FIELD OF THE INVENTION

The invention relates to the detection of leaks in liquid-filled pipelines, such as buried natural gas pipelines, subterranean oil pipelines and other pipelines requiring the testing of long lengths, often characterise by conditions of difficult access." BACKGROUND ART

In the testing of such pipelines in the prior art, the section of pipeline under test is filled with water o other fluid, which is pressurized, and the pressure in th pipeline monitored so that a drop in pressure indicative of a leak, may be detected.

The pressure in the pipeline is significantly affected by temperature variations, and in many, cases the pipeline temperature will vary in a diurnal cycle, so monitoring of the pressure over a twenty-four hour period is normally required.

In order to determine the average temperature of suc pipelines, measurements must be taken at intervals along the length, and apart from the difficulty and expense of this, such measurements may not be truly representative o the whole pipeline and this can lead to incorrect conclusions regarding the absence of leaks in the section of pipeline under test. SUMMARY OF THE INVENTION

In accordance with the present invention, the need for prolonged multiple measurements of temperature and th monitoring of pressure over a long period are avoided, by a technique in which measurement of the acoustic velocity in the pipeline is substituted for temperature measurement.

In one broad form, the invention resides in a method of testing for leaks a length of liquid filled pipeline comprising the steps of (a) obtaining at first and second times first and second measurements of the hydrostatic pressure at a

selected location in the pipeline,

(b) obtaining at said first and second times first and second measurements of the average acoustic velocity in said length, and (c) examining the degree of correlation between the change in hydrostatic pressure between said first and second measurements and the change in acoustic velocity between said first and second measurements.

Preferably the method comprises the further steps of (d) obtaining for the said pipeline the relationship between the pipeline temperature, and the rate of change of acoustic velocity with hydrostatic pressure,

(e) obtaining an estimate of the mean pipeline temperature, (f) deriving from said change in hydrostatic pressure the correlative change in acoustic velocity, and

(g) comparing said correlative change with the change in acoustic velocity between said measurements.

In another form, the invention broadly resides in a method of testing for leaks a length of liquid-filled pipeline comprising the steps of

(a) obtaining at first and second times first and second measurements of the hydrostatic pressure at a selected location in the pipeline, (b) obtaining at said first and second times first and second measurements of the average acoustic velocity in said length,

(c) deriving from said first and second measurements of average acoustic velocity first and second average pipeline temperatures; and

(d) examining the degree of correlation between the change in hydrostatic pressure between said first and second measurements and the change in average pipeline temperature between said first and second measurements. In preferred embodiments of the invention, the acoustic velocity is determined by measuring the transit time of an acoustic wave transmitted from one end of the length under test and returning to the point of

transmission after reflection from the other end of the length of pipeline.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is a schematic illustration of a pipeline tes set-up embodying the present invention given by way of example only;

Fig. 2 shows a curve plotting the relationship between the rate of change of pressure with acoustic velocity against temperature, for a typical pipeline; Fig. 3 shows the waveform of the transmitted acousti signal in an example of operation of the apparatus of Fig

1;

Fig. 4 shows the waveform of the reflected signal;

Fig. 5 shows the curve correlating the predicted source and reflected signals; and

Fig. 6 shows the predicted and actual reflected signals. DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

In the embodiment illustrated in Fig. 1, a water- filled pipeline 10 is fitted with a dead-weight tester 11 for the purpose of measuring the hydrostatic pressure within the pipeline, a low frequency oscillator 12 in communication with the interior of the water-filled pipeline, and an acoustic transducer 13, these components being located at one end of the pipe section under test.

The acoustic velocity in liquid may be expressed as

t÷) 1/2

(1)

where 7" is the ratio of specific heats for the liquid, K is the fluid bulk modulus of the liquid, and p is the density of the liquid.

When water is contained within a pipe the acoustic velocity is reduced by the elasticity of the pipe wall an may be expressed as

-- 1 X//2L

» a

^■ . (-¥^ ^{♦ 1} ) ( 2 )

where D is the pipe inside diameter, C is a factor based on the restraint conditions of the pipe and is discussed below, E is Young's Modulus of elasticity of the pipe, and t is the pipe wall thickness.

The factor C in equation (2) is dependent on the restraint conditions of the pipeline. Restrained pipelines, typically buried pipelines, have zero longitudinal strain, and for such pipelines C is given by ° the expression

2 t (3)

(1 + U) + D + t (1 - V )

where Λλ, is the Poisson's ratio for the pipe material.

For unrestrained pipelines, typified by offshore pipelines laid uncovered on the seabed and free to expand longitudinally,

° - ¥ ^{2 t »ι »■} • ^> ir υ i -i 2- O \ ( ₄ )

It will furthermore be appreciated that pressure and temperature in such pipelines are related, and it can be shown that

. f = ^R _D ^ - 2 ^«) (5)

where Kp is an equivalent bulk modulus for a liquid filled

pipe, given by

and Cλ an β are respectively the coefficients of linear and thermal expansion of the pipe material.

Acoustic velocity in liquid changes with temperature and for the case of a liquid-filled pipe equation (1) may be written as

Considering Kp and p to be functions of temperature, and relating the density term to the expansion coefficients, it can be shown that

da dK

+ (β - 2α) dT I dT (8)

From equations (5), (8) and (6) and it can be shown that the differential of acoustic velocity with respect t pressure can be expressed as follows

dP 2 K da K_ IXC (9)

Uβ - 2 ) dT ⁺ Et

This relationship is illustrated in Fig. 2, where dP/da is plotted against temperature, for the case of a water-filled pipeline having a ratio of internal diameter to thickness D/t of 75.6.

The variation of dP/da with temperature is mainly du

to the influence of the term dK/dT, the change in bulk modulus with temperature. For an unrestrained pipeline, the values are approximately 1% higher than for the restrained pipeline at the same temperature. If then an acoustic wave is transmitted from the oscillator 12 and the time taken for the passage of that wave to the end of the pipe section and its return to the transducer 13 is determined, the acoustic velocity derived from that transit time will be a measurement resulting from contributions from the whole of the test length, and this will permit leak testing to be carried out without concern for thermal stabilisation of the pipeline, and without the need for testing over a 24 hour duration to eliminate diurnal affects on exposed pipe portions.

In the preferred embodiment of the invention, the transmitted signal generated by the oscillator 12 is detected by the transducer 13 and recorded for subsequent processing, and the output of the transducer after the transmission of the source signal is recorded and subsequently scanned to identify the reflected signal.

Fig. 3 shows a typical source signal waveform, generated by a solenoid driven oscillator of the kind which is described in applicant's co-pending Australian Patent Application entitled 'Low Frequency Pressure Oscillator', while Fig. 4 shows the end reflected signal received by the transducer 13 in a typical pipeline measurement.

In order accurately to relate the reflected signal to the source signal, since the latter is significantly modified by the transmission characteristics of the pipeline, it is preferred to produce a prediction for the waveform of the reflected signal based upon the known characteristics of the pipeline. The curve correlating the predicted source and echo signals for the example in question is shown in Fig. 5, and the predicted and measured echo signals are compared in Fig. 6. The degree of correspondence between these signals may be improved by modifying the factors used in deriving the predicted signal.

In this way, an accurate measure of the acoustic velocity in the pipeline may be obtained, with sensitivities of the order to 0.05 m/s. Over the period between velocity measurements it will be expected that th hydrostatic pressure indicated by the dead-weight tester 11 will change, as will the temperature and the acoustic velocity. Providing the mean pipeline temperature can b estimated with reasonable accuracy, the rate of change of pressure with velocity for that pipeline may be determine as described above. From the measured change of velocity, the equivalent pressure change which would be expected may then be calculated, and this compared with the change in pressure actually measured. If the difference between those figures is less than a previousl determined allowable difference, then the test will be regarded as having produced acceptable results, while a difference which is greater than the predetermined allowable difference will indicate a leak.

As in the prior art, some figure of acceptable liqui loss must be adopted, and typically in the prior art a loss of around 100 litres in 24 hours from a pipeline pressurised to 10,000 to 20,000 kPa may be regarded as acceptable, as this will represent a leak which is so small as to be self-repairing and impractical of location As in the prior art, the volume/pressure characteristics of the pipeline, may be used to convert that liquid loss to an allowable pressure difference for the purposes of the present invention. For example, a fall of 20 kPa in pressure over the test period may be accompanied by a measured fall of acoustic velocity of 0.1 m/s. At 15°C, for this pipeline dP/da may be 150 kPa/m/s so the pressur change equivalent to the measured change of velocity woul be a fall of 15 kPa. The difference between the measure and predicted pressure loss is 5 kPa which would be well within a typical allowable difference of 35 kPa.

An estimate of the effective pipeline temperature for the purpose of the method of the present invention may be obtained by calculation from the initial measured acousti

velocity and pressure. Alternatively, a temperature probe may be inserted into the ground and a ^' stable temperature reading obtained prior to testing (for example, while strength testing of the pipe is being carried out) .

It is to be understood that, since the measurement of the acoustic velocity enables the determination of the mean temperature of the pipeline, in the practice of the present invention, instead of working from the observed change in acoustic velocity, the thus measured temperature change may be used, and correlated with the change in pressure, as in the prior art.

As indicated above, the acoustic wave may be generated by means of an oscillator of the kind described in the above mentioned co-pending application, although of course any suitable transducer, capable of producing a repeatable acoustic wave within the pipeline at the temperatures and pressures employed, may be used.

The choice of frequency employed for the acoustic wave is governed by the absorption which takes place in the pipeline. As discussed by Kinsler and Frey, Fundamentals of Acoustics, John Wiley & Sons Inc. , New York, 1962, absorption occurs in a liquid filled pipe due to viscous resistance at the walls, and the attenuation varies according to the square root of the angular frequency. Since this attenuation increases with frequency, the length of the pipeline and the sensitivity of the detection equipment will impose an upper limit on the useable frequency range. For a 50 km pipeline with a 40 dB attenuation of the reflected signal, the maximum frequency which can be used is of the order of 16 Hz. Higher frequencies can of course be used for shorter pipeline lengths, and for pipelines in the range of 50 km to 1 km, frequencies in the range of approximately 10 to 200 Hz permit echo detection where a 40 dB attenuation is acceptable.

The transducer 13 is preferably an electronic pressure transducer. Such a device capable of providing

a minimum pressure resolution of 10 kPa or less, will be found satisfactory. Preferably, the source of acoustic wave and the signal detector should be attached at the same location as the dead-weight tester or other hydrostatic pressure measuring device. The signal detector may be remotely located provided any influences of the connecting pipe or hose on the measured acoustic velocity, are assessable.

Considerable care should be taken during the fillin of the pipe, to minimise the presence of free air bubbles or air pockets within the pipeline. Normally, the leakage testing will be carried out immediately following the strength testing of the pipeline under water pressure and the presence of air in the pipeline will be revealed during pressurising prior to testing. Care must be exercised in the control of the pig during liquid filling if the effects of free air are to be avoided. It is significant to note that solution of air during the test period will produce a drop in pressure and an increase in velocity, and this may be initially interpreted as a leak due to the fact that the pressure will be dropping while the acoustic velocity is increasing. Such solution of air is therefore not detrimental to the technique since i will not mask any leak which may exist, but rather produc a spurious indication of the existence of a leak.