Transient Electromagnetic Method for Directly Detecting Corrosion on Conductive Containers

The present invention relates to a non-destructive method for directly detecting corrosion on electrically conductive containers such as pipelines, storage vessels, pressure vessels and the like.

Oil and gas pipelines located at Alaska's Prudhoe Bay are wrapped with a jacket of insulating material to prevent the rapid cooling, and provide better transportability, of oil and gas fluids. The outer surface of the insulation is covered by a metal jacket for keeping out moisture. The metal jacket is typically provided in two half portions with each portion having flanges for aiding in the retention of the jacket on the pipeline. The two half portions of the jacket are joined together at the flanges which form seams. Water occasionally enters through the jacket seams and travels through the insulation to the pipe where it causes corrosion.

Prior art methods of detecting pipeline corrosion have proven inadequate. For example, pigs with corrosion detection equipment can only be used on pipelines that have access locations; many pipelines lack such locations. Ultrasonic detection methods require removal of the metal jacket and insulation, a timely and expensive procedure. Radiography detection methods are potentially hazardous and the equipment is cumbersome, requiring impractical or inconvenient adjacent vehicular support. Furthermore, with radiography methods it is often difficult to distinguish between corrosion pits filled with corrosion products and uncorroded portions of pipe walls. What is needed then is a method of detecting corrosion through insulation and the surrounding jacket, and which method can be practiced with portable equipment.

Electromagnetic probing techniques provide such a method for detecting corrosion through insulation. In the prior art, frequency domain electromagnetic probing techniques are used to detect corrosion in aircraft fuel tanks. Frequency domain electromagnetic probing techniques utilize a small number of frequencies and measure magnitude and phase differentials between the transmitted signals and the received signals. However, because frequency domain techniques, as a practical matter, utilize only a small number of frequencies, the amount of information obtained is inherently limited, thus detracting from the accuracy of the techniques.

The application "METHOD FOR DETECTING CORROSION ON CONDUCTIVE CONTAINERS" by Brian Spies (the inventor herein) and the application "METHOD FOR DETECTING CORROSION ON CONDUCTIVE CONTAINERS HAVING VARIATIONS IN PROTECTIVE COVERING THICKNESS" by Pedro Lara, which applications are assigned to the assignee of the present invention and which applications are filed the same day as the present application disclose time domain electromagnetic probing methods for use in detecting corrosion in conductive containers.

The present application discloses a method of directly detecting corrosion on conductive containers. Such a method is particularly useful on extruded pipe, on which it is difficult to indirectly detect corrosion using wall thickness measurement methods because the manufacturing process of extruding pipe results in larger variations of pipe wall thickness over short distances than are found in rolled and welded pipe.

It is an object of the present invention to provide a method for directly detecting corrosion on insulated conductive containers.

FIG. 1 is a schematic diagram showing a typical situation in which the method for detecting corrosion in a container in accordance with a preferred embodiment of the present invention can be practiced, together with typical testing apparatus.

FIG. 2 is a schematic diagram showing a transverse cross-section of the pipeline of FIG. 1.

FIG. 3 is a schematic cross-sectional view showing the antenna means of FIG. 2 in detail.

FIG. 4 is a graph showing the time domain response curves of various conductors, obtained by the transient electromagnetic probing (TEMP) method of the present invention.

FIG. 5 is a graph of the response curve of a pit in a pipe wall, with the response curve obtained by computing the ratio of the "corrosion" to the "no corrosion" response curves of FIG. 4.

FIG. 6 is a graph showing a longitudinal cross-sectional TEMP profile of the pit of FIG. 5, with the profile being obtained by averaging the late time responses at each antenna means location.

FIG. 7 is a graph showing the effects of the jacket flanges and of variations in antenna means height on time domain responses of pipe walls.

FIG. 8a is a circumferential map of a portion of a pipe showing both the location of corrosion and the ultrasonic wall thickness measurements.

FIG. 8b is a graph showing transverse TEMP profiles of the unjacketed pipe of FIG. 8a, taken along line A--A.

FIG. 8c is a graph showing transverse TEMP profiles of the jacketed pipe of FIG. 8a, taken along line A--A, with the

FIG. 8d is a graph showing the same TEMP profiles as in FIG. 8c, but uncorrected for the effects of the jacket flanges.

FIG. 9 is a graph showing a plurality of time domain response curves for various pipes, there being shown a plurality of response curves for each pipe taken at different locations on each pipe, with the response curves corrected for variations in distance between the antenna means and the pipe walls.

FIG. 10 is a graph showing a plurality of response curves for a pipe taken at different locations on the pipe, with the response curves corrected for variations in distance between the antenna means and the pipe wall, there being shown a nomogram superimposed on the corrected response curves.

FIG. 11 is a graph showing time domain response curves of various conductors, obtained by the transient electromagnetic probing (TEMP) method of the present invention.

FIG. 12 is a graph showing processed response curves, the response curves having been processed by determining the rate of decay of the response curves of FIG. 11.

FIG. 13 is a graph showing time domain response curves for a portion of a pipe wall.

FIG. 14 is a graph showing processed response curves, the response curves having been processed by determining the rate of decay of the response curves of FIG. 13.

In FIGS. 1-3 there is schematically shown a typical situation in which the method of detecting corrosion in electrically conductive containers 11 can be practiced, together with typical detecting apparatus 25. The method of the present invention utilizes transient electromagnetic probing (TEMP) to detect corrosion.

The conductive container shown in FIGS. 1-3 is a portion of a pipeline 11, which is of course made up of a plurality of individual pipes 13. The pipes 13 have a diameter and the pipe walls 15 have a thickness. The pipe walls 15 are made up of an electrically conductive material such as steel.

In Alaska's Prudhoe Bay region, pipelines wrapped with insulating material 17 are used to transport oil and gas fluids. The insulation 17 is provided to prevent rapid cooling of the oil and gas fluids in the pipeline and thus provide better transportability of these fluids in the pipeline. In refineries, pipelines and vessels are commonly wrapped with insulation as a safety measure in protecting personnel from high temperatures. The insulation 17 on pipelines is typically a thermoplastic foam such as polystyrene, and has a radial thickness. Surrounding the insulation 17 is a metal jacket 19 which is provided to keep out moisture. The jacket 19 has a thickness which is much less than the thickness of the pipe wall. The metal jacket 19 has two half portions that extend longitudinally along the pipeline. Each jacket half portion has seam means in the form of flanges 21 that extend radially outward. When the jacket half portions are assembled onto the pipeline, the respective flanges 21 abut one another to form seams. The half portions are retained in place on a pipeline by securing the respective flanges together with suitable means.

In FIG. 3, the pipe wall 15 is shown to have a corrosion pit 23 adjacent to the insulation. The corrosion acts to reduce the thickness of the pipe wall, wherein it forms the pit and fills the pit with corrosion products. The corrosion that has pitted the pipe wall is caused by water that has entered the insulation between the jacket flanges 21.

Detecting apparatus 25 is provided near that portion of the pipe wall which is to be tested for corrosion and includes antenna means 27, a transmitter 29, a receiver and amplifier 31, and a digital computer 33.

The antenna means 27 include a transmitting antenna coil 35, a receiving antenna coil 37 and core means 39. In the preferred embodiment, the transmitting and receiving antenna coils 35, 37 are wound onto the same core means 39, an arrangement which is hereinafter referred to as coincident (see FIG. 3). The core means 39, which is in the shape of a spool, is made of a non-magnetic and non-conductive material such as plastic. The number of turns of the transmitting antenna coil are kept to a minimum to minimize the inductance of the transmitting antenna and to provide for an abrupt switching off of the transmitting antenna coil. In the preferred embodiment, the transmitting antenna coil 35 is made up of 120 turns of 20 to 24 gauge wire. The receiving antenna coil 37 is made up of 400 turns of 34 to 40 gauge wire. The transmitting and receiving antenna coils 35, 37 are connected to the transmitter 29 and receiver 31 by respective pairs of wires 41, 43.

The transmitter 29 which is conventional, generates a train of pulses having magnitudes of 1 to 5 amps. As discussed in more detail below, a plurality of pulses are transmitted for each location of the antenna means 27 for data enhancement purposes. The pulses have abrupt fall times on the order of 10 to 100 microseconds. The pulses of the transmitter pulse train alternate polarity to eliminate dc bias in the instrumentation. The duration of each pulse is sufficiently long to stabilize the pulse magnitude so that there are no induced currents in the pipe wall before the end of the pulse. The transmitter 29 repeats the pulses at a repetition rate that allows all of the necessary data to be obtained for each pulse. For example, a thick pipe wall requires more time to obtain data than does a thinner pipe wall because the induced current takes longer to diffuse in the thick pipe wall. Thus, the repetition rate of pulses will typically be slower for thick pipe walls than for thinner pipe walls.

The receiver and amplifier 31 is a broad band instrument with a wide (5 or 6 orders of magnitude) dynamic range. The receiver 31, which has an A/D converter, samples the signal at a constant rate and integrates the signal over a time window or channel. The duration of the time windows increases with time. The transmitter 29 and the receiver and amplifier 31 are conventional. In practice it has been found that the SIROTEM transmitter, receiver and amplifier unit manufactured by Geoex Pty. Ltd. of Adelaide, Australia, works well. The battery operated SIROTEM unit is portable, a characteristic which allows ease of use when surveying pipelines in the field.

The digital computer 33 is a conventional portable computer with sufficient memory capacity to record the data.

The method of detecting corrosion on a conductive container of the present invention will now be described. As mentioned earlier, the method of the present invention utilizes transient electromagnetic probing (TEMP). TEMP allows the remote probing of a conductor by inducing a current into the conductor and then analyzing the decay of the current.

First, the antenna means 27 is placed on the jacket 19 so as to be in proximity with the near surface 45 of the portion of the pipeline 11 that is to be investigated. Suitable means (not shown) are used to secure the antenna means 27 in position so as to minimize any motion of the antenna means over the investigated pipe wall portion. The transmitting antenna coil 35 is then energized by the transmitter 29 with a pulse. As described above, the transmitting antenna coil 35 is energized for a sufficient period of time to stabilize the pulse magnitude, thereby insuring no eddy currents are induced into the pipeline 11. Then, the transmitting coil 35 is abruptly de-energized by the transmitter by having the pulse fall off rapidly to zero magnitude. This abrupt de-energization of the transmitting antenna coil 35 induces eddy currents into the conductors located near the coil; namely the jacket 19 and the pipe wall 15. The eddy currents, which decay ad diffuse away from the antenna means 27 inside of the respective conductors, create a magnetic field that is detected as a time-varying voltage in the receiving antenna coil 37. As soon as the transmitting antenna coil is de-energized, the receiver 31 then switched on. The receiving antenna coil 37 detects the presence of and the decay of the induced eddy currents in the conductors. The eddy currents are gradually dissipated within the conductors by resistive heat losses. The rate of diffusion is dependent on the conductivity and thickness of the conductor. The receiver 31 samples the signal as detected by the receiving antenna coil 37, whereupon it is amplified to a suitable level and sent to the digital computer 33 for storage and processing. The receive 31 measures the signal from the time the eddy currents are first induced into the conductors until the signal becomes indistinguishable from noise. The level of noise is reduced by minimizing any motion of the receiving antenna coil 37 relative to the conductors. The received signal is unprocessed data and forms a record in the computer 33 of the decay of the induced currents in the conductors. The transmitting and receiving procedure is repeated many times with the antenna means 27 in the same location to increase the signal-to-noise ratio.

The data is then processed by computer data processing means into a suitable format for interpretation. The first steps in the processing of the data involve the normalization of the received signals and the summing and averaging of the received signals. Because the transmitter 29 in the preferred embodiment is battery operated, the magnitude of the transmitter current is subject to variation. The effects of variation in magnitude in the data are removed by normalizing the received voltage to the transmitted current. The summing and averaging of the received signals for a particular antenna means location serves to increase the signal-to-noise ratio. In particularly noisy environments, as an alternative to summing and averaging, selective stacking can be used to eliminate noisy transients. The result of this initial data processing is a time-varying response curve such as shown in FIG. 4. (FIG. 4 illustrates response curves for various conductors.)

The response curves may be interpreted in accordance with methods which will now be described, with reference to FIGS. 4-8d. Referring in particular to FIG. 4, the presence or absence of corrosion on a conductor wall is inferred by examining the shape of the various response curves which have been taken over the area of interest. The shape of each response curve depends in part on the thickness of the conductor wall. For example, the magnitude of the response curve of an infinitely thick conductor wall decays at a fairly even rate (on a log-log graph), resulting in a fairly straight response curve, whereas the response curve of a conductor having a finite wall thickness begins to break at some point into a more pronounced downward direction than before and decays at a faster rate. This breaking phenomenon is attributed to the induced currents diffusing to and reaching the far surface 47 of the conductor wall. Response curves for thin conductor walls break at earlier times than do response curves for thicker conductor walls.

Because corrosion reduces the thickness of a conductor wall, the presence or absence of corrosion can be inferred by comparing the shape of the response curve for the investigated pipe wall portion to the shape of the response curve for an uncorroded portion of the same type of pipe. For example, in FIG. 4, the two response curves labeled "corrosion" and "no corrosion" are taken from the same pipe. The "no corrosion" response curve is taken from an uncorroded portion of the pipe and is used as a reference, while the "corrosion" response curve is taken from a different portion of the same pipe, which different portion has a pit to simulate corrosion (with the antenna means located at the same distance from the pipe wall, for both response curves). At about 17 ms (milliseconds), the "corrosion" response curve breaks into a more pronounced downward direction and begins to decay at a faster rate than before. The "corrosion" break point occurs at an earlier time than does the "no corrosion" break point (at about 25 ms), indicating that the conductor wall represented by the "corrosion" response curve is thinner than the conductor wall represented by the "no corrosion" response curve.

Referring now to FIG. 5, the "corrosion" and "no corrosion" response curves of FIG. 4 are compared by plotting the ratio of the two curves as a percent response curve, using the "no corrosion" response curve as a reference. The percent response curve highlights the differences between the "corrosion" and the "no corrosion" response curves. By examining the late time portions of the percent response curve (from about 17-20 ms on, which is about when the "corrosion" response curve of FIG. 4 begins to break sharply downward), one can see that the "corrosion" response curve deviates 20 to 30 percent from the "no corrosion" response curve. This 20 to 30 percent difference clearly indicates a difference in wall thickness between the corroded portion of the pipe and the uncorroded portion of the pipe.

In FIG. 4, the response curve labeled "jacket only" is that taken from the metal jacket 19, without the pipe 13. The "jacket only" response curve decays very rapidly so that by the relatively late time of 20 ms, the jacket 19 contributes very little to the total response. This is because the wall thickness of the jacket is much smaller than is the thickness of the pipe wall, so the currents diffuse much more rapidly in the jacket. Thus, for those portions of the "jacket and pipe" response curves that are of interest in locating corrosion (that is the later times), the effect of the jacket can be ignored.

Responses measured near jacket flanges are affected quite strongly by the jacket flanges at all times, as shown in FIG. 7. A response measured near jacket flanges can be corrected to remove the effects of the jacket flanges by normalizing the affected response curve to a reference response curve obtained away from the jacket flanges. As shown in FIG. 7, an effect of the jacket flanges on the response curve is a generally parallel shift in a downward direction in the intermediate and late time ranges (later than about 4 ms). That is to say that in the intermediate and late time ranges, the affected response curve is generally parallel to the reference response curves. The affected response curve is corrected by normalizing the affected response curve to the reference response curve in the intermediate time range.

FIG. 7 also serves to illustrate the effect that variations in distance between the antenna means and the pipe wall at one location on the pipe and between the antenna means and the pipe wall at another location on the pipe can have on responses. Such variations in distance result from non-uniform thicknesses of the insulation between the pipe wall and the jacket. Increasing the distance of the antenna means from the pipe wall causes the magnitude of the response to decrease at intermediate and late times, which decrease in magnitude shows up as a generally parallel shift. The responses can be corrected to remove the effects of variations in distance by normalizing the response curves to a reference response curve obtained with the antenna means at some known distance, in the intermediate time range.

The antenna means 27 gives a reading of the average conductor wall thickness over a search area. The size of the search area depends upon antenna size, antenna configuration and the duration of the receiver measuring time after each transmitter pulse. The search area of the antenna means increases with larger antenna sizes or with longer measuring times. In the preferred embodiment, the antenna means 27 has a diameter of about 3 inches. For a 10.5 inch pipe, the search area is about 12 inches in diameter.

In the usual case, the portion of the pipeline that is to be investigated for corrosion is much larger than the search area of the antenna means. Therefore, a typical pipe survey requires the antenna means to be moved to new locations to complete the survey. In FIGS. 8a through 8d there are shown a corrosion map of a pipe section and corresponding TEMP surveys or profiles along line A--A of the pipe section. In obtaining the TEMP profiles of FIGS. 8b through 8d, the antenna means was positioned at various locations-along line A--A. In FIG. 8a, the numbers along line 8a indicate ultrasonic point measurements of the wall thickness- (in inches) and the shaded areas indicate heavy corrosion, wherein the thickness of the pipe wall is less than for the unshaded areas. The map shows that the pipe wall along line A--A is thickest around 180.degree. and gets thinner moving towards 0.degree. and 360.degree..

FIG. 8b shows TEMP profiles of the pipe of FIG. 8a along line A--A, without a metal jacket. In FIG. 8b only those values of the response curve at selected discrete instances of time for each antenna means location are plotted. The response curve values at equivalent instances of time are then connected together to form a TEMP profile. Thus, for each antenna means location, the response curve values at time=8.5 ms, 32.8 ms, 67 ms, 79 ms, 92 ms, and 105 ms are plotted, forming respective TEMP profiles of pipe wall thickness. Each TEMP profile is normalized to the TEMP response obtained over the thickest portion of the pipe. As can be seen in FIG. 8b, the TEMP profiles show that in moving away from 180.degree. in either direction (towards 0.degree. and towards 360.degree.) the pipe wall thickness lessens and is thinnest around 0.degree. to 60.degree. and 320.degree. to 360.degree.. The late time TEMP profiles (67 ms and greater) in particular clearly show the reduced wall thickness, corresponding with the pipe corrosion map of FIG. 8a.

In FIG. 8c, there are shown TEMP profiles of the pipe of FIG. 8a along line A--A, but with a metal Jacket. The TEMP profiles of FIG. 8c were obtained in the same manner as the TEMP profiles of FIG. 8b. The jacket flanges, which are located at approximately 95.degree. and 270.degree., have caused reductions in the amplitudes of the TEMP profile portions near the flanges. The TEMP profiles of FIG. 8c have been corrected to reduce the effects of the jacket flanges by normalizing the responses measured near the jacket flanges to a response measured away from the jacket flanges. The responses are normalized in the intermediate time range (3-6 ms) and the late times (32 ms and greater) are then analyzed. (In FIG. 8d there are shown the TEMP profiles of FIG. 8c before the profiles have been corrected for the effects of the jacket flanges.) There is a good correlation between the TEMP profiles of FIG. 8c and the corrosion map of FIG. 8a. The TEMP profiles of FIG. 8c show that the pipe wall is reduced in thickness around 0.degree. to 60.degree. and 320.degree. to 360.degree., thus leading to an inference of corrosion at those locations.

FIGS. 8a through 8d illustrate an advantageous difference of the TEMP method over the ultrasonic method. The ultrasonic method makes point measurements, requiring a large number of measurements, whereas the antenna means of the TEMP method has a large search area requiring fewer measurements. While the ultrasonic measurements in FIG. 8aare essentially confined to line A--A, the TEMP measurements encompass portions of the pipe extending for a few inches to either side of line A--A. Furthermore, ultrasonic measurements must be made on the bare pipe, while TEMP measurements can be made on the jacket.

For TEMP profiles such as are shown in FIGS. 8b-8d, the effects on the responses due to the variations in distance between the antenna means and the pipe wall, which variations are caused by moving the antenna means from one location on the pipe to another location, can be corrected for by creating reference response curves with the antenna means placed at a number of known distances from the pipe wall. The intermediate times of the response curves having distance error are then normalized to the intermediate times of the respective reference response curves.

In FIG. 6, there is shown a TEMP profile of the corrosion pit of FIG. 5. The TEMP profile is obtained by moving the antenna means to a plurality of locations and averaging the responses for the 25 to 52 ms time window at each antenna means location. The physical extent of the corrosion pit is indicated at the bottom left corner of the graph, which shows the pit to have a radius of about 8 inches. The TEMP profile of FIG. 6 shows a good correlation to the physical profile. From about 17 inches on, the TEMP profile shows a slight decrease in magnitude due to the induced currents interacting with the nearby pipe end.

Another method of interpretation of the response curves of FIG. 4 involves examining the time at which the far surface 47 of the pipe wall is initially manifested in the response curve. This time is referred to as the "critical time", and is that point where the response curve begins to break into a more pronounced downward direction than before, as discussed hereinbefore (see FIG. 4). The wall thickness of the pipe is proportional to the square root of the critical time. The constant or factor of proportionality is dependent on the geometry and the conductivity of the pipe, and may be determined by making a determination of the critical time of a particular thickness of the pipe.

The method of the present invention can be used to make quantitative measurements of wall thickness, once the instruments and data have been calibrated on pipes of known thickness and conductivity. Once the actual wall thickness of the investigated pipe is known, comparison to the manufactured wall thickness leads to a determination of wall loss due to corrosion on the investigated pipe.

Another method for correcting error in the responses due to variations in distance between the antenna means and the pipe wall from one location along the pipe to another location will now be described, with reference to FIGS. 9 and 10.

In FIG. 9 there are shown a plurality of TEMP response curves (plotted as discrete values instead of as continuous values) that have been corrected for effects due to variations in distance between the antenna means and the pipe walls. FIG. 9 illustrates the corrected response curves for a forty-two inch diameter pipe (with a pipe wall thickness of 0.438 inches), a twelve-inch diameter pipe (with a pipe wall thickness of 0.406 inches), and a thirty-inch diameter pipe (with a pipe wall thickness of 0.344 inches). For each pipe, numerous TEMP surveys were taken at different locations along the pipe. The TEMP surveys were obtained through insulation 17 and a metal jacket 19. The thickness of the insulation along each pipe varied as much as fifty percent, thus causing the distance between the antenna means and the pipe walls to vary by the same amount.

The record of the decay of induced current in a pipe wall (as shown by the respective designated response curves of FIG. 4) is corrected for the variations in distance by determining with respect to time the rate of decay of the induced current. Correcting for variations in distance by determining the rate of decay of induced current recognizes the phenomenon that variation in distance between the antenna means and the pipe wall affects the magnitude of the response, but does not affect the rate of decay of the response. In FIG. 9, the corrected response curves shown illustrate the respective rates of decay of uncorrected TEMP survey responses obtained from the pipes. The rates of decay were obtained by taking the logarithmic derivative (d(ln V)/d(ln t); where V is the received voltage and t is time) of the uncorrected TEMP survey responses. (Central difference numerical methods were used to obtain the graphs of FIGS. 9 and 10.) As can be seen, in spite of the variations in insulation thickness, the vertical scatter among the corrected TEMP surveys for each pipe is small, and is in fact due to variations in pipe wall thickness. Thus, by determining the rates of decay of the responses, the effects of variation in distance between the antenna means and the pipe wall have been corrected for. The presence or absence of corrosion in the pipe wall is inferred by the interpretation methods discussed hereinabove.

The rate of decay correction method also allows the quantitative determination of wall thickness. As the induced currents diffuse through the conductor walls from the near surface 45 (see FIG. 3), the responses decay at a constant logarithmic rate of about -1.5. Then, as the induced currents begin to interact with the far surface 47 of the conductor wall, the responses decay at increasing rates. The time of departure of a response decay rate from the constant logarithmic rate of about -1.5 is a function of the pipe wall thickness and the pipe diameter. The corrected responses of thinner pipe walls break downwardly at an earlier time than do the corrected responses of thicker pipe walls. Increases in pipe diameter cause the responses to break downwardly at later times. After the induced current reaches the far surface of the pipe wall, the rate of decay of the response approaches asymptotically a constant second derivative. This asymptotic portion of the rate of decay of the response is independent of the thickness of the pipe wall or the pipe diameter and has been determined empirically to be:

where A is a function of pipe wall thickness, pipe diameter, and pipe metallurgy. The asymptotic characteristic of the rate of decay of the response, coupled with the time of departure of a response decay rate being dependent upon the pipe wall thickness allows the construction of nomograms that can be superimposed on the corrected response curves. The nomograms are constructed from numerous reference records which are created by inducing current into pipe walls of known thickness and substantially similar diameters. In FIG. 10, there is shown a nomogram consisting of straight lines superimposed on corrected response curves (TEMP surveys A, B, and C) obtained from various locations along a corroded eight-inch diameter pipe. Thus, it can be seen by extrapolatory comparisons between the individual TEMP surveys and the nomogram that the pipe wall portion that was probed with TEMP survey A had a thickness of about 0.46 inches, the portion that was probed with TEMP survey B had a thickness of about 0.43 inches, and the portion that was probed with TEMP survey C had a thickness of about 0.41 inches. From these quantitative pipe wall thickness measurements, the presence or absence of corrosion on the pipe wall can be inferred.

A method of directly detecting corrosion on pipe walls will now be described with reference to FIGS. 11-14. Such a method is particularly useful for detecting corrosion on the walls of extruded pipe. Due to the extrusion process used during manufacturing, the pipe wall thickness of extruded pipe varies over relatively large tolerances, and cannot always be used to directly infer the presence of corrosion.

The method of directly detecting corrosion utilizes my discovery that the corrosion scale or products on steel pipes exhibits superparamagnetic properties. Superparamagnetism, or viscous magnetization as it is also known, is a property wherein the magnetic susceptibility of a material is effectively time-dependent. Most materials are not superparamagnetic; the magnetic flux density B reacts almost instantaneously to changes in the applied magnetic field H. However, in superparamagnetic materials, the magnetic flux density B reacts somewhat sluggishly to changes in the magnetic field. Superparamagnetism has been observed in magnetite, maghemite and hematite. Analysis of corrosion scale taken from NGI gas pipelines reveal that the scale is predominantly composed of magnetite and hematite, with some maghemite and manganese dioxide, and exhibits TEMP behavior indicative of superparamagnetism.

The detection of corrosion scale having superparamagnetic properties begins with the utilization of the TEMP methods described hereinabove. In FIGS. 11 and 12 there are shown respective TEMP response curves for a pipe and a bag of corrosion products. The response curves for the pipe behaves as described hereinabove. The response curve for the corrosion products decays at a constant rate described as a 1/t time relationship (a slope of -1 on a log-log scale). The constant (1/t) rate of decay of the response curve for the corrosion products results from the superparamagnetic property of the corrosion products. At relatively early and intermediate times, the magnitude of the response curve of the corrosion products is much smaller than the magnitude of the response curve of the pipe. At relatively late times, the response curve of the pipe decays at a much faster rate than does the response curve of the corrosion products, causing the magnitude of the pipe response curve to first approach the corrosion products response curve in magnitude and then to diminish in magnitude relative to the corrosion products response curve Thus, for a corroded portion of pipe, the TEMP response of the pipe portion has a response component due to the pipe wall and a response component due to the corrosion, with the pipe wall component overwhelmingly predominating at relatively early and intermediate times and the corrosion component detectable at relatively late times.

FIGS. 13 and 14 illustrate an application of the method of the present invention, in accordance with a preferred embodiment. FIGS. 13 and 14 show respective TEMP response curves obtained from a 10.5 inch diameter NGI pipe. The "corrosion" response curve was obtained with the antenna means placed over a patch of corrosion scale on the pipe with an approximate thickness of 1/8 inch (corresponding to a wall thickness loss of about three percent). The "no corrosion" response curve was obtained with the antenna means placed over an uncorroded portion of the pipe. FIG. 14, in particular, shows the difference between the two response curves. In FIG. 14, the response curves of FIG. 13 were processed to enhance the corrosion component or lack thereof by determining the respective rates of decay. At relatively late times (from about 80 ms on) the "corrosion" response curve has a decrease in the magnitude of the rate of decay, shown as an upward inclination of the "corrosion" response curve, whereas the "no corrosion" response curve continues to decay at an increasing rate. This slowing of the decay rate of the "corrosion" response curve at relatively late times is caused by the corrosion scale component of the response curve which becomes detectable in the overall response of the corroded portion of pipe. At relatively late times, the magnitude of the pipe wall component of the "corrosion" response curve first approaches the magnitude of the corrosion scale component and then diminishes relative to the corrosion scale component since the pipe wall component magnitude decreases at an increasing rate of decay. If the TEMP system signal-to-noise ratio is sufficiently large, the TEMP response of the pipe can be detected for a sufficient length of time so that the corrosion scale component predominates over the pipe wall component.

The above described method can be used on insulated and uninsulated pipes or containers.

An important aspect of the method of the present invention is that corrosion on a container means wall can be directly detected, regardless of variations in the thickness of the container means wall. The method utilizes my discovery that corrosion scale on steel pipes has superparamagnetic properties.

Although the method of the present invention has been described for use in detecting corrosion on pipelines, the method may also be used to detect corrosion on the electrically conductive walls of other types of container means such as storage vessels and pressure vessels. In addition, the method of the present invention can be used on uninsulated as well as insulated container means.

The antenna means can have the transmitting antenna and receiving antenna configured in arrangements other than the coincident arrangement described herein. One such arrangement has the transmitting antenna separate but coplanar with the receiving antenna. Another arrangement has a plurality of receiving antennas located within a large transmitting antenna loop.

Although this invention has been described with a certain degree of particularity, it is understood that the present disclosure is made only by way of example and that numerous changes in the details of construction and the combination and arrangement of parts may be resorted to without departing from the spirit and the scope of the invention, reference being had for the latter purpose to the appended claims.