in the speed of sound in the uid have the effect of changing the angle of the beam in the uid and hence the sensitivity of the owmeter as shown in Eq. (1). For processes where the process uid temperature is likely to change it is important that a owmeter with speed of sound compensation is employed. Process temperature also has an impact on the selection of transducers and the couplant. Special transducers are required for low and high temperatures. The transducers often use potting to protect the piezoelectric crystals. Temperature can reduce the effective transparency to sound of some potting materials. Temperature can cause expansion of the materials at different rates causing them to split apart and either destroy them, or form air bubbles. If the temperature is too high, the piezoelectric crystals will reduce effectiveness and eventually stop working. A typical temperature range for a standard transducer is 40C to +100C, although transducers are available which enable measurements to be made from 190C to 500C. For high temperature applications the transducer may be coupled using a metal couplant and buffered from the process by being mounted on the ends of long buffer rods. Care must also be taken with the use of couplants with temperature. Supplier of couplants should be consulted for abnormal operation. At high temperature water based couplants will evaporate. At low temperature water based couplants may freeze and change characteristics. At high temperature some oil based couplants may become runny. At low temperatures some oil based couplants change their characteristics. 5.9. Pipe size and transducer frequency All suppliers support a large range of pipe sizes. Different transducers are often used to cover the range. 1 MHz transducers appear to be the standard. These cover approximately the pipe size range of 50 mm to 2000 mm. 2 MHz transducers are used for smaller pipe sizes, below 100 mm diameter and down to 10 mm.

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

139

0.5 MHz transducers are used for large pipe diameters, approximately 500 mm to 5000 mm diameter. Transducers operating at 4 MHz are offered in very small sizes. In general where there are bubbles or particles in the ow, a lower frequency is recommended. Also if there are problems with getting sound through the pipe, there is a tendency to use lower frequencies. 5.10. Pipe diameter and wall thickness measurement The internal pipe area is required to calculate the volumetric ow rate from the measured velocity. As the internal pipe diameter cannot be measured directly, it is often inferred from the outside diameter and the wall thickness. The uncertainty of internal diameter measurement is a major contributor to the ow uncertainty. Pipe wall thickness is also required, in combination with the material, to work out the transducer positions. Thus wall thickness will also affect the velocity measurement. The recommended procedure for pipe internal diameter measurement is as follows: Measure the circumference of the pipe using a traceable and accurate tape, or pipe tape. Check the ovality using calipers. Determine the pipe outside diameter. Measure the wall thickness. Subtract twice the wall thickness from the pipe outside diameter. The wall thickness may be obtained by a number of methods: From a set of drawings or pipe specication. By using an ultrasonic wall thickness gauge. If the pipe is lined it may not be possible to tell whether the thickness includes the liner using this method. There are devices that could show the different reections which could be of some assistance to the skilled user. If there is build-up on the inside of the pipe, it may not give the correct reading. Drilling a small hole or using a tapping to insert a thickness gauge or direct diameter measuring device. At the same time, the pipe walls can be checked for build-up. The best method is often to nd an off-cut and measure it. 6. Manufacurer specic 6.1. Measurement of Time difference Most manufacturers use a variant of the leading edge measurement technique shown in Fig. 17a. In this con-

Fig. 17.

a Leading edge measurement. b Phase measurement system.

guration the transducers are pulsed and the time is measured from the arrival of the receive pulse. In some designs a phase measurement system, see Fig. 17b, is employed in which a sinusoidal burst is sent out from both transducers and the phase of the received signals compared. Although such a technique enables multiple measurement of the phase over several cycles within the receive burst, it may lead to limitations as to the range of velocities or pipe sizes over which the measurements

140

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

can be made before phase ambiguity occurs as a consequence of exceeding 360 of phase shift. 6.2. Correction for Reynolds Number

suppliers as to the range of diagnostic functions they provide.

7. Future work As has been shown in Section 2, clamp-on TTUFs are sensitive to Reynolds Number. Manufacturers may or may not compensate for this. For those manufacturers who do compensate, the compensation is generally based on a knowledge of the particular liquid being monitored. For products whose temperature is changing signicantly during the measurement, then compensation for Reynolds Number on-line is required. Enquiries should be made of manufacturers as to whether or not they provide on-line correction. 6.3. Correction for speed of sound All manufacturers require the user to input either the speed of sound or identify the liquid (and perhaps its temperature) in the initialisation stage of the measurement as part of the process to determine transducer separation. Some manufacturers continue to use this value for the ow measurement itself whereas other manufacturers provide on-line measurement of the speed of sound in the liquid under process conditions. For liquids for which the temperature is signicantly changing, it is necessary to enquire of the manufacturer what means of temperature compensation is provided. 6.4. Diagnostics Nearly all manufacturers provide the user with an indication of the signal strength being received by the transducers and incorporate some means of automatic gain control to compensate for attenuation of the ultrasound in the pipe wall and in the uid. The signal strength indicator may be used in situations where the separation of the transducers does not appear to be the correct one and the pipe is then scanned to nd the correct separation. Additional diagnostics during the measurement phase may be employed. These diagnostics may measure the modulation of the received signal (indicating the presence of a second phase such as air bubbles or solids and the need to switch operation from TTUF mode to DUF mode) or the signal to noise ratio of the received signal which provides an indication of the likely quality of the measurement. Since most of the measurement systems are microprocessor based they will include a range of diagnostic tools which will enable the hardware and the software of the system to be checked and error codes agged which would enable the specic error to be identied. These diagnostic functions will vary from manufacturer to manufacturer and enquiries should be made of potential These current guidelines have identied a number of areas where at the present time there is insufcient data available to provide rmer advice to the users of this technology. A programme of work has been identied, for inclusion in the next Department of Trade and Industry Flow Programme, including the following:The evaluation of the performance of clamp-on meters in sizes above 600 mm with water as the owing medium. The performance of clamp-on meters on small sizes below 50 mm on water since there is a signicant area of application to these devices in water and energy metering. The performance of small meters on liquids other than water since there is a widespread use of small clampon meters in the oil and food sector. The temperature characteristics of clamp-on owmeters because of their application in process measurements where the uid temperature can vary. The effect of installation conditions on small bore clamp-on owmeter measurements. The effect of pipe roughness on the performance and operability of clamp-on owmeters since these meters are often required to work in conditions where there is corroded pipework. The sensitivity of meter output to incorrect set-up information being supplied to the meters. It is clear that different manufacturers employ different algorithms to compute the owrate and that the effect of such errors will be different for different manufacturers. This work will enable the best algorithms to be identied and hence the overall performance of such meters to be improved.

8. Conclusions These Guidelines have been produced to improve the use of clamp-on TTUFs. As extent of user experience increases and further calibrations are undertaken under a wider range of conditions, a greater understanding of the performance of such devices is likely and clamp-on TTUFs with improved performance are likely to emerge.

Acknowledgements The authors are grateful to the Department of Trade and Industry for its nancial support in the production

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

141

of the Guidelines and its permission for them to be reprinted here. A CD version of the Guidelines, including speed of sound data for pipe materials and liquids and densities and viscosities of liquids, are available from the Department of Process and Systems Engineering on CD at the address of the authors, e-mail address: pase@craneld.ac.uk.

[21]

[22]

[23]

Further reading
[1] R.F. Brummer. Theoretical and experimental assessment of uncertainties in non-intrusive ow measurement. NBS SPECIAL Publication 84, Proc.Symp. On ow in open channels and closed conduits. 1977, pp 277291. [2] T. Cousins, The Doppler Ultrasonic Flowmeter, Flow Measurement of Fluids, North-Holland Publishing Co., 1978 pp 513-518. [3] L.C. Lynnworth, Ultrasonic owmeters, in: W.P. Mason, R. Thurston (Eds.), Physical Acoustics, 14, 1979, pp. 407525. [4] D.E. Morris et al. Improve SiCl4 production with clamp-on owmeters that avoid corrosion problems. Chemical Engineering Processing (1981) 1617. [5] M.L. Sanderson, J. Hemp. Ultrasonic owmetera review of the state of the art. Proc.Int. Conf in Advances in Flow Measurement Techniques, BHRA Fluid Engineering, Bedford, UK. 1981. pp. 157178. [6] H. Yada, A clamp-on ultrasonic owmeter for high temperature uids in small conduits, in: W.W. Durgin (Ed.), FlowIts Measurement and Control in Science and Industry, 2, Instrument Society of America, South Triangle Park, NC, 1981, pp. 546553. [7] R. Fell. Systematic errors in cross-correlation velocity measurements. Acta Meko (1982) 205214. [8] R. Keech, The KPC multichannel correlation signal processor for velocity measurements, Trans. Inst. M&C 14 (1) (1982). [9] L.C. Lynnworth, Ultrasonic owmeters, Trans. Inst. M & C 14 (1) (1982) 224. [10]R. Thorn, et al. Non-intrusive methods of velocity measurement in pneumatic conveying, J. Phys. E 15 (1131) 1982. [11] J. Coulthard. Cross correlation owmeters: a history and the present state of the art. Measurement and Control, 16 June 1983 [12] M.L. Sanderson, B. Torley. A self-calibrating clamp-on transit time ultrasonic owmeter. Proc. FLOMEKO 85, Melbourne, Australia. 1985. pp 163170. [13] K. Spendel. On non-invasive ultrasonic owmeters. PhD thesis, Craneld Institute of Technology, 1985 [14] M.S. Beck, A. Plaskowski, Cross Correlation FlowmetersTheir Design and Application, Adam Hilger, 1986. [15] E. Hayes. Evaluation of Ultrasonic Flowmeters in Two Phase Air/water Flows. CIT Short Course Notes, Department of Fluid Engineering and Instrumentation, Craneld Institute of Technology, Craneld, Bedford, UK, 1986. [16] P. Hojolt. Installation effects on single and dual beam ultrasonic owmeters. Proc. Int. Conf. on Flow Measurement in Mid 80s. NEL, Glasgow, UK, Paper 11.2, 1986. [17] M.L. Sanderso, B. Torley. Error assessment for an intelligent clamp-on transit time ultrasonic owmeter. Proc. Int. Conf. on Flow Measurement in Mid 80s. NEL, Glasgow, UK, Paper 11.1, 1986. [18] J. Heritage. Performance of commercial ultrasonic owmeters. DTI/NEL Contract No:RD119/02, Craneld Institute of Technology, Craneld, Bedford, UK, 1987. [19] R.C. Baker et al. Installation effects electromagnetic and ultrasonic owmeters. FLOMIC Consortium Report, Craneld Institute of Technology, Craneld, Bedford, UK, 1988. [20] L.C. Lynnworth. Buffer rod designs for ultrasonic owmeters at

[24]

[25] [26] [27] [28]

[29] [30]

[31]

[32] [33]

[34]

[35]

[36] [37]

[38]

[39]

[40]

[41]

[42]

[43]

[44] [45]

cryogenic and high temperatures +/-200C. Proc 34th Int. Instrumentation Symp. Alburqueque, NM. 1988. pp 697702. M.L. Sanderson et al. Non-intrusive owmetering. FLOMIC Consortium Report, Craneld Institute of Technology, Craneld, Bedford, UK, 1988. J.E. Heritage. The performance of transit time ultrasonic owmeters under good and disturbed ow conditions. Flow Measurement and Instrumentation (1989) 2430. R.C. Mottram, J.L. Sproston. The effect of pulsation on the performance of owmeters. FLOMIC Consortium Report, Craneld Institute of Technology, Craneld, Bedford, UK, 1989. J. Halttunen. Installation effects on ultrasonic and electromagnetic owmeters: a model-based approach. Flow Measurement and Instrumentation (1990) 287292. R. Motegi et al. Widebeam ultrasonic owmeter. IEEE 1990 Ultrasonics Symp. 1990. pp. 331336. J. Gatke, Problems of transducer coupling with acoustic clampon owmeters, MSR 34 (5) (1991) 211217. J. Szebeszczyk, S. Pietrasz, Clamp-on owmeter for homogeneous liquids, MSR 34 (5) (1991) 208211. J. Sweetland. An assessment of clamp-on transit time ultrasonic owmeters. CIT Report DJS/18/508/117, Craneld Institute of Technology, Craneld, Bedford, UK, 1992. M. Shortall. Ultrasonics are advancing. Control and Instrumentation (1993). J. Baumoel. Use of clamp-on transit-time ultrasound owmeters in aircraft mass fuel, hydraulic leak detection. Proc. 40th Int. Instrumentation Symp. 1994. pp 243260. F. Cascetta. Application of a portable clamp-on ultrasonic owmeter in the water industry. Flow Measurement and Instrumentation (1994) 191194. S.T. Lange. Making a choice. Which owmetering technology? Water Engineering and Management (1994) 1819. J.M. Szebeszczyk. Applications of clamp-on ultrasonic owmeter for industrial ow measurement. Flow Measurement and Instrumentation (1994) 127134. H.P. Vaterlaus, H. Gabler. A new intelligent ultrasonic owmeter for hydropower applications. Int. Water Power & Dam Construction (1994) 8488. L.A. Xu et al. Clamp-on ultrasound cross-correlation owmeter for liquid/solid two-phase ow measurement. Flow Measurement and Instrumentation (1994) 203208. B. Funck, Clamp-on owmeter, Proc Sensors 95 (1995) 8590. B. Funck, A. Mitzkus. Acoustic transfer function of the clampon owmeter. Proc 1995 IEEE Ultrasonics Symp. 1995. pp. 569575. Lynch, E. Horciza. Flow measurement using low cost portable clamp-on owmeters. SCE Waterpower, Proc.Int.Conf. On Hydropower. 1995. pp. 766773. L.C. Lynnworth. Clamp-on transducers for measuring swirl, cross ow and axial ow. Proc 1994, IEEE Ultrasoncis Symp. 1995. pp. 13171321. H.P. Vaterlaus. New intelligent ultrasonic owmeter for closed conduits and open channels. Proc 1995 Int. Conf. On Hydropower. 1995. pp. 9991008. B. Funck, A. Mitzkus. Acoustic transfer function of the clamp-on owmeter. IEEE Trans Ultrasonics, ferroelectrics & Frequency Control. 1996. pp. 569575. V. Pavlovic et al. Ultrasonic pulse-phase method applied in uid ow measurements. IEE Proc. Science, Measurement and Technology. 1996. pp. 327333. M. Vestrheim, S. Vervik. Transit time determination in a measurement system with effect of transducers. Proc 1996 IEEE Ultrasonics Symp. 1996. pp. 665668. G. Vass. Ultrasonic owmeter basics. Sensors (1997) pp. 7378. H. Eren. Accuracy of real time ultrasonic applications and transit

142

M.L. Sanderson, H. Yeung / Flow Measurement and Instrumentation 13 (2002) 125142

[46]

[47] [48]

[49] [50] [51]

[52]

time owmeters. IMTC/98 Conf. Proc. 1998. Vol. 1, pp. 568 572. B. Svensson, J. Delsing. Application of ultrasonic clamp-on owmeters for in-situ tests of billing. Flow Measurement and Instrumentation (1998) 3341. A. Worch. A clamp-on ultrasonic cross-correlation owmeter for two-phase ow. Proc FLOMEKO 98. 1998. pp. 121126. A. Worch. Clamp-on ultrasonic cross correlation ow meter for one-phase ow. Measurement Science and Technology (1998) 622630. L.C. Lynnworth. High Temperature ow measurement with wetted and clamp-on ultrasonic sensors. Sensors (1999) 3652. L.C. Lynnworth, V. Magori, Industrial process control sensors and systems, Physical Acoustics 23 (1999) 275470. NEL. Velocity Distribution Effects on Ultrasonic Flowmeters Part 1Theoretical Analysis. NEL Report No 357/99, NEL, Glasgow, UK, 1999. NEL. Velocity Distribution Effects on Ultrasonic Flowmeters Part 2. NEL Report No 348/99, NEL, Glasgow, UK, 1999.

[53] NEL. Research into Clamp-on Ultrasonic Meter. NEL Report No 359/99, NEL, Glasgow, UK, 1999. [54] M.L. Sanderson. Industrial ow measurement by ultrasonics, insight: non-destructive testing and condition monitoring. 1999. pp. 1619. [55] S. Aikainen, J. Halttunen. Experiences of clamp-on ultrasonic owmeters in small pipes. Proc. FLOMEKO2000, Salvador, Brazil, 2000. [56] Brown et al. Modelling of transit time ultrasonic owmeters in theoretical asymmetrical ow. Proc. FLOMEKO2000, Salvador, Brazil, 2000. [57] J.D. da Mata et al. Applications of ultrasonic ow meters in the petroleum industry. Proc FLOMEKO2000, Salvador, Brazil, 2000. [58] E. Van Bokhorst, M.C.A.M. Peters. The impact of low frequency pulsations on a dual-beam ultrasonic owmeter. Proc. FLOMEKO2000, Salvador, Brazil, 2000. [59] K. Zanker. Installation effects on single and multi-path ultrasonic meters. Proc.FLOMEKO2000, Salvador, Brazil, 2000.