GB2406654A

GB2406654A - Method and apparatus for determining the sizes of leaks in water distribution networks

Info

Publication number: GB2406654A
Application number: GB0418988A
Authority: GB
Inventors: Charles Gerard Harris
Original assignee: Metrika Ltd
Current assignee: Metrika Ltd
Priority date: 2003-08-26
Filing date: 2004-08-26
Publication date: 2005-04-06
Anticipated expiration: 2024-08-26
Also published as: GB0319888D0; GB2406654B; GB0418988D0

Abstract

The size of a leak in a buried water pipe forming part of a water distribution network is assessed and evaluated by a method which includes the measurement of the surface vibration level(s) on the ground surface above the approximate or known position of the leak.

Description

DETERMINING THE SIZES OF LEAKS IN WATER

DISTRIBUTION NETWORKS

Field of the Invention

This invention relates to a method of and means for determining the sizes of leaks in water distribution networks.

Background to the Invention

Until now there has been no reliable method for estimating the size of a leak in a buried water distribution pipe before excavation. This has caused problems because much effort, expense and inconvenience is incurred by excavating and repairing leaks that turn out to be too small to warrant the excavation costs.

Furthermore, inadvertently repairing small leaks diverts effort away from repairing bigger ones that may be waiting in the repair queue. These larger leaks need to be repaired first but there has not been a reliable method for identifying them.

Water escaping from leaks in buried water distribution pipes causes vibrations on ground surfaces above the leaks and on connected pipes and fittings.

It is an object of the present invention to provide a method of, and means for, quantifying the size of these hidden leaks.

It is a further object of the present invention to provide a leakage manager with a system for determining, in a batch of leaks, the ones which need repairing first so that the minimum amount of water is wasted.

Summarv of the Invention According to a first aspect of the present invention there is provided a method of assessing the size of a leak in a buried water distribution pipe which includes measuring the surface vibration level on the ground surface above the suspected or known position of the leak in the buried water distribution pipe.

According to a second aspect of the present invention there is provided means for assessing the size of a leak in a buried water distribution pipe which includes means for measuring the surface vibration level on the ground surface above the suspected or known position of the leak in the buried water distribution pipe.

For convenience the measurement of the surface vibration level on the ground surface above the leak is called the Over Leak Signature and this measurement may be used on its own or combined with one or more of the following data:

Background Signature,

Local water pressure and leak power, Over leak vibration spectral characteristics, Stop Tap Signatures, Leak Noise Correlation Data, Pipe Materials estimates, Pipe Dimension estimates, Failed component identity estimates, Prevailing local leakage rates, Any other data.

Each of these elements is now described in turn starting with the Over Leak Signature, Background Signature, local water pressure and leak power.

1) Over Leak Signature 1.1) Raw Over Leak Signature The operator measures and records ground surface movement versus time over the suspected or known position of a leak in a buried water distribution pipe. Ground surface movement may be recorded using an accelerometer attached to a recording apparatus with appropriate signal amplification and conditioning.

Other motion detection methods may be used. This measurement is called the Raw Over Leak Signature.

The Raw Over Leak Signature is recorded in units proportional to acceleration. The signature is recorded for a period of typically 30 seconds. Longer or shorter periods may be used.

The signal is sampled at a rate sufficient to enable component frequencies in the signal to be recovered in the range of from near OHz to at least 4KHz. Other frequency ranges may be specified.

1.2) Processing the Raw Over Leak Signature The Raw Over Leak Signature will include not only vibrations caused by the leak but also ambient vibrations caused for instance by road traffic. In most practical instances, road traffic noise is intermittent. The measuring means extracts, for example, the quietest one second's worth of data from the raw signature by calculating a rolling Root Mean Squared (RMS) value for contiguous or over-lapping or non-overlapping one second segments of the raw data. The segment with the lowest RMS value is extracted and stored. The lowest RMS value is also recorded and is called Min_OverleakRMS. Different segment lengths may be used. The segments may or may not overlap. The RMS calculation may be performed on overlapping segments and the stored segment with the minimum RMS value is called the Stored Over Leak Signature.

Instead of the lowest RMS value being used to detect the quietest portion of the raw signal, any other indicator of signal magnitude may be used and stored.

1.3) Measuring and processing the Raw Background

Signature.

In a manner similar to that described above for recording the Raw Over Leak Signature and extracting and saving the Stored Over Leak Signature, the measuring means may measure and record ground surface movement versus time at a position typically 3 metres away from the Over Leak position. The purpose of this is to measure the minimum continuous ambient or background surface vibration that is not being caused by the pipe leak. This data may be recorded using a second accelerometer attached to a recording apparatus with appropriate signal amplification and conditioning. Other motion detection methods may be used and this measurement is called the Raw Background Signature.

1.4) Processing the Raw Background Signature.

The Raw Background Signature may be recorded

simultaneously with the Raw Over Leak Signature. The measuring means may extract the one second's worth of data from the Raw Background Signature which is coincident with the segment extracted from the Raw Over Leak Signature or it may select a non-coincident lowest value segment. This segment is stored and is called the Stored Background Signature. Segment lengths other than those of one second may be used and the segments may be contiguous, overlapping or non-overlapping. As with the Stored Over Leak Signature, the RMS value of the selected background

segment is recorded and is called BackgroundRMS.

Instead of measuring the Raw Background Signature

simultaneously with the Raw Over Leak Signature using two vibration transducers or other measuring means, the background signature may be measured just before or just after measuring the Over Leak Signature using the same transducer and recording apparatus.

1.5) Local Water Pressure and Leak Power Let the volume of water escaping from a leak in a given time be called the leak size Q. Let the Pressure in the pipe network at or in the region of the leak be called the pressure P. Then the mechanical power of the Leak (i.e. the rate at which the leak is dissipating energy) is the product Q X P. One would expect to see features related to the leak such as surface vibration being proportional to the leak's power.

Theoretically, the more power the leak has, the greater will be the vibration. If it is possible to do so, it is useful to measure pressure in the region of a leak because a small leak at high pressure may have a similar power to a large leak at low pressure. Everything else being equal, both will be causing the same amount of vibration.

Thus, taking pressure into account enables the different leak sizes to be distinguished even though they have the same vibration signatures.

For the purposes of the present invention, water pressure may be defined as a true pressure i.e. a force per unit area measured in, for example, Newtons per square metre or alternatively as a static head H of water, measured typically in metros.

1.6) Relationships between Over Leak Surface Vibration,

Background vibration, Pressure and Leak size

Let the quantity Min_OverleakRMS minus BackgroundRMS be called the Net_SurfaceRMS.

Then, following observations of a large number of actual leaks, there are approximately linear relationships ranked roughly in increasing linearity between the following quantities: a) Min_OverleakRMS versus Leak size Q. b) Net_SurfaceRMS versus Leak size Q. c) Min_OverleakRMS divided by Pressure P versus Leak size Q. and d) Net_SurfaceRMS divided by Pressure P versus Leak size Q. By measuring and calculating any of the left hand quantities in the above relationships a) through d) for a particular leak and knowing the appropriate constants of proportionality or other mathematical functions describing the relationships, a leak size indication is provided.

1.7) Refinement of leak size indication.

The relationships described in 1.6) above are complicated by a large number of unknowns: For instance, for a given size of leak: i) The depth of the pipe at the leak will affect the level of vibration that is measured at the surface. The deeper the pipe, the less the vibration that may be expected. It is very difficult to measure pipe depth accurately before excavation and reliable records are rarely kept.

ii) The mechanical properties of the pipe and the backfill surrounding it will affect the amount of vibration the leak will transmit to the surface. It is not normally possible to measure these properties without excavating.

iii) The orientation of the jet of water escaping from the leak relative to the pipe will affect surface vibration. A jet escaping in a direction normal to the surface may be expected to generate more surface vibration than a non-normal jet. Jet orientation - if indeed there is a welldefined 'jet' - cannot normally be determined without excavating.

iv) The position of the leak may have been miscalculated.

The surface signature may not have been recorded where it is the maximum or over the leak.

These complications cause some large leaks to have unexpectedly small vibration signatures and vice versa. Thus the method of the present invention provides for the interpretation of the relationships discussed in 1.6) statistically in order to calculate the most probable leak size associated with a particular set of vibration and pressure data and/or other diagnostic indicators.

1.8) Probabilistic estimation of leak size The relationships described in section 1.6) show how various diagnostic quantities are related to leak size. It is possible to take any one of these relationships such as, for instance, Min_OverleakRMS versus Leak size Q and draw up a table such

as shown in Table 1.

Table 1 - Illustrative table of likelihoods I Diagnostic Leak Size litre/sec eg Min_ OverleakR M S 0 to <0.05 0.05 to 0.1 to <0.5 0.5 to 2.0 0.0 to <2.5 0.70 0.4 0.1 0.01 l 2.5 to <5.0 0.20 0.3 0.2 0.02 5.0 to <7.5 1 0.08 1 0.2 1 0.3 0.12 7.5 to 10.0 1 0.02 1 0.1 0.4 0.85 1 Sum 1 1 1 1 1 1 The column headings show the leak range split into four hypothetical sub-ranges 0 to <0.05 L/sec, 0.05 to <0.1 L/sec etc. The row headings show the diagnostic value split into four hypothetical sub- ranges 0.0 to <2.5, 2.5 to <5.0 etc. The cell entries show how, for each flow range, the diagnostic indicator values are hypothetically distributed. For example, taking the flow range column 0 to <0.05 as an example: 70% of leaks in the range 0 to <0.05L/sec have Min_OverleakRMS values in the range 0.0 to <2.5, 20% of them have Min_OverleakRMS values in the range 2.5 to <5.0, 8% occur in the range 5.0 to <7.5 and 2% occur in the range 7.5 to 10.

The measuring means of the present invention includes means for inputting and storing likelihood tables for one or more of the relationships described in section 1.6). The column distributions of likelihoods may be stored as explicit values or they may be described by mathematical distribution functions.

In the stored tables, any number ≥2 of column or rows may be used. The ranges need not be evenly divided.

1.9) Display of estimated leak size A further feature of the present invention is that having recorded and analysed a diagnostic feature for a particular leak, it is possible to display the relative probabilities of the leak being in each category of flow range. For each row of likelihoods, the relative probabilities may be obtained by dividing each row entry by the sum of the row entries. The display may be graphical, e. g. a bar graph on an LCD, with the leak size being categorized numerically in appropriate units or descriptively, e.g. tiny, small, medium, big etc. Where a suitable quantity of data has been accumulated, each cell entry in the likelihood tables, e.g. Table 1, can be associated with a mean leak value or a statistical range of mean values (e.g. 95% confidence limits). For each row, the expected mean leakage flow will be the sum of the products of each cell's relative probability and mean flow. The measuring means may display the expected value or expected value range.

1.10) Adjusting Surface RMS values for depth.

There is an inverse relationship between leak depth and size of surface vibration. Making a compensating adjustment for leak depths would improve diagnosis based on surface signatures but estimating depths manually has been found to be error-prone. The measuring means may accordingly carry out a cross-correlation between two synchronized surface vibration signatures A and B. Signature A is collected on the ground surface above the leak and signature B is collected at a known distance along the surface from signature A. The cross-correlation enables the time of flight difference between the noise patterns received by both sensors to be measured. Coupling this information with data about average ground vibration transmission velocities enables the depth of the leak to be estimated. This estimate may be used to adjust surface RMS values.

2) Over leak vibration spectral characteristics 2.1) Surface_Spectral_ratio The Stored Over Leak Signature may be analysed further as follows: A frequency spectrum of the Stored Over Leak Signature is calculated showing the distribution of energy or power in the signature over frequencies in the range (e.g. 0 - 4KHz).

Average values of the spectrum may be calculated for sub- frequency ranges within the spectrum for instance as follows: 0- 99Hz, 100 to 199Hz, 200 to 299Hz, 300 to 399Hz, ...3900 to 4000Hz.

Diagnostic indicators may be constructed from the ratios of the average spectrum value of one sub-frequency range to the average spectrum value of any other frequency range. For convenience we may call such a diagnostic indicator a surface_spectral_ratio.

The frequency limits of the spectrum may be different to those shown above. The spectrum may be divided into different frequency ranges. The frequency ranges need not be uniformly sized.

Likelihood tables may be input and stored showing the distribution of leak sizes versus ranges of Surface_spectral_ratio similar to that shown in Table 1 and described for surface vibration data in sections 1.8) and 1.9) above.

These likelihood tables may be used independently to indicate leak size or may be used in combination with other likelihood tables, such as that described for Surface RMS data in section 1.8).

2.2) Surface_spectral_match The measuring means may calculate average normalised spectra for the Stored Over Leak Signatures for a representative sets of differently categorised leaks (e.g. a set for leaks <0.05L/sec, another for leaks in the range from 0.05 to 0.1 L/sec etc. These averaged normalised spectra may be input and stored.

When a new estimation is being made, the measuring means calculates a normalised spectrum for the Stored Over Leak Signature for the leak being investigated and carries out least squares fit calculations between this spectrum and the stored representative spectra. The stored category that shows the closest match to the signature under test is used to indicate the likely leak size.

It is convenient to use normalised spectra in the above process, but it is possible to use un-normalised spectra.

Closeness of fit calculations may be carried out using procedures other than the least squares method. Complete spectra may be used in the matching process or just parts of the spectra.

Likelihood tables may be input and stored showing the distribution of leak sizes versus ranges of Surface_spectral_match results similar to that shown in Table 1 and described for surface vibration data in sections 1.8) and 1.9) above.

These likelihood tables may be used independently to indicate leak size or used in combination with other likelihood tables such as that for Surface RMS data described in section 1.8).

3) Stop Tap Vibration Levels In addition to causing vibrations of ground surfaces, water escaping from leaks on buried water distribution pipes causes vibrations on connected pipes and fittings.

Of particular, but not exclusive, interest are the vibrations that may be easily detected on accessible fittings on the communication and service pipes connecting usually larger diameter mains and sub-mains to customer properties. A common fitting is the stop tap. Bearing in mind that it is possible to record vibrations on other types of fitting, for convenience, we shall call the vibrations measured on such fittings 'Stop Tap Signatures'. A Stop Tap Signature is recorded as follows: 3.1) Raw Stop Tap Signature The operator measures and records vertical or near vertical stop tap vibration versus time on the fitting closest to and connected by pipework to the buried leaking pipe. Vibration may be recorded using an accelerometer attached to a recording apparatus with appropriate signal amplification and conditioning.

Other motion detection methods may be used and it is possible to record simultaneously or otherwise vibration on axes other than the vertical. Let the vibration measurement be called the Raw Stop Tap Signature.

The Stop Tap Signature may be recorded in units proportional to acceleration. The signature is recorded for a period of typically 30 seconds. Longer or shorter periods may be used.

The signal is sampled at a rate sufficient to enable component frequencies in the signal to be recovered in the range near OHz to at least 4KHz. Other frequency ranges may be specified.

3.2) Processing the Raw Stop Tap Signature The Raw Stop Tap Signature will include not only vibrations caused by the leak but also ambient vibrations caused, for instance, by road traffic. In most practical instances, road traffic noise is intermittent. The system extracts, for example, the quietest one second's worth of data from the raw signature by calculating a rolling Root Mean Squared (RMS) value for contiguous or overlapping or non-overlapping one second segments of the raw data. The segment with the lowest RMS value is extracted and stored. The lowest RMS value is also recorded and is called Min_StopTapRMS. Different segment lengths may be used. The segments may or may not overlap. The RMS calculation may be performed on overlapping segments. The stored segment with the minimum RMS value is called the Stored Stop Tap Signature.

3.3) Relationships between Over Stop Tap Vibration, Pressure and Leak size Following observations of a large number of actual leaks, there are relationships between the following quantities: a) Min_StopTapRMS versus Leak size Q. and b) Min_StopTapRMS divided by Pressure P versus Leak size Q. By measuring and calculating any of the left hand quantities in the above relationships a) or b) for a particular leak and knowing the appropriate constants of proportionality or other mathematical functions describing the relationships, it is possible to indicate a leak size.

3.4) Refinement of leak size indication The relationships described in 3. 3) above are complicated by unknown factors: For instance, for a given size of leak: i) The distance from the stop tap to the leak will affect the level of vibration that may be measured at the stop tap. The further away the leak, the less the vibration that may be expected. It is very difficult to characterize transmission losses without accurately knowing the true subterranean distance to the leak, and the composition and structure of the pipe.

ii) The mechanical properties of the pipe and the backfill surrounding it will affect the amount of vibration the leak will transmit to adjacent fittings. It is not normally possible to measure these properties without excavating.

iii) The orientation of the jet of water escaping from the leak relative to the pipe will affect the transmitted vibration. Jet orientation - if indeed there is a well-defined 'jet' - cannot normally be determined without excavating.

These complications cause some large leaks to have unexpectedly small vibration signatures and vice versa. Thus the measuring means interprets the relationships discussed in 3.3) statistically in order to calculate the most probable leak size associated with a particular set of vibration and pressure data or other diagnostic indicators.

Where the pipe material and/or subterranean distance to the leak are known accurately or can be estimated, the Min_StopTapRMS may be adjusted using this information to compensate for attenuation losses.

Likelihood tables may be input and stored showing the distribution of leak sizes versus ranges of Min_StopTapRMS or Min_StopTapRMS divided by pressure similar to that shown in Table 1 and described for surface vibration data in sections 1.8) and 1.9) above.

These likelihood tables may be used independently to indicate leak size or used in combination with other likelihood tables such as that described for Surface RMS data in section 1.8).

3.5) StopTap_spectral_match It is possible to calculate average normalised spectra for the Stored Stop Tap Signatures for representative sets of differently categorised leaks (e.g. a set for leaks <0.05L/sec, another for leaks in the range from 0.05 L/sec to 0.1 L/sec etc. These averaged normalised spectra may be input and stored.

When a new estimation is being made, the measuring means calculates a normalised spectrum for the Stored Stop Signature for the leak being investigated and carries out least squares fit calculations between this spectrum and the stored representative spectra. The stored category that shows the closest match to the signature under test is used to indicate the likely leak size.

Likelihood tables may be input and stored showing the distribution of leak sizes versus ranges of StopTap_spectral_match results similar to that shown in Table 1 and described for surface vibration data in sections 1.8) and 1.9) above.

Such likelihood tables may be used independently to indicate leak size or used in combination with other likelihood tables such as that described for Surface RMS data in section 1.8).

4) Leak noise correlation data Leak noise correlators are used to locate leaks on buried water pipes. Two accelerometers or hydrophores are attached or inserted into the pipeline as close as possible to and either side of the suspected leak position. Each sensor records synchronized noise signals and these two sets of signals are cross-correlated. It has been proposed by others (PCT/US98/09068) that the maximum value of the cross correlation function (suitably adjusted for distance attenuation) is proportional to leak size.

However, experience suggests that the relationship between the maximum value of the cross correlation function and leak size is statistically noisy. Big leaks may have small cross correlation function magnitudes and vice versa. In accordance with the present invention, the interpretation of cross correlation function information is improved by constructing likelihood tables showing the distribution of leak sizes versus ranges of maximum cross correlation function magnitudes similar to the one shown in Table 1 and described for surface vibration data in sections 1.8) and 1.9) above.

The cross-correlation magnitudes may be adjusted for distance attenuation.

Further, the cross-correlation function results may be adjusted for the power in the leak (see section 1.5) above) by dividing the maximum value of the cross correlation function by the local pressure. This latter operation is distinctly different to that suggested in PCT/US98/09068) where pressure is used only to adjust the distance attenuation coefficient.

In accordance with a further feature of the present invention further processing of the cross-correlation function may take place as follows: The energy spectral density of the cross-correlation function is calculated by Fourier Transforming the cross-correlation function (see Page 242, Digital Signal Processing, E.C. Ifeachor et al, Addison Wesley). This gives the distribution of energy density in the noise signals that is common to both sensors and thus likely to have originated from the leak. The energy at any frequency may be obtained from the density function by multiplying the density value by its respective frequency.

Features from either the energy spectral density plot or energy spectral plot may be related to leak size. Diagnostic features may include peak energy or energy density values and frequencies at which these values occur. All the foregoing features may be adjusted for distance attenuation effects and leak power. The diagnostic information in the foregoing features may be interpreted by constructing and referring to likelihood tables similar to the one shown in Table 1 and described for surface vibration data in sections 1.8) and 1.9) above.

The foregoing likelihood tables may be used independently to indicate leak size or used in combination with other likelihood tables such as that described for Surface RMS data in section 1.8).

5) Incorporating Estimates of Pipe Material Types Experience suggests that, when certain types of pipe fail, they tend to leak more seriously than others. For instance, certain types of plastic pipe exhibit this behaviour. To estimate leak size it is possible to store a likelihood table showing categories of leak size versus estimated pipe material. Where a user is confident that they know the pipe material and that the leak is on the pipe itself and not on a joint or fitting, it is possible to use the material table in forming a judgement about the size of the leak.

The foregoing table may be used independently to indicate leak size or used in combination with other likelihood tables such as that described for Surface RMS data in section 1.8).

6) Incorporating Estimates of Pipe dimensions Experience suggests that the magnitude of Surface RMS data as described in section 1, are significantly less across all categories of leak on large diameter mains pipes than for the same categories of leak on small diameter pipes.

It is relatively easy to determine whether a leak being investigated is on a large diameter main or on a smaller communication or service pipe. It is accordingly possible to maintain more than one set of likelihood tables describing the relationship between the various forms of overleak RMS data described in section 1.6 and/or other data and leak size category for different sizes of pipe. Whichever is appropriate for the leak being investigated is then selected.

Alternatively, a single likelihood table may be used, but the measured RMS values and/or other data may be adjusted to compensate for the effect of pipe diameter or indeed any other known characteristic of the pipe.

Pipe dimensions also affect the likely rate of leakage once failure occurs. A likelihood table describing this effect may be used independently or in combination with other likelihood tables such as that described in section 1.8 to relate pipe diameter to leak size.

7) Incorporating Estimates of identity of Failing Component Where a large body of data has been collected, reliable trends may be observed between types of failed pipeline component and leak size. Such classes of component may include ferrules, in-line joints, joints at stop tap boxes,and failures on the pipe in mid section.

To estimate leak size, it is possible to store a likelihood table showing categories of leak size versus likely failed component type. Where a user is confident that they know what the failed component type is, this table will be used in forming a judgement about the size of the leak.

8) Naive Bayes Classification To calculate the probability of a leak being in any particular leak size category, it is possible to use the well-known procedure of Naive Bayes Classification. When a diagnostic feature is measured, the likelihoods of the leak belonging to each leak size category are calculated by consulting the stored likelihood tables.

These likelihoods may be combined with likelihoods associated with other diagnostic features to give the overall relative probabilities of the leak belonging to each size category.

Means are preferably provided for updating the stored likelihoods or adding new diagnostic likelihood tables.

In certain circumstances, the prevailing leakage rate in a particular geographic area may be known. In other words, all other factors being equal, the user may know from experience that leaks in this area are either usually small or big. This information may be incorporated in a likelihood table and combined with the other likelihood data to produce estimates of the relative probabilities of the leak size categories.

9) Ranking Leaks in order of severity A number of leak size assessments and associated measured data arising from different leakage incidents may be stored. An operator such as a manager may retrieve these and the system may then rank them in order of leakage severity.

Means may be provided to rank leaks in order of increasing or decreasing leakage severity. Ranking may be based on the magnitude of the probability of the leak being in a particular leak size category or group of categories. Alternatively, ranking may be by statistically expected leak volume. Where ranking by probability assigns more than one leak to a particular category, the leaks within that category may be further ranked by reference to the magnitudes of any quantity discussed in section 1.6 and/or any other diagnostic feature. Ranking may also be by direct reference to the magnitude of any of the quantities discussed in section 1.6 and/or the magnitude of any other diagnostic features either in isolation or in combination with any other diagnostic features. For such ranking, the magnitudes of different features may be combined mathematically.

10) Dry hole avoidance The process of locating underground leaks is errorprone.

Frequently excavations are made in the wrong place. These are called 'Dry Holes'. Means may thus be provided for alerting the operator in certain situations when a dry hole is likely to be dug.

One possible set of dry hole alert signs are: a) Low surface vibration at suspect leak position but medium to high vibration on stop tap and/or, b) Low surface vibration at suspect leak position but medium to high energy in correlation signature.

Other measurands may be used to indicate dry hole likelihood.

Based on the experience of a large number of site surveys and excavations, a Dry Hole likelihood table can be stored showing the likelihood of dry holes for particular combinations of measurands.

Table 2 shows a hypothetical example.

Table 2 lilustrative table of Dry Hole likelihoods Min_Overleak Likelihood of dry hole RMS when Ranges Min_StopTapRMS >Alert Limit 0.0 to <2.5 0.65 2.5 to <5.0 0.20 5.0 to <7.5 0.1 0 7.5 to 10.0 0.05 Sum 1 The dry hole likelihood can thus be displayed to the operator.

The measuring means of the present invention may comprise two parts:

a) A field instrument used by a leakage technician

b) Software on a computer used by a leakage manager

The field instrument

The field instrument may be embodied in a number of forms, for instance, and not exclusively as: a standalone purpose built instrument designed solely for leak sizing, part of a ground microphone system used for leak detecting and/or locating, part of a leak noise correlator system used for leak detecting and/or locating.

The field instrument can be used to carry out a standalone assessment of leak size based on data collected at the site. The instrument may also tell the technician whether the leak is big enough to warrant excavation. This latter function may be governed by settings on the instrument specified by the technician's manager.

The field instrument may store raw data, processed data and/or leak size assessments of more than one leakage incident. The instrument may transfer this data to an external device such as a computer. The transfer may be by any means including wireless telephony. The instrument may provide a means for recalling the details and data from past assessments for review.

The field instrument may receive updated or new likelihood tables or other operating instructions from an external device such as a remote computer. The transfer of such data may be by any means including wireless telephony. The data may also be input manually using appropriate controls on the instrument.

In addition to measuring data and storing data such as described in the foregoing sections, the instrument may also record and transmit information such as site location, other site information such as site photographs and/or sketches and administration details. Site location may be by a Global Positioning System or similar.

The field instrument may comprise more than one module.

For example, there may be a first module containing the measuring means and a second module containing the analysis and display means. Communication between the modules may be effected by any suitable means, for example, by wireless communication such as "Blue Tooth".

The instrument may have controls and displays to guide the operator through the process of collecting one or more of the data sets and other data required for making a leak assessment.

Monitoring levels of vibration on the ground surface above a buried leaking pipe is useful in tracking the precise location of a leak. The most likely location is usually below the point where the ground vibration is the maximum. Listening sticks and ground microphones exploit this feature. Listening sticks require a considerable degree of skill. Ground microphones are perceived to be difficult to use in normal street conditions where the operator needs to be alert to the sounds around him. Simple ground microphones also only give point measurements requiring the operator to remember loudness profiles as he moves the head over the ground surface. This can be tricky when there are multiple peaks caused by, for instance, uneven pipe depths. Both listening sticks and ground microphones require good hearing.

In accordance with the present invention, the listening head may act like a ground microphone, but the measured values may be displayed on a simplified map of the ground surface on a screen or other display on the field unit. Each measurement may be seen in the context of its neighbours. No headphones are needed although facilities may be provided for direct listening to the leak noise. Measurements will be always consistent as there is no need for interpretation by hearing.

The user will put the listening head on the ground over the suspected leaking pipe. His field unit screen may show a line representing this pipe. The line may have scaling marks drawn on it representing typically 0.5 metre increments over the ground surface. The line may also show a user-defined datum point such as a stop tap chamber, chalk mark or other real world feature.

This datum point and other such features may be 'drawn' on the line by placing ready-made pictures or'icons' copied from an array of such icons pre-drawn on the display.

Let us assume that the user has placed the listening head over the pipe 1. 5 metros from the stop tap. To take a measurement, he may simply tap his pointer on the screen display three divisions up the line from the stop tap symbol to highlight the third division and then press a function button. The system takes a measurement by recording up to, for example, 5 seconds of surface vibration data. This record may be analysed in a similar manner to that described above for processing the Raw Over Leak Signature in section 1.2 and the minimum RMS value may be calculated. This value may be plotted as a point on a two-axis graph in which one axis (the x-axis) runs parallel to and is scaled identically and matches point for point the drawn pipeline. In the example being considered, the measured value will be placed at a point three divisions up from the stop tap position on the x-axis and 50% along the "y" axis.

The user now moves the listening head to a new position along the surface over the pipe, e.g. 2 metros from the stop tap.

To take this measurement, he highlights the 4th division from the stop tap symbol and presses the function button. The new measurement is drawn one division further along the graph x axis and if this measurement is bigger than the first, it will be set to a value of 90% and the first measurement will be reset from 50% to 10%. These values will be reversed if the second measurement is smaller than the first measurement.

Further measurements may be carried out in like manner.

Automatic y-axis scaling may be provided to always ensure that the biggest measurement is recorded as 90% and the smallest as 10% with other measurements being linearly scaled between these two limits.

When the ground has been probed along the pipe route, a graph drawn on the field unit screen will show the relative magnitudes of the minimum RMS surface vibration or other measurements along the pipe. Peak positions on this graph will be points of interest for the technician in his search for leaks. The field unit may record this graph and include it with the other site information gathered during leak sizing activities.

The leak location system may be equipped with adjustable filters to, for instance, reject low frequency noise.

Automatic Paperwork Generation Using the information held on or transmitted by the field unit in conjunction with data held on or available to the remote computer, the remote computer or the field unit may be programmed to automatically generate leak repair administration paperwork such as job numbers, street works notices, location

maps, leak description and site sketches.

Software on a computer used bv a leakage manager The software may collate data collected from one or more field instruments so that, for instance, a manager can prioritise leak repairs based on the field assessments of number of leaks in his area. Prioritisation may be based on the ranking procedures discussed above. When the details of a fresh leak arrive at the computer, the software will shuffle it into the appropriate position on a ranked list. The manager will then be able to see at a glance the top ten, twenty, thirty, etc. leaks in his patch. The software may incorporate likely repair costs and may have a facility for re-ranking the list using this repair cost data to optimise the manager's leakage recovery versus repair costs.

The software may also be able to interrogate field

instruments to ensure that the repair management system is being applied correctly.

The software may record and update statistical data relating to leakage incidents. It may use this information to calculate new or updated likelihood tables for use on field instruments. It may program field instruments with such tables and the appropriate methods of combining them.

Claims

Claims: 1. A method of assessing the size of a leak in a buried water

distribution pipe which includes measuring the surface vibration level on the ground surface above the suspected or known position of the leak in the buried water distribution pipe.
2. A method as claimed in Claim 1, which includes

measuring the Background Signature.
3. A method as claimed in either of the preceding claims, which includes measuring the local water pressure and leak power.
4. A method as claimed in any one of the preceding claims, which includes measuring the over-leak vibration spectral characteristics,
5. A method as claimed in any one of the preceding claims, which includes measuring the stop tap signatures.
6. A method as claimed in any one of the preceding claims, which includes obtaining leak noise correlation data.
7. A method as claimed in any one of the preceding claims, which includes obtaining pipe material and/or pipe dimension data.
8. A method as claimed in any one of the preceding claims, which includes the use of failed component identity estimates.
9. A method as claimed in any one of the preceding claims, which includes the use of data concerning prevailing local leakage rates.
10. A method of assessing the size of a leak in a buried water distribution pipe substantially as hereinbefore described.
11. Means for assessing the size of a leak in a buried water distribution pipe which includes means for measuring the surface vibration level on the ground surface above the suspected or known position of the leak in the buried water distribution pipe.
1 2. Assessment means as claimed in Claim 11, which

includes means for measuring Background Signature.
13. Assessment means as claimed in Claim 11 or Claim 12, which includes means for the utilisation of data concerning local water pressure and leak power.
14. Assessment means as claimed in any one of Claims 11 to 13, which includes means for measuring over-leak vibration spectral characteristics.
15. Assessment means as claimed in any one of Claims 11 to 14, which includes means for the use of data concerning stop tap signatures.
16. Assessment means as claimed in any one of Claims 11 to 15, which includes means for utilising leak noise correlation data.
17. Assessment means as claimed in any one of Claims 11 to 16, which includes means for utilising data concerning pipe materials and/or pipe dimensions.
18. Assessment means as claimed in any one of Claims 11 to 17, which includes means for utilising data concerning failed component identity estimates.
19. Assessment means as claimed in any one of Claims 11 to 18, which includes means for utilising data concerning prevailing local leakage rates.
20. Means for assessing the size of a leak in a buried water distribution pipe substantially as hereinbefore described.