GB2059064A - Method and apparatus for acoustically investigating a casing in a borehole penetrating an earth formation - Google Patents

Abstract

Methods and apparatus for acoustically investigating a casing 12 in a borehole to derive the quality of a cement bond 32 behind the casing and casing thickness are described. The techniques employ an acoustic pulse source 56, 36 having a frequency spectrum selected to stimulate a selected radial segment of the casing into a thickness resonance. The selected frequency spectrum enhances the reverberations between the inner and outer walls of the casing which traps the thickness reverberations with significant amplitudes for a duration depending upon the amount of acoustic energy leaked into adjacent media. Acoustic returns are formed by the reflections from interfaces between media of different acoustic impedances and acoustic energy leaked into the bore of the casing from the acoustic thickness reverberations stimulated within the casing walls. The acoustic returns are detected to generate a reflection signal which is processed (at 348 to 364) to determine casing thickness or to evaluate (at 21) the cement bond. The acoustic pulse has a frequency spectrum which is particularly effective in discriminating different cement bond conditions caused by small cement separations 30, known as micro-annuli, around the casing. Several signal processing techniques and tools are described to provide accurate and high resolution cement bond evaluation and casing thickness determination by processing a gated portion of the reflection signal representative of the thickness reverberations. <IMAGE>

Description

(12)UK Patent Application (19)G13 (11) 2 059 064 A (21) Application No

8035388 (58), (60), (7 f)1- (72) and (74) (22) Date of filing 4 Jug 1980 Date lodged 4 Nov 1980 (30) Priority data (31) 814588 911016 (32) 11 Jul 1977 30 May 1978 (33) United States of America (US) (43) Application published 15 Apr 1981 (51) INT CL3 G01 B 17/02 continued overleaf (54) Method and apparatus for acoustically investigating a casing in a borehole penetrating an earth formation (57) Methods and apparatus for acoustically investigating a casing 12 in a borehole to derive the quality of a cement bond 32 behind the casing and casing thickness are described. The techniques employ an acoustic Pulse source 56, 36 having a fre- ERRATUM

SPECIFICATION NO 2059064A

Front page, heading (22) Date of filing delete 4 Jul 1980 insert 4 Jul 1978 THE PATENT OFFICE 25 June 1981 51.5 5.1. z 519 fA are formed by the reflections from interfaces between media of different acoustic impedances and acoustic energy leaked into the bore of the casing from the acoustic thickness reverberations stimulated within the casing walls. The acoustic returns are detected to generate a reflection signal which is processed (at 348 to 364) to determine casing thickness or to evaluate (at 21) the cement bond. The acoustic pulse has a frequency spectrum which is particularly effective in discriminating different cement bond conditions caused by small cement c#.n.qrntions 30. known as micro-an- d f f Bas 83595111 7 71R I- 64 f3Z IrEFZ rcrlolv v f&LSE J -99 W R/P fx L 24 (-f 15(4 1 GB 2 059 064A 1 SPECIFICATION

Method and apparatus for acoustically investigating a casing in a borehole penetrating an i earth formation This invention relates to methods and apparatus for acoustically investigating a borehole penetrating an earth formation, and is more particulary concerned with a method and apparatus using an acoustic pulse echo technique to determine the thickness of a casing lining the borehole.

According to one aspect of the present invention, there is provided a method for determining 10 the thickness of a casing cemented in a borehole penetrating an earth formation from a reflection signal derived from an acoustic investigation of the casing with an acoustic pulse directed at a radial segment of the casing and formed of acoustic waves at frequencies selected to stimulate a thickness resonance inside the casing walls, the method comprising the steps of:

generating a spectrum signal representative of the frequency spectrum of a reverberation segment of the reflection signal wherein said reverberation segment is substantially representative of acoustic reverberations between the casing walls at said radial segment; and determining the frequency of components in said spectrum signal contributing to a peak value thereof and providing a thickness signal representative of said measured frequency as indicative of the casing thickness at said radial segment.

According to another aspect of the invention, there is provided apparatus for determining the thickness of a casing cemented in a borehole penetrating an earth formation from a reflection signal derived from an acoustic investigation of the casing with an acoustic pulse directed at a radial segment of the casing and formed of acoustic waves at frequencies selected to stimulate a thickness resonance inside the casing walls, the apparatus comprising:

means for selecting from the reflection signal a reverberation segment substantially representative of acoustic reverberations between the casing walls; means for generating a spectrum signal representative of the frequency spectrum of said reverberation segment; means for determining the frequency of components in said spectrum signal contributing to a 30 peak value thereof and producing a thickness signal representative thereof as the casing thickness.

Methods and apparatus in accordance with the invention for determining the thickness of a casing cemented in a borehole in an earth formation will now be described, by way of example only, with reference to the accompanying drawings, of which:

Figure 1 is a schematic representation of a related apparatus, which can be used to evaluate the quality of the cement bond; Figure 2 is a waveform representation of a preferred acoustic pulse generated in the apparatus shown in Fig. 1; Figure 3 is a plot of the frequency spectrum of the acoustic pulse shown in Fig. 2; Figures 4A, 4B and 4C are illustrative waveforms representative of acoustic reflections obtained in a pulse-echo investigation technique conducted as described herein; Figure 5 is an amplitude response curve useful in specifying the performance requirement of a transducer preferred for use in an acoustic borehole investigation as described herein; Figures 6A-6C are illustrative spectra of acoustic reflections observed with the acoustic 45 investigation apparatus described herein; Figure 7 is a block diagram of a signal processing apparatus for evaluating the cement bond; Figure 8 is a block diagram of another form for a signal processing apparatus for evaluating the cement bond; Figure 9 is a schematic representation of another cement bond evaluation tool; Figure 10 is a block diagram of a signal processor for use with a cement bond evaluation tool of a type such as shown in Fig. 9; Figure 11 is a timing diagram of signals generated in the signal processor shown in Fig. 10; Figures 12 and 13 are top views in partial section of transducers for use in a tool such as shown in Fig. 9; Figure 14 is a partial side view in elevation of an acoustic investigation tool employing transducers as shown in Figs. 12 and 13; Figure 15 is a schematic representation of an apparatus for determining the thickness of a casing; Figure 16 is an amplitude frequency plot of several spectra obtained with the apparatus of 60 Fig. 15; Figure 17 is a block diagram of a signal processing apparatus for determining the quality of the cement bond and casing thickness; Figure 18 is a block diagram of part of an apparatus for detecting casing thickness; and Figure 19 is a sectional view of an acoustic borehole investigating tool employing a rotating 65 2 GB 2 059 064A 2 reflector for scanning of the borehole.

Figs. 1, 2, 3, 4 and 5 With reference to Figs. 1 through 3, a system 10 is illustrated for acoustically investigating the quality of the cement bond between a casing 12 and an annulus of cement 14 in a borehole 16 formed in an earth formation 18. An acoustic pulse producing tool 20 is suspended inside the casing 12 with a cable (not shown) having signal paths along which signals for control of tool 20 and for its observations are transmitted between a signal processor 21 in tool 20 and surface located controls and signal processing equipment such as shown at 22. A depth signal, representative of the depth in borehole 14 of toot 20, is derived on a line 24 with a conventional depth monitor (not shown) coupled to the cable with which the tool 20 is moved along casing 12.

The cylindrical casing 12 is shown in partial section as well as the surrounding cement annulus 14. The shape of the borehole 16 is shown as uniform and the casing correspondingly illustrated as equidistantly spaced from the borehole wall. In practice, however, the borehole 15 wall is likely to be irregular with crevasses and cracks. Hence, the cement annulus 14 may vary in thickness and the spacing between the casing 12 and the formation 18 may vary.

The cement 14 is shown with various bond states frequently encountered. At region 26 the cement is shown as adhering to the casing 12 while at 28 a micro-annulus, ga, 30, which is hydraulically secure, occurs. In the region 32 the annulus 30 is shown enlarged to a thickness 20 with which vertical zone separation is no longer obtainable while at region 34 the cement is entirely absent. The cement-free regions at 28, 32 and 34 normally are filled with water or a combination of water and mud. These cement conditions do not necessarily occur as illustrated and are shown here for purposes of illustrating the invention. Suffice it to note that the cement conditions at regions 26 and 30 are to be evaluated as good bonds while those at regions 32 and 34 must be detected as bad.

Casing 12 is further shown with externally corroded segments 33.1, 33.2 and an internally corrodied segment 33.3 where the casing wall has been reduced in thickness. Such corrosions may occur at other regions and can be particularly harmful when one occurs in a region leading to hydraulic communication between zones which must remain isolated from each other. The 30 illustrated corroded segments 33.1 -33.3 may appear as actual gaps or occur as scaly segments which present a rough surface appearance and may even partially separate from the good parent metal. The scaly segments become saturated by the borehole fluid segments so that acoustic investigation of the good parent metal beneath the scaly segments can still be made.

The tool 20 fits within the casing 12 which normally is filled with water or a mixture of water and mud. The tool 20 is kept central in the casing 12 with appropriate centralizers (not shown) as are well known in the art. In the practice of the invention the tool 20 preferably is kept parallel to the casing wall, though the tool may be displaced relative to the central axis of the casing 12. As will be further explained with reference to Fig. 1, some compensation for tilt conditions, i.e. when the tool 20 forms an angle with the casing axis, is obtained with the 40 system 10.

Tool 20 is further provided with a transducer 36 functioning as a pulse transmitter and receiver. In some instances the transmitter and receiver functions can be produced by separate devices. The transducer 36 is oriented to direct an acoustic pulse onto an acoustic reflector 38 and then through a window 40 onto a selected radial segment of the casing 12. The acoustic pulse is partially passed through casing 12 and partially trapped in casing 12 with reverbera tions occurring in the radial segment at the thickness resonance of the casing.

The term -radial segment- as used herein means the segment of the casing extending between its walls and surrounding a given radius which extends generally normal to the casing wall from the center of the casing.

The nature of the window 40 may vary and preferably is formed of such material and so inclined relative to the direction of travel of the acoustic pulses from transmitter 36 that the acoustic returns can pass through with a minimum of attenuation and source of reflections.

Windows 40 can be made of polyurethane such as sold by the EmersonCummings Company as CPC-41 having an acoustic velocity of about 1,700 meters/second and a density of about 1.1 graMS/CM3. Such material exhibits a similar acoustic impedance as a fluid placed in the spqce between source 36, reflector 38 and window 40 to equalize pressure across window 40.

The fluid with which the space inside the tool between the transducer 36, and window 40 is filled is preferably selected for low or minimum attenuation and an acoustic impedance which will not contrast too widely from that of the borehole fluid in the frequency range of interest. An acceptable fluid may, for example, be ethylene glycol.

Window 40 is inclined at an angle 0 which is defined as the angle between the direction of propagation of the initial acoustic pulse from transducer 36 and the normal 41 to the window surface area upon which this acoustic pulse is incident. Such inclination serves to deflect secondary transmissions such as 43.1 in a direction which avoids window produced interfer- 3 GB 2 059 064A 3 ence. Suitable annular acoustic absorbing surfaces such as baffles 45 may be used inside the tool to trap and absorb acoustic reflections 43.2 from the inside wall of window 40. The size of the angle 0 may be of the order of 20 to 3C as suggested in the U.S. Patent 3,504,758 to Dueker.

Although the inclination of window 40 could be in a direction measured relative to the 5 incident beam travel path, as shown in the U.S. Patents 3,504,758 to Dueker, or 3,504,759 to Cubberly, the preferred orientation is as illustrated in Fig. 1 herein to enable use of a larger reflector 38.

The size of reflector 38 is significant in that the reflector surface area influences focusing of the acoustic energy onto the casing 12 and the capture of a sufficient acoustic return for 10 improved signal to noise ratio.

If the reflectors of Dueker or Cubberly are enlarged, the internal reflections from their windows are likely to be intercepted by the reflectors and redirected onto the receiver transducer in interference with the desired acoustic returns from the casing. When a window inclination as illustrated in Fig. 1 herein is employed, however, a large reflector 38 can be used, within 15 effective dimensions sufficient to either focus or preserve the beam shape of the acoustic energy directed onto casing 12 and provide a significant acoustic return to receiver transducer 36.

The inclination of window 40 can be clearly distinguished from that employed in Dueker or Cubberly with reference to the orientation of the internal window normal 41' relative to the point of incidence of the acoustic beam along its travel path D2 from reflector 38. When as 20 shown in Fig. 1, the normal 41' lies between the beam travel path D2 and the acoustic receiver function of transducer 36, the inclination angle and also the angle of incidence, can be considered as positive. This angle would also be positive when the internal normal lies between the beam travel path and a separate acoustic receiver such as employed in the acoustic borehole apparatus illustrated in the Russian Patent SU 405,095.

In case of a window orientation as shown in the Dueker or Cubberly patents, the inclination angle or angle of incidence can be construed as negative because the internal window normal is on the other side of the acoustic beam travel path and points away from the receiver transducer.

With the window inclination as illustrated in Fig. 1, care should be taken to avoid directing reflections such as 43.2 onto the transducer 36; the inclination angle, therefore, should be positive and sufficiently large. However, the inclination angle should be not so large that reflections such as 43.2 fail to be either absorbed or intercepted by baffles 45.

A portion of the acoustic pulse is passed through casing 12 and, in turn, is partially reflected by the next interface, which in region 26 would be cement material, while at the regions 28, 32, would be the annulus 30 and water-mud at region 34.

In the embodiment of Fig. 1 the acoustic transducer 36 is selectively located so that its effective spacing (the travel time for an acoustic pulse) to the casing 12 is sufficiently long to permit isolation of interference from secondary transmission caused when the strong acoustic casing reflection is again partially reflected by either a window or the transducer 36 back to casing 12 to produce new reverberations and secondary acoustic returns. A desired total spacing 40 D is obtained by locating the transducer 36 generally at an axial distance D, from reflector 38, which in turn is spaced a distance D2 from the casing 12.

The total distance D = D, + D2 between transducer 36 and casing 12 is further selected sufficiently long so that the desired acoustic returns including those attributable to reverbera tions trapped between the casing inner and outer walls 13 and 13' respectively can be detected. 45 The total distance D is thus sufficiently long to include those acoustic returns prior to their decay to some small value as a result of leakage into adjoining media. On the other hand, the total spacing D is kept sufficiently small to avoid undue attenuation by the mud external to tool 20 and the fluid inside tool 20.

In addition to these spacing considerations, the distance D, between transducer 36 and 50 reflector 38 has been found to affect the sensitivity of the system to tool positions away from a concentric relationship with the central axis 47 of casing 12. It should be understood that tool is provided with suitable centralizers, not shown, as are generally well known. Despite the presence of such centralizers some tool displacement, shown as an eccentricity distance e between the casing axis 47 and tool axis 49, may arise from a number of conditions inside casing 12. The distance D, for this reason is selected to tolerate a maximum amount of tool eccentricity e.

The optimum value for the spacing D, depends further upon such factors as the effective dimensions of surface 37 of transducer 36 such as its diameter in case of a disk transducer 36.

For a disk transducer having a diameter of the order of about one inch to produce a pulse 60 such as 50 in Fig. 2 with a frequency spectrum such as 52 in Fig. 3, the total distance D, is generally of the order between about 2 to about 3 inches.

A basis for selecting the total distance D is thus to assure sufficient time to receive all those acoustic returns which significantly contribute to an accurate judgment as to the quality of the cement bond in the presence of a small casing cement annulus. The total distance D should be 65 4 GB2059064A 4 long enough to enable the portion in the acoustic returns attributable to a bad cement bond to be received free from interference.

The acoustic returns include acoustic reflections arising as a result of the interaction of the initial acoustic pulse with various media. A first acoustic casing reflection arises from the interface between the water or mud inside the casing 12 and the inside casing wall 13. This 5 first reflection tends to be consistently the same, varying with mud consistency, inside casing wall condition, and tilts of tool 20. Subsequent acoustic returns arise as a function of reflections from successive media as well as the leakage of acoustic reverberations entrapped inside the casing.

Thus, after the first casing reflection, the acoustic portion transferred into casing 12 is now 10 reverberating inside the casing walls 13-13' and leaking energy at each reflection. The energy lost depends upon the coefficients of reflections r. (the reflection coefficient between the fluid inside casing 12 and the casing) and r, (the reflection coefficient between casing 12 and the next layer which may be cement as in region 26 or water as in region 32). The duration over which significant reverberations last inside the casing walls 13-13' is a function of the casing 15 thickness. Since casing of greater thickness tend to cause longer lasting reverberations, the total spacing D between the casing and receiver-transducer should be correspondingly increased.

When a window, which is normal to the direction of travel of the acoustic pulse, as suggested in dotted line at 42 in Fig. 1 is employed, the casing reflection and other acoustic returns produce reflections at the interface between window 42 and the mud inside casing 12. Such 20 reflections appear as secondary transmissions which are returned to the casing to produce a second casing reflection with subsequent reverberations in the casing and thus also secondary acoustic returns. These secondary acoustic returns disturb the cement evaluation, particularly in case of a good cement bond when the formation also has a smooth surface. In this latter situation reflections caused by secondary reverberations mix with a significant reflection from the 25 formation, giving an overall erroneous impression of a bad bond.

Hence, another criterion for determining an acceptable casing to receiver distance may involve selecting a distance D3, between a window 42 and casing 12, such that secondary acoustic returns decay below a preselected percentage of their initial value. Thus, it can be shown that the number N, of reverberations in the steel casing 12 in such range is given by the relationship 1 n (x) N, = 1 n (1 r.rl 1) where x is the percentage fraction.

The distance D3 can then be shown as given by the relationship D3>N, L C.

c, where L is the thickness of the casing 12, C. the velocity of sound of the material inside the 45 casing, mainly water, and C, the velocity of sound in the casing, namely steel.

As a numerical example to arrive at an acceptable total casing to receiver distance, one may assume the values for the materials employed in the following Table 1.

TABLE 1

Acoustic Impedance Density Velocity of Sound Z in g/CM2 sec water Z. = 1.5 X 105 steel Z, = 4.6 X 10' cement Z, = 7.7 X 10' and Z, = Z. in case of a bad bond.

p in g/CM3 P>11 p, 7.8 P2 1.96 C in ft/sec C. = 4920 C, = 19,416 C2 = 12,000 Using these constants the values for the reflection coefficients can be determined as r. = 0.937 r1G = -.731 (for a good bond) r,, = -.937 (for a bad bond).

Y GB 2 059 064A 5 0 15 The casing to receiver distance or D3 can be determined from the above constants and time setting constraints. For example, if the reverberations in the casing are to decay to about five percent of their initial value, the distance D3 can be from about one and one-quarter inch to about three inches for a normally occurring range of casing thicknesses L from about.2" to about.65". By relaxing the final value of decay of the casing reverberations the source to casing distance can be decreased, though about one inch is likely to be a lowest possible limit for D3. Since the largest casing thickness is preferably accommodated, the distance from the transducer 36 to either window 40 or 42 is chosen such that there is no secondary transmission interference over the time interval of interest. The distance D3, when applicable, is chosen such that secondary reflections attributable to the window do not present signal interference. When 10 the tool 20 employs a window such as 40, secondary reflections from such window are no longer a consideration in selecting the transducer to casing spacings.

In the selection of the transducer 36, a disk transducer having a diameter to wavelength ratio of greater than unity is employed. In practice, a disk transducer having a diameter of about one inch has been found useful. The transmitter pulse is formed of such duration and frequency as 15 to stimulate a selected radial segment of the casing upon which the pulse is incident into a thickness resonance. Acoustic energy is transferred into the casing and reverberates in an enhanced manner with the duration and magnitude of reverberations highly sensitive to the layer of material adjacent the external surface of casing 12. Such sensitivity, however, should not include hydraulically secure micro-annuli such as at region 28.

In the selection of the frequency spectrum of the acoustic pulse from transducer 36, a primary basis is determined by the fundamental thickness resonance frequency of casing 12. Such resonance enables a trap mode with which enhanced acoustic energy is trapped in the casing. The subsequent reduction of trapped energy in the casing may be considered the result of leakage attributable to the degree of acoustic coupling to adjacent media. The frequency spectrum of the acoustic pulse should preferably include either the fundamental or a higher harmonic thereof. Expressed in mathematical terms, the stimulating frequency in the acoustic pulse is given by cl f,, = N 21--- where C, is the casing compressional velocity and L is the casing thickness measured normal to the casing wall and N is an integer.

An upper limit of the frequency spectrum of the acoustic pulse is set by practical considerations such as casing roughness, grain size in the steel casing and mud attenuation.

Furthermore, the hydraulically secure micro-annulus must appear transparent.

In practical cement bond applications a casing-cement annulus equal or smaller than.005" (. 1 27mm) represents a good cement bond and thus prevents hydraulic communications 40 between zones intended to be separated. When annuli larger than this value occur, these should be construed as bad cement bonds. Furthermore, as long as any annulus is less in thickness than about 1 /30 of a wavelength of an acoustic wave traveling in water, such annulus is effectively transparent to an acoustic wave of such wavelength. Hence, in terms of casing- cement annuli, the frequency spectrum of the acoustic pulse should be selected such that CO f.< (tia,) X 30 where C. is the velocity of sound in water and ga, is the thickness of the annulus.

In practical terms, casing thicknesses L normally encountered are from about.211 (5.08 mm) to about.65" (16.51 mm). Hence, with an effective frequency of from about 300 KHz to about 600 KHz for the acoustic pulse, the casing 12 can be stimulated into a trap mode which is insensitive to hydraulically secure micro-annuli. This frequency spectrum is selected so that the 55 trap mode can be stimulated with either the fundamental frequency or its second harmonic for the thicker casings.

Within such frequency spectrum, the duration of the reverberations inside the steel casings become sensitive to both good and bad micro-annuli. For an acceptable micro-annulus the casing reverberations (and their observed leakage) decay more rapidly than for an excessively 60 large micro-annulus.

The acoustic transmitter pulse is thus formed with characteristics as illustrated in Figs. 2 and 3. The transmitter pulse 50 shown in Fig. 2 represents a highly damped acoustic pulse of a duration of the order of about eight microseconds. The frequency spectrum of such pulse 50 is shown in Fig. 3 with a frequency-ampfitude curve 52 showing a 6 db (one- half power) GB 2 059 064A 6 bandwidth extending from about 275 KHz to about 625 KHz with a peak at about 425 KHz.

The thick casings having a trap mode below 275 KHz are driven into resonance primarily with a higher harmonic such as the second which occurs with significant amplitude in the bandwidth of the spectrum 52.

The transmitter 36 can be formed of a variety of well known materials to produce pulse 50 5 with the frequency spectrum 52. For example, an electrical signal having these characteristics can be formed and amplified to drive a suitable piezoelectric transducer 36 capable of operating as a transmitter and receiver.

Preferably transducer 36 is formed with a piezoelectric disk crystal which is backed with a critically matched impedance such that an acoustic pulse is formed at the resonant frequency of 10 the disk. The backing material has an impedance selected to match that of the crystal while strongly attenuating the acoustic pulse to avoid reflections from the back. In some applications a protective front layer may be employed integrally mounted on the front of the transducer 36.

Such front layer is preferably made of a low attenuation material having an acoustic impedance which is approximately the geometric mean between the crystal impedance and the expected 15 borehole fluid impedance. Such front layer has a quarter wavelength thickness as measured at the center resonant frequency of the crystal.

Since the disk is critically matched, the acoustic output pulse has a wide frequency bandwidth. Excitation of such transducer 36 may then be achieved with an electrical pulse of very short duration. For example, an impulse having a rise time of from about 10 to about 100 nanoseconds and a fall time of 0.5 to about 5 microseconds can be used.

In the transmitter mode transducer 36 may be actuated in a repetitive manner at a pulse rate, say, of the order of a hundred pulses per second. At such rate a circumferential region around casing 12 can be scanned as tool 20 is moved upward along the casing by making reflector 38 and its associated window 40 a rotatable mounting as illustrated for rotation in the direction of arrow 53.

Fig. 5 defines the performance criteria for a suitable transducer 36, The transducer has a center acoustic frequency at about 425 KHz with a 6 db bandwidth of 300 KHz. The Fig. 5 illustrates an acceptable received amplitude response curve 55 when transducer 36 is energized with a pulse drive signal of about two microsecond duration and directed at a water/air interface spaced from the transducer at a distance equivalent to about 100microseconds of twoway acoustic wave travel time, T, The output signal from transducer 36 as a result of the echo from the interface preferably should have an appearance as illustrated where the first echo, formed of the three main peaks 57.1, 57.2 and 57.3, should be of no greater total duration, T2, than approximately six microseconds. The level A2 of the noise immediately after the first echo 35 should be about 50 db below the level A, of the peaks 57 and have a duration T3 of less than about 30 microseconds. The noise level A3 following internal T. preferably should be at least 60 db below Yhe level A, of peaks 57.

The controls and circuitry necessary for firing of the transducer may originate from above ground equipment or from a suitable clock source located in tool 20. In either case, recurring synch pulses are produced on a line 54 of Fig. 1 to activate a pulse network 56 which generates a suitable pulse on line 58 to drive transducer 36 while simultaneously protecting the input 60 to amplifier 62 with a signal line 64.

The transducer 36 responds to the pulse from network 56 with an acoustic pulse of the type as shown in Figs. 2 and 3. The acoustic pulse is directed onto reflector 38 which acts to direct the acoustic energy at the wall of casing 12. The effect of reflector 38 aids in compensating for variations in alignments of the acoustic pulse out of the plane normal to the casing wall. The reflector 38 can be a flat surface at an angle a of about 45 to the acoustic energy from transducer 36 or a slightly concave or convex surface.

When the acoustic pulse 50 impinges upon casing 12, some of the energy is reflected and 50 some transferred into the casing 12. The reflected energy is returned to transducer 36 via reflector 38 and is reproduced as an electrical signal and applied to input 60 of amplifier 62.

The energy transferred into casing 12 reverberates, causing in turn further acoustic returns to transducer 36. The resulting received output from transducer 36 may have the appearance as illustrated with reflection signal waveforms 64, 66 and 68 in Figs. 4A, 413 and 4C.

The initial segment 70 of each reflection signal waveform represents the strong initial casing reflection whose duration is of the order of about five microseconds. The remainder 72 is characterized as a reverberation segment in that it represents a large number of cycles of pulses representative of acoustic reverberations whose magnitudes decay over a period of time. The decay period varies as a function of the type of cement bond, as can be observed for waveforms 64, 66, 68 obtained with respectively differently sized annuli 30 around casing 12.

Except for the initial casing reflection segment 70, the reflection signals 64, 66, 68 do not have a highly predictive pattern wherein the peaks are precisely defined and extractable. Accordingly a prior art technique such as shown in the previously identified U. S. Patent to

Norel et al for comparing adjacent peaks to ascertain decay time constants for the waveforms is 65 7 GB 2 059 064A 7 6 difficult to implement.

Instead, the signal processing segment 21 of the apparatus 10 operates on each reflection signal by separating the reverberation segment 72 from the initial strong acoustic casing reflection segment 70 and subsequently integrating the reverberation segment 72 over a 5 particular time span to determine the energy therein.

In the embodiment of Fig. 1, the reflection signals from transducer 36 are amplified in amplifier 62 whose output is applied to a full wave rectifier 76 to produce on line 78 a DC signal representative of the amplitude of the received acoustic wave. The DC signals are filtered in a filter 80 to provide on line 82 a signal representative of the envelope of the waveforms from transducer 36.

The envelope signal on line 82 is applied to a threshold detector 84 which initiates subsequent signal processing by detecting the start of the initial casing reflection segment 70 (see Fig. 4). The amplitude at which the threshold detector 84 operates can be varied with a selector control applied to line 86 and can be automatically set.

The output on line 88 of threshold detector 84 is applied to activate an enabling pulse on 15 output 90 from a pulse producing network 92. The pulse from this network 92 is selected of such duration that the envelope segment on line 82 and attributable to the initial casing reflection 70 is gated through an amplifier 94 as a casing reflection signal.

The duration of the enabling pulse on output 90 is selectable so that the entire casing reflection segment 70 can be selected in the event its duration varies. The DC form of the casing reflection signal is applied to an integrator network or peak amplitude detector 96 to produce a signal representative of the amplitude of the casing reflection 70 on line 98. This casing amplitude signal is stored such as with a sample and hold network 100 actuated by an appropriate pulse derived on line 102 from network 92 at the end of the pulse on line 90.

The output 88 from the threshold detector 84 is also applied to a reverberation segment selection network 103 including a delay 104 which produces an enabling pulse to pulse producing network 106 at a time after the initial casing reflection 70 has terminated. Network 106 generates a segment selection pulse on line 108 commencing at the beginning of the reverberation segment 72 and having a duration sufficient to gate the entire envelope form of the reverberation segment 72 (see Fig. 4) through gating amplifier 110 to integrator 112. The 30 segment selection pulse on line 108 commences after the initial casing reflection and terminates after the desired number of acoustic returns of interest have been received but before secondary transmission interference arises. A typical pulse would start about six microseconds after the initial casing reflection is detected and would last for a period of about 40 microseconds after an acoustic pulse issued such as shown in Figs. 2 and 3 and with a spacing D of the order of about 35 three inches.

The integrator 112 integrates the envelope form for a time period determined by the pulse on line 108. At the end of this latter pulse a signal on line 114 from pulse producer 106 activates a sample and hold network 116 to store a signal representative of the energy in the reverberation segment 72.

The outputs from sample and hold networks 100, 116 are applied to a combining network in the form of a divider 118 which forms a quotient by dividing the signal representative of the energy in the reverberation segment 72 by the normalizing signal indicative of the amplitude of the casing reflection 70 to generate a normalized energy bond signal on output line 120. The normalized energy signal on line 120 can be transmitted to above ground for recording 45 reflection energy as a function of the depth on a plotter 122. The normalized energy signal may also be applied to a comparator 124 for comparison with a reference signal on line 126 derived from a network 128 and representative of the threshold level between good and bad cement bonds. The output 130 from comparator 124 indicating the presence or absence of a good cement bond can also be recorded on plotter 122 as a function of depth.

With the signal processing embodiments, the bond signal on line 120 is made less sensitive to tool tilts and attenuation in the fluid whereby the acoustic energy is directed at casing 12 along a plane which is skewed relative to the axis of the casing 12. When such condition occurs, the received acoustic returns are reduced in amplitude and may be interpreted as good cement bonds when, in fact, the cement bond may be bad. By employing the amplitude of the 55 initial casing reflection as a gauge of tool tilt and mud conditions, the bond signal on line 120 provides a reliable indication of the cement bond quality.

There may in certain cases arise a need to obtain a bond signal which has not been normalized or which may be normalized at a later time. In such case the output 117 of the sample and hold network 116 is the bond signal which may be transmitted to above ground 60 equipment for recording such as on a tape recorder or on plotter 122 or in the memory of a signal processor 130 after conversion to a digital form.

After a bond signal has been generated and a new synch pulse occurs on line 54, the synch pulse is applied to several reset inputs of sample and hold network 100, 116 and integrators 6 5 9 6, 112. The reset of the sample and hold networks 110, 116 can be delayed for a smoother 65 8 GB 2 059 064A 8 output until such time at the outputs from integrators 96, 112 are ready for sampling.

The selection of a signal representative of the acoustic reverberation return 72 is obtained with a pulse produced on line 108 as can be determined with a segment selection network 132. This network controls the length of the delay 104 and the width of the enabling pulse from pulser 106. As previously described with reference to Figs. 4A, 413 and 4C, the reverberation segment 72 is selected in such manner that the casing 'reflection 70 is effectively excluded.

This exclusion can be advantageously achieved by the signal processor 21 since it is activated by the detection of the strong casing reflection 70 as sensed by threshold detector 84. The resulting integration of the remaining envelope provides a sharp discrimination between a good bond signal and a bad bond signal. For example, the integration of the reverberation segment 10 72.1 of the waveform 64 in Fig. 4A will be greater than the integration of the reverberation segment 72.3 of waveform 68 in Fig. 4C by a factor of about 3. When the area of the envelopes are compared for an example as set forth in Table 1, with the resulting reflection coefficients for r. and r, for good and bad cement bonds, an integration ratio of about 3.8- to-1 between bad and good signals occurs. Hence, an extremely sharp good-to- bad bond contrast is 15 obtained which is likely to be obtained even in the presence of a dense mud inside the casing 12. With certain types of cement one may wish to construe a micro-annulus of a thickness of the order of about.0 10 inches (. 25mrn) as a good cement bond. In such case, the frequency 20 spectrum 52 of the acoustic pulse 50 may be adjusted to investigate the cement. One may, for 20 example, employ two types of acoustic pulses of different frequency spectrum, one having the fundamental frequency and the other acoustic pulse having a harmonic. If the results from these pulses do not give the same reading, a hydraulically secure micro-annulus can be concluded to be present. 25 Theoretically a bond will appear as good for a micro-annulus having a thickness of the order 25 of half wavelength (about 0.08 inches). However, in practice such large annulus is unlikely to arise and other conventional cement quality investigation techniques can be employed to identify such unlikely large annulus as a poor cement bond.

Figs. 6A-6C and 7 Figs. 6A through 6C illustrate the effectiveness of tool 10 when a frequency spectrum is made on the observed entire acoustic returns such as illustrated in Figs. 4A4C. The spectra 140 of Figs. 6A-6C represent respectively a bad bond with a large annulus, an intermediate bonding situation such as with an annulus of.005" and a good cement bond. The spectra 140 when originally obtained may have varied in absolute magnitude because the reflection changes 35 in the tool eccentricity e and the coupling of acoustic energy to the cement 14 behind the casing 12 varies. Thus for a good cement bond, the absolute amplitude of the acoustic returns is lower than for a bad cement bond. The relative size of dips 142, however, varies with a larger dip for a bad cement bond and a smaller dip 142 for a good cement bond. For convenience, the spectra 140 are shown in Figs. 6A-6C with generally equal amplitudes so that their dips 142 40 can be evaluated by a visual comparison with each other. The significance of dips 142 should be determined in light of the overall energy spectrum ofthe reflection signal.

The sharp dips 142 in spectra 140 are centered at the trap mode or thickness resonance of the casing from which the reflections came. In the spectra 140 the dips 142 occur at.5 MHz (500 KHz) for a.23 inch thick casing and resemble the effect of a narrow bandwidth energy 45 trapping filter. In the case of a bad bond, such as for spectrum 140.1 in Fig. 6A, the dip 142.1 is deep, indicative that a relatively substantial amount of energy at the thickness resonance has been trapped inside the casing walls 13-131.

The improvement of the cement bond is evidenced in spectrum 140.2 by a correspondingly smaller amount of energy being trapped inside the casing walls 13-13'. Hence, dip 142.2 in 50 Fig. 613 is smaller in comparison with dip 142.1 in Fig. 6A while dip 142. 3 in Fig. 6C is the smallest for a good cement bond.

Fig. 7 illustrates an embodiment 150 for evaluating the cement bond utilizing the sharpness of the dips 142 in spectra 140 of Figs. 6A6C. The output 63 of amplifier 62 in network 21 is applied to two passband filters 152 and 154. Filter 152 is a passband filter tuned to the casing 55 thickness resonance frequency of the casing 12 under acoustic investigation. The passband for filter 152 preferably is narrow with sharp rising and failing slopes. The filter 152, however, should be sufficiently wide in its frequency band to overlap the frequency range of dips 142 for the expected tolerance variations in casing thickness. Generally, a filter 152 with a passband of about 10% to 15% of the center frequency would suffice, though a smaller passband of about 60 5% may provide a dip amplitude indication on line 156. A digital as well as analog filter 152 may be used.

Filter 154 preferably is tuned to a separate non-overlapping segment of the spectrum of the signal on line 63 to provide a reference signal on line 158 indicative of the amplitude of the spectrum of the signal on line 63. Other devices can be employed to derive such reference 65 9 GB 2 059 064A 9 1 f signal such as the peak detection technique described with reference to the embodiment of Fig. 1. The dip amplitude signal on line 156 is thereupon normalized by dividing this signal by the reference on line 158 with a divider network 160. A normalized dip value signal is then available on the output 162 of divider 160 to provide an indication of the quality of the cement 5 bond for recording or plotting as the case may be.

Fig. 8 Fig. 8 illustrates another embodiment for determining the cement bond. The output from transducer 36 on line 63 from amplifier 62 (see Fig. 1) is applied to a high speed analog to digital (A/D) converter 172 which is actuated a specified time after an acoustic pulse. This 10 produces a digitized reflection signal formed of sequential numerical samples representative of the amplitude of the reflection signal. The A/D converter may be deactivated a certain time period following generation of an acoustic pulse.

A/D converter 172 is located downhole in tool 10 and is capable of operating at a very high speed and is provided with sufficient storage capacity to initially store and subsequently transmit 15 the samples at a slower rate to a surface located signal processor 174. The latter could, if desired be also located in tool 10, but this would depend upon the scope of operations the signal processor 174 must perform.

The sampled digital reflection signal is stored in a memory 176 which may be a solid state memory or a magnetic memory. The memory 176 can be an integral part of processor 174 for 20 immediate processing of the samples or be a peripheral device which is accessed at a later time after logging of the borehole 16.

Signal processor 174 may be programmed to select, at 178, those reflection samples, A,, representative of the casing reflection 70 (see Fig. 4). The procedure can be similar to that illustrated in analog form in Fig. 1. Thus the reflection samples are scanned to detect the first sample which exceeds a predetermined threshold and this first sample becomes the arrival time of the casing reflection. A certain number of samples following this first sample are then selected as representative of the casing reflection 70 (see Fig. 4).

A certain number of reflection samples, A, following the casing reflection samples Ac, are selected at 180 as representative of the reverberation segment 72 in the reflection signal (see 30 Fig. 4).

Integration of the reverberation samples is done by summing the absolute values of the samples at 182. This summing step could be carried out as the reverberation samples are selected at 180. However, for purposes of clarity, the summing operation is shown as a separate step. The integrated sum ER is stored.

Integration of the casing reflection samples Ac is obtained at 184 by summing the absolute sample values and storing the result, E..

A normalized bond value, C13, representative of the quality of the cement bond may then be obtained at 186 by dividing the integral ER by the integral Ec at 186. The bond value CB may be recorded in memory or plotted as desired at 188.

Figs. 9, 10 and 11 Fig. 9 illustrates still another embodiment for investigating the quality of the cement bond. A tool 210 suspended from a cable 211 is provided with a plurality of transducers, such as 36, but arranged circumferentially around the tool 210 to provide sufficient circumferential cement bond evaluation resolution. The transducers 36 are axially spaced to accommodate the large number. A practical number of transducer 36 may be eight which are circumferentially spaced at 45 intervals. The axial spacing is selected commensurate with the size of the transducer 36.

Figs. 10 and 11 relate to a signal processor 215 for operating a tool such as 210 shown in Fig. 9. The signal processor 215 as described is useful for a tool 210 employing eight transducers 36; however, a greater number of transducers can be accommodated. The signal processor 215 has an adjustable clock 212 on whose output 214 are pulses 216 (see Fig. 11) at a rate selected to determine the resolution of the cement bond investigation. The clock source may be derived from above ground devices or from a suitable oscillator located in the tool 210.

The clock pulses 216 are applied through a delay network 218 to a transducer selector 220 and a transmitter pulse multiplexer 222. The transducer selector 220 provides a discrete output enabling signal on line 224 to identify each different transducer 36 in succession. Hence, multiplexer 222 is enabled to sequentially fire pulsers 226 coupled to transducers 36.

The transducers 36 also act as receivers and produce signals on output lines 228 for amplification in pre-amplifiers 230 operatively associated with each transducer 36. The output 60 of amplifiers 230 are connected to a receiver multiplexer 232 which is controlled by the transducer identifying signals on line 224 from transducer selector 220. In addition, a segment selection network 234 is activated with each transducer firing to generate enabling signals 236, see Fig. 11, on an output 238 to effectively enable multiplexer 232 to select the desired segment from the transducer outputs while rejecting or blanking out the initial transmitter 65 GB 2 059 064A 10 segments. The output 240 from multiplexer 232 will have an appearance as illustrated at 244 in Fig. 11. A small noise signal 242 precedes the reflection signal 244 which has an appearance generally as illustrated in Figs. 4A-4C.

Returning to Fig. 10, the reflections on output line 240 are amplified by two variable gain amplifiers (VGA) 246, 248. Amplifier 246 has its gain controlled by a signal on line 249 and derived from either above ground equipment to adjust for mud attenuation effects or from a down hole automatic gain control. The second amplifier 248 has its gain automatically controlled in tool 210 to compensate the eccentering of tool 210 as will be further explained.

The output 250 from amplifier 248 is rectified in a network 76.1 and applied to a casing reflection sensing network formed of gated amplifier 94, integrator 96 and sample and hold 10 network 100 as described with reference to Fig. 1.

The output on line 250 from amplifier 248 is further amplified in an amplifier 252 by a sufficient amount to compensate for the approximate difference in signal amplitude between the casing reflection and the acoustic returns indicative of subsequent reverberations. An acceptable compensation may be a gain factor of about 20 db for amplifier 252. The reflections of interest 15 are then applied to a full wave rectifier 76.2 for subsequent integration with devices as described with reference to Fig. 1.

Control over the gating amplifiers 94, 110 is derived generally as described with reference to Fig. 1 with a threshold detector 84 responsive to the output on line 78 from full wave detector 76.1. A reference threshold value is derived on line 86 as a result of a similar previous cement 20 bond investigation made with the particular transducer as shall be further explained.

The output 88 from threshold network 84 is applied to the set input of a latch network 256.

Network 256 has a reset input 258 responsive to the clock pulses on line 214 (before the delay from network 218). When the threshold detector senses a signal on line 78 greater than the reference value on line 86, a pulse is applied to network 256 which thereafter is inhibited from 25 responding to further inputs from the threshold detector until network 256 is reset by a pulse on line 214. The output on line 260 will have the appearance as shown with pulses 262 (Fig. 11) having an active state upon the occurrence of the large casing reflection.

The integration times, T1 and T2 (see also Fig. 11), for signals representative of the casing reflection and the reverberations are derived with pulse networks 92 and 106 respectively, whose outputs 90, 108 are applied to enable gating am plifiers 94, 110. The duration and occurrence of the integration periods T1 and T2 are respectively about 8 microseconds for the casing reflection and about 30 microseconds for the reverberations.

Subsequent integration of the casing reflection signal by integrator 96 and the reverberation segment by integrator 112 are terminated at the end of pulses T, and T2 when the output from 35 amplifiers 94, 110 go back to zero. The integrator outputs are sampled at the end of pulse T2 and the samples made available for further processing with a suitable multiplexer 266 for transmitting the samples to above ground equipment. Transmission of information may employ an analog to digital converter 267 and suitable telemetry equipment 269 for transmission up cable 211. The integrators 96, 112 are reset by pulses on line 219 and the sample and hold 40 network by pulses on line 214 from transfer logic 271 at the time of clock pulses 214.

As previously mentioned, the gain control for amplifier 248 is automated by sensing the peak value of the casing reflection on line 78 with a peak detector 270. The peak value is then converted to a digital value with A/D converter 272 and this value placed in a storage network 274 in a location associated with the transducer from which the reflection was obtained. The 45 next time this transducer is energized, the transducer selector 220 provides an appropriate address signal to a read-in read-out logic network 275 to apply the previously stored peak value to a gain control network 276 and a threshold reference signal producing network 278.

For gain control the digital peak value is converted to an analog signal and an appropriate bias applied to control the gain of amplifier 248. In a similar manner, the threshold reference so value on line 86 is maintained at the appropriate level for each transducer 36.

The techniques employed in evaluating the cement bond as described herein advantageously enable accurate measuring of the eccentricity of the tool as it moves along the casing. This technique as shown in Fig. 10 involves a timer 280 which is energized each time a transducer 36 is initially fired. The timer 280 is deactivated to store a measured time interval when a casing reflection is detected by the threshold detector 84 as evidenced by the -signal on line 260. The measured time intervals for the various transducers should be the same and any difference may then be attributed to an off-center position of the tool. The output of timer network can be recorded or plotted and suitable processed to measure and locate the eccentricity of tool 210.

The vertical resolution of the tool 210 is a function of the repetition rate with which the transducers 36 are energized and produce detectable casing reflections and reverberations. A repetition rate as high as 100 per second can be accommodated to yield a resolution as small as about every one-tenth of an inch when the tool is moved at a logging rate of about 10 inches per second along the casing. A signal on line 213, see Fig. 9, representative of the depth of 65 11 GB2059064A 11 tool 210 is obtained to enable a signal processor 215 to adjust for the difference in levels of transducers 36.

Figs. 12, 13 and 14 Figs. 12, 13 and 14 illustrate an acoustic energy source and detector 300 for multiple use on a tool such as shown and described with reference to Fig. 9. The detector/source 300 is radially mounted to a cylindrical housing 302 with a mounting bracket 304 having a central aperture 306 to receive a cylindrical or disk transducer 36. The mounting bracket 304 extends past the emitting surface 37 of transducer with a slightly outwardly expanding aperture wall 308.

Bracket 304 may be directly mounted to housing 302 such as shown in Fig. 12 or with an intermediate spacer 310 as shown in Fig. 13. In the mounting of Fig. 12, the transducer to casing spacing D can accommodate smaller casings, say from about 5 1 /2 inch diameter. The arrangement of Fig. 13 can accommodate larger casings.

The radial orientation of transducers 36 preferably involves no window or intermediate materials. Furthermore, the spacing D between the transducer face 37 and casing 12 is kept as small as possible.

Since too small a spacing D enables secondary transmissions to interfere with the reflections of interest, the spacing D cannot be too small. On the other hand, if the spacing D is too large, mud attenuation effects can be too large as well. Hence, a compromise spacing D may be 20 selected based upon expected attenuations.

The attenuations may vary depending upon the type of mud used. For example, a heavy or dense mud may cause an undesirably high attenuation. Hence, in the selection of an acceptable spacing D, it may be necessary to also specify an upper mud density limit. With such upper limit, the maximum attenuation may be about 4 to 5 db per inch in contrast with a heavy mud 25 attenuation of about 8 to 10 db per inch.

With these general constraints, an acceptable spacing D may be of the order of about one to about two inches for most casings.

The described arrangements for tool 20 with a rotatable reflector 38 may be varied in a number of ways. For example, in some instances it may be desirable to mount the reflector 38 30 in a pad near the wall of casing 12 to reduce the attenuation effect of a dense mud fluid. Care should be exercised to assure that the reflector 38 remains sufficiently spaced from the wall of the casing 12.

Figs. 15 and 16 and 6 Casing thickness is measured by analyzing the frequency spectrum of the reverberation segment 72 (see Fig. 4A) representative of acoustic returns attributable to reverberations between the casing walls 13-13'. When an acoustic pulse such as 50 is directed at the casing 12, a substantial amount of acoustic energy at the resonance frequency is trapped inside the casing walls.

The reverberation segment 72 has predominant components in a frequency portion 320 (see Figs. 6A-6C) generally in frequency alignment with dips 142. The dips 142 increase in depth as the quality of the cement bond decreases, but the amount of energy trapped in between the casing wallsincreases with poorer bonding between the cement and the casing. Hence, the actual amplitude of the acoustic returns in the frequency portion 320 will vary. Generally, the 45 actual amplitude of the acoustic reverberations within the frequency portion 320 reduce as the acoustic coupling between casing 12 and cement 14 becomes more efficient; i.e. as the cement bond becomes better.

This is illustrated in the spectrum plot of Fig. 16 with curves 322 and 324 which respectively illustrate the frequency spectrum of a frequency portion 320 for a bad cement bond and a good 50 cement bond.

When thin spots develop in casing 12 such as at 33.1 and 33.2 in Fig. 15, they are likely to affect the cement bond evaluation. The effect of such thin spots upon the cement bond is not easy to predict and appears likely to be a function of such factors as size and cement conditions.

For example, there is no cement bonding behind the thin spot 33. 1, but since the casing is substantially thinner here, less acoustic energy remains trapped inside the casing walls 13-13' than in case of a normal thickness so that the thin spot 33.1 may appear as a good bond. On the other hand, if an isolated external thin spot such as 33.2 occurs at a well bonded area the casing 12 may appear as poorly bonded. Hence, it is advantageous to be able to correlate a casing thickness measurement with an evaluation of the cement bond to remove ambiguities. 60 The measurement of casing thickness is done in the apparatus 326 of Fig. 15 by forming a frequency spectrum of the reverberation segment as derived on line 63 of Fig. 1. The frequency spectrum is characterized by one or more peaks of which the largest occurs at a fundamental frequency whose wavelength is twice the thickness of the casing. Other peaks occur at frequencies which bear a whole multiple relationship to the fundamental frequency.

12 GB2059064A 12 Fig. 16 illustrates several frequency spectra 322, 324 of several reverberation segments 72 selected from different signals. It should be understood that in the presentation of the various spectra in Fig. 16, there is no intent to set forth an amplitude relationship between the spectrum 52 of the acoustic pulse 50 (see Figs. 2 and 3) and the other spectra 322, 324; rather, it is only intended to show a frequency relationship in that the spectra 322, 324 occur within the frequency bandwidth of the incident acoustic pulse. In practice, the absolute amplitudes of the acoustic spectra would be quite small in comparison with that of the transmitted pulse.

Of particular interest is the relative frequency shift between the spectra peaks 328, 330. The frequency difference between peaks 328, 330 can be attributed to a change in the thickness, IL, of casing 12. Hence, by determining the frequency of the peaks predominantly attributable to 10 acoustic returns from the reverberations between the casing walls, an indication of the casing thickness can be obtained.

The casing thickness, L, can be derived from the following relationship L= N-, 2f, where fp is the frequency of the peak in the spectrum, C the compressional velocity in the casing 12 and N is an integer depending upon whether the measured peak is for the fundamental 20 frequency (N = 1) or a higher harmonic.

Since the frequency spectrum 52 of the acoustic pulse 50 has a bandwidth of from about 300 to 600 KHz for use with casings 12 over a wide range of thicknesses, from about.2" to about.75", the second harmonic (N = 2) is likely to produce the largest peak in the reverberation spectra for the thicker casings while N = 1 for the thinner casings. The value for 25 N, therefore, can be determined prior to an acoustic investigation from a knowledge of the type of casing installed in the borehole.

For example, an installed casing is known to have a nominal thickness of. 362 inches, so that its fundamental thickness resonance occurs at about 331 KHz for a value of C of 20,000 ft/sec.

As actually measured, the spectrum 322 may present a peak 328 at a frequency Of fP2 of about 30 348 KHz corresponding to an actual casing thickness of.345 inches in one radial segment of the casing. Spectrum 324 presents a peak 330 at a frequency fp, of about 303 KHz corresponding to an actual casing thickness of.395 inches. These measurements illustrate the resolution of the technique by detecting a casing thickness variation of about 7% due to manufacture from the nominal value of.362 inches.

In apparatus 326 of Fig. 15 the casing thickness is measured by selecting the reverberation segment 72 on a line 332 with a selection network 334 coupled to the reflection signal on line 63. The selection network 334 employs a casing reflection detector 336 to provide on output 338 a pulse whose leading edge is representative of the start of the casing reflection 70 (see Fig. 4). Detector 336 may be formed of a threshold detector 84 for rapid response or as shown 40 in Fig. 1 of a full wave rectifier 76, filter 80 and threshold detector 84.

The pulse on line 338 is delayed by a delay 340 for a time period commensurate with the duration of the strong initial casing reflection 70 to then actuate a pulse network 342. The latter produces a reverberation segment selection pulse on line 344 to enable an analog gate 346 for a duration corresponding to the time needed to select the portion of the reflection signal 45 predominantly representative of reverberations inside the casing walls.

A spectrum analyzer 348 is responsive to the reverberator segment on line 332 to provide on line 350 a signal representative of the amplitude, A, of the frequency components in the reverberation segment 72 while output line 352 carries a corresponding frequency signal, f, representative of the frequency of the amplitude components on line 350.

The amplitude and frequency signals on lines 350, 352 are individually applied to analog to digital converters 354, 356 which produce and store in a memory 358 of a signal processor 360, the digital signals representative of the amplitude A,, and frequency, f,, of the frequency spectrum of the reverberation segment 72.

The operation of spectrum analyzer 348 and A/D converters 354, 356 is initiated by the reverberation segment selection pulse generated on line 344 from pulse network 342. During the latter pulse, a local oscillator, internal to spectrum analyzer 348, is repeatedly swept through a frequency range to produce the amplitude spectrum on line 350. Each time the local oscillator is swept through its frequency range, spectrum analyzer 348 generates a spectrum field of amplitude, A,, and frequency, f,, signals. Hence, during the selection of a single reverberation 60 segment 72 a plurality of spectrum fields are generated and stored in memory 358.

For a non-recurring reverberation segment 72, a discrete multiple of sweeps of the local oscillator in the spectrum analyzQr 348 can be sufficient to derive an indication of the frequency spectrum. The A/D converters 354, 356 are of such type that an adequate number of conversions can be made during each sweep of the local oscillator.

13 GB 2 059 064A 13 1 5 Once the spectrum fields formed of frequency, f,, and amplitude, Ai, signals are stored in memory 358, signal processor 360 is actuated to search for a peak amplitude value, AP, at 362. This is done by searching all of the stored amplitude values, A,, and comparing each with the next amplitude value and retaining the larger amplitude value for the next comparison. By preserving the frequency value, f,, associated with each each retained amplitude value, the frequency, fp, of the peak AP can be found and both are appropriately stored at 364.

In certain instances several peaks may occur in the stored spectrum samples. Although the largest peak is used to derive a thickness determination, one may also employ both peaks for this and select the casing thickness measurement which is closest to the nominal value as the proper measurement.

The detected peak values, both amplitude, AP, and frequency, fp, may then be recorded such as on plotter 122. The frequency, fp, may be recorded directly as an indication proportional to casing thickness, L, or the latter may be computed on the basis of the previously described relationship and then recorded. Other information may be simultaneously recorded on plotter 122 such as well depth on line 24, the cement bond signal on line 120, azimuth of a rotational15 scanning reflector on line 37 to identify the depth and circumferential location of the radial casing segment whose thickness was measured.

Fig. 17 20 In an alternate embodiment for determining casing thickness as shown in Fig. 17, the entire 20 reflection signal on line 63 is digitized as -described with respect to Fig. 8 for the evaluation of the cement bond. The digitizing process is commenced upon the detection of the arrival of the casing reflection by detector 336 which is described with reference to Fig. 15. The output pulse on line 338 from detector 336 is a pulse of sufficient duration to enable 25 digitizing of an entire reflection signal such as 64 (see Fig. 4A). This pulse activiates a network 25 370 which generates a pulse on line 372 with a duration generally about equal to that of the casing reflection segment 70 shown in Figs. 4. The pulse on line 372 in turn closes an analog casing logic 374 for this time period to pass the casing reflection segment 70 onto A/D converter 172. The latter digitizes the casing reflection segment 70 and stores the samples in a suitable memory (not shown).

When the casing reflection segment has passed, the pulse on line 372 goes inactive which, in turn, activates a network 342 to provide an enabling pulse on line 344 to permit analog reverberation gate 346 to pass a reverberation segment 72 through an amplifier 376, having a gain controlling input 378, to A/D converter 172.

The amplifier 376 permits amplification of the normally weak reverberation segment 72 for 35 more precise signal processing. The digitized reflection signal may be processed downhole or transmitted up the cable with a suitable telemetry device 380.

A signal processor 382 is provided to operate on the digitized reflection signal from A/D converter 172. The processor 382 provides a casing thickness determination at 384 and a cement bond evaluation signal, C13, at 386.

The casing thickness is determined by selecting the reverberation samples AR at step 388 and generate a spectrum thereof at 390 with a fourier transformation. The spectrum is formed of amplitude values A, and associated frequency values Fi.

The spectrum is then scanned to select the maximum peak value. This may be done by setting, at 392, a counter equal to the number, DN, of reverberation samples, a constant K = 1 45 and the values of AMAX and FIVIAX equal to zero.

A test is made at 394 whether the amplitude value A for the sample K is greater than AMAX.

If so, then the values for AMAX and FIVIAX are made equal to A(K) and F(K) at 396. The next samples may then be examined by incrementing K and decrementing the counter by one at 398 and testing for whether the counter is equal to zero at 400.

If not all of the samples have been scanned, the counter is not equal to zero and the search for a maximum spectrum value is repeated at 394. Once all of the samples have been scanned, the maximum values, AMAX and FIVIAX can be plotted at 384 or the casing thickness, L, derived from the formula L = N c 2 (FMAX) A cement bond evaluation can be conveniently made by signal processor 382 utilizing the 60 steps as described with reference to Fig. 8.

The cement bond signal CB varies as a function of casing thickness. This variation can be substantially removed from the cement bond signal at 402. This involves dividing the cement bond signal CB by a casing thickness signal L as determined at 404 from the frequency measurement FIVIAX using the casing thickness relationship as previously explained.

14 GB 2 059 064A 14 This normalization of the cement bond signal removes variations due to directly proportional casing thickness changes leaving lesser second order casing thickness effects. The cement bond for a particular radial segment can thus be advantageously evaluated in a manner which is substantially insensitive to the casing thickness at the same radial segment. Cement bond normalization relative to casing thickness may also be carried out directly with a cement bond 5 signal such as available at 182 in Fig. 17 or on line 117 in Fig. 1 before normalization by the casing reflection signal. The latter signal may then be employed to further normalize the cement bond evaluation as described.

Fig. 18 Fig. 18 shows an alternate embodiment for deriving the frequency of a peak in the spectrum of a reverberation segment 72. The outputs 350, 352 from spectrum analyzer 348 (see Fig.

15) are recorded on continuous tracks 410.1 410.2 of a storage medium 412 such as a magnetic disc or drum. After recording the output from analyzer 348 for a reverberation segment 72, the information is played back for analysis for an associated signal processing network 414 to detect, store and record the amplitude and frequency peak values, AP and fp.

The spectrum analyzer outputs 350 and 352 are shown coupled through logic amplifiers 416, 418 to record/playback heads 420, 422 operatively disposed with respect to magnetic storage disc 412. The amplifiers 416, 418 are enabled by the segment select pulse on line 344 (see Fig. 15). The amplitude, A, and frequency, f, signals are recorded on separate continuous 20 tracks 410.1, 410.2 which have sufficient recording length to record an entire reverberation segment 72.

After recording of the reverberation segment, logic playback amplifiers 424, 426 are enabled, by virtue of the removal of the disabling effect of the pulse on line 344 through inverter 428.

This then permits playback of the previously recorded amplitude, A, and frequency, f, signals. 25 j A peak detector 430 is provided to scan for the peak value in the amplitude signals played back through amplifier 424. The detected peak value is then applied to a comparator 432 together with another playback of the previously recorded amplitude signals on track 410.1 to enable the determination of the frequency, fp, at the time the peak occurs.

When comparator 432 recognizes equality between its inputs, a pulse is produced on output 30 line 434 to activate a sample and hold network, 436 coupled to sample the played back frequency signal, f, from amplifier 426. The frequency, fp, of the amplitude peak value is then stored and made available on output line 438 for recording and use as an indication of the thickness of the casing 22 as previously described.

The recording, peak scanning and peak frequency selection are carried out in sequence under 35 direction of control signals on line 440 from a control logic network 442. This network is initiated by the pulse on line 344 and subsequently by the playback of a recording of like pulses derived from a control track 410.3 on magnetic storage medium 412.

Fig. 19 illustrates another form 460 for an acoustic cement bond and casing investigating tool, wherein as in Fig. 1, a rotating reflector 38 is employed. The tool 460 is provided with a 40 stationary transducer 36 and a longitudinal cylinder 462 centrally and rotatably mounted relative to tool 460 about a rotational axis 464 which in this embodiment is preferably coincident with the central tool axis.

The tool 460 has an annular acoustically transparent window 466 mounted between an upper tool section 468 and a lower tool section 470. The cylinder 462 internally bridges the window 45 466 and rotationally engages the upper and lower sections 468, 470 through bearings 472.

The cylinder 462 has a tubular section 474 into which transducer 36 projects through an open end at 476. The tubular section 474 terminates at reflector 38 from where the cylinder 462 preferably is solid down to its end 476. Cylinder 462 is provided with a pair of annular radially extending flanges 478.1 and 478.2. Bearings 472 are clamped against flanges 478 50 with annular bushings 480 affixed to tool sections 468, 470 with screws such as 482. Bearings 472 fit in axially open annular grooves 484, 486 in flanges 478 and bushings 480 respectively. Bearings 464 provide both thrust and radial low friction support. Additional bearings and flanges can be employed if needed.

Cylinder 462 is of rugged strong construction to reinforce the lower tool section 470 to which 55 a load producing device, such as an externally mounted centralizer (not shown), can be applied.

The cylinder 462, thus serves as a strong reinforced bridge over acoustic window 466. The ability to employ a centralizer below the rotating reflector 38 enables a precise. placement of the rotational axis 464 relative to the casing 12 and thus preserve an accurate spacing of reflector 38 from casing 12.

The acoustic reflector 38 has a reflection angle a of a magnitude necessary to enable acoustic communication through a side-located opening 490 in tubular section 474. In front of opening 490 and contiguous with the outer wall of upper tool section 468 is the acoustic window 466 formed of a material having a predetermined acoustic impedance and provided with a shape selected to minimize undesirable acoustic reflection.

GB2059064A 15 of:

The acoustic window 466 is formed of a material whose acoustic impedance closely matches the acoustic impedance of a fluid; such as described with reference to Fig. 1, and which is placed in the space between source 36, reflector 38 and window 466. The acoustic temperature and pressure coefficients, i.e. the change in acoustic impedance as a function of temperature and pressure for both the fluid and the window 466 are selected as close as practically possible. 5 The acoustic window 466 can be made of a material as described with reference to window 40 in Fig. 1 or of polysulfone, a material sold by the Union Carbide Corporation under the trade name RADEL and having an acoustic velocity of about 2200 meters/second. Hence, as an acoustic pulse is generated from source 36 towards reflector 38, the acoustic energy passes through the fluid/window interface 492 with a minimum of reflection.

In order to further reduce the effect of acoustic reflections from a window interposed between the source 36 and casing 12, the window is conically shaped with an inclination angle 0 relative to reflector 38 as described with reference to Fig. 1 to permit use of a large reflector 38 and also to deflect secondary transmissions away from the casing 12.

Transducer 36 in Fig. 19 is mounted to a bracket 494 attached to the wall of tool section 15 468. An electrical cable 496 connects transducer 36 to electronic circuitry (not shown).

A rotational drive for cylinder 462 is provided by an electrical motor 498 mounted inside tool 460 and having an output shaft 500. A gear coupling 502 interconnects the motor shaft 500 to the cylinder 462.

The gear coupling 502 may take a variety of different forms and is, for illustrative purposes, 20 shown composed of a pair of pinions 504, 506, with the latter mounted to a shaft Q08 journaled in a bushing 510 on bracket 494. A bevel drive, formed of 45 bevel gears, 512, 514, is used to interconnect the shaft 508 with cylinder 462.

With a tool 460 as shown in Fig. 19, the structural integrity of the tool is extended to below the annular window 466. This provides additional strength below the window and permits its 25 centralization relative to casing 12 with a centralizer. Window 466 can be made sufficiently strong to withstand such twisting forces as may be coupled through from the rotating cylinder 462.

Attention is directed to our co-pending UK Patent Application No. 28819/78, from which the subject matter of the present application was divided.

Claims (24)

1. A method for determining the thickness of a casing cemented in a borehole penetrating an earth formation from a reflection signal derived from an acoustic investigation of the casing with an acoustic pulse directed at a radial segment of the casing and formed of acoustic waves at frequencies selected to stimulate a thickness resonance inside the casing walls, the method comprising the steps of:

generating a spectrum signal representative of the frequency spectrum of a reverberation segment of the reflection signal wherein said reverberation segment is substantially representa tive of acoustic reverberations between the casing walls at said radial segment; and determining the frequency of components in said spectrum signal contributing to a peak value thereof and providing a thickness signal representative of said measured frequency as indicative of the casing thickness at said radial segment.

2. The method of claim 1, wherein the reflection signal is formed of digital samples, said generating step produces a spectrum signal formed of samples indicative of amplitudes and 45 associated frequency values and said measuring step further includes scanning said amplitude samples for said peak value and selecting the associated frequency of sample contributing to said peak as representative of the casing thickness.

3. The method of claim 1, comprising:

generating an acoustic pulse from inside the casing in a radial direction towards the formation, wherein the acoustic pulse has a frequency bandwidth selected to stimulate a thickness resonance with acoustic reverberations inside the walls of a radial segment of the casing; detecting acoustic returns arising from the interaction of the acoustic pulse with materials in the path of the acoustic pulse and producing a reflection signal indicative thereof; and selecting from the reflection signal a portion which includes acoustic returns produced by said acoustic reverberations inside the walls of the casing, said frequency spectrum being formed from the selected reflection signal portion.

4. The method of claim 3, wherein said frequency determining step further includes the steps of:

digitizing the frequency spectrum to form samples thereof with associated frequency values for the samples; and scanning the samples to determine a peak value thereof.

5. The method of claim 3, wherein said frequency spectrum forming step includes the steps 16 GB2059064A 16 applying said selected portion to a spectrum analyzer to generate an amplitude signal representative of the amplitude of the frequency components in the selected portion and a frequency signal representative of the frequency of the components in said amplitude signal; and storing said amplitude and frequency signals; scanning said stored amplitude and frequency signals to detect a peak value of the amplitude signal with its associated frequency signal as an indication of the thickness of the casing.

6. The method of claim 3, wherein said frequency spectrum forming step further includes the steps of:

digitizing said selected portion to form digital samples thereof; and forming a fourier transform of the digital samples of the selected portion.

7. The method of claim 1, comprising steps of:

generating a highly damped acoustic pulse from inside the casing in a radial direction towards a radial segment of the casing with said acoustic pulse being generated with acoustic wave frequencies in a bandwidth selected to stimulate the casing into a thickness resonance with 15 acoustic reverberations between the walls of the casing; detecting acoustic returns arising from the interaction of the acoustic pulse with materials in the path of the acoustic pulse and producing a reflection signal indicative thereof; and converting the reflection signal to digital samples, the converting step comprising forming a frequency spectrum of samples representative of 20 casing reverberations occurring subsequent to samples representative of an initial acoustic reflection off the inner wall of the casing with the frequency spectrum composed of amplitude samples with associated frequency values.

8. The method of claim 7, further including step of summing absolute values of samples representative of casing reverberations to provide a bond 25 signal indicative of the quality of the bond between the casing and the cement.

9. The method of claim 8, further including steps of:

summing absolute values of samples representative of the initial casing reflection; and forming a quotient between said respectively summed samples to provide a normalized bond signal.

10. An apparatus for determining the thickness of a casing cemented in a borehole penetrating an earth formation from a reflection signal derived from an acoustic investigation of the casing with an acoustic pulse directed at a radial segment of the casing and formed of acoustic waves at frequencies selected to stimulate a thickness resonance inside the casing walls, the apparatus comprising:

means for selecting from the reflection signal a reverberation segment substantially representative of acoustic reverberations between the casing walls; means for generating a spectrum signal representative of the frequency spectrum of said reverberation segment; and means for determining the frequency of components in said spectrum signal contributing to a 40 peak value thereof and producing a thickness signal representative thereof as the casing thickness.

11. The apparatus of claim 10, wherein the reflection signal is formed of digital samples and said spectrum generating means includes means for generating a fourier transform of samples representative of the reverberation segment as said spectrum signal.

12. The apparatus of claim 10, wherein said means for determining the peak value further includes means for producing samples of the spectrum signal with associated values of the frequency of the samples; means for scanning said spectrum samples for a peak value thereof and selecting the 50 associated frequency value as an indication of the thickness of the casing.

13. The apparatus of claim 12, wherein said portion selection means further includes means responsive to the reflection signal for detecting a signal therein representative of an initial acoustic reflection signal from the casing; and means responsive to the directed casing reflection signal for selecting said portion following 55 the initial casing reflection.

14. The apparatus of claim 10, comprising:

means for generating a highly damped acoustic pulse from inside thecasing. in a radial direction towards a radial segment of the casing with said acoustic pulse being generated with acoustic wave frequencies in a bandwidth selected to stimulate an acoustic resonance between 60 the walls of the casing with acoustic reverberations and providing a reflection signal representa tive of acoustic returns caused by the acoustic pulse; and means for generating digital samples of the reflection signal; wherein said selecting means comprises means for selecting samples representative of said casing reverberations and occurring subsequent to samples representative of an initial casing reflection; 65 Z 17 GB 2 059 064A 17 said generating means comprises means for generating a spectrum of the selected reverberation samples and form amplitude samples with associated frequency values; and said determining means comprises means for determining a maximum amplitude sample and its associated frequency value as an indication of the thickness of the casing.

15. The apparatus of claim 14, further including means for summing the absolute value of the selected samples representative of the reverberations in the casing as a measurement of the quality of the bond between the casing and cement.

16. The apparatus of claim 14, including:

means for selecting samples representative of an initial acoustic casing reflection of the inner 10 wall of the casing; means for summing the absolute values of the samples representative of the initial acoustic casing reflection; means for summing the absolute values of the selected samples representative of the casing reverberations as a measurement of the quality of the bond between the casing and cement; and 15 means for forming a quotient between the respective sums generated by the summing means to normalize said measurement of the quality of the cement bond.

17. A method for determining the thickness of a casing cemented in a borehole, substan tially as hereinbefore described with reference to Fig. 15 of the accompanying drawings.

18. A method for determining the thickness of a casing cemented in a borehole, substan- 20 tially as hereinbefore described with reference to Figs. 15 and 17 of the accompanying drawings.

19. A method for determining the thickness of a casing cemented in a borehole, substan tially as hereinbefore described with reference to Figs. 15 and 18 of the accompanying drawings.

20. A method for determining the thickness of a casing cemented in a borehole, substantially as hereinbefore described with reference to Figs. 15 and 19 of the accompanying drawings.

21. An apparatus for determining the thickness of a casing cemented in a borehole, 30 substantially as hereinbefore described with reference to Fig. 15 of the accompanying drawings. 30

22. An apparatus for determining the thickness of the casing cemented in a borehole, substantially as hereinbefore described with reference to Figs. 15 and 17, of the accompanying drawings.

23. An apparatus for determining the thickness of a casing cemented in a borehole, substantially as hereinbefore described with reference to Figs. 15 and 18.

24. An apparatus for determining the thickness of a casing cemented in a bo.rehole, substantially as described with reference to Figs. 15 and 19 of the accompanying drawings.

Printed for Her Majesty's Stationery Office by Burgess & Son (Abingdon) Ltd-1 98 1. Published at The Patent Office, 25 Southampton Buildings, London, WC2A lAY, from which copies may be obtained.