JP4371364B2 - Automatic ultrasonic flaw detector and automatic ultrasonic flaw detection method for thick structure - Google Patents

Description

  The present invention relates to an automatic ultrasonic inspection device and an automatic ultrasonic inspection method for a thick-walled structure. More specifically, the present invention relates to an improvement in an automated technique of measuring crack defects generated in a thick structure using a conventional probe by applying both the TOFD method and the phased array method.

  The aging of existing thermal power generation facilities has progressed year by year, and in Europe and the United States, there is inherent heat damage in heat-affected zones (HAZ) that are difficult to detect by existing non-destructive inspection in high-temperature steam piping of aged thermal power plants. Experienced blast accidents caused by defects. In Japan, there is a strong concern for defects inherent in such thick-walled structures such as high-temperature steam piping.

  In addition to the weld heat affected zone of such high-temperature steam pipes, there are inherent defects such as cracks that occur in the welded parts of thick-walled structure materials, particularly welded parts of this thick-walled structure due to slag entrainment and poor penetration (this specification Then, as a non-destructive inspection for detecting such an internal defect such as a crack as a “crack defect” or simply “a crack”), an ultrasonic flaw detection test has been carried out. Application research such as TOFD (Time of flight Diffraction) method and phased array method (PA method) has been promoted for the purpose of early detection.

  In the TOFD method, ultrasonic waves are emitted between two probes for transmission and reception (probes), and the position of a crack defect in a thick structure material that is a flaw detector is identified. Here, the “position” of the crack defect means a three-dimensional position including both the position in the front-rear and left-right directions and the position (or height) in the thickness direction of the material. For example, flaw detection ultrasonic waves are transmitted from one of the two probes (probes) 101 and 102 shown in FIG. 14 (transmission probe 101) to the material 103 such as a pipe to be flaw detected, The lateral wave LW propagating on the surface of the material 103, the diffracted wave at the upper end 104a and the lower end 104b of the crack 104, and the reflected wave BW on the bottom surface of the material 103 are received by the other probe (receiving probe 102). The height and position of the crack 104 are identified from the phase difference (see FIG. 15). Incidentally, the diffracted wave at the tip of the defect such as the upper end 104a and the lower end 104b of the crack 104 is also called an end echo. Also called “end echo method”. Then, according to the TOFD method as described above, the height and position of the crack 104 can be identified with high accuracy. In addition, in this TOFD method, the height of the crack 104 and the like can be adjusted only by performing a so-called line scan (also referred to as a D scan) in which the probes 101 and 102 are linearly moved in the longitudinal direction in parallel with the weld line 105. There is also a feature that it can be identified (see FIGS. 16 and 17).

  On the other hand, in the phased array method (PA method), a plurality of pulse receivers 201 and piezoelectric elements (piezo elements) 202 are arranged on a one-to-one basis (see FIG. 19), and the timing of the pulse voltage is shifted to shift the ultrasonic beam. It is possible to detect flaws by changing the direction and changing the depth of focus (see FIGS. 20 and 21). Previously, a single pulse receiver 201 vibrated a single piezoelectric element 202 through a cable 203 to generate ultrasonic waves (see FIG. 18), but this made it difficult to change the direction freely. Therefore, a flaw detection technique has been proposed that separates each of the plurality of pulse receivers 201 and the piezoelectric elements 202 to change the direction of the ultrasonic waves without changing the direction of the probe. According to such a phased array method, the direction of the ultrasonic wave can be changed by shifting the timing of voltage application to the piezoelectric element 202 without needing to change the direction of the probe (see FIGS. 20 and 21). In the case of the phased array method, there is also an advantage that three-dimensional information can be obtained by moving the probe having the above-described configuration only once in parallel with the weld line.

  Furthermore, a composite flaw detection technique in which flaw detection is performed by combining the above-described TOFD method and phased array method has also been proposed (see, for example, Patent Document 1).

JP 2001-50938 A

  However, in the case of the TOFD method, the height and position of the crack 104 can be identified with high accuracy, but on the other hand, it is possible to completely identify where the crack 104 is located with respect to the weld line 105. There is a problem that it is difficult. For example, a crack 104 may be generated between the weld line 105 and the base material (material portion other than the weld line) in piping or the like (see, for example, FIG. 16). In some cases, it may not be possible to distinguish between an originally existing flaw called a weld defect or the like and a potentially fatal crack 104 that has occurred over the course of time. There is a problem that high flaw detection results cannot be obtained.

  Further, in the case of the TOFD method, since a two-dimensional scan is required, a method of moving the probe 101 (102) in a zigzag manner as shown in FIG. 22, for example, is required. When two-dimensional scanning is performed in this way, three-dimensional information is obtained, and high-accuracy crack defect information is still obtained, but it takes too much time to scan. In addition, another problem arises that an expensive driving device for performing two-dimensional scanning is also required.

  In addition, in the case of the above-described end echo method, since it is a flaw detection method of determining the end echo from a subtle change in the waveform accompanying the operation of the probe, it greatly depends on the skill of the inspector, and the inspection result There is also an aspect that individual differences are likely to occur.

  On the other hand, the phased array method has the advantage that the direction of the ultrasonic beam can be changed without moving the probe, but has the problem that the accuracy of estimation of the crack height is worse than that of the TOFD method. In addition, when a crack defect is inherent in a place other than the focal point of the ultrasonic beam, there is a problem that it is difficult to detect a reflected wave from the defect.

  In addition, as described above, there are flaw detection technologies that combine the TOFD method and the phased array method, but these flaw detection technologies are simply technologies that focus on the hardware that combines the TOFD method and the phased array method. The software is not specially processed. For example, in Patent Document 1 described above, there is no disclosure of a technique such as signal processing for supporting accurate time measurement. For this reason, in addition to the fact that the measurement accuracy is not so high for the time required for ultrasonic flaw detection, waveform measurement is necessary for crack measurement, and there is a problem that measurement error by the measurer is large. Furthermore, cracks may be difficult to detect due to noise caused by the metal structure of the target material.

  Furthermore, when a crack defect is inherent in the material, for the time being, the actual situation is that the part where the crack exists or the equipment is removed or replaced. However, if the inherent defect such as a crack generated in the material is relatively small compared to the size of the equipment, the presence of the crack does not immediately lead to damage to the equipment. In this case, there is sufficient time to cause damage. Therefore, in the present situation, it is desirable to more accurately estimate and grasp the remaining life from the occurrence of a crack to failure, and to develop a more rational maintenance plan. It is required to measure the crack dimensions as accurately as possible. And in order to measure a crack dimension with high accuracy in this way, an automated technique for improving the reproducibility of the inspection result is important as well as optimizing the probe and measurement conditions.

  Therefore, the present invention is capable of acquiring information on rough flaw detection and precision flaw detection within the inspection time required for conventional rough flaw detection, and is capable of obtaining highly accurate defect dimensional measurement results. An object of the present invention is to provide an ultrasonic flaw detection apparatus and an automatic ultrasonic flaw detection method. Furthermore, the present invention is capable of discriminating originally existing scratches and cracks that have occurred with the passage of time, such as HAZ defects and weld defects, from quantitative position measurement results and dimension measurement results. An object of the present invention is to provide an automatic ultrasonic flaw detection apparatus and an automatic ultrasonic flaw detection method for a thick-walled structure capable of obtaining a flaw detection result.

  In order to achieve this object, the present inventor has conducted various studies. With the TOFD method, it is possible to measure the size of internal defects such as HAZ with higher accuracy than with the phased array method, but it is also difficult to clearly distinguish between existing scratches and crack defects that have occurred over the course of time. . Therefore, in the case of the conventional weld inspection technique, the crack defect position in the direction perpendicular to the weld line (depth direction) cannot be determined only by the result of the rough scanning of the D scan normally performed by the TOFD method. In addition, it was necessary to carry out a B-scan at the defect detection site for precise flaw detection (see FIG. 11). In this case, since an expensive driving device capable of two-dimensional scanning is required and the inspection time becomes long, the present inventor considers this and instead of the B scan of the TOFD method or the raster scan of the pulse echo method, We thought it desirable to use the phased array method together with the D scan of the TOFD method.

  Furthermore, the present inventor targets that measurement accuracy is not so high for the time required for conventional ultrasonic flaw detection, and that waveform measurement is necessary for crack measurement, but the measurement error by the measurer is large. With regard to the fact that cracks may become difficult to detect due to noise caused by the metallographic structure of the material, the accuracy of measurement can be improved while maintaining the conventional flaw detection equipment by further processing the obtained measurement result information. Focusing on the technology to improve, as a result of various examinations and tests, it came to know the technology that can achieve this.

  The present invention is based on such knowledge, and the automatic ultrasonic flaw detector for a thick structure according to claim 1 applies both the TOFD method and the phased array method to crack defects generated in the thick structure. In an ultrasonic flaw detector with a thick structure that detects flaws with ultrasonic waves, it propagates in the material from the transmitted ultrasonic waveform and the waveform of the reflected echo such as the bottom surface and internal crack defects of the material of the thick structure. Sound velocity / distance measurement means that measures the actual sound speed of ultrasonic waves and the distance to the reflector for each material using at least one of the zero cross method, correlation function method, and phase spectrum method, and shape echo in the received waveform Or noise due to backscattered waves, the noise intensity is reduced by signal processing using either or both of the spatial averaging method and wavelet transform method. In addition to the noise removal means that emphasizes the defect signal in exchange, the measurement result of the relationship between the frequency and attenuation rate of the ultrasonic wave depending on the size of multiple types of crystal grains in the material, or the size of the crystal grains The size of the crystal grain is evaluated using the measurement result of the amplitude and frequency characteristics of the ultrasonic backscattered wave depending on the thickness, and the optimal flaw detection frequency for the crystal grain is selected based on the evaluation result. An evaluation means, and an ultrasonic wave propagation characteristic prediction means for analyzing and visualizing the wave propagation when the wave propagation behavior of the ultrasonic wave is complicated, and predicting the ultrasonic wave propagation characteristic of a material having a coarse crystal. It is characterized by having.

  The invention according to claim 2 is an ultrasonic flaw detection method for a thick-walled structure in which crack defects generated in a thick-walled structure are flawlessly detected by applying both the TOFD method and the phased array method. The zero-crossing method and the correlation between the transmitted ultrasonic waveform and the waveform of the reflected echo from the bottom surface of the material of the thick-walled structure and internal crack defects, as well as the actual sound velocity of the ultrasonic wave propagating through the material and the distance to the reflector. A sound velocity / distance measurement step that measures each material using at least one of the functional method and phase spectrum method, and when noise due to shape echoes or backscattered waves appears significantly in the received waveform A noise removal step of reducing the noise intensity by signal processing applying either one or both of the average method and the wavelet transform method and emphasizing the defect signal in exchange for this, The measurement result of the relationship between the frequency and attenuation rate of the ultrasonic wave depending on the size of multiple types of crystal grains in the material, or the amplitude of the back scattered wave of the ultrasonic wave depending on the size of the crystal grains as well as this The measurement result of frequency characteristics is used to evaluate the size of the crystal grain, and based on this evaluation result, the crystal grain evaluation step for selecting the optimum flaw detection frequency and the behavior of ultrasonic wave propagation are complicated. In this case, the wave propagation is analyzed and visualized, and an ultrasonic wave propagation characteristic prediction step that makes it possible to predict the ultrasonic wave propagation characteristic of a material having a coarse crystal is characterized.

  In the automatic ultrasonic flaw detector for a thick structure according to claim 1, the flaw detection method is supplemented by applying both the TOFD method and the phased array method (more specifically, the dimension of the defect by the TOFD method, By evaluating the position of the defect by the phased array method), it is possible to measure the crack defect position and size with high accuracy even in one-dimensional scanning (scanning) (see FIG. 11). In addition, in this case, since two-dimensional or three-dimensional scanning is not required, there is an advantage that an expensive driving device is not required, and an inspection time is shortened.

  In addition, according to the present invention, since the obtained measurement information is subjected to data processing by various functions, a more accurate flaw detection result can be obtained. In other words, (i) the sound velocity / distance measurement function that determines the sound velocity and distance from the transmission waveform, etc., makes it possible to size defects in the material with higher accuracy, and (ii) shape echo and backscattering in the obtained signal. When noise caused by waves is significant, these unnecessary noises can be removed. (Iii) The optimum flaw detection frequency can be selected based on the evaluation result of the crystal grain size in the material. If the wave propagation behavior of the system is complex, the wave propagation can be analyzed and visualized. By processing the measurement information further, the existing flaw detection equipment remains intact and more accurate than before. A flaw detection result is obtained. In this case, information on rough flaw detection and precision flaw detection can be acquired within the inspection time required for conventional rough flaw detection, and highly accurate defect dimension measurement results can be obtained.

  According to the automatic ultrasonic flaw detection method for a thick-walled structure according to claim 2, the flaw detection method is supplemented by applying both the TOFD method and the phased array method, as in the above-described automatic ultrasonic flaw detection device (more Specifically, by measuring the defect dimensions by the TOFD method and the defect positions by the phased array method, it is possible to measure the crack defect position and dimension with high accuracy even in one-dimensional scanning. Yes (see FIG. 11). In addition, in this case, since two-dimensional or three-dimensional scanning is not required, there is an advantage that an expensive driving device is not required, and an inspection time is shortened.

  In addition, according to the present invention, since the obtained measurement information is subjected to data processing by various functions, a more accurate flaw detection result can be obtained. In other words, (i) the sound velocity / distance measurement function that determines the sound velocity and distance from the transmission waveform, etc., makes it possible to size defects in the material with higher accuracy, and (ii) shape echo and backscattering in the obtained signal. When noise caused by waves is significant, these unnecessary noises can be removed. (Iii) The optimum flaw detection frequency can be selected based on the evaluation result of the crystal grain size in the material. If the wave propagation behavior of the system is complex, the wave propagation can be analyzed and visualized. By processing the measurement information further, the existing flaw detection equipment remains intact and more accurate than before. A flaw detection result is obtained. In this case, information on rough flaw detection and precision flaw detection can be acquired within the inspection time required for conventional rough flaw detection, and highly accurate defect dimension measurement results can be obtained.

  Hereinafter, the configuration of the present invention will be described in detail based on embodiments shown in the drawings.

  1 to 13 show an embodiment of the present invention. The automatic ultrasonic flaw detector for a thick structure according to the present invention (hereinafter also simply referred to as “automatic ultrasonic flaw detector”) is also referred to as a material for a thick structure (hereinafter simply referred to as “material”). The ultrasonic wave oscillated from the probe (or probe; indicated by reference numeral 3 in FIG. 5) by applying both the TOFD method and the phased array method to the crack defect (denoted by reference numeral 2 in FIG. Flaw detection.

Furthermore, the automatic ultrasonic flaw detector according to the present embodiment is based on the result of the crack inspection based on the assumption that the conventional probe 3 used when performing the conventional flaw detection technique (for example, TOFD method) is used. And allows defect dimensions to be measured. In other words, the hardware surface of the conventional probe 3 and the like is left as it is, and the data processing peculiar to the software surface is performed, and the defect position and the defect size can be measured while using the conventional device as it is. The main part of data processing on the software side is the “waveform analysis function” that analyzes the waveform during crack inspection and the “visualization function” that visualizes the ultrasonic propagation phenomenon by numerical analysis. More specifically, the four functions described below, namely:
(i) A function that calculates the speed of sound or the distance to the reflector from the waveform of the transmission and the waveform from the reflector such as the bottom or defects (sound speed / distance measurement function)
(ii) A function that removes noise caused by shape echoes and backscattered waves in the obtained signal by signal processing (noise removal function)
(iii) Function for evaluating crystal grain size using the frequency, velocity dispersion, and attenuation of backscattered waves (crystal grain evaluation function)
(iv) Function to predict the ultrasonic propagation characteristics of material 1 with coarse crystals by the finite element method (ultrasonic propagation characteristics prediction function)
It consists of each function. In the following, first, each of these functions will be described in order, and then a measurement technique on the hardware side for improving the welded part inspection technique will be described. Each of the above functions is mounted on, for example, a control processing device or a control processing circuit built in the automatic ultrasonic flaw detector.

(i) “Sound velocity / distance measurement function” First, in this embodiment, the sound velocity of ultrasonic waves propagating in the material 1 (“sound velocity” generally means the velocity at which sound propagates in the air. However, in this specification, the propagation speed of the ultrasonic wave propagating in the material 1 is also measured in real time, and the defect in the material 1 is more accurately determined based on the measured sound speed. Trying to size. That is, in addition to the fact that I) the sound velocity (propagation velocity) of the ultrasonic wave in the material 1 has a great influence on the sizing accuracy of the crack defect 2, II) the sound velocity in the target material 1 is the material 1 Based on the two points of change due to deterioration over time, the actual propagation speed in the material 1 is grasped for each material 1 so that more accurate ultrasonic flaw detection can be realized. In addition, as a method for actually obtaining the ultrasonic sound velocity, 1) a zero cross method, 2) a correlation function method, and 3) a phase spectrum method are provided, and any one of these methods is used.

In the case of the “zero cross method” of ₁ ), the propagation time difference between points of the same phase for each of the i-th reflected wave (time T ₁ ) and the i + 1-th reflected wave (time T ₂ ) from the bottom of the plate-shaped specimen Δt (= T ₂ −T ₁ ) is obtained, and the sound speed is calculated based on this (see FIG. 1). The zero-crossing method calculates the speed of sound by calculating the propagation time from the horizontal axis position where 0 (zero), more specifically, based on the measurement results at two adjacent measurement points, the phase between the two measurement points. The position (time) at which the horizontal axis becomes zero is obtained, and the sound speed is accurately calculated from this position (see FIG. 1).

However, since the zero cross method cannot be applied when the similarity of adjacent waveforms is reduced, the “correlation function method” of 2) is used in that case (FIGS. 2 and 2). 3). In the correlation function method, two waveforms that maximize the correlation function R _xy (these two waveforms are not limited to those adjacent to each other as in the above-described zero cross method), and a propagation time difference between them (here, τ). In the above-described zero cross method and this correlation function method, it is possible to measure the sound speed with an accuracy of about 0.2% by obtaining the propagation time up to the minimum recording time interval by interpolation (also referred to as interpolation). . For example, a value at a certain time t is x (t) (≠ 0), and a value at the next time t + Δt is x (t + Δt) (≠ 0). A time T of x (T) = 0 (t <T <t + Δt) is obtained by interpolation. In digital recording, the recording time is 0, Δt, 2Δt, 3Δt, 4Δt,..., NΔt, (n is a natural number). Therefore, if nothing is processed, the propagation time is always an integer multiple of Δt. However, as described above, an accurate time that becomes 0 can be obtained by interpolation, and a propagation time up to a minimum recording time interval can be obtained.

Further, when the material 1 is a dispersible material, the phase velocity of the ultrasonic wave changes depending on the frequency. In this case, the sound velocity is obtained by the phase spectrum method of 3) (see FIG. 4). That is, two different waveforms are subjected to FFT transform (fast Fourier transform) to calculate the phase spectrum of each waveform, and the phase velocity is obtained from Equation 1 from the phase difference Δφ between the two waveforms.
[Equation 1]
V _p = 2πfL / (− Δφ)
Here, L is a propagation time, Δφ is a phase difference, and f is a frequency.

  As described above, the automatic ultrasonic flaw detector of the present embodiment having the sound speed / distance measurement function obtains the sound speed and the propagation time with high accuracy in real time while measuring the sound speed and propagation time online, that is, (I) Material 1 is obtained. Even if it is thick, it becomes possible to measure the height of the defect with an accuracy within an error range of at most about 1%, and (II) the distance between the corner echo and the end echo can be measured more accurately. Therefore, the depth of the defect can be obtained with higher accuracy than before, and (III) when the phase velocity is obtained by the phase spectrum method of 3) described above, the characteristics of the material 1 can be estimated online. It becomes like this.

  The “corner echo” refers to an echo reflected at the corner of the defect, and the “end echo” refers to a diffracted wave at the tip of the defect in the material 1 such as the upper end or the lower end of the crack 2. Refers to that. The “edge echo method” refers to a technique for estimating the height of the defect from the difference in propagation time between the corner echo and the edge echo. For example, as shown in FIG. 5, when an ultrasonic beam represented by an arrow hits a corner portion of the crack 2, the reflected corner echo returns to the probe 3, so that the diffracted wave (end echo) and the propagation of this corner echo are transmitted. A time difference can be obtained (see FIG. 5).

(ii) “Noise-removing function” Austenitic steel has larger crystal grains than ferrite, and the weld has larger crystal grains than the rolled structure of the base metal. An echo appears. This echo is called a “forest echo” because it looks like a forest. When the flaw detection specimen is this austenitic steel, the welded portion has a coarse columnar crystal structure having elastic anisotropy. Therefore, when the welded portion is flawed, the forest echo is the end of the crack defect 2. The echo may be masked and the sizing accuracy of the defect may be reduced. Further, in the case of the material 1 having a crystal grain equal to or greater than the wavelength of the incident ultrasonic wave, noise due to the backscattered wave may reduce the SN ratio and the flaw detection accuracy may be reduced.

  In contrast, in the automatic ultrasonic flaw detector of the present embodiment, even if there is a noticeable forest echo or noise due to backscattered waves, these are removed by online signal processing to remove defect echoes. Since it can be identified, it is effective for ultrasonic flaw detection. Furthermore, in the present embodiment, by providing a “noise removal function” as described below, the noise obtained by such shape echoes and backscattered waves appears remarkably in the obtained signal. These can be removed by signal processing (see FIG. 6).

That is, when performing noise processing (step 1 in FIG. 6), the noise phase changes faster than the phase of the defect echo when it is slightly distant from a certain position. (Step 2), the intensity of noise is reduced (step 3), and the defect signal is enhanced in exchange for this. To explain this, first, the wave can be simply expressed as Aexp (ωt + kx) (A is the amplitude, ω is the angular frequency, t is the time, k is the wave number, and x is the distance from the source to a certain spatial position. , Ωt + kx is the phase). Although the distance from the crack defect to the vicinity of a certain spatial position does not change very much, the distance from the noise source to the spatial position varies widely, so the phase of the noise echoes the defect echoes even slightly away from a certain position. It changes faster than the phase. The spatial averaging method performed in the present embodiment is to average received waveforms at several positions centered on a certain position, thereby reducing the noise intensity by suppressing variations due to the position. A specific example will be described below. If the signal at the center point c of a certain region is s _c (k) (k represents time), and the signal at the point (i, j) around the point c is s _ij (k), a spatially averaged point The signal s _c (k) _{avr at c} can be calculated as in Equation 2 below.

Also, focusing on the fact that defect echoes do not change much in intensity of each frequency component of echoes compared to forest echoes, UT (Ultrasonic testing) signals subjected to spatial averaging are subjected to wavelet transform (steps 4 and 5). . That is, in the band f _b = [f _L , f _H ] localized in the flaw detection frequency band of the probe 3, the band f _b is divided into n, and the time-frequency component s _i (t ) (Where i = 1, 2, 3,..., N). Normalization is performed on each s _i (t), and correlation processing is performed using Equation 3 below.

Here, E _c (t _j ) is a correlation coefficient corresponding to the UT signal at time t _j , and S _i is a normalized value of s _i . Since the intensity of the forest echo varies considerably with the change in frequency, the correlation coefficient corresponding to the forest echo is small. On the other hand, since the intensity of the frequency component of the defect echo does not change much, the correlation coefficient corresponding to the defect echo is larger than the forest echo if the SN ratio of the UT signal is greater than a certain value. Therefore, if there is a threshold e _c or correlation at time t _j, it can be determined that the echo localized in its time is defect echo.

  Note that the noise removal function in the present embodiment selects and uses the above-described spatial averaging method and wavelet transform method appropriately according to the case. That is, in performing noise removal processing (step 1), it is first determined whether or not to apply the spatial averaging method (step 2). If it is determined that sufficient noise removal is possible only by the spatial averaging method, the process proceeds to step 3. If not, the process proceeds to step 6 to apply the wavelet transform method (step 6). Further, after performing the spatial averaging method in step 3, it is determined whether or not the wavelet transform method is further applied (step 4), and after performing the wavelet transform method in step 6, whether or not the spatial average method is further applied. (Step 7), and when it is determined that both noise removal processes should be performed, the remaining one process can be continued (see FIG. 6). That is, if only one noise removal process is sufficient as determined at the beginning, the process is completed with a smaller number of steps. On the other hand, if it is determined that one noise removal process is not sufficient as a result of referring to the noise process. At the same time, the other noise removal process is performed so that the optimum noise removal process is performed on each material 1 on a case-by-case basis.

  When displaying an image, the A / scan waveform at each position is subjected to spatial averaging and wavelet transform of forest echoes and backscattered noise, thereby improving the SN ratio and facilitating defect discrimination. Become.

(iii) “Crystal Grain Evaluation Function” In the present embodiment, even if the crystal grains of the material 1 are large, the defect detection performance is prevented from deteriorating and high sizing accuracy is maintained. From the point of view, the monitoring of crystal grains is planned. That is, for example, a weld metal or cast product of austenitic steel such as stainless steel has large crystal grains, so that a scattered wave is generated between the crystal grains with respect to high-frequency ultrasonic waves for flaw detection, and the signal is attenuated violently. For this reason, there may be a phenomenon that the SN ratio of the flaw detection signal is lowered and the defect detection performance is lowered. On the other hand, if the frequency of the flaw detection ultrasonic wave is lowered in order to avoid such a situation, the directivity of the ultrasonic beam is deteriorated and the sizing accuracy is lowered. Therefore, in this embodiment, the crystal grains of the material 1 are monitored, and the size of the crystal grains is accurately grasped to select the optimum flaw detection frequency for the crystal grains, thereby improving the defect detection performance and the sizing accuracy. ing. In addition, since the size of the crystal grains depends on the environment and the deterioration state, for example, it is said that the crystal grains become larger when the thermal aging treatment is performed. The lifetime can be predicted by evaluating the degree of deterioration and the degree of damage, thereby preventing accidents and extending the life of the equipment.

A specific example of crystal grain monitoring in the present embodiment will be described. First, the attenuation coefficient α of the polycrystal is expressed by the following formulas 4 to 6 from the average crystal grain size D and the wavelength λ (where A, B, and C are constants).
[Equation 4]
(When λ / D >> 1) α = AD ³ f ⁴ (Rayleigh scattering)
[Equation 5]
(When λ / D≈1) α = BDf ² (Stochastic scattering)
[Equation 6]
(When λ / D << ¹ ) α = CD ⁻¹ (diffuse scattering)
Note that the crystal of material 1 does not change significantly during use. Therefore, when comparing the magnitudes of “λ / D” as described above, it is only necessary to determine which formula is used based on the value of the crystal grain at the time of manufacture.

When the wave having the amplitude A ₁ propagates through the distance X to become the amplitude A ₂ , the attenuation coefficient α is obtained by Equation 7. If the crystal grain is large, the attenuation coefficient α also increases.

  In this way, the attenuation of the signal used for flaw detection depends on the frequency, and it is recognized that the higher the frequency, the greater the degree of attenuation. This relationship is expressed as an attenuation spectrum as shown in FIG. (However, if the wavelength is sufficiently smaller than the crystal grain, the wave cannot pass through the medium, so the frequency f is not necessary in Equation 6.) That is, FIG. 7 shows the relationship between the frequency and the attenuation factor in the case of a certain crystal grain size, and this relationship curve naturally changes as the crystal grain size changes. Therefore, if the relationship between the frequency and the attenuation rate is measured in advance for a plurality of types of crystal grain sizes, the size of the crystal grains can be estimated with at least some accuracy from the frequency and the attenuation rate ( Hereinafter referred to as “attenuation spectrum method”).

  Further, since the back scattered wave also depends on the size of the crystal grain, it is possible to adopt a method of evaluating the size of the crystal grain using the back scattered wave as shown in FIG. In other words, the amplitude and frequency characteristics of the backscattered wave shown in FIG. 8 also depend on the crystal grain size, as in the above-described relationship curve between frequency and attenuation rate. Therefore, by analyzing the backscattered wave, the crystal grain size can be estimated with a certain degree of accuracy (hereinafter referred to as “backscattered wave method”).

  Furthermore, if the size of the crystal grains is estimated by the above-described methods and corrected and complemented with each other, more accurate crystal grain sizes can be grasped, and defect detection performance and sizing accuracy can be further improved. Is preferable in that it becomes possible. For example, the estimated values of crystal grains obtained by the above-described “attenuation spectrum method”, “attenuation method”, and “backscattering wave method” (respectively D1, D2, and D3) are more accurate. Can be obtained (see FIG. 9). Specifically, the weighted average D = f (D1, D2, D3, f, V) obtained by determining the weights of D1, D2, and D3 from the frequency f and the sound velocity V is set as a more accurate estimated value. can do.

(iv) “Ultrasonic propagation characteristic prediction function” Furthermore, in this embodiment, from the viewpoint of further improving the accuracy of flaw detection identification, defect signal identification, etc., wave propagation is analyzed and visualized by the finite element method. ing. For example, the welded part of austenitic steel has characteristics that the crystal is coarsened and has elastic anisotropy in the welded part, so that the propagation behavior becomes complicated and the experimental results are difficult to interpret. Thus, if the wave propagation is analyzed and visualized by the finite element method, the characteristic prediction accuracy is improved.

  A specific example of wave propagation analysis / visualization by the finite element method in this embodiment will be described (see FIG. 10). First, the shape of the test body and the shape of the welded portion are converted into a CAD file format (for example, .dxf format) and input to a finite element method program (steps 21 and 22). As a result, it is possible to easily perform analysis using the shape of the actual specimen or the welded portion. Also, by incorporating crystal information such as crystal size and orientation obtained by EBSP (Electron Backscatter dif-fraction Patterns) into the finite element method model as it is (steps 23 and 24). Thus, it is possible to establish a model that is closer to reality, and as a result, it can be expected that an analysis result that is more realistic will be obtained (steps 27 and 29). In step 25 and step 26, for example, judgment elements such as “step time”, “element size”, and “how many seconds to calculate” are input as parameters for analysis in the wave analysis (step 27) (step 25, step 26). ).

  If the above analysis result is visualized (step 28), for example, the situation of the scattered wave between the grain boundaries will be clarified by this visualization, and the result will be applied to the estimation of the crystal grain size by the back scattered wave. As a result, the propagation characteristics of ultrasonic waves can be further clarified and used as an effective basis for interpreting experimental results and selecting optimal experimental conditions. For the displacement at each position of the model at each time obtained by the finite element method, for example, the distribution of the displacement at each time is imaged using an AVS or Micro AVS visualization environment. Wave propagation can be visualized by converting the displacement image at each time into a movie (step 28).

  Further, in performing wave analysis in step 26 and step 27, in this embodiment, “explicit solution” is performed (see FIG. 10). That is, if the mass distribution approximation that considers that the mass distributed to the elements is concentrated at the nodes, the mass matrix becomes a diagonal matrix having values only in the diagonal components. At this time, since the inverse matrix is obtained by the inverse of the diagonal component, the nodal acceleration can be easily obtained by multiplying the inverse of the diagonal component of the mass matrix. In this case, the nodal acceleration can be obtained without solving simultaneous equations (explicit method).

  The above is a description of the data processing function peculiar to the software side. Next, the measurement technique on the hardware side for improving the welded part inspection technique will be described.

  First, regarding the sizing accuracy by the TOFD method and the phased array method, for example, regarding the intrinsic defect of the HAZ, the TOFD method has a characteristic that the crack defect 2 can be measured with higher accuracy than the phased array method, but at the time of welding construction, There is a problem that it is difficult to distinguish the generated weld defect (for example, slag entrainment or poor penetration) from the HAZ defect, that is, it is difficult to clearly identify both. For example, as in the case of manufactured specimens, there are many slag entrapments and penetration defects in the vicinity of the first layer of the outer surface welding in actual machine welds. Therefore, if the site of natural defects can be more accurately evaluated, crack defects 2 It is possible to accurately evaluate the properties of. Here, when briefly inspecting the conventional welded part inspection technique, when inspecting the crack defect 2 in the welded part, the position of the crack defect 2 in the direction perpendicular to the weld line (depth direction) is usually measured by the TOFD method. In order to discriminate between the crack 2 and the natural defect, it is necessary to perform a B-scan at the defect detection site as a precise flaw detection in addition to this (see FIG. 11). . However, an expensive driving device capable of two-dimensional scanning is required, and the inspection time is inevitably increased (see FIG. 11).

  Based on the above, in this embodiment, the phased array method is used in combination with the D scan of the TOFD method instead of the B scan of the TOFD method or the raster scan of the pulse echo method. In such a case, an inexpensive one-dimensional scan is sufficient for the driving device, and even for a minute crack 2 having a height of 10 mm or less, the RMS (root mean square); An average amplitude given by the square root of the average value) is 0.9 or more, and a highly accurate inspection can be performed (see FIG. 11). Therefore, according to this, only one-dimensional mechanical scanning can be performed, and the crack defect 2 can be measured, the inspection time including the rough flaw detection and the precision flaw detection can be shortened, and the driving device necessary for the scanning can be made inexpensive. (See FIG. 11). For the above-mentioned RMS, the mean square error was determined by the following formula 8.

Here, x _i is a measured value, X _i is an actually measured value, and n is the number of data.

  The evaluation method using the TOFD method and the phased array method in this way will be described more specifically (see FIG. 12). Evaluation is started (step 31). First, when the crack defect 2 is detected by the TOFD method (step 32), the defect size and position # 1 (test body) are determined from the B-scan image of the TOFD method. Thickness direction and direction parallel to the weld line) (step 33). If it can also be detected by the phased array method (PA method), the defect position # 1 is read as a reference value from the B-scan image (steps 34 and 35). A defect that can be recognized from both methods is defined as a detection level A, and a defect that is recognized only by the TOFD method is defined as a detection level Ba (see FIG. 13). When the material 1 is thick and the crack defect 2 near the inner surface cannot be detected, the size and position # 1 of the crack defect 2 are determined from the B-scan image of the phased array method, and such a defect is detected at level B-b. It is defined as In the case of detection levels A and B-b, position # 2 (direction perpendicular to the weld line) is determined from the top view (projected image in the thickness direction of the specimen) of the phased array method (step 36), and the defect size is determined. Then, the nature of the defect is specified from the position and the detection level (step 37). Subsequently, Nd is used as the total number of defect instructions, and the magnitude is compared with the counter value N (step 38). When Nd is larger than N, 1 is added to N (step 39), and the process returns to step 32. On the other hand, if not, the dimension, position, detection level and defect properties of the defect are described (step 40), and the evaluation is terminated (step 41). As described above, by evaluating the dimension of the defect by the TOFD method and the position of the defect by the phased array method, the flaw detection test of the welded portion can be performed with higher accuracy than conventional.

As described so far, in the present embodiment, for the purpose of early detection of the crack defect 2 in the welded portion of the material 1, techniques such as the TOFD method and the phased array method are simultaneously applied with a single mechanical scan. This makes it possible to acquire information on rough flaw detection and precision flaw detection in the inspection time required for conventional rough flaw detection, and to perform highly accurate defect dimension measurement. Further, it is possible to distinguish HAZ defects and weld defects from quantitative position and dimension measurements. That is,
(1) In order to improve nondestructive inspection of high-temperature steam pipe welds and the like, measurement is performed using the conventional TOFD probe 3 simultaneously with the phased array probe 3 in one automatic flaw detection.
(2) By using the phased array probe 3 in comparison with the conventional precision flaw detection using a single transducer, the time required for measurement is significantly reduced (specifically, for example, about 1/60). Is possible.
(3) The three-dimensional position of the defect is obtained by flaw detection (diagonal flaw detection) using the phased array probe 3, and the position measurement accuracy is high with a mean square error in each of the axial direction and the circumferential direction (for example, 0.64 mm or less and 2.27 mm or less), and circumferential crack defects 2 with an interval of 2 mm in the axial direction can be sufficiently identified.
(4) By the way, when the defect indications do not overlap, the TOFD method has better height measurement accuracy than the phased array method with respect to the height of the crack 2, and the phased array method has a higher measurement accuracy with respect to the length. Is good. Further, when the crack defect 2 exists in a direction perpendicular to the D-scan direction of the TOFD method and the defect instruction is superimposed, the phased array method is effective.
(5) By using the TOFD method and the phased array method in combination, the defect position is measured from the detected defect indication to identify the crack 2 and the weld defect in the heat affected zone, and the defect dimension is measured with high accuracy. It becomes possible to do.

  The above-described embodiment is an example of a preferred embodiment of the present invention, but is not limited thereto, and various modifications can be made without departing from the scope of the present invention. For example, in the present embodiment, as a specific example of signal processing in the “noise removal function” of (ii), a method of applying one or both of the spatial averaging method and the wavelet transform method has been described (see FIG. 6). These two methods are merely examples of suitable methods at the time of filing of the present application, and can be replaced with other signal processing methods that perform the same processing as these.

In one embodiment of the automatic ultrasonic flaw detection of a thick-walled structure according to the present invention, the zero-cross method is used to detect points between the same phase for each of the i-th reflected wave and the i + 1-th reflected wave from the bottom of the plate-shaped specimen. is a diagram showing calculation contents of obtaining the propagation time difference Δt of ₍₌ T 2 -T _1). It is a figure which shows the calculation content of selecting the two waveforms with the largest correlation function _Rxy and calculating | requiring the propagation time difference (tau) between both by the correlation function method, when the similarity of adjacent waveforms is falling. . It is a figure which shows the phase spectrum of each waveform at the time of selecting two waveforms by correlation function method and calculating | requiring the propagation time difference (tau) between both. It is a figure which shows the example of a relationship between the frequency at the time of calculating | requiring a sound speed with a phase spectrum method. It is a figure which shows simply a mode that a reflected wave called a corner echo returns to a probe from an ultrasonic beam hitting the corner part of a crack. It is a flowchart which shows an example of the process sequence in the case of performing signal processing on-line even when there exists noise by a remarkable forest-like echo and a backscattered wave by a noise removal function. It is the figure which showed an example of the attenuation spectrum showing that attenuation of the signal used for a flaw detection is dependent on the frequency. It is a figure which shows the image of a backscattered wave, and the amplitude and frequency characteristic of the said backscattered wave. It is a figure which shows directly the method of obtaining a more exact estimated value from the estimated value of the size of the crystal grain obtained by each of "attenuation spectrum method", "attenuation method", and "backscattering wave method". It is a flowchart which shows the specific example of the analysis and visualization procedure of the wave propagation by the finite element method in this embodiment. It is a table | surface which compares and shows the conventional welding part inspection technique and the inspection technique concerning this embodiment. It is a flowchart which shows the procedure of the dimensional property evaluation of the crack defect in this embodiment. It is a table | surface which shows the example of an evaluation result of the dimensional property evaluation of the crack defect in this embodiment. It is a figure which shows simply the mode of the ultrasonic flaw detection of the thick structure in the past. It is a figure which shows an example of the waveform obtained in the case of the conventional ultrasonic flaw detection. It is a figure which shows simply the mode of the ultrasonic flaw detection of the thick structure in the past. It is a figure which shows a mode that a probe is linearly moved to a longitudinal direction in parallel with a welding line. It is a figure which shows the prior art example of the probe structure which vibrates a single piezoelectric element with a single pulse receiver through a cable, and produces an ultrasonic wave. It is a figure which shows the structural example of a phased array which arrange | positions a some pulse receiver and a piezoelectric element on one-to-one basis. It is a figure which shows the phased array method which changes the direction of an ultrasonic beam by shifting the timing of a pulse voltage, and changes a depth of focus, and performs a flaw detection. In FIG. 20, it is a figure which shows simply the mechanism in which the direction of an ultrasonic beam changes by shifting the timing of a pulse voltage. In a TOFD method, it is a figure which shows a mode that a probe is moved to a zigzag and a two-dimensional scan is performed.

Explanation of symbols

1 Materials for thick-walled structures 2 Cracks (crack defects)
3 Probe

Claims (2)

  In an ultrasonic flaw detection apparatus for a thick-walled structure in which crack defects generated in a thick-walled structure are flawlessly detected by applying both the TOFD method and the phased array method, the transmitted ultrasonic waveform and the thick-walled structure The actual sound speed of the ultrasonic wave propagating in the material and the distance to the reflector from the waveform of the reflected echo such as the bottom surface of the material or an internal crack defect, and at least one of the zero cross method, the correlation function method, and the phase spectrum method are used. Sound velocity / distance measuring means that measures each material using one, and if noise due to shape echo or backscattered wave appears remarkably in the received waveform, these noises are either spatial averaging method or wavelet transform method Noise removal means for enhancing the defect signal in exchange for reducing the noise intensity by signal processing using one or both, and a plurality of types of connections in the material. Measurement result of the relationship between the frequency and attenuation rate of the ultrasonic wave depending on the size of the grain, or measurement result of the amplitude and frequency characteristics of the backscattered wave of the ultrasonic wave depending on the size of the crystal grain similarly to this The crystal grain size is evaluated using the above-described evaluation results, and the crystal grain evaluation means for selecting the optimum flaw detection frequency for the crystal grain based on the evaluation result, and the behavior of the ultrasonic wave propagation is complicated. Ultrasonic flaw detection of a thick-walled structure characterized by comprising ultrasonic propagation characteristic predicting means that analyzes and visualizes wave propagation and makes it possible to predict the ultrasonic propagation characteristic of the material having a coarse crystal apparatus. In an ultrasonic flaw detection method for a thick structure in which a crack defect generated in a thick structure is ultrasonically detected by applying both the TOFD method and the phased array method, a transmitted ultrasonic waveform and the thick structure The actual sound speed of the ultrasonic wave propagating in the material and the distance to the reflector from the waveform of the reflected echo such as the bottom surface of the material or an internal crack defect, and at least one of the zero cross method, the correlation function method, and the phase spectrum method are used. The sound velocity / distance measurement step that measures each material using one, and if noise due to shape echo or backscattered wave appears remarkably in the received waveform, these noises are either spatial averaging method or wavelet transform method A noise removal step of reducing the noise intensity by signal processing applying one or both and emphasizing a defect signal in return, and a plurality of noise removal steps in the material Measurement result of the relationship between the frequency and attenuation rate of the ultrasonic wave depending on the size of the crystal grain, or the amplitude and frequency characteristics of the backscattered wave of the ultrasonic wave depending on the size of the crystal grain as well as this The size of the crystal grain is evaluated using the measurement results of the above, and the crystal grain evaluation step for selecting the optimum flaw detection frequency for the crystal grain based on the evaluation result, and the behavior of the wave propagation of the ultrasonic wave are complicated Ultrasonic wave of a thick-walled structure, characterized in that it comprises an ultrasonic wave propagation characteristic prediction step that makes it possible to analyze and visualize the wave propagation and predict the ultrasonic wave propagation characteristic of the material having a coarse crystal Flaw detection method.