EP0886759A1 - Compton backscatter pipe wall thickness gauge employing focusing collimator and annular detector - Google Patents

Abstract

This invention is directed toward the measure of the thickness of material using nuclear techniques. More particularly, the invention is directed toward apparatus and methods for measuring the thickness of pipe wall, from the outside of the pipe (70), using an annular detector (50), concentric conical detector collimation (26), and backscatter gamma radiation. A source of gamma radiation (14), preferably at energy 279 keV from mercury-203, is collimated into a pencil beam and impinged perpendicularly upon the outside of the pipe wall to be measured. Compton backscatter radiation from the pipe wall is then measured with an annular, concentric photon detector which preferably comprises a scintillation crystal, such as sodium iodide, optically coupled to a single photomultiplier tube. Pipe thickness is determined from detector response using a predetermined relationship obtained during system calibration. Conical, concentric detector collimation is used to obtain accurate and precise measurements when the pipe is filled with liquid.

Description

COMPTON BACKSCATTER PIPE WALL THICKNESS GAUGE EMPLOYING FOCUSING COLLIMATOR AND ANNULAR DETECTOR

BACKGROUND OF THE INVENTION

FIELD OF THE INVENTION

This invention is directed toward the measure of the thickness of material using nuclear techniques. More particularly, the invention is directed toward apparatus and methods for measuring the thickness of pipe wall, from the outside of the pipe, using an annular detector, concentric conical detector collimation, and backscatter gamma radiation.

BACKGROUND OF THE ART

In December 1986, four workers at a nuclear power plant in the United States were killed, and another four injured, when an 18 inch secondary suction line carrying 175° C water burst. The cause of this tragedy was erosion- corrosion that reduced the pipe's wall thickness from 0.5 inch to only 0.06 inch. The cost of the accident besides the loss of life was estimated at $2 million for pipe replacement and related repairs, and 76 days of lost power generation while the plant was shutdown. The income loss for the utility company due to the shutdown was approximately $40 million. Subsequently, there has been an increased focus on non-destructive measurement of pipe walls in all power plants including fossil plants. In fact measurements of this type are performed at any facility where steam, or high temperature, turbulent flow streams are used, such as chemical processing, waste processing, and the like.

Currently, the state-of-the-art technique for pipe wall inspection uses ultrasound. The major drawback to this is the requirement for removal of pipe scuff guard and insulation. In all installations this adds an extra labor component to the process. In nuclear facilities, particularly boiling water reactors (BWR), significant radiation exposure to workers can result. In addition, the use of ultrasound is made more complex when fluids are flowing in a given pipe, or when the pipe is at high temperature. Therefore, performing ultrasonic measurements can reduce the availability of the system. Nuclear techniques have been used to measure the thickness of the pipe wall. One common type of measurement is to place a gamma ray source on one side of the pipe wall, and a gamma ray detector on the opposite side of the pipe wall, and to measure the attenuation of transmitted gamma radiation. The amount of attenuation of gamma radiation can be related to the thickness of the pipe. An obvious disadvantage of this technique is that either the gamma ray source, or the gamma ray detector, must be placed within the pipe. This can be extremely impractical, and sometimes virtually impossible, if the pipe contains flowing fluids, especially when at high temperature and high pressure. Other nuclear techniques have been used to measure the thickness of pipe from within the pipe. Such techniques have been used in the oil and gas industry to measure the thickness of steel casing using probes lowered within the borehole. These measurements are neither precise nor accurate, and are adversely affected by the material immediately outside the pipe (such as cement sheathing and/or earth formation), and by the material within the pipe such as gas, oil and water.

Nuclear backscatter techniques have been used in the prior art to measure the wall thickness of pipe from the outside of the pipe. The most common of these systems is based upon the measure of scattered radiation at approximately 90° with respect to the incident beam. For scattering angles ranging fro 90° to 180°, Houlong. Lee in "A High-Speed Wide Aperture Compton Scatter Imaging Technique - A Computational Study With Application to a Pipe Inspection System", The Pennsylvania State University. Ph. D. Thesis, 1991 , showed that scattering probability is near a minimum at 90° and a maximum at 180°. Using the so-called "wide aperture" method wherein the detector subtends a relatively large solid angle with the portion of pipe being measured, poor sensitivity to pipe wall thickness was observed when the pipe was filled with water or any other dense substance. These findings were disclosed by D. R. Wood, "Nondestructive Examination of Pipe Wall Erosion Using Compton Scatter Spectral Analysis" The Pennsylvania State University,

Master's thesis, 1990.

An object of the present invention is to provide apparatus and methods for measuring the thickness of pipe wall from the outside of the pipe.

An additional object of the invention is to provide apparatus and methods for measuring the thickness of a pipe wall while fluid is flowing within the pipe. Yet another object of the present invention is to provide apparatus which can be used to scan the wall thickness of the pipe, and to provide methodology for converting these scans into images of the thickness of the pipe wall thereby immediately exposing any dangerous wall thinning. Another object of the invention is to provide apparatus and methods for measuring wall thickness of pipe without having to remove any insulation or scuff guard on the outside of the metal pipe.

An additional object of the invention is to provide apparatus for which the sensitivity and spatial resolution can be varied depending upon operational conditions including whether the pipe is filled with liquid, or filled with gas, or filled with a combination of liquid and gas, or contains or does not contain outer insulation, or contains liquid of varying density.

Still another object of the present invention is to provide an apparatus which is relatively insensitive to positioning with respect to the wall of the pipe to be measured.

There are other objects of the invention which will become apparent in the following disclosure.

SUMMARY OF THE INVENTION This disclosure is directed toward apparatus and methods for measure of the thickness of pipe wall, from the outside of the pipe, using nuclear techniques. The disclosed measurement system employs a source of gamma radiation, a gamma ray detector, and Compton backscattering as the basic nuclear reaction in order to obtain a pipe thickness measurement. While Compton backscattering for thickness measurement is well known in the prior art, the present invention system is novel in the application of a unique detector and detector collimator to improve sensitivity in water filled pipes. More specifically, without the detector and detector collimation arrangement of the present invention, the measure of pipe thickness in an operating flow system is highly impractical. The detector collimation can be changed, and can be removed to yield accurate thickness measurements of gas filled pipes with improved statistical precision. By moving the apparatus around the periphery of the pipe and along the axis of the pipe, a complete image of the pipe thickness can be generated. Unlike prior art systems, the present invention utilizes Compton scattering at a scattering angle range near 180° for improved efficiency. It was also shown by Lee that the use of a dual detector arrangement provides excellent insensitivity to positioning errors. Such errors can dramatically affect results in a conventional Compton imaging system For even greater efficiency the invention employs a single annular scmtillator crystal. This also reduces system complexity by using a single photomultiplier to collect scintillation pulses. With only one detector element, calibration is simplified.

As stated previously in the above referenced systems of Lee, and in the early work of the inventors of the present measurement system, it was found that the wide aperture backscatter approach yields poor sensitivity to pipe wall thickness when the pipe is filled with water or any other dense substance. The use of a focusing detector collimator has been used in the prior art and allows the detector to sense scatters from only within a relatively small inspection volume.

The present invention uses an annular gamma ray detector coupled to a conical detector focusing collimator. The gamma ray detector is preferably a scintillation detector crystal made of thallium activated sodium iodide (Nal(TI)), bismuth germinate (BGO) or any other suitable material which can be formed into the annular geometry. The system has been designed for the use of a relatively low energy gamma ray source to generate the desired backscatter

203 radiation An isotopic source such as mercury-203 (Hg ), which emits a photon in the gamma ray energy of 279 keV, is preferred. The terms photon and gamma rays will be used interchangably in this disclosure, but it should be understood that the referenced photons are within the gamma ray energy range

203 of the spectrum. Gamma radiation emitted by the Hg source is collimated using a source collimator with an aperture to produce a "pencil" beam of 279 keV radiation which preferably impinges upon the outer pipe wall perpendicular to the major axis of the pipe and perpendicular to the wall of the pipe. The high efficiency, annular detector geometry allows the use of a relatively low activity

203

Hg source even with the use of the conical detector collimator. These two factors give rise to a very portable pipe thickness survey system weighing less than 20 pounds.

The axes of the annular gamma ray detector and the conical detector collimators (when used) are preferably coincident with the axis of a source collimator aperture, and are, therefore, perpendicular to the major axis of the pipe and to the surface of the pipe upon which the collimated beam impinges The system is, therefore, configured to detect backscatter radiation at essentially 180 degrees.

This detector and collimator arrangement is unique, effective, and straightforward, and can be used with disclosed measurement methods to image the wall thickness of pipes containing air, steam, any type of liquid and combinations of gas and liquid. Because of the nature of Compton imaging, physical contact with the pipe wall is unnecessary and the measurement can be made outside of the pipe while fluid is flowing within the pipe. With the proper choice of detector collimator geometry, the disclosed system can image pipe walls without removing insulation or scuff guards. The wide aperture design allows acceptable counting rates to be obtained with relatively low intensity sources thereby reducing the weight of required collimation and personnel shielding and thereby reducing the overall weight of the apparatus. With the use of available robotic motion control systems, the system can be used to make automated scans or images of the wall thickness of a pipe in a matter of minutes.

BRIEF DESCRIPTION OF THE DRAWINGS

So that the manner in which the above recited features, advantages and objects of the present invention are attained and can be understood in detail, more particular description of the invention, briefly summarized above, may be had by reference to embodiments thereof which are illustrated in the appended drawings.

Fig. 1 illustrates the basic Compton scatter reaction which is utilized in the invention;

Fig. 2 shows a prior art Compton scatter measurement system; Fig. 3 illustrates the normalized differential scattering coefficient for Compton scattered photons, for an incident photon energy of 279 keV;

Fig. 4 shows the relationship between gross count rate recorded by the system detector as a function of detector position, for three different pipe wall thicknesses;

Fig. 5 is a cross sectional view of the thickness measurement instrument; Fig. 6 is a perspective sectional view of the detector and detector collimator elements of the system; Fig. 7 is an illustration of a single backscattering by a slab of material; Fig. 8 illustrates the geometry used to calculate the response of the system;

Fig. 9 shows the geometric response of the system;

Fig. 10 is a graphical calculation of pipe thickness of bare pipe measured by the system, using a source which emits 279 keV gamma radiation;

Fig. 11 shows system response to insulated, empty pipe using a 279 keV gamma ray source;

Fig. 12 shows system response to water filled pipe using a source emitting 279 keV gamma radiation; Fig. 13 illustrates the system and pipe geometry used in Monte Carlo calculation of the system response;

Fig. 14 illustrates system response as a function of pipe thickness for pipes filled with varying densities of fluids, and using detector collimation and no detector collimation; Fig. 15 shows pipe thickness measurements made with the system compared with known pipe thicknesses, where the pipe is both insulated and bare;

Fig 16 a shows the response of the system to the thickness of a pipe which is partially filled with water; Fig. 16b is a cross sectional illustration of the pipe partially filled with water;

Fig. 17 illustrates the system response to the thickness of aluminum plate;

Fig. 18 illustrates a means for conveying the measurement system around the circumference of a pipe to be gauged;

Fig. 19 illustrates apparatus for measuring an image of pipe wall thickness; and

Fig. 20 illustrates a hypothetical pipe wall thickness image measured with the invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

Before presenting a detailed description of the apparatus and methods of the invention, a brief background of the principles of the Compton scattering reaction will be presented. BASIC PRINCIPLES

Referring to Fig. 1 , the essence of the pipe thickness measurements utilizing Compton backscattering is given by the equation:

where mo is the rest mass of an electron 10, E' is the energy of the scattered photon, E_γ is the incident photon energy, c is the speed of light and θ is the scatter angle 12. The remaining terms in equation (1) are also shown in Fig. 1. Using equation (1), if the incident energy E_γ is known, and θ , the detection angle 12 is fixed, returning photons at the predicted energy can be counted. If scatter angle θ is set and the ratio of scattered to incident photons is compared, the thickness of an object can be measured as will be illustrated subsequently. Usually, the scattered photons at energy E' are detected using one of two techniques. The first technique allows the detector to measure the gross gamma ray flux, and the flux at energy E' is spectroscopically separated from the gross flux. The second technique coliimates the detector such that it only "sees" a focal point of scatter activity at the angle θ , therefore (ideally) all gamma radiation detected is at a single energy, namely E' . The present invention need not always employ spectroscopy thereby simplifying the system, reducing cost and weight, and increasing reliability. Simple spectroscopy with relatively simple electronics is one useful option.

Equation (1) and Fig. 1 show the basis for all Compton scattering measurements. Generally, when using Compton scattering for thickness measurements, a gamma ray source 14 is used to form a collimated or "pencil" beam by means of a collimator 16 with an aperture 18. This beam is aimed at an object 20, and a detector 24 is placed to allow measurement of the photons that scatter through a photon scatter angle θ = 90° as shown in Fig. 2. A focusing collimator 26 is used to further define the volume 22 of the object 20 from which the measured scatter events occurs, defined as the "inspection" volume. It is not obvious that the use of a θ= 90° scattering angle is somewhat inefficient. Using the Klein-Nishina equation, we can predict the relative probability (differential scattering coefficient) of the photon scattering into a solid angle dΩ through an angle θ : σ, a² ( 1 - cos 0)²

(2) cos 0 + dΩ 2mlc^* | + a( 1 - cos 0) J 1 -r a( 1 - cos β) J

where ø and mo are the charge and mass of an electron, c is the speed of light_, and where hf is the incoming photon energy A plot of this equation normalized, and for a photon energy of 279 keV is shown as the curve 32 in Fig 3, where θ is plotted along the x axis and the normalized scatter probability is plotted along the y axis It is usually impractical to perform Compton imaging at anything less that 90°, because the object to be measured would be positioned between the source and the detector In pipe thickness measurements, it would be necessary to position either the source or the detector within the pipe For purposes of discussing the principles of this invention, the portion of the curve 32 to the left of a line 30 (i.e. θ approximately equal 90°), will be ignored In Fig 3, it can be clearly seen that the scattering probability is nearly equal at θ 90° and 180°

In the previously referenced work of Lee, it was shown that in reality the total scattering probability at a practical detector angle θ is near a minimum at 90°, and πses to a maximum at 180° This non-intuitive result is due to a vaπety of factors including the attenuation length of the scattered gamma rays, the energy dependence of the attenuation coefficients of material traversed by the gamma radiation, such as the attenuation coefficients of water and steel in the case of water filled steel pipe, the influence of photon interaction processes other than Compton scattering (Rayleigh scattering, photoelectric effect), and the effect of multiple scatters A Monte Carlo study performed by Lee illustrated this, as shown in Fig 4, which is a plot of gross counts recorded by the detector

(y axis) as a function of scatter angle θ (x axis) The curves 34, 36 and 38 represent pipe wall thicknesses of 1/8, 1/4 and 1/2 inches, respectively While increasing backscatter angle θ gives significant improvement in counts measured by the detector and thereby increases the statistical precision of the measurement, two major problems are still encountered, namely (1) when the pipe is filled with liquid, the conventional detector system is insensitive to pipe wall thickness, and (2) the system measurement is subject to positioning error The pipe wall thickness measurement system disclosed below overcomes these two problems and presents numerous other advantages over the prior art as outlined in the summary of the invention APPARATUS

A cross sectional schematic representation of the major elements of the measurement system is shown in Fig. 5. The backscatter instrument is identified as a whole by the numeral 100. A preferably Hg ²⁰³ source 14 is mounted within a source housing 40 which is affixed to a source plug 42 which can be removed from a shield 44. The shield 44 is preferably lead (the preferred heavy metal) and functions primarily to shield personnel from the source 14. A portion of the gamma radiation at 279 keV, emitted by the source 14, passes through a passage 48 in the shield 44 and through a source shutter 46, through an aperture 18 in the source collimator 16, which is positioned within the annulus of an annular detector 50, thereby forming a beam of incident gamma radiation indicated schematically by the arrow 15. The source shutter 46 is made of gamma ray absorbing material and can be used to block the passage 48 thereby terminating the beam 15. The pencil beam 15 emerges from the aperture 18 and impinges upon the pipe to be scanned. A pipe 70 is illustrated with a layer of insulation 62 and a scuff guard 60. The system is designed such that scatter events occurring within an inspection volume 22 will be detected by the annular detector 50. The detector is preferably a scintillation crystal such as Nal(TI) or bismuth germinate (BGO) and is hermetically enclosed by a housing 51 which usually incorporates a reflecting material. The annular detector preferably comprises a disk shaped crystal with a concentric, cylindrical annulus. It should be understood, however, that variations of this geometry can be employed, such as an annular detector which does not encompass the entire 360 degrees, but perhaps only 300 or 320 degrees. The system as shown employs a scattering angle range near

180° for improved efficiency. A single photomultiplier tube 52 or other light sensing device is optically coupled to the detector crystal 50. Signals from the photomultiplier tube 50 are input into preferably a computer (not shown in this Figure). Power is supplied to the photomultiplier tube 52 by a suitable power supply (not shown in this Figure).

As mentioned previously, so-called wide aperture systems exhibit good sensitivity to pipe wall thickness when the pipe is filled with gas, but poor sensitivity to pipe wall thickness when the pipe is filled with water or any other dense substance. Still referring to Fig 5, the use of a focusing collimator allows the detector to sense scatters from only within a relatively small inspection volume 22 and thereby greatly increases the response of the system to pipe thickness when the pipe is filled with water. The focusing collimator comprises a plurality of conical annular cones 26 which are aligned around the annular crystal 50. The apex of each cone is truncated and faces the pipe to be inspected. The base of each cone is positioned against the face of the crystal 50. The function of the conical focusing collimator is unique, effective, and straightforward, as will be described in more detail in following sections. The cooperating annular detector and conical cone detector collimator permits pipes containing air, steam, liquid, or a combination of gas or liquid to be effectively imaged for thickness. The flexibility, accuracy and precision of the measurement system significantly advances the art of pipe thickness imaging. Furthermore, the system has been designed for the use of a relatively low

203 gamma ray energy source 14, preferably Hg with a gamma ray energy of 279 keV. This low energy combined with a high efficiency detector 50 with annular geometry, allows the use of a relatively low activity source (as low as 100 milliCuries (mCi)), even with use of the detector collimator 26. These factors require minimum shielding 44 which, in turn, give rise to a very portable system weighing less than 20 pounds.

Fig. 6 is a perspective, cutaway view of the detector collimatmg cones 26, the detector 50, the source collimator 16 and the source aperture 18. and perhaps better illustrates the arrangement of these elements with respect to the photomultiplier tube 52 and the pipe 70.

SYSTEM RESPONSE Basic Reactions The system is designed such that the gross count rate, which is greater than the count rate in a specified energy region corresponding to 180° backscattering, can be employed. A gross counting rate method has several advantages when compared to the spectral measurement technique. These advantages include reduced requirements for system components and simpler data analysis. All subsequent discussion of the system response assumes that the data comprise gross count rate data from the detector 50.

It is helpful to take a look at a few simple scatter cases from the view of photon transport through material in order to fully understand the response of the measurement system. Some basic features in scatter measurement can be seen clearly. This part of the discussion presents the conceptual background of the system. Issues like detector symmetry, photon energy, backscatter saturation, and three basic effects in the scatter process are presented in an intuitive manner, thereby avoiding a complicated mathematical development in this section.

An ideal case is considered where an object is in the path of a pencil gamma ray beam. This object is so small in size that its geometry can be ignored. The photons scattered by the object have a certain distribution. This distribution can be described using equation (3).

(3)

where dΩ ' is an element of solid angle, and E and E' are the incident photon energy and the scattered photon energy, respectively, Ω' • Ω is the cosine of the scattering angle θ , and r₀=e /m₀c²=2.81794 x10 cm is called the classical electron radius. For simplicity, the influence of electron binding energy on the scattering process is ignored in this study.

The directional preference of a scattered photon suggests that, for high energy incident photons, more Compton scattered photons appear in the forward direction, at a scattering angle 0of less than 90 degrees, than at greater degrees of scattering. Solely from the point view of a scattering distribution, it is more efficient to set a detector to measure scatter radiation at relatively small or "forward scatter" scatter angles θ. However, the coherent scattering and primary beam are also forward oriented thereby creating additional gamma rays which would be detected by the detector, and which would be considered as "noise". Stated another way, the forward scatter gamma radiation would have to be separated from these other types of radiation in order to obtain the desired results. Therefore, a measurement can be easily complicated if a detector is set up to measure the forward scattering. In addition, in many cases one-side access to a test object is preferred and even necessary as discussed previously. Therefore forward direction scatter measurements are not used in the present invention.

The symmetric distribution of scattered photons about an incident beam direction is indicated in equation (3). In the case of pipe wall thinning, the test objects and pipes also have a certain geometric symmetry. These features suggest that a detector of a symmetric shape about the incidental beam can achieve better efficiency, stability, flexibility and simplicity. Therefore, an annular shape detector becomes a natural selection for the system in order to reduce the sensitivity of the measurement on the relative position between the 5 measuring device and the pipe. This kind of detector is one of the most unique parts in the present Compton imaging system.

Besides the directional preference of scattered photons, the attenuation and geometry of a system have a great impact on detection efficiency. Reference is again directed toward Fig. 5. Generally speaking, the closer a I 0 detector 50 is to a scattering site 22, the more efficiency and more scattering information the detector can obtain. This is simply because the detector 50 covers or subtends a greater range of scattering angle, and thereby interacts with more scattered gamma rays. As a result, one should install the detector as close as possible to the pipe wall being inspected in order to achieve the

1 5 highest detection efficiency. This spacing does, however, affect other properties of the measurement, and sometimes requires "trade-offs" as will be discussed in subsequent sections.

Unlike the most conventional prior art Compton scatter system, which employ a collimator in front of the detector to define a certain sensitive region 0 and to eliminate scattering from other positions, the wide-aperture detector embodiment of the present invention accepts all the radiation which reaches the detector front surface. The effects of the concentric, conic collimator used in another embodiment of the invention will be discussed in great detail in subsequent sections. For the moment, however, the present invention, 5 embodied with no detector collimation, will be used to better understand the geometric factors associated with the present measurement. The removal of the prior art type collimator from the apparatus of the present invention has a great impact on the measuring process. A simple example can be used to demonstrate some characteristics of scatter measurements. Attention is

3 0 directed to Fig. 7. Suppose that a wide-aperture detector element 80 at a scattering angle of 0 = 180° is used to observe the backscattering from a uniform slab 82 with thickness s denoted by the dimension 84, as shown in Fig. 7. The slab 82 is illuminated by a pencil beam 15 perpendicular to the slab surface. Also assume that a solid angle ΔΩ, which the detector element 3 5 subtends from a point along the radiation beam in the slab, changes very little. The detected singly scattered photons can be expressed as equation (4) W -77 = -7 -; TrT^^{1 ~} exp[-(A/(£ ₀ ⁾ + /i(£ ^{' @}0 = 180")⁾*])

/ _QΔΩ μ(E_Q) + μ(E ) ^υ

where F is the scatter cross section, and the remaining terms have been defined previously. It is obvious that the detected intensity / is not directly related to the slab density when [μ(E₀)+μ{E')]s is large enough, since the exponent containing this term is much, much less than 1.0. Stated another way l/lrjΔΩ = F/l [μ(E₀)+/i(E')]. Instead, the ratio of the Compton scattering kernel to the sum of attenuation coefficients turns out to be an important parameter of the process of the invention. In addition, a "saturation" of detected counts is expected when the slab thickness s is large enough. This saturation is determined predominantly by one property of the test material, namely, the ratio of Compton scattering to the total cross section of the tested material for a given energy. This feature can be conveniently used to identify an unknown material as, for example, in the fields of oil or coal well logging, since the scatter measurement is made within a pipe surrounded by earth formation, and the physical extent of the earth formation is "infinite" with respect to the depth of investigation of the measurement. Note that the selection of photon energy becomes very important since the cross section ratio is energy dependent. In a conventional prior art Compton system, the electron density in the region defined by the system is proportional to the number of observed singly scattered photons, and the Compton scattering cross section is usually the dominant factor affecting recorded counts. However, with a wide-aperture detector, the Compton-to-attenuation ratio turns out to be a more important factor than the Compton scattering cross section alone. In addition, multiple scattering will have a more dramatic influence on the measurement process. Because of these factors, the response of the wide-aperture detector used in the present invention is quite different from that of conventional prior art systems. As a direct result of equation (4), one can calculate the saturation thickness s of a steel slab for different incident photon energies. Here the saturation thickness is assumed to be the thickness at which the detected intensity reaches its 90 percent maximum intensity. As the incident photon energy increases, the saturation thickness increases. It can be shown, however, that the maximum backscattered photon energy is about 250 keV. The saturation thickness will also reach a limit when the incident photon energy is high enough, because the backscattered photons are absorbed within the material and therefore can not reach the detector. For a very low photon energy, this effective range is too small to cover certain desired pipe thicknesses as noted in this disclosure. One can also calculate the amplitude of the relative saturation intensity

(; — ) as a function of energy for a steel slab by using equation (4), which is l ₀ ΔΩ another factor influencing the detection efficiency of the system. Low energy scattered photons have smaller probabilities of being detected by the detector 80 shown in Fig. 7, because of their smaller Compton-to-attenuation ratio. The utilization efficiency of the low energy photons is therefore very low. This is one of the reasons why the lower energy component in previously referenced system of Lee is difficult to use.

In the case of singly scattered photons, the observed intensity of singly scattered photons as shown in equation (4), is the accumulated result of scattering and attenuation. These two effects involve the differential scattering cross section and the exponential attenuation factor through which the photon is transported. The scattering term F, of course, presents the scattered photons. The attenuation term for the incident beam of energy Eo has the effect of reducing the incident beam intensity along its path, while the one for scattered photons is seen to reduce the possibility of the photons being observed by the detector 80. Another factor governing the detector response is the system geometry. The system's geometric influence on a measurement is implicitly expressed in the terms of integral paths and the solid angles that a detector subtends. This influence will be identified as the "geometric response" or "geometric effect".

Each of the previously discussed terms describing photon interaction with traversed material, which is a uniform slab 82 in Fig. 7, behaves in a different fashion. These processes obviously complicate the measurement of the pipe thickness measurement system. On the other hand, they may also provide some possible new avenues for Compton scatter imaging. New techniques can be found by exploiting these terms to enhance various features in the measuring processes. It has been found that this enhancement can be realized by suitable detector design under certain circumstances.

A first technique for determining material thickness using the previously described apparatus, which is suggested in a simple slab geometry case shown in Fig. 7 and as discussed above, is to fully accumulate scattering information. The data measured is related to the material thickness and the ratio of Compton-to-attenuation coefficients. A second technique for determining material thickness using the previously described apparatus is to use the attenuation effect in the scattering process. Since scattered photons traverse various paths to the detector element 80, and these paths are relatively independent of incident beam path 15, although their origins are the primary beam path 15. These scattered photons experience attenuation as they traverse material on their way to the detector element 80. This process can be seen to be comparable to conventional radiography in certain aspects, wherein photon attenuation is directly related to the density of the material being traversed by the photon. The scattering medium then can be regarded as a radiation source which is activated by the incident beam 15. A detector, represented by the detector element 80 in Fig. 7, can then be used to observe the attenuated photons arising from this activation, which can be regarded as emerging from the medium 82. By emphasizing one factor or the other, both techniques can work well for a system in specific situations for measuring the thickness of material such as the thickness of a pipe wall.

Source Selection

The energy of the incident radiation beam is very important for Compton scatter pipe thickness measurement system presented in this disclosure. For such a system, choosing proper photon energy, which falls within the gamma ray energy portion of the photon energy spectrum, involves many system considerations, such as the source preparation and manufacturing, the behavior of the selected energy photons in the system, and system weight. There are several constraints on the selection of radiation sources, such as the limited available suitable radioactive isotopes, the low weight shield required for system portability, and the higher energy photons desired for a better Compton-to-attenuation ratio as discussed above. There always exist pros and cons for a selected source. Selecting the best source for the disclosed system required much study, with the selection criteria being outlined below.

Among the candidate isotopes that nature offers, one suggested source is mercury-203 (Hg²⁰³). The isotope Hg²⁰³ emits single-energy gamma rays (E=279 keV, yield=77.3%) with a half-life 47 days. The advantages of using

Hg source are that it yields a good gamma ray energy for Compton imaging and for reduction of system shielding weight, has suitable half-life for generating high intensity gamma rays, and emits a simple energy emission spectrum with the predominant gamma ray emission at 279 keV. Calculations of relative saturation intensity as a function of source energy also suggests that higher photon utilization efficiency can be reached using 279 keV gamma ray emission from Hg²⁰³. There are additional benefits produced by Hg²⁰³ that will be discussed in subsequent sections of this disclosure . The disadvantages of

203 using Hg as a source are that it is unavailable commercially and it requires a long irradiation time or a high neutron flux to make the source due to the low abundance of Hg²⁰² (29.9%) and small neutron absorption cross section (4 barns)

Response Model

The following model was used to calculate the response of the pipe thickness measuring system. The model was used to optimize design parameters of the system, and was further used to calibrate the operating thickness imaging system as will subsequently be described in detail

The geometry of the model is shown in Figure 8. An annular detector 50, with inner radius r, as identified by the dimension 82, and outer radius r₂, as identified by the dimension 81, is used to detect scattered photons, which are preferably in the gamma ray energy region. A collimated incident gamma ray flux with 279 keV energy from a Hg²⁰³ source 14 is defined by a source collimator 16 and passes through the aperture 18 (see Fig. 5) thereby forming a beam 15 which is treated as a pencil beam along the detector center line. The beam 15 penetrates the regions associated with the inspected pipe. These regions can be any combination of pipe 70, water 80 filling the interior of the pipe, heat insulation 62, and scuff guard 60 which is typically aluminum, ail being aligned as shown in Fig. 8. These regions are simplified as slabs for modeling purposes, but yield results which closely match the response of the measuring device shown in Figs 5 and 6. The surface of the detector 50 is parallel to the surfaces of the slabs 60, 62, 70 and 80. The above simplification has the effect of producing a large reduction of computation time because of the symmetry of the model. The distance from the detector to the front surface of the objects is Tair, over which only air exists. The slab thicknesses of heat insulation, pipe wall and water are noted as Tjns. Tpjpe, and Tw. respectively. Typical composition parameters of the regions are listed in Table A, and are used in computing attenuation coefficients and other reaction parameters required for the model predictions.

Table A

The singly scattered photon intensity can be computed. There are integral equations which link the calculated quantities with those to be measured. In a measurement using the apparatus of Figs 5 and 6, it is difficult to determine the order of scattered photons (i.e. single scattered or multiple scattered). However, theoretical calculations can provide a more insightful view of the measurement process. In this study, the contribution by higher order scatterings (beyond order 2) will not be considered, because of the complexity of computation and the relatively minor influence of such scattering on the measured results. The actual calculations are numerical in nature, using the Romberg integration method, but will not be presented for reasons of brevity.

Computed results, using the analytical-numerical method, are presented in the categories of geometric responses, empty pipe and water filled pipe; uninsulated and insulated. The numerical approach was applied to the cases of empty pipes and water filled pipes, respectively. The relative intensity of output to input gamma ray intensity, lo/lj, where I₀=lι+l₂. is used in the following discussion without loss of any physical meaning. The term "relative counts" used in the following disclosure is defined as lo/lj. The intensity and l₂ are the first and second order scatter intensities, respectively. The incident photon energy of 279 keV from Hg²⁰³ is used in the calculations.

Geometric Response of the System

The first question to ask, when testing a pipe, is where one should place the detector relative to a test object. Lee recommended to position the detector(s) at a high scattering angle in order to obtain higher efficiency. This is just a part of the answer. The complete geometric response is associated with all geometric factors in the system. To understand this effect, it is helpful to consider a simple case, as shown in Fig. 9. An annular detector 50 has again an inner radius r_t and outer radius r₂ identified by the numerals 82 and 81 , respectively. The distance from a point P, which can be thought of as the previously defined inspection volume 22 of the measurement, on the central or "beam" line 15 to the detector surface is defined as T. The solid angle that the detector front surface subtends is expressed as

(5) ΔΩ =

This equation can be used to estimate the distribution of geometric response functions for various distance T, inner detector radius r, and outer detector radius r₂.

Because of the anisotropic distribution of Compton scattering and the dependency of scattered photon energy on scattering angle, the actual geometric response is somewhat more complicated than the above discussion. The analytical model discussed above was applied to a 0.9525 cm (3/8 inch) thick bare steel slab which is positioned in front of the detector 50. The distance between the detector front surface to the slab, Tair. was varied. Also the outer radius of the detector, r₂, was assigned different values. A calculation of relative counting rate versus Tair. at ^=0.75 cm, showed that the detection system has a geometric efficiency weight distribution which varies for different positions 22 where scattering occurs. This positional preference or geometric response can be appropriately arranged to enhance any portion of the signal of interest. The detector dimensions and Tair are seen to be the basic parameters for the system arrangement and optimization, and these components were optimized using the model calculations. Results of the design parameters of the system will be summarized below. For a bare pipe with T_air set at around the maximum of the geometric response, the measurement will have better stability and better efficiency, and the scattering from the pipe wall can also be enhanced. This enhancement can also reduce the necessary gamma ray source strength thereby reducing shielding requirements. The maximum geometric response for bare pipe occurs at a value of T_aj_r about 1 cm, a practical distance for a detector setup for measuring pipes. Keeping Tair unchanged at 1 cm and r_t at 0.75 cm, the detector responses in terms of relative counting rate to the pipe wall thickness were computed and are shown in Fig. 10 for different detector outer radii ^rz of 1.385 cm, 2.655 cm, and 4.56 cm represented by curves 110, 112, and 114, respectively. It is clear that the responses demonstrate that the measurement system has good sensitivity to the pipe wall thickness ranging up to about 1.4 cm, with the best sensitivity being obtained with the largest outer detector radius. Increasing the detector annular area or "window size" influences the measurement of the system in several ways in that: (1) increases geometric efficiency; (2) spreads the scattered radiation spectrum; (3) increases the detection of multiply scattered radiation; and (4) changes the detectable radiation field. A point should be made that the geometric response function is changed when a larger detector size is applied. More exactly, the deeper portions of the pipe wall gain more positional weight than the outer surface portions. That is why better sensitivity and resolution appear for the larger detector setup. In other words, the geometric response function is used to improve system resolution to the pipe thickness as well as detection efficiency. For a pipe covered with heat insulation, the optimal position for the insulated pipe appears to be as close as possible to the detector, and a large detector window size has been found to improve both geometric efficiency and resolution of the measurement. For a suitable detector setup, say Tair at 1 cm and r, at 0.75 cm, the detector responses to the pipe wall thickness, with 2.54 cm and 5.08 cm thick insulation, were computed and are shown as curves 122 and 120, respectively, in Fig. 11. Although the sensitivity for insulated pipes is not as good as that for bare pipes, it is still reasonably good for measuring pipe wall thickness. The heat insulation used in power plants usually has low density and limited thickness. While it generates scattered photons, and attenuates the primary beam and scattered photons, the most severe influence on the measurement is that it keeps the inspected pipe away from the detector a minimum distance equal to the insulation thickness. In other words, the geometric response plays its role through the detector window size and the distance of the inspected pipe to the detector. The sensitivity and resolution to the pipe thickness are obviously improved by increasing the detector window size. Figs. 10 and 1 1 also indicate that this system has better sensitivity variation for thinner pipes than for thicker pipes. Since most industrial applications are more concerned with the residual pipe thickness of thinner pipes, this feature will provide better accuracy for thinner pipe measurements.

The filling water in an uninsulated or insulated pipe has a severe impact on the scattering process. Water has a high Compton-to-photoelectric ratio. This is a good effect since fewer photons are lost by photoelectric absoφtion. The region over which the filling water exists is generally large. All of these features result in less resolution to the pipe thickness. However, these features of the filling water can be exploited if one adopts a slight modification of the previously discussed technique. As already noted, the geometric factor has a strong effect in the measurement system. The photons scattered by the water can be made to gain more importance by judiciously reshaping the geometric response. Fig. 12 shows the system response to the pipe thickness for water filled pipes, with the h = 0.75 cm, X2 - 4.56 cm. Curve 132 represents the system response as a function of pipe thickness, when the pipe is bare. Curve 130 shows the response of the system as a function of pipe thickness, where the pipe is wrapped with a 1 inch thick layer of insulation. Good resolution to the pipe thickness is obvious in the curves, but these graphs are very different from those for empty pipes as shown in Figs. 10 and 11. The mechanism behind the above phenomena is not as obvious as that for empty pipes. To explain this behavior it should be noted that as the radiation beam illuminates the target, the Compton scattered photons can be viewed as a distributed radiation source along the path of the beam. Although this source is quite different from a conventional radiography source in the aspects of spatial distribution, directional anisotropy and energy spectrum, it is similar in the sense of where the radiation originates. The view of the detector clearly has different weights for different scattered source positions along the beam due to geometric response functions. For the current setup, the pipe wall is positioned in a lower geometric weight position in direct view. However, the weight for the filling water is increased as if there was a stronger radiation source behind the pipe wall, in the water. Moreover, the high Compton-photoelectric ratio and small attenuation coefficient of water enhance its role as a radiation source. The problem then becomes similar to conventional radiography, meaning we are left with an attenuation measurement along the scatter path to the detector. It can now be seen that the Compton scattering method and the attenuation of conventional radiography join together to produce a new technique, which can be applied to many situations in industrial NDT in addition to pipe wall thinning. For the Hg²⁰³ source used as a source of radiation, the energy of the backscattered photons from water is about 130 keV, which is quite adequate for resolving steel pipe wall thicknesses out to about 1 cm. An additional safety benefit arising from this technique is that it avoids the danger of higher energy radiation and its potential radiation exposure to workers as in conventional radiography.

The computed results discussed above suggest that one can use two different approaches to resolve the pipe wall sensitivity problem by fully exploiting the inherent geometric responses of this system. Geometric responses inherently exist in many measurement systems. Similar to the output of a system being affected by its transfer function, the measurement is modulated by its geometric response function. It can be said that beyond specific system setups such as embodiments utilizing concentric, conic detector collimation, geometric response functions define various system behaviors.

For the conventional prior art Compton scatter system, of the type shown in Fig. 2, its geometric response is defined by the two collimators 16 and 26 in front of the source 14 and the detector, 24 respectively. Its geometric function is quite like a delta function which samples a certain point. For the wide aperture embodiment of the present system, the response function can be shaped by varying the aperture 18 size of the source collimator 16 (see Fig. 5) so that scattering information enhanced by the response function can be obtained.

Detector Collimation A desirable geometric function for a system can be further shaped in addition to varying the source collimator aperture. For the large area annular detector of the present invention, it is possible to implement significantly different geometric response functions by implementing the detector collimation in front of the detector surface. This detector collimator embodiment provides increased potential for applications of this system design to various inspection or imaging situations. The present system is designed such that no detector collimation can be used, or alternately different types of detector collimation can be used. As an example, system performance with empty pipes is quite good without any type of detector collimation, and the absence of any collimation increases the measured count rate and therefore improves the statistical significance of the measurement. On the other hand, water within the pipe being measured introduces serious problems in uncollimated, wide aperture measurements, and the introduction of detector collimation greatly improves the measurement. It is possible to extend this idea to generate a variety of geometric response functions using adaptive detector collimators. These collimator " masks" can shape various geometric response functions, just like a camera, which can zoom in and out over a relatively large range of distances. This type of adaptive collimation mask can produce more diverse thickness measurement images. Designing or reshaping that type of geometric functions requires a feasible and flexible detector and collimator system. For purposes of further illustration, concentric, conic detector collimators will be discussed. It should be understood, however, that different configurations of detector collimators can be used to obtain different thickness measurement results.

Detector collimation design was optimized using the MCNP (Monte Carlo Neutron Photon) computer code, which is a widely accepted code utilizing the

Monte Carlo method. This code was primarily written by Dr. Thomas Godfrey between 1975 and 1989. MCNP Version 4A was used for runs presented in this disclosure. This code deals with defining planes in 3 dimensions and filling volumes in space with desired material. In this manner one is able to duplicate the geometric and physical setup of an actual system to a system in the computer. Once the geometry and material has been specified, particles, either neutrons, photons or both can be created with varying energies and directions. By specifying a region of tally, one is able to simulate an actual detector response. The actual nature of interaction between particles and matter is determined from information provided by additional files. For example, the file xsdlr provides material cross-sections for neutrons and gammas. Additional files provide information in order to select interaction types as well as resulting energy and direction of the ensuing particle or particles resulting from the interaction. As MCNP is a Monte Carlo code, all such selections are done on a random number or probabilistic basis. Fig. 13 is a illustrates a typical MCNP geometric layout used to optimize detector collimator design. The source 14 is Hg²⁰³ and the source collimation 44 is lead. A pencil beam 15 is formed and impinges upon a pipe 70 filled with water 80 and wrapped with insulating material 62. The plurality of shielding cones are conic, as illustrated previously, each with a truncated apex directed toward the pipe, and a base coaxially aligned with the center axis of the angular detector 50. The length of each cone is identified by the dimension 190, the spacings between the cones at the detector face is identified by the dimension 192, and the distance between the truncated apex of the cone and the surface of the insulation around the pipe is defined by the dimension 192. Cone lengths are preferably, but not necessarily equal. Cone spacings are preferably, but not necessarily, equal.

In order to accurately optimize the detector collimator design and the positioning of the measuring device with respect to the pipe, it was necessary to predict how certain parameters of the system design affect data quality. An extremely important aspect of the system design is the actual dimensions of the detector focusing collimator. In an early test to determine the collimator's effectiveness, various collimators were built with varying numbers of focusing cones, cone spacing and collimator length. It was found that certain collimators produced more superior data than others. Each collimator did, however, produce superior results to results obtained without using a collimator. Rather than build additional collimators to fully investigate the effect of collimator design on data quality, it was considerably more efficient to simulate the system in a computer code and adjust parameters in the model. The MCNP code proved to be ideal for such an investigation.

Various assumptions were made in this computer model in order to save computer time and run as simple a model as possible. No assumptions were made which would significantly alter the quality of data, nor were any made which would qualitatively alter results. Among the assumptions made were the production of photons which are the principal particles to be followed in the setup. Again referring to Fig. 13, the photons were assumed to be unidirectional and mono-energetic. Photons produced by Hg²⁰³ source 14 are predominantly 279 keV, so the mono-energetic assumption is a reasonably accurate one. As will be later seen, this assumption does not noticeably affect the results of the MCNP runs. While the Hg²⁰³ 14 source is in the form of an oxide pellet of length and diameter roughly equal to 0.5 cm, it can be assumed to behave like a point source. The shield design 44 collimates the gamma rays through an aperture 18 which is 3/16 in. in diameter. This aperture is 3 inches long and is seen to extract roughly a pencil beam 15 from the isotropic gamma emissions source 14. This pencil beam 15 is assumed to be 3/16 in wide in the MCNP simulation with uni-directional gammas. Both these approximations are shown to produce accurate results when compared to the actual experiments run in the laboratory setup.

There are numerous design parameters which were investigated with an MCNP model. These parameters include collimator dimensions such as cone length, cone spacing, the number of cones, and the like. In addition to these collimator parameters, there are other design parameters which may be investigated. These parameters include the annular detector 50 dimensions such as diameter, width, inner hole diameter, and the like. Additional investigations may also be performed on varying the distance between the collimator and the pipe wall. For brevity, however, the results of the computations used to optimize the detector collimator design will be given. In order to scan pipes with thermal insulation, the collimator length would typically need to be greater than the insulation thickness. Referring again to Fig. 13, the collimator can then be placed at a distance 192 from the insulation equal to the difference between the collimator focal region and the insulation thickness, where again the focal region is defined as the position of primary response that the system "sees". This would then place the focal point of the collimator at the wall of the pipe 70. The theoretical location of the collimator is defined as the point where the concentric cones 26 are designed to focus. However, in practice the focal length is located between the theoretical focal length of the collimator and the focal length (the region of maximum recorded counts) of the bare crystal without a collimator. The region of maximum counts for this detector is seen to occur roughly 1/2 inch from the detector. With the addition of a collimator to this detector, the focal length will shift to some value between 1/2 inch and the focal length of the collimator. With a cone spacing 194 of 0 (i.e. an infinite number of cones), the practical focal length will equal the theoretical focal length. With an infinite cone spacing, that is essentially 0 cones, the focal length will equal that without a collimator, in this case roughly 1/2 inch. The smaller the cone spacing 194, the closer the experimental focal length will be to the theoretical value. Similarly a large cone spacing would lessen the effect of a collimator and the focal length would be closer to that of a detector with no collimator. The collimator length 190 is also expected to change the focal length. As the collimator length is reduced, the effect of the collimator is minimized. The focal length is therefore expected to be closer to a detector 50 without a collimator as a result of the reduced collimating effect. Conversely, it is also logical that the longer the collimator, the greater the collimating effect, and the closer the practical focal length will be to the theoretical focal length. By adjusting the collimator length 190 and performing runs in MCNP for different lengths, the variation in focal length as a result of changing collimator length was investigated. The cone spacing as well as the number of cones was kept constant. Similar to varying cone spacing, a decrease in collimator length is seen to increase the number of recorded counts as well as a shift in focal length. A longer collimator length results in a greater focal length while a shorter collimator gives a shorter focal length. It is clear that as the collimator length is decreased, backscatter from closer locations are able to reach the detector. This tendency therefore shifts the focal length closer to the detector.

The addition of a focusing collimator to the front of the detector 50 has proved to be invaluable when scanning water filled pipes. While the effect of multiple scattering is greatly diminished, the number of total counts recorded by the detector is also seen to be significantly reduced. This reduction in the number of recorded counts can be attributed to two major components. The collimator cones 26 naturally block a fraction of counts which are meaningful to the count spectrum (backscatter from the pipe wall) as well as counts which are insignificant, like backscatter from insulation and water. In addition to blocking out meaningful counts from the spectrum, the addition of the collimator cones 26 to the detector 50 naturally means the detector will be further away from the pipe wall 70. This greater distance 192 plus 190 results in a smaller solid angle of backscatter which can interact with the detector, hence a lower number of counts. Using the MCNP model, the pipe wall thicknesses was varied from 1/16 to 1/2 inches. In each case it was seen that there is a significant drop in recorded counts recorded by the detector 50 as the pipe wall 70 was moved away from the detector. It was also seen that the maximum number of counts were obtained with the detector roughly 1/2 inch from the pipe wall. In order to minimize this inherent loss of recorded counts, one may use a collimator of reduced length 190 so as to lessen the distance between the detector and the pipe wall. One may also lessen the spacing 194 between the collimator cones, hence causing less interference for pipe wall backscatter. Increasing the size of the detector 50 to provide a greater solid angle to record backscatter photons should also increase system counting efficiency. However, it was found that there is a trade off between the number of recorded counts and system resolution. Lowering the collimator length 190 and increasing the cone spacing 194 increases the number of counts recorded by the detector, but also brings the system resolution closer to that of a system without a focusing collimator. Increasing the detector size does not reduce the resolution but makes the entire system more cumbersome and expensive to produce.

In addition to the trade off between resolution and recorded counts, it is also desirable to make the collimator as insensitive to distance 192 as possible. This effect would provide a greater margin for error when scanning pipes as the distance 192 + 190 between the pipe wall and the detector can not be kept perfectly constant. The degree of this distance variation will depend on the physical uniformity of the thermal insulation 62 as well as the scanning system connected to the detector. Using the MCNP model, it was found that the collimator with cone spacing 194 of 3/16 inch shows the greatest sensitivity to distance and will likely generate significant errors should it be used to conduct a pipe scan. The collimator with cone spacing 194 of 3/4 inch displays the greatest insensitivity to position errors but also gives the worst resolution. It is apparent that an optimized collimator needs to be designed in order to maintain a desired resolution as well as obtain the maximum counting efficiency. Maximum efficiency is essential for counting statistics. It will also allow us to use the radioactive source for greater lengths of time. It is also desirable to use a collimator that does not show great variation to distance changes from the pipe wall as distance sensitivity will likely increase expected error. All these effects are investigated in the MCNP code by changing design parameters and comparing the resulting data.

Still referring to Fig. 13, the length 190 of the collimator is an extremely important design parameter for two reasons. It plays an important role in the trade off between system sensitivity and the number of recorded counts.

Secondly, as the collimator is placed in front of the detector 50, the collimator length 190 naturally determines the closest possible distance between detector and pipe wall. This effect is extremely important to maximize the total number of detected counts. As the detector is placed further away from the pipe wall, solid angle of backscatter counts becomes increasingly small. This reduced solid angle translates to a lower number of detected counts. This parameter was investigated in the MCNP code with collimator lengths between 0.5 and 1 inches. While the collimator length was varied, the remaining collimator design parameters were kept constant. The cone spacing 194, for example was kept at 3/8 in. and the outer radius of the detector 50 was kept at 1.612 inches. In order to fully investigate the effect of changing collimator length 190, two types of runs were performed. The first run deals with the resolution of the backscattered photons to pipe wall thickness. This was accomplished by simulating an insulated, water filled pipe and varying the pipe wall thickness. This simulation showed how backscatter counts change for different collimators as a result of pipe wall change. The second run performed dealt with taking counts as collimator distance was varied. This simulation was performed by placing a 3/16 inch metal sheet at different collimator distances and measuring the amount of detected backscatter. This experiment was performed with different collimator lengths, hence providing information on how detected counts varied as collimator length 190 was changed. It also showed the sensitivity of counts to changes in distance from the collimator. As previously mentioned, sensitivity to changes in pipe to detector distance will likely result in greater uncertainty in the measured pipe thickness and should therefore be kept to a minimum. As a result of these calculations, it was found that the longer the collimator length 190, the sharper the change of counts as distance from the collimator changes. The total number of counts was also found to increase as the collimator length 190 decreased. This increased number of counts is expected as a result of the geometric effect of a shorter collimator, i.e. the solid angle of detected counts increases resulting in more photons interacting with the detector. It would, therefore, seem reasonable to keep the collimator length a short as possible as this would be favorable for both the higher counts as well as insensitivity to collimator distance. The calculations also show, however, that these effects lead to reduced resolution to changes in pipe wall thickness. With the shortest collimator length 190 of 0.5 inches, it was found that the change in counts is considerably less than that for a collimator 0.625 inches in length. It is evident that even small changes in collimator length 190 results in diminished resolution to pipe wall thickness. Based upon these calculations and the above discussion, it has been concluded that an acceptable trade off with resolution versus, counts and insensitivity to distance would therefore be expected between a collimator length 190 between 0.85 and 1.0 inches. Attention is again directed to Fig. 13. The effect of cone spacing 194 essentially provides an additional parameter where a trade off exists between the number of counts and the system resolution. In this respect the change in cone spacing 194 gives similar results to changes in collimator length. However there are certain differences between these two design parameters. For example, the distance between the collimator and detector is not dependent on cone spacing. The number of counts is therefore not expected to vary as dramatically as with collimator length 190, as there is no change in solid angle with varying cone spacing. With focusing cones 26 placed closer together, the expected focal region becomes smaller, hence causing greater sensitivity to distance changes. However, this smaller focal region would also more effectively eliminate multiple scatters from the water, thereby increasing the system resolution. A smaller cone spacing 194 would also be expected to eliminate a greater number of scatters from the water as well as scatters from the pipe 70. A lower count rate is therefore expected. In order to investigate the effect of cone spacing, all other design parameters were kept constant. These parameters included the collimator length 190 which was kept at 1.0 inch. The number of cones was kept as the maximum number allowed by detector crystal 50 with X2 = 1.612 inches. With cone spacing 194 of 3/8 inches, for example, this translates to a maximum of four cones 26. Calculations were conducted for cone spacings 194 between 3/16 and 3/4 inches. It was found that the greater cone spacing gives both a larger number of counts and lower sensitivity to changes in collimator distance. However, the calculations also indicated a large loss of sensitivity to pipe wail thickness 70 as the cone spacing 194 is increased. Once again, this is to be expected as the greater cone spacing allows the use of fewer cones 26 for a given detector radius. As a result, fewer cones obstruct fewer multiple scatters and the system behaves more like the embodiment without the addition of a detector collimator, hence the decreased resolution. In order to attain reasonable efficiency and resolution, cone spacing of roughly 3/8 inch gives acceptable results.

The MCNP code using the geometry shown in Fig. 13 was also used to determine the optimum number of detector collimator cones 26. It may at first be thought that the number of cones would be determined by the cone spacing 194 as greater cone spacing allows fewer cones 26. This dependency is, however, only valid for fixed detector size. By maintaining the same cone spacing 194 and collimator length 190, adding additional cones 26 would naturally require a larger detector radius X₂- Unlike the design parameters of collimator length 190 and cone spacing 194, the number of cones does not depend on a trade off between counts and system resolution. As additional cones 26 are added, it was found that the system resolution does not drastically change. Additional cones 26 would however reduce the size of the focal region and therefore cause greater sensitivity to change in distance from the collimator. As additional cones are added resulting in a larger detector, the expected number of counts increases, but at the expense of a more cumbersome system pipe thickness measurement system. MCNP calculation were made maintaining the previously discussed optimum cone spacing 194 of 3/8 inch and the previously discussed optimum collimator length 190 at 1.0 inch. These values required a detector radius of 4.05 inches for 4 cones, 5 inches for 5 cones and slightly over 6 inches for 6 cones. Calculations showed that as the number of cones 26 is increased, the number of counts increases. The smallest detector 50 considered, which is one using a collimator with 3 cones, gave the greatest counts per unit area. This tendency is expected as smaller detected solid angles provide a greater photon backscatter density. Once these data were adjusted for detector area, the detector 50 with the largest area was found to provide the maximum number of counts. The resolution to pipe wall thickness was found to be insensitive to the number of cones used. Using 6 cones would prove to be impractical as the sensitivity to distance 192 from the collimator would likely cause errors during a pipe scan. The minimum detector size of x_∑ = 6 inches would also make the system excessively cumbersome. From the MCNP calculations, it was determined that both 4 and 5 cones offer a compromise between the number of counts, acceptable detector size and insensitivity to changes in distance. Therefore, using 4 or 5 cones would offer acceptable results for the overall system design.

OPERATION OF THE SYSTEM The above sections have been directed toward basic principles behind the invention, the apparatus of the thickness scanning system, methods used to optimize the elements of the system, and the mechanisms governing the basic of the measurement system. Attention is now directed to specific applications of the measurement system, results of the applied measurements, and various embodiments of the system directed toward field applications. Effects of Fluid Density Upon Pipe Thickness Measurements

The density of fluid within a pipe, or more specifically the water density within the pipe, plays an extremely important role in the resolution of pipe wall thickness. As discussed previously, the invention embodied without detector collimation can be effectively used to measure the wall thickness of low density, gas filled pipe. Conversely, without the addition of the detector focusing collimator, changes in pipe wall thickness with water filled pipes is virtually undetectable. Even with liquid such as water of relatively low density, for example 0.7 specific gravity, the change in pipe wall thickness is clouded by multiple scatter in the water. While the addition of a focusing collimator greatly diminishes the effect of multiple scatters, it does not completely block out scatters from the water. As a result, changes in water density are seen to affect the total counts as well as resolution to pipe wall thickness.

In order to investigate the effect of water density, the model shown in Fig. 13 was again set up in the MCNP code using a collimator with cone spacing 194 of 3/8 inch and lengths 190 of 1.0 inch. The detector 50 outer radius X2 was maintained at 1.613 inches, the water 80 within the pipe 70 was varied from 0.7 to 1.0 specific gravity, and the wall thickness of the pipe 70 was varied between 1/16 and 1/2 inches. The results using this collimator were then compared to a similar MCNP calculations using a bare detector with no collimation. in both cases, the pipe was covered with 1.75 inches of thermal insulation 62. This setup would indicate the change in counts as well as the loss in resolution to pipe wall thickness that is expected from denser water.

Results of these calculations are shown in Fig. 14, which is a plot of measured counts as a function of pipe wall thickness. The family of curves identified by the numeral 202 represents counts measured with detector collimation for varying densities of fluid within the pipe 70 ranging from 0.1 to 1.0 grams/cubic centimeter (gm/cc). Curves 206, 208, 210 and 212 represent results with no detector collimation for fluid within the pipe 70 of densities 0.1 , 0.8, 0.9 and 1.0 gm/cc, respectively. As can be seen, the collimator increases thickness resolution by blocking out counts which are not a result of Compton scattering in the pipe. The total recorded counts with the detector collimated are, as would be intuitively expected, lower than with a bare detector counts. Examining curves 206, 208, 210 and 212 for the bare detector, as the water density is increased, the total counts increase dramatically. All of these additional counts are the result of increased scatter in the water 80 and are, therefore, meaningless in determining pipe wall thickness. Although each of these bare detector curves exhibits some sensitivity to pipe thickness, the density of the fluid within the pipe is the dominating factor thereby rendering the bare detector embodiment essentially useless, unless the density of the fluid is known apriori. With detector collimation, sensitivity to pipe thickness is quite good. Furthermore all curves within the family of curves 202 essentially overlay at pipe thicknesses greater than 0.2 - 0.3 cm, with the exception of curve 204 which represents a low density fluid such as light gas. Using the collimated detector embodiment, pipe thickness can be determined, with acceptable accuracy, without knowing the density of the fluid within, with the possible exception of gas. Fig. 14 presents a graphical illustration of the previously stated facts that either a bare or collimated detector can be used to effectively measure the thickness of gas filled pipe, but only a collimated detector yields acceptable results if the pipe is filled with liquid. For gas filled pipe, the bare detector embodiment is possibly preferred in that the measured counts are greater thereby yielding a more statistically precise answer.

Collimator Distance from the Pipe Wall

Referring again to Fig. 14, the distance 192 of the collimator cones 26 from the wall of the pipe 70 is an extremely important thickness scanning parameter. In order to maximize the number of recorded counts, the detector 50 should be placed as close to the pipe wall as possible. As the detector is moved closer to the pipe wall, the detected solid angle increases, hence increasing total recorded counts. However, as the collimator 26 is moved closer to the pipe wall, the focal point moves into the pipe interior. With water filled pipes, this interior region consists of water 80. A larger amount of scattering within the water is therefore expected, along with a loss in resolution from single and multiple scatters. As a result of these extra recorded counts, the total number of counts is expected to increase. A large portion of these counts is meaningless when investigating pipe wall thickness, and only serve to reduce the system resolution. It is therefore desirable to minimize this effect. In addition to the extra counts from scatters in the water 80, the geometric effect of detecting a greater solid angle adds to the total counts. Unlike the counts from the scatters in water, these extra counts also consist of scatters from the pipe wall and are therefore not harmful to the system resolution. MCNP calculations show that the distance 192 is clearly a trade off between counting statistics and system resolution. When an actual eroded pipe 70 is to be scanned, the distance 192 chosen between the collimator and the pipe wall will depend on the source strength and the time that is allowed to perform a complete pipe scan. Both higher source strength and longer counting time will provide a greater number of recorded counts, hence providing better counting statistics.

Examples of Thickness Measurements

Attention is directed to Fig. 15 which illustrates thickness measurements of an actual eroded pipe specimen as a function of the azimuthal angle q around the circumference of the pipe. Curve 267, which connects "diamond" measured data points, represent thickness measurements of the specimen with no insulation. Curve 269, which connects rectangular measured data points, represents thickness measurements of the same specimen through 1 inch of insulation. Agreement between the two sets of measurements is quite good thereby demonstrating the invention's ability to accurately determine the thickness of both insulated and uninsulated pipes. The data shown in Fig 15 also demonstrates precision in the measurement in that repeatability is quite good. The solid data points 268 represent "true' or "actual" pipe thickness values at the indicated angles q. These measurements were taken with a caliper, and were made difficult by the complex fine structure in the eroded pipe sample. These points agree well with the values obtained with the backscatter system indicating that the accuracy of the invention's measurement is good. Finally, the good spatial resolution of the invention is clearly demonstrated by the resolution of fine structure of the remaining pipe wall. Figs. 16a and 16b demonstrate the system's performance in a partially filled pipe situation. The pipe 70 is shown in Fig 16b in cross section, and is partially filled with water 80 and air 315. The azimuthal angle q is measured from the q = 0.00 as illustrated. A notch 302' was machined in the pipe at q = 20 degrees, and another notch machined in the pipe at q = 110°. It is apparent that the air water interface, as shown in the insert, at about q = 100° and q = 200°.

The counts recorded by the detector are plotted as a function of the angle q in Fig. 15a. The counts recorded around the air filled portion of the pipe from approximately q = 0.0° to q = 100°, defined by the region 310, produces approximately 3000 counts per measure point. The notch at 10° is clearly indicated by the drop in count rate 302 to approximately 2500 counts per measure point. At q = 100°, the counts per measure point increase to approximately 3400 clearly indicating the air/water interface, and remains at this level over the arc 314 until an air water interface is again encountered at q = 200°, where the count then drops to 3000 per measured point again indicating the presence of air on the opposite side of the pipe wall. The machined notch at q = 110° is again clearly indicated by the drop in counts at 316. It is important to note that this increase in counts is quite small in going from air to water. This is a result of the use of the detector focusing collimator as discussed in detail in previous sections.. This increase is, however, obvious as shown in Fig. 16a, and can be corrected or "normalized" to the air filled geometry by the simple application of a different calibration curve for the water filled pipe. System calibration will be discussed in a subsequent section.

As mentioned previously, the use of thickness measurement system is not limited to measuring the wall thickness of steel pipe, but can be used to measure the thickness of any material which possess nuclear characteristics falling within the bounds previously discussed. One such example is aluminum plate or sheet material. Fig. 17 shows the system response for thickness measurement of aluminum plate. Curve 322 represents the counts per measurement interval recorded in the system detector as a function of the thickness of the aluminum plate. These data were generated to illustrate applicability of the invention to aircraft inspection. Commercial aircraft are constructed primarily of aluminum alloys. A major maintenance problem is corrosion of the aluminum skins from the inside of wing and fuselage structures. It is apparent that the system is quite sensitive to changes in aluminum thickness in the range from 0 to 4 cm. This response is even better than that demonstrated for steel pipes. In addition, the absence of insulation on aircraft structures, and the absence of liquid behind the aluminum skin, further enhances the system's performance.

Based upon the above disclosure, it is apparent that the backscatter thickness measurement system can be used to make single, "spot" type measurements, or alternately can be used to make thickness measurement scans of pipe wall or other structure. Fig. 18 illustrates the system configured to make scans around the circumference of a pipe 70. The backscatter instrument 100 is connected to a track or rail 381 by a bracket 379. The track 381 fits around the circumference of the pipe 70. The track is positioned around the pipe 70 by expanding about a hinge 383, and subsequently clamping by means of a clamp 377. The backscatter instrument 100 is then conveyed circumferentially around the pipe on the track 381 by a suitable drive mechanism (not shown). Counts in the system detector can be taken with the instrument 100 positioned at interval stations around the pipe, or can be collected and recorded as the instrument is moved continuously around the circumference of the pipe 70. It should be understood that the mounting and instrument conveyance means shown in Fig. 18 is only one embodiment, and that there are numerous alternate means for mounting and conveying the instrument about the structure to be measured.

Previous disclosure has also stated that the present invention can be used to generate images of the thickness of a structure. The system embodied to produce an image of pipe wall thickness is shown in Fig. 19. As in Fig. 18, the instrument 100 is mounted so that it can scan a pipe circumference, at an axial position along the z axis of the pipe, following a path indicated by the broken line 270 by rotation about an angle q. Count data from the photomultiplier tube 52 are measured as a function of the angle q and the axial position along the z axis and input into a computer 290. The instrument is then moved axially along the z axis, and another circumferential measurement is made by rotation about the angle q following a path 272. Again, counts recorded as a function of q and z are input into the computer 290. This process is repeated sequentially for the paths 274, 276, and other paths not shown. The resulting count data recorded as a function of q and z are then used to generate an image or map of the thickness of the pipe wall, and output to a suitable recording device 292 which can produce a hard copy, or a digital recording of the image. Power is supplied to the photomultiplier 52 by a power supply 284. A clock 286 and control electronics 288 also cooperate with the power supply

284 so that counts can be recorded for a predetermined time interval, if desired. As mentioned in the previous paragraph, the system can be embodied to continuously circumferentially scan the pipe 70, wherein the clock 286 is used to record count rate. Fig. 19 illustrates the instrument as moving around the circumference of the pipe in a series of discrete, circular paths 270, 272, 274, and 276. An alternate embodiment of the system would employ a mounting and drive mechanism (not shown) such that the instrument is conveyed along the z axis in a helix path.

Fig. 20 illustrates a hypothetical pipe wall thickness scan generated with the system embodied as shown in Fig. 19. For purposes of illustrating the a imaging concept, discrete peripheral counts are made at eight positions spaced at q = 45°, at each of nine positions along the z axis of the pipe. This yields a total of 72 station measurements which are used to generate pixels for the pipe wall thickness image. It should be understood that the resolution of the system allows much higher peripheral resolution measurements, such as shown in Fig. 16, and the eight point peripheral measurement is used only as a hypothetical example for purposes of illustration. The counts recorded at each position are converted to an absolute thickness measurement using calibration methods discussed in the following section. The results can be displayed as a two dimensional 8 x 9 measurement array of numerical thickness measurements. Alternately, a "gray" scale can be used to convert these measurements to a more visual scan as shown in Fig. 20. Using this method, a shade of "gray" is assigned to each (q,z) pixel 282 generated by the scan. As an example, the darkest shade can be used to represent the maximum or "gauge" pipe thickness. Thicknesses less than the maximum gauge thickness can be assigned proportionally lighter shades of gray, where no metal (i.e. zero thickness) is represented by white. In the hypothetical example of Fig. 20, a region of helical thinning of the pipe wall is indicated, starting at z = 0.5, q = 90 and extending to z = 3.5, q = 270. It should be understood that the data can be displayed in alternate ways, such as a three dimensional cylinder. Furthermore, the gray scale can be replaced with a color scale, wherein the color indicates pipe thickness.

System Calibration

It is necessary to calibrate the thickness measurement system such that the measured quantity, namely counts recorded by the detector, can be converted to the quantity of interest, namely material thickness. The relationship between the measured count data and quantities of interest is, as discussed in detail above, a function of numerous factors such as the type of material being measured, the fluid type behind the material being measured, the source strength, the detector efficiency, the type of source and detector collimation used, count intervals, and the like. As in most nuclear measurement systems, "environmental" calibrations, "computed" calibrations, or a combination of both types of calibrations, can be employed. Stated simply, environmental calibrations use the response of the instrument as measured in known environmental conditions. A calibration relation is then derived to relate instrument response to thickness measurements made under these known conditions. Computed calibrations use various models, such as the previously discussed MCNP code, to calculate calibration relations. Since it is usually impractical to environmentally simulate every condition in which an instrument is to be used, a combination of environmental and computed calibrations are usually used, wherein a relatively large number of computed calibrations are "normalized" to a relatively limited number of environmental calibrations. The combination of environmental and computed calibrations is the preferred means for calibrating the present thickness measuring system.

SUMMARY

A system has been developed to measure the thickness of pipe wall, from the outside of the pipe, using nuclear apparatus and methods. The system comprises a photon source, such as Hg²⁰³, which emits 279 keV gamma radiation, and an annular gamma ray detector which are positioned on the outside of the pipe wall. Pipe wall thickness is determined from the response of the annular detector to gamma radiation from the source which first passes through the annulus of the detector, and which is subsequently Compton backscattered by the pipe wall. Concentric, conical detector collimation is used in one embodiment of the system to enhance the thickness measurement of the walls of pipe which are filled with relatively dense fluids such as water. The system can also be used to measure, nondestructively, the thickness of materials in other types of structure. In measuring pipe wall thickness, the system has two major benefits over conventional, prior art ultrasonic pipe wall thickness measurement systems. More specifically, the system eliminates the need to remove insulation which results in great savings in the person-hours required to perform pipe inspections. In addition, the system can image pipe walls in empty, steam-filled, or liquid-filled pipes even while in operation, e.g. at normal elevated operating temperatures or flow velocities.

There are other embodiments and applications of the invention which will be apparent to practitioners of the art. While the foregoing is directed to the preferred embodiments, the scope thereof is determined by the claims which follow.

What is claimed is:

Claims

C LAIMS

1. A system for measuring thickness of material comprising: (a) a source of photons; and (b) an annular photon detector, wherein material thickness is determined from the response of said photon detector to photons

(i) which are emitted by said source and first pass through an annulus of said detector, and

(ii) which are subsequently Compton backscattered by said material into to said detector.

2. The system of claim 1 further comprising a source collimator which is positioned within said annulus of said annular detector thereby forming a pencil beam of photons emerging from said source collimator perpendicular to a face of said detector which faces said material.

3. The system of claim 2 wherein said annular detector is collimated by a plurality of cones, wherein:

(a) axes of each of said cones and said pencil beam of photons are coincident;

(b) apexes of each of said cones are truncated and face said material; and

(c) bases of each of said cones abut said detector face.

4. The system of claim 2 wherein:

(a) said pencil beam of photons impinges upon said material at an angle of about 90 degrees, and

(b) the scatter angle of said detected Compton backscatter radiation is about 180 degrees.

5. The system of claim 2 wherein said response of said annular detector comprises counts recorded by said detector over a known time interval, and said material thickness is computed from said counts using a predetermined relationship. 6. The system of claim 2 wherein said annular detector comprises;

(a) a scintillation crystal, and

(b) a photomultiplier tube optically coupled to said scintillation crystal.

7. The system of claim 2 wherein said photon source comprises mercury- 203 and said pencil beam of photons comprises 279 keV gamma radiation.

8. The system of claim 6 wherein said scintillation crystal comprises sodium iodide.

9. A system for measuring the wall thickness of pipe, comprising:

(a) a source of photons;

(b) an annular photon detector in the form of a disk with a concentric annulus; and (c) a source collimator positioned within said annulus, wherein

(i) photons emitted by said source pass through a passage in said source collimator thereby forming a pencil beam of photons,

(ii) said pencil beam impinges perpendicularly upon the surface of the outside wall of said pipe, (iii) photons are Compton backscattered into said detector by said pipe wall, and

(iv) the thickness of said pipe wall is determined from the response of said detector to said Compton backscattered photons.

10. The system of claim 9 wherein said annular detector is collimated by a plurality of cones, wherein:

(a) axes of each of said cones and said pencil beam of photons are coincident;

(b) apexes of each of said cones are truncated and face said pipe wall; and

(c) bases of each of said cones abut a face of said detector.

1 1. The system of claim 9 wherein said response of said annular detector comprises;

(a) counts resulting from Compton backscatter at about 180 degrees; and (b) counts recorded by said detector over a time interval, wherein

(c) said pipe wall thickness is computed from said counts using a predetermined relationship.

12. The system of claim 11 including means for conveying said system around the periphery of said pipe, wherein said wall thickness determinations are made at a plurality of azimuthal positions around the periphery of said pipe.

13. The system of claim 12 further comprising a computer in which said predetermined relationship is stored, and having an input for said recorded counts and determining said pipe wall thickness.

14. The system of claim 13 including means for conveying said system along the wall of said pipe parallel to the axis of said pipe, wherein said wall thickness determinations are made at a plurality of axial positions along said pipe.

15. The system of claim 14 wherein said azimuthal pipe wall thickness measurements and said axial pipe wall thickness measurements are combined within said computer to form a map of the wall thickness of said pipe.

16. The system of claim 15 further comprising a recorder which cooperates with said computer, wherein said map is output from said recorder as a hard copy display or as a digital recording.

18. The system of claim 9 wherein said annular detector comprises a sodium iodide scintillator and a photomultiplier tube optically coupled to said sodium iodide scintillator.

19. The system of claim 9 wherein said photon source comprises mercury- 203 ana said pencil beam of photons comprises primarily gamma rays of energy 279 keV. 20. A method for measuring thickness of material comprising:

(a) irradiating said material with photons emitted by a source; and

(b) detecting photons which are Compton backscattered by said material, wherein (i) said Compton backscattered photons are detected with an annular detector,

(ii) said photons emitted by said source first pass through an annulus of said detector and are subsequently Compton backscattered by said material into to said detector, and (iii) said material thickness is determined from said detected backscattered photons.

21. The method of claim 20 further comprising the step of collimating said photons emitted by said source by means of a source collimator positioned within said annulus of said annular detector thereby forming a pencil beam of photons emerging from said source collimator pθφendicular to a face of said detector which faces said material.

22. The method of claim 21 further comprising the step of collimating said detector with a plurality of cones, wherein:

(a) axes of each of said cones and said pencil beam of photons are coincident;

(b) apexes of each of said cones are truncated and face said material; and (c) bases of each of said cones abut said detector face.

23. The method of claim 21 comprising the additional steps of:

(a) impinging said pencil beam of photons upon said material at an angle of about 90 degrees, and

(b) detecting said Compton backscatter radiation at an angle of about 180 degrees.

24. The method of claim 21 comprising the additional steps of recorded counts in said detector over a known time interval, and computing from said recorded counts said thickness of said material using a predetermined relationship.

25. The method of claim 22 wherein said detector and said collimator cones measure the thickness of said material when said material is backed by a liquid.

26. The method of claim 21 wherein said annular detector comprises;

(a) a scintillation crystal and a photomultiplier tube optically coupled to said scintillation crystal; and

(b) said photon source comprises mercury-203 and said pencil beam of photons comprises 279 keV gamma radiation.

27. A method for measuring the wall thickness of pipe, comprising the steps of:

(a) irradiating said pipe wall with a source of photons;

(b) providing an annular photon detector in the form of a disk with a concentric annulus; and

(c) positioning a source collimator within said annulus thereby

(i) foiming a pencil beam of photons from photons emitted by said source and passing through said source collimator,

(ii) impinging said pencil beam perpendicularly upon the surface of the outside wall of said pipe,

(iii) measuring photons which are Compton backscattered into said detector by the said pipe wall, and

(iv) determining the thickness of said pipe wall from the response of said detector to said Compton backscattered photons. 28. The method of claim 26 comprising the additional step of collimating said annular detector with a plurality of cones, wherein:

(a) axes of each of said cones and said pencil beam of photons are coincident; (b) apexes of each of said cones are truncated and face said pipe wall; and

(c) bases of each of said cones abut a face of said detector.

29. The method of claim 27 wherein pipe wall thickness is computed from the response of said detector, wherein said detector response comprises:

(a) counts resulting from Compton backscatter at about 180 degrees; and

(b) counts recorded by said detector over a known time interval, wherein said pipe wall thickness is computed from said counts using a predetermined relationship.

30. The method of claim 29 including the step of conveying said system around the periphery of said pipe, wherein said wall thickness determinations are made at a plurality of azimuthal positions around the periphery of said pipe.

31. The method of claim 30 further comprising the step of storing said predetermined relationship in a computer, inputting said recorded counts into said computer, and using said predetermined relationship and said input recorded counts to compute said pipe wall thicknesses at said plurality of azimuthal positions.

32. The method of claim 31 including the step of conveying said system along the wall of said pipe parallel to the axis of said pipe, wherein said wall thickness determinations are made at a plurality of axial positions along said pipe.

33. The method of claim 32 including the step of combining said azimuthal pipe wall thickness measurements and said axial pipe wall thickness measurements to form a map of the wall thickness of said pipe. 34. The method of claim 33 including the steps of providing a recorder which cooperates with said computer, and outputting said map from said recorder as a hard copy display or as a digital recording.

35. The method of claim 20 wherein said annular detector comprises a sodium iodide scintillator and a photomultiplier tube optically coupled to said sodium iodide scintillator.

36. The method of claim 20 wherein said photon source comprises mercury- 203 and said pencil beam of photons comprises primarily gamma rays of energy 279 keV.

37. The method of claim 21 wherein said detector and said collimator cones measure the wall thickness of pipe filled with liquid.