|Publication number||CA2612835 A1|
|Application number||CA 2612835|
|Publication date||28 Dec 2006|
|Filing date||19 Jun 2006|
|Priority date||21 Jun 2005|
|Also published as||CA2612835C, EP1896842A1, EP1896842A4, EP1896842B1, US7328618, US7475596, US20060283251, US20080078247, WO2006136013A1|
|Publication number||CA 2612835, CA 2612835 A1, CA 2612835A1, CA-A1-2612835, CA2612835 A1, CA2612835A1, PCT/2006/996, PCT/CA/2006/000996, PCT/CA/2006/00996, PCT/CA/6/000996, PCT/CA/6/00996, PCT/CA2006/000996, PCT/CA2006/00996, PCT/CA2006000996, PCT/CA200600996, PCT/CA6/000996, PCT/CA6/00996, PCT/CA6000996, PCT/CA600996|
|Inventors||Osama Hunaidi, Marc Bracken, Alex Wang|
|Applicant||National Research Council Of Canada, Osama Hunaidi, Marc Bracken, Alex Wang|
|Export Citation||BiBTeX, EndNote, RefMan|
|Classifications (17), Legal Events (1)|
|External Links: CIPO, Espacenet|
Non-Destructive Testing of Pipes FIELD OF THE INVENTION
100011 This invention relates to the non-destructive condition assessment of pipes, and in particular buried pipes or pipes that are otherwise inaccessible, such as fire sprinkler systems and pipelines in the oil and gas industry.
BACKGROUND OF THE INVENTION
 Buried pipes, such as pipes used in municipal water systems, lose metal resulting in a reduction in pipe wall thickness, as a result of internal and external corrosion. Sustained wall thickness loss eventually causes the pipe to fail. Such pipes have to be periodically inspected and evaluated for structural adequacy. Accurate information about pipe wall thickness is needed to determine the structural capacity of pipes. Also, monitoring of pipe metal loss is needed to estimate the remaining service life of pipes, which is an important part of maintenance management systems of large pipe networks (e.g., urban water or gas distribution systems). It is therefore desirable to be able to assess the condition of buried pipes in a non-destructive and non-intrusive way, that is without damaging the pipe and without taking it out of service.
 One way of determining pipe wall thickness is to obtain pipe coupons (or samples) and then measure the thickness off-site using thickness callipers. Also, wall thickness can be measured using an ultrasonic gauge at locations where the pipe is naturally exposed or intentionally excavated. Both methods provide only a discrete measurement of wall thickness. An alternative method for obtaining a continuous measurement of pipe wall I
thickness employs ultrasonic or magnetic flux leakage or remote-field eddy current devises which are launched inside pipes using robots (known as pipeline pigs).
100041 Excavating pipes for the purpose of measuring wall thickness is costly, as it requires excavating the pipe at several locations. Also, thickness values based on the coupons may not be statistically representative of the overall condition of the pipe unless a very large number of locations are excavated. Pipeline pigging requires taking the pipe out of service.
It is strictly suitable for large diameter pipes with simple geometry (i.e., no abrupt bends).
Data acquisition and analysis is very intensive and expensive. The method is used mostly for oil and gas transmission pipelines. It is not suitable for water distribution systems, which are geometrically complex and extensive networks comprised of small-diameter pipes that are usually filled with debris and have no pig launching stations.
SUMMARY OF THE INVENTION
100051 In accordance with the invention, the pipe wall thickness is determined from the average propagation velocity of low-frequency acoustic disturbances in the pipe (e.g., pressure fluctuations). Low-frequencies typically have a value of less than about 1000Hz.
The propagation velocity depends on the internal diameter and circumferential thickness profile of the pipe, density and bulk modulus of elasticity of the fluid, and Young's modulus of elasticity and Poisson's ratio of the pipe wall material. The velocity can be calculated using the equations derived theoretically. In the proposed method, the pipe wall thickness is back calculated from these equations using the measured value of the propagation velocity and the above pipe and fluid parameters, which are usually known or easily found.
r 11/20/07 17: 33 FAX 61~ 230 8$21 MAR~S & CLER.R (j004 .. ~ .PCTICA ZoOb/ Q40996 ' . ~ ~ NOV=~ 2007 2 ~
. oiI . Qx  According to the preserit ir~vention therc is provided a method for the non-deStructive condition assessrnent Qf a pipe carrying afluid, comprising detecting an at,oustic disturbance propagating past two 1ongitudinally separated points on said pipe;
. determining an actual value representative of the propagation vei*ocity of said acoustic disturbance; cornputing e cortesponding predicted value for the propagation velocity as a .funrtaon af at least one wall thickness parameter of said pipe by using a theoretical model fnr the propagatfon of acoustic waves hi said pipe that assumes said pipe has a finite wall thickness with a predeterrmined non-uniform circumferential thickness profilc and takes I
. into account stresses across the thickness of the pipe walI, and wherein said wa11 thickness pararneters are representatYve of wail thickness at different circurrrfcrential locations on said pipe; and compu#ing said af least one waii thickness parameter by matching said actuai value with said predicted value.
(00071 The acoustic di sturbance can propagate in the in-bracket mode, where the source Qf tbe dasturbAnce lies between the two points and the distarbance propa,gates in opposite :
., directions ljram the source to the separatcd points, or in the out-of-bracket tnodc, where the a.coustic source lies oeitside the two points, afld the disfiutbance propagates from the source and sequentially past the two points in the saune direction.
 The propag,ation velocity, usually the average prnpag$tion rrelacity, can be detearmmined onrisite non-destructively and non-intrvsively by measuring the time delay between acQustic signals measured at two poEnts Qn the pipe that a,re a known distance apart (e.g., ccntrol valves or fire hydrants). Ambient noise in the.pipe can be iased as the acoustic source. .
10009J Alternativeiy, acoustic noise created by releasing pressurized fluid or by a specially designed device at an in-bra,cket 4r out-of-bracket potnt (e.g., a fire hydrant in the case of water distribution pipes)Or acoustlc noise can be used. Acoustfic signals can l,e measured using a pair of'vibration sensors (e.g., accelerometers) or hydropttonGs. The . AMENDEDSHEET
time delay between the two measured signals is determined using the well-known time-of-flight method or the cross-correlation function (traditional or enhanced) calculated in the time or frequency domains as described in U.S. patent 6,453,247, the contents of which are herein incorporated by reference. The average propagation velocity in the pipe can then be calculated by dividing the difference between the distances of the sensors from the acoustic source by the time delay.
100101 In addition to determining the average thickness, the acoustic damping capacity of the pipe material can also be determined based on the attenuation of the acoustic signals described above. Alternatively, the damping capacity can be determined based on the width of the frequency spectrum peak that corresponds to the fundamental ring frequency of the pipe or based the logarithmic decrement of transient ring vibration. The frequency spectrum and logarithmic decrement can found using a simple mechanical impact test at a location where the pipe is exposed, e.g., in an access manhole. The damping capacity of the pipe material can be used to obtain information about its tensile strength, especially for cast iron pipes, based on published relationships. This is helpful in view of the fact that the tensile strength of cast iron pipes varies widely, depending on their source of supply and vintage.
For cast iron pipes, the damping capacity increases as their strength decreases, which depends on the carbon content and cooling rate. Carbon content and cooling rate affect the matrix and graphite structure of cast iron and in turn its stress-strain non-linearity. Non-linearity in the form of a hysteresis loop leads to high damping capacity.
100111 The method in accordance with the invention has the advantage of being non-destructive and non-intrusive. All required measurements can be made from the ground surface. The length of the pipe section represented by the average profile can be arbitrarily chosen. Initially, a 100 metres long section, which is the usual distance between fire hydrants in urban areas, can be chosen. Subsequently, if higher resolution is needed, small access holes to the pipe may be drilled, e.g., using keyhole vacuum excavation technologies, to measure the thickness over shorter pipes sections. Alternatively, arrays of closely spaced hydrophones may be inserted into pipes thru fire hudrants or corporation stops. The measurement and calculation of the pipe wall thickness can be easily made using Windows-based PC software. The proposed method is easy to implement (e.g., in conjunction with routine leak detection surveys). Also, it does not require a high level of operator skill.
BRIEF DESCRIPTION OF THE DRAWINGS
100121 The invention will now be described in more detail, by way of example only, with reference to the accompanying drawings, in which:
 Figure 1 a shows an arrangement for measuring propagation velocity of buried pipes in accordance with a first embodiment;
100141 Figure lb shows an arrangement for measuring propagation velocity of buried pipes in accordance with a second embodiment;
 Figures 2a and 2b show PC screens displaying the results of a wall thickness measurement in accordance with an embodiment of the invention;
 Figure 3 shows the variation of predicted pipe thickness as a function of bulk modulus of water; and  Figure 4 shows the frequency spectrum of impact-echo tests on ductile iron sample of reference pipe; and 100181 Figure 5 shows the predicted wall thickness of reference pipe using adjusted bulk modulus of elasticity of water of 1.95 GPa at 20.8 C.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
100191 In order to assist in understanding the invention, it is helpful to consider the underlying theory.
100201 The propagation velocity of acoustic waves in an unbounded fluid body is defined by the following equation:
vo = yK/ p (1) where K is the bulk modulus of elasticity of the fluid, p is its density, and y is the ratio of the specific heats of the fluid (i.e., the ratio of the fluid's heat capacity in a constant pressure process to the heat capacity in a constant volume process).
 The derivation of this equation can be found in Pearsall, I.S. "The velocity of water hammer waves", Proceedings of the Institution of Mechanical Engineers, Vol. 180, Part 3E, pp. 12-20, 1965, the contents of which are herein incorporated by reference.
100221 For a nearly incompressible liquid, which is the case under normal temperatures and pressures in water distribution pipes, y is approximately equal to unity.
Then, Eq. (1) becomes:
vo = K l P (2) 100231 The elasticity of the wall of a pressurized fluid-carrying pipe reduces the velocity of acoustic waves in the fluid. The amount of the reduction of the velocity depends on the size and shape of the cross-sectional area of the pipe and the elastic modulus of the material of the pipe. The general equation of the velocity of acoustic waves in the fluid is defined by the following equation:
v=vo 1+yK~~ (3) where A is the cross-sectional area occupied by the fluid, p is pressure inside the pipe, and vo is defined by Eq. (1). For a nearly incompressible liquid, yis approximately equal to unity. This equation is explained in Wylie, E. B., and Streeter, V. L. Fluid Transients, McGraw-Hill, New York, 1982. the contents of which are herein incorporated by reference.
[00241 For such fluids, Eq. (1) becomes:
v=v~ rl+K~~l IA (4) where vo is defined by Eq. (2).
100251 For accurate estimation of the velocity, it is necessary to consider the variation of K and p with the temperature, air content, and pressure of the fluid. For pipes having very thick walls or for gas-filled pipes, ~A is negligible and therefore v~ vo. On the other P
hand, for very flexible pipes, the acoustic velocity may be approximated as:
v = a (5) p  The cross-sectional area strain of the pipe, aA/A, is caused by the change in total circumferential strain, 8cB, which in turn is caused by the change in pressure, ap. The total circumferential strain is equal to:
as _ a6e ~ca6, (6) a E
where (r, and 6Q are circumferential and axial stresses in the pipe wall, respectively, and E
and is the elastic modulus and Poisson's ratio of the pipe material.
100271 For a circular pipe, and considering a small angular part of the cross-sectional area defined by the angleaB, the partial change in the area is:
aAe = (RaB)auR (7) where R is the internal radius of the pipe and auR is the change in the radial displacement at radius R. Dividing Eq. (7) by (A c'p) and integrating over 9leads to:
l aA 1 2 JauR d9 (8) A ap ;rRap o where the change in the radial displacement at radius R, OUR, is equal to:
au _ Ra9 + ase RaB _ R
R ae (9) = 8CB R
100281 For a pipe that is thin-walled, that is where the ratio between its diameter and thickness is large, stresses in the pipe wall are nearly uniform across its thickness.
Consideration of the equilibrium of forces acting on half of the cross-sectional area of the pipe wall shows the circumferential stress to be:
6B = Rp and a6e = Rap (10) t t where t is the thickness of the pipe wall defined by:
t= f(B) (11)  For a pipe that's free to move axially, e.g., with expansion joints throughout its length, or for a pipe material having a high modulus of elasticity, e.g., steel or concrete, the change in axial stress, a6, , is negligible and can be assumed to be zero.
For other fixation conditions, the change in the axial stress can be significant and should be taken into account. For example, for a pipe constrained against axial movement throughout its length, ct = 0 and hence: a6, =P'06B (12)  Substituting Eq. (10) and (12) in (6), Eq. (6) in (9), and finally Eq.
(9) in (8) leads to (assuming that the wall thickness profile is symmetric about the pipe's centreline):
laA r 2ird9 A ap ,rE O f(e) (13) D dO
_ C-;,rE j.f(B) where D is the diameter of the pipe and c is a factor that takes into account the fixation method of the pipe and is given by:
rl for a pipe having expansion joints throughout its length or having a high modulus of elasticity c= jl 1-~ll z for a pipe constrained against axial movement throteghottt its length (14)  Pipes buried in corrosive soils may not loose their thickness evenly around the circumference. Most likely, this is due to the formation of corrosion cells on the outer surface of the pipe as a result of differential aeration. This happens when the pipe's cross-section is partially in contact with two soils , for example a porous and hence well-aerated soil, such as soils used for backfill, and an impermeable and hence poorly aerated soil, such as native clay soils. Differential aeration may also occur on the internal surface of the pipe due to the accumulation of deposits at the bottom of the pipe. To take this phenomenon into account, the wall thickness of the pipe may be assumed to vary linearly with the angular coordinate as follows:
tmax tmin B+t for - f9r < B< f9r t = ~ min (15) t nax for fir <_ B<_ - fir \
wheref is the fraction of the pipe cross-section having non-unifonm thickness, and tmin is the minimum wall thickness at e= 0 and t,,,ax is the maximum wall thickness at B= 7T.
The mean thickness, tmean, is then equal to:
t _ f tmax 2 tinin + l(1 _ f)t,nax (16) mean  A linear circumferential thickness profile was selected in the example above for its simplicity and ease of demonstration. However, if more accurate results are required, more representative profiles can be used. Substituting Eq. (15) in (13) and evaluating the integral leads to:
1 aA = c D (1- f) + ln W tmax + ll - J) tinin 1- ln(t,nin ) (17) A ap E tmax tinax - tinin and substituting Eq. (16) in (4) leads to the following equation for the pipe's acoustic velocity:
V- V 1+ CD K (1 - f)+ 1nLJ tmax + ll - J) tmin 1- ln(tmin ) (1 g) tmax tmax - tmin where c is defined by Eq. (14).
 The validity of Eq. (18) can be checked by considering the special case of a pipe having uniform thickness profile, i.e., tmin = tmax = t= Since lim in(x) - ln(x - a) - 1 (19) a--->o x-(x-a) x Eq. (18) becomes:
v=vo 1+c~ K (20) t E
The velocity of acoustic waves given by Eq. (20) is the same as that derived in standard textbooks for fluid-filled cylindrical thin-walled pipes with a uniform cross-section.
100341 The errors resulting from assumption that the pipe is thin-walled, as is the case in US patent no. 6,531,032, are small for the purpose of locating pipe leaks.
However, for thickness measurements, the error may not be sufficiently small. For example, for a 152 mm cast iron pipe that is 10 mm thick with v = 1218 m/s, the back-calculated uniform thickness is equal to 9 mm, assuming that the pipe is thin-walled. This corresponds to a spurious loss of 10%. For more accuracy, full account must be taken of the variation of stresses across the thickness of the pipe wall.
 The radial displacement at the inner radius for an unconstrained pipe with expansion joints throughout its length, taking into account the variation of stresses across its thickness, is given by:
au ap D3 +D(1+~C> (21) R E 4t(D+t) 2 and for pipes with constrained axial movement throughout it's given by: ap [
D' auR = ~. 4t(D+r)(1-P2)+~(l+,u) (22) Halliwell, A.R. "Velocity of a water hammer wave in an elastic pipe", Proceedings of the American Society of Civil Engineers, Journal of the Hydraulics Division, Vol.
HY4, pp. 1-20, 1963.
100361 Substituting Eq. (15) in (21) and (22), Eq. (21) and (22) in (8), Eq.
(8) in (4) and evaluating the integral leads to the following velocity equation for a pipe with expansion joints throughout its length:
v=v,, F+E D F+2(1+,u) (23a) ;z and the following equation for pipes with constrained axial movement throughout its length:
D F(1-,u2)+2(1+fc) (23b) v=vo F+E
where F- (1-f);T + 1 In 2nfa + b - b2 -4ac -ln (b-lb2 -4ac (24) tmax (D + tmax ) b2 4ac 2nfa + b+ b2 - 4ac b+ b2 - 4ac a tinax - tinin (25a) b = ~ tmax tmin (2tmin + D) (25b) C = tinin + tminD (25c)  For a thick-walled pipe having a uniform thickness, Eq. (23) becomes:
v=vo I+D EK (~(1+~)+D+t) (26a) for a pipe with expansion joints throughout its length, and v=vo 1+D E(~(l+u)+D+t(1_P2)) (26b) //~ t for a pipe with constrained axial movement, which are in agreement with expressions derived in standard textbooks for fluid-filled cylindrical thick-walled pipes with a uniform cross-section.
 The above formulations for thin and thick-walled pipes with non-uniform circumferential thickness profiles do not take into account shear stresses in the pipe wall that are expected to develop as a result of having unaxisymetric radial displacement. In the formulation for thin-walled pipes, it is assumed that the cross-section of the pipe remains circular in order to simplify the calculation of the circumferential stress, i.e., Eq.
10. Similarly, to simplify the formulation for thick-walled pipes, radial displacements for pipes with uniform thickness profiles are used. A rigorous formulation, taking into account shear stresses in the pipe wall, is possible. However, such formulation would be too complex and the small gain in accuracy is unwarranted.
[00391 The above formulations for both thin and thick-walled pipes do not take into account the inertia of the pipe wall and the surrounding soil. It is assumed that the pipe responds statically to propagating dynamic pressure waves. At low frequencies, well below the ring frequency of the pipe, errors caused by this assumption are small as the pipe's response is dominated by the stiffness of its wall. For example, considering only the inertia of the pipe wall, the velocity equation for a pipe having uniform cross-section and with expansion joints throughout its length, the velocity equation is given by:
v=vo 1+D K 2 21 (27) t E 1-D CV Ppipel4E
where co is the frequency of the propagating pressure wave (in radians per second) and pp.pe is the density of the pipe wall material. The derivation of this expressions is explained in Muggleton, J.M, Brennan, M.J., and Pinnington, R.J. "Wavenumber prediction of waves in buried pipes for water leak detection", Journal of Sound and Vibration, Vol. 249, No. 5, pp. 939-954, 2002, the contents of which are herein incorporated by reference.
 For the case of a cast iron pipe that is 152 mm in diameter and 12 mm thick, the maximum frequency of acoustic noise signals is typically 800 Hz. The maximum error in velocity resulting from ignoring the inertia of the pipe wall is a miniscule 0.07% (E = 128 GPa, ppipe 7100 kg/m3, K = 2.157 GPa at 16 C). This is also the case for other metallic pipes.
100411 Change in the nominal internal diameter, due to wall thickness loss, is not taken into account. However, such changes can be easily incorporated, although the small gain in accuracy makes the additional effort unwarranted.
100421 In deriving the general form of the velocity equation, i.e., Eq. 3, based on conservation of mass in a infinitesimal pipe section, small terms are neglected. It is assumed that this has a negligible effect on the acoustic velocity. Even if the effect is not so, determination of the bulk modulus of elasticity of the fluid carried by the pipe as described later amounts to calibrating the whole thickness measurement method.
The back calculated bulk modulius implicitly includes corrections for approximations made in the derivation of the velocity equation, e.g., neglecting small terms and inertial effects.
 The minimum thickness of the pipe, tm;n, can be back calculated using Eqs. (18) or (23) for pipes with linearly varying wall thickness. The maximum thickness, tmax, is assumed to be equal to the original thickness of the pipe wall, which is usually known or can be measured using a single pipe sample. This is justified based on the observation that a certain part of the pipe's cross-section normally retains its original thickness. This part corresponds to the cathode of the corrosion cell created by differential aeration of the pipe's surface. The minimum thickness can also be calculated using Eq. (16) from the mean thickness back calculated using Eq. (20) or (26) for pipes having uniform wall thickness. This will lead to less accurate results than would be obtained using Eq. (23) for a thick-walled pipe with linearly varying wall thickness. For the same fixation condition, using Eqs. (18), (20), or (26) underestimates t,n;n. If the pipe can be assumed to have a uniform thickness profile (as may be found from experience for a particular area), the average thickness can be back calculated using Eq. 26.
100441 The velocity of acoustic waves in the pipe, v, can be measured by correlating acoustic noise signals, which can be ambient or created by intentionally by an acoustic source at a known location. The internal diameter of the pipe, D, bulk modulus of elasticity of the fluid, K, and its density, p, elastic modulus of the pipe material, E, and its Poisson's ratio, p, are usually known or can be easily found.
 In order to practice the method in accordance with the invention, it is necessary to know the velocity of sound in unbounded fluid, vo or the bulk modulus of the fluid, and its density. Considering the case of water distribution pipes, the velocity of sound in an unconfined body of pure water is given by:
v = 1.402385 x 103 + 5.038813 T 5.799136 x 10-2Tz +3.287156 x 10-4 T3 (28) - 1.398845 x 10-6 T4 + 2.787860 x 10-9 T5 where T is the water temperature in C. This expression is based on Marczak, W. (1997), Water as a standard in the measurement of the speed of sound in liquids.
Journal of the Acoustical Society of America, Vol. 102, No. 5, pp. 2776-2779, the contents of which are incorporated herein by reference.
 Table 1 lists the velocity of sound, density and the corresponding bulk modulus of elasticity for temperatures between 0 and 40 C. Density values are based on the Handbook of Chemistry and Physics, CRC Press, 85'h edition, 2004-2005, the contents of which are incorporated herein by reference. Bulk modulus values were obtained by using Eq. 2.
Table 1: Velocity of sound, density and bulk modulus of elasticity of water Temperature Velocity of sound Density of water Bulk modulus ( C) (m/s) (kg/m3) (GPa) 0 1402 999.8426 1.966 1 1407 999.9015 1.980 2 1412 999.9429 1.994 3 1417 999.9672 2.008 4 1422 999.975 2.021 1426 999.9668 2.034 6 1431 999.943 2.046 7 1435 999.9043 2.059 8 1439 999.8509 2.071 9 1443 999.7834 2.083 1447 999.7021 2.094 11 1451 999.6074 2.105 12 1455 999.4996 2.116 13 1459 999.3792 2.127 14 1462 999.2464 2.137 1466 999.1016 2.147 16 1469 998.945 2.157 17 1473 998.7769 2.166 18 1476 998.5976 2.176 19 1479 998.4073 2.185 1482 998.2063 2.194 21 1485 997.9948 2.202 22 1488 997.773 2.210 23 1491 997.5412 2.218 24 1494 997.2994 2.226 1497 997.048 2.234 26 1499 996.787 2.241 27 1502 996.5166 2.248 28 1504 996.2371 2.255 29 1507 995.9486 2.261 1509 995.6511 2.268 31 1511 995.345 2.274 32 1514 995.0302 2.280 33 1516 994.7071 2.285 34 1518 994.3756 2.291 1520 994.0359 2.296 36 1522 993.6883 2.301 37 1524 993.3328 2.306 38 1525 992.9695 2.311 39 1527 992.5987 2.315 1529 992.2204 2.319 (0047] Limited available information about the compressibility of lake waters indicates that it is not significantly different from that of pure water, see for example Dorsey, N.E.
(1940). Properties of Ordinary Water-Substance. Reinhold Publishing Corporation, New York, U.SA. However, the velocity of sound given by Eq. 28 is based on experimental results that have been obtained for pure water at MHz frequencies. Therefore, this equation may not be applicable for pure or ordinary water at frequencies below 1000 Hz, which dominate acoustic signals in water distribution pipes. It was shown by O'Donnell M., Jaynes E.T., and Miller J.G. (1981), Kramers-Kronig relationship between ultrasonic attenuation and phase velocity, J. Acoust. Soc. Am. Vol 69, No 3, pp 696-701that a system that exhibits attenuation, which is the case for water, must exhibit dispersion, i.e., dependence of phase velocity on frequency. Also, they showed that acoustic velocities at low frequencies are typically lower than those at higher ones. Consequently, the bulk moduli of water in distribution pipes may be lower than those listed in Table 1.
100481 Bulk moduli used for water hammer analysis in some texts and handbooks are the same as those listed in Table 1; while they are slightly lower in others.
Small discrepancies in the bulk modulus have a minor effect on water hammer calculations.
However, when bulk moduli listed in Table I were used with the proposed non-destructive testing method, the predicted thickness loss was overestimated.
The dependence of predicted wall thickness on the bulk modulus of water is shown in Figure 3 for a 12 mm thick, 152 mm 0 pit cast iron pipe having an acoustic velocity of 1247 m/s.
For example, using a bulk modulus equal to 2.2 GPa, as used by some texts at 20 C, predicted thickness would be 8.3 mm, while it would be 10.4 mm using a modulus of 2.03 GPa, as used by other texts at 20 C.
100491 The uncertainty regarding the appropriate value of the bulk modulus to be used for thickness calculations can be minimized in accordance with an embodiment of the invention by careful measurement. First, the acoustic velocity is measured for a "reference" pipe of a known wall thickness, diameter and Young's modulus.
Preferably, a recently installed pipe of a known class should be used and its Young's modulus should be measured dynamically for an exhumed or leftover sample. Then, the bulk modulus is back calculated from the appropriate theoretical acoustic velocity equation or a numerical model for water-filled pipes together with other already known or measured pipe parameters. Measurement of the acoustic velocity of the reference pipe and other pipes should be performed within few days of each other. This is to ensure that temperatures of the fluids in the pipes are similar.
 Alternatively, the bulk modulus may be determined from a relationship established from a set of measurements of the fluid's bulk modulus and temperature for the reference pipe at different times of the year. Determination of the bulk modulus as described here amounts to calibrating the whole thickness measurement method. The back calculated modulus implicitly includes corrections for approximations made in the derivation of the velocity equation, e.g., neglecting small terms and inertial effects.
100511 Following is an example of the determination of the relationship between the bulk modulus of elasticity of the fluid and its temperature. Acoustic velocity was measured for a recently installed 152 mm 0 Class 52 cement-lined ductile iron water distribution pipe having the following temperatures: 20.8, 11.7, and 5.3 C.
[00521 For a cement-lined pipe of circular cross-section, the acoustic wave velocity can determined by the following equation, given in Fluid Transients in Pipes and Tunnels:
Speed of Propagation of Pressure Waves. Engineering Sciences Data Item No.
83046, ESDU International, 1983, London, U.K:
v = KwlP,v (29) 1+K (D-ep)(I-P Z)[(DZ +ep)(1-f.c)+2,uDep]
EpDep(D-ep)(1-,u)+Ele,[(DZ +epz)(1- ft)+2,uDeP]
where v is velocity of acoustic waves in the pipe, K,v and p,,, are bulk modulus of elasticity and density of water, respectively; Ep, D, and ep are Young's modulus, internal diameter, and metal thickness of the pipe, respectively; El and el are Young's modulus and thickness of the cement lining, respectively; and p is Poisson's ratio, assumed to be the same for ductile iron and the cement lining.
 The thicknesses of the metal section and cement lining for 152 mm 0 Class 52 ductile iron pipes are 7.9 and 2 mm, respectively. The thickness of an "equivalent" ductile iron pipe with no cement lining is equal to 8.2 mm, as found from Eq. 29. The equivalent pipe has the same acoustic wave velocity as the cement-lined one. Young's modulus of ductile iron is equal to 169 GPa, as found from the resonance frequency of a rod sample using an impact-echo test that can be seen in Figure 4; Poisson's ratio was taken equal to 0.3; and Young's modulus of the cement lining was taken equal to 24 GPa. The impact-echo test and modulus calculation was performed in accordance with ASTM
1876-01: Standard test method for dynamic Young's modulus, shear modulus, and Poisson's ratio by impulse excitation of vibration, 2001.
100541 The bulk modulus of water was back calculated by adjusting its value until the predicted pipe wall thickness was equal to 8.2 0.1 mm. The modulus was found to be 1.95 GPa at 20.8 C, 1.875 GPa at 11.7 C, and 1.81 GPa at 5.3 C. A straight line was fitted to the backcalculated moduli and is shown in Figure 5 together with moduli based on the velocity of sound given by Eq. 28 for pure water at MHz frequencies.
Over the 5 to 25 C temperature range shown in Figure 5, which is expected for water in distribution pipes, the modulus for pure water at high MHz frequencies almost varies linearly with temperature. This lends support to the adequacy of the fitted straight line for estimating the bulk modulus of water in distribution pipes.
 Typical setups in accordance with embodiments of the invention are shown in Figures la and lb. A buried water pipe 10 has hydrants 12a, 12b, 12c. Sensors 1,2 are mounted on two of the hydrants, while the third provides a source outside the hydrants including the sensors. In Figure la, the sensors are mounted on hydrants 12a and 12b, while the hydrant 12c acts as a source. In this case the propagation velocity is determined by measuring the propagation time between hydrants 12a and 12b. The propagation velocity is given by the expression v = D / AT, where OT is the time delay between the signals sensed by sensors 1 and 2, and assuming equal lengths for pipes connecting fire hydrants 12a and 12b to the main pipe 10.
 In Figure lb, the hydrant 12b acts as the source, and the disturbance is detected by the sensors 1, 2 at hydrants 12a, 12c on either side of the source. In this case the propagation velocity is given by the expression v = (L2 - Lj)/ AT, assuming equal lengths for pipes connecting fire hydrants 12a and 12c to the main pipe 10.
100571 Signals are sent remotely from transmitters 14 attached to the sensors 1, 2 to a receiver 15 connected to PC 16 running a program transferred to its hard drive from a storage medium, such as a CD ROM 18. The computer 16 is WindowsTM based (although it will be appreciated that any other suitable operating system can be used) and displays the results in windows shown in Figures 2a and 2b. Figure 2a shows the results of a measurement. Figure 2b is a screen allowing the user to enter relevant parameters, such as the pipe material, the temperature etc. The signals may be transmitted in analogue or digital forms. In an alternative setup, signals from the sensors 1, 2 can be sent to the PC
via cables. Alternatively, dedicated hardware can be used instead of a PC to calculate the time delay between acoustic signals and determine the pipe's wall thickness.
 When sensors are mounted on fire hydrants that are connected to the pipe being tested, the connecting pipes should be traced on the surface and their lengths measured accurately and incorporated in the sensors' distances from the acoustic source. If it is not possible to trace the connecting pipes and hence their lengths are based on a "best guess' of their layout, the lengths should be checked for accuracy. One way of doing so is to measure the propagation velocities of acoustic noise created by at least two sources at different locations. For example, two out-of-bracket sources on opposite sides of the bracket can be used. If distances of the sensors from the acoustic sources are inaccurate due to errors in the lengths of connecting pipes between fire hydrants and the main pipe, the separately measured propagation velocities of acoustic noise induced by the two sources will be different.
[00591 A system of linear equations can be formed based on the relationship between time delays of acoustic signals induced by acoustic sources at different locations, acoustic propagation velocity in the pipe, and lengths of connecting pipes. Time delays between measured acoustic noise can be determined using the time-flight method or the cross-correlation function (either traditional or enhanced). The acoustic velocity can then be found by solving the system of linear equations. The number of acoustic sources at different locations should be equal to the number of unknowns. Two or three acoustic sources are sufficient in most cases.
[00601 As an example, consider the case in which the sensors 1 and 2 are mounted on fire hydrants that are connected to the main pipe with perpendicular connecting pipes having unknown lengths of L, and L2, respectively. The distance between the two points where the connecting pipes to the fire hydrants branch out from the main pipe is D
and can be measured accurately. The time delays Ot, and At2 between acoustic signals induced by out-of-bracket sources at the side of sensors I and 2, respectively, are related to the acoustic velocity, v, and LI and L2 by the following equations:
Ot, (D+L2)-L, (30) v At2 - L2 -(D+L,) (31) -v which can be solved to determine the propagation velocity as follows:
v - - 2D (32) (At2 - Ot,) 100611 The pipe wall thickness determined by the method described represents an average value for the pipe section between the two points where acoustic signals are measured. This is not a limiting aspect of the method. Generally, pipes will have a roughly uniform thickness profile over significant lengths, say 100 metres, as soil and bedding conditions are unlikely to change significantly over wide areas. Even for situations where the longitudinal thickness profile varies, an average thickness profiles is more appropriate and meaningful than discrete thickness values for evaluating the residual life of pipes for the purpose of long-term planning of rehabilitation and replacement needs.
100621 Virtually every water utility in the industrialized countries of the world is facing major challenges in securing capital dollars to replace aging underground water distribution pipes. This non-destructive technology in accordance with embodiments of the present invention to measure pipe metal loss, i.e., to determine the remaining thickness of pipe walls, provides these utilities with a reliable and accurate way to obtain data that allows them to calculate the remaining service life of pipes. This allows the formulation of capital budget plans that are based on technically sound engineering data, which in turn enhances the decision making process.
 The invention also has applicability to the oil and gas industry as well as the building industry (e.g., pipes of fire sprinkler systems, which are known to suffer severely from the action of corrosion).
|International Classification||G01M3/02, G06F17/10, G01N29/07|
|Cooperative Classification||G01M3/243, G01N2291/02827, G01N2291/2634, G01N29/4418, G01N2291/02854, G01N29/222, G01N29/07, G01N2291/011, G01B17/02|
|European Classification||G01N29/44B1, G01N29/07, G01N29/22F, G01B17/02, G01M3/24B|