Pipe wall thinning inspection using EMAR

In piping systems of nuclear power plants, pipe wall thinning is caused by ageing-related flow-accelerated corrosion (FAC) and liquid droplet impingement (LDI) erosion. Piping maintenance is based on residual lifetimes estimated by periodic measurements of thickness. High accuracy and reliability of these measurements are mandatory in the nuclear power industry. Currently, ultrasonic (UT) thickness gauges are used to inspect pipe wall thinning [1]. Although UT thickness gauges provide measurements with high accuracy of about 0.1 mm, these are sensitive to surface rusts and paint, and require highly skilled workers to operate

Electromagnetic acoustic transducers (EMAT), which is one ultrasonic measurement method, provide non-contact measurements, since ultrasonic waves are transmitted and received electromagnetically [2]. For this reason, the method is often applied in high-temperature environments [3-5]. The drawback of EMAT is its low electromechanical conversion efficiency, which leads to low sensitivity and low signal-to-noise (S/N) ratio. One of the ways to overcome EMAT's drawback is to introduce an electromagnetic acoustic resonance method (EMAR) [6]. Since the principle of the method is based on the through-thickness resonances of bulk waves, that is, thickness oscillations, targets are usually limited to ones with simple geometry such as plates. The authors have proposed a signal processing method, called the superposition of nth compression (SNC) method, to extract thickness of pipes with wall thinning from EMAR spectral response data. This method was applied to a measurement of piping cut from a mock-up test loop, and the high capability of evaluating thickness was confirmed [7].

This paper presents the method of EMAR developed by the authors [4, 7] and two applications of EMAR with the SNC method to evaluation of pipe wall thinning: online monitoring [8] and inspection of pipe wall thinning [9].

EMAT operations

The typical EMAT consists of a coil and a permanent magnet. When a radiofrequency pulse current is applied to the coil placed on a nonmagnetic conductor specimen, an eddy current is induced near the specimen surface. When a static magnetic field perpendicular to the specimen surface is further introduced by the permanent magnet, an Ampere force parallel to the specimen surface is generated and a transverse wave having a frequency equal to that of the radiofrequency pulse applied to the coil propagates through the specimen. In the case of a magnetic material specimen, its dimensions change cyclically due to the magnetostrictive effect caused by the magnetic field induced by the coil and the static magnetic field induced by the permanent magnet; ultrasound waves are emitted as a result. Reception of ultrasound waves is conducted in a reverse way by detecting the voltage generated in the coil.

Because the sources of vibration and reception exist within the specimen surfaces, it is possible to conduct non-contact measurement. Therefore, a contact medium, which is required in ultrasonic thickness gauges using piezoelectric devices, is not required. In addition, EMATs are less susceptible to coating and rust on specimen surfaces.

An ultrasound burst wave that enters a specimen is reflected repeatedly on both ends of the specimen. When the wavelength is an integral multiple of the propagating distance in the specimen, the phase of the incident wave matches with that of the reflected wave and resonance occurs. The EMAR measurement method uses this phenomenon to improve the signal-to-noise (SN) ratio of the received signals by amplifying echo waves whose amplitude has been attenuated by scattering.
The relationship between resonance frequency f and specimen thickness d is given by: (1)

F_n = n x f₁ = n x v/2d

where fn is the nth resonance frequency, f1 is fundamental resonance frequency, n is order of resonance, and v is sound velocity in the specimen.

The SNC method

Figure 1 shows a simulated wall-thinned specimen made of carbon steel SS400. The bottom surface of a plate 150 mm long, 50 mm wide, and 10 mm thick was processed two-dimensionally to form an R-shaped dent 2 mm deep and 70 mm wide as a simulated thinned wall. The sound velocity is 3240 m/s at room temperature.

Figure 2 shows the result of EMAR measurement at a location 12.5 mm distant from the centre of the dent. Because of the sloped nature of the dent bottom surface, signal intensity is lowered by ultrasound scattering, and resonance frequency peaks are unclear. Although the fundamental resonance frequency is 195 kHz in theory, there are several peaks around this value so it is difficult to determine the fundamental resonance frequency from peak distances.

The SNC method is an analytical method based on the cyclic appearance of resonance frequencies at integral multiples. This means that by reducing the frequency measurement results obtained by the EMAR method by a factor of n, the intensity of nth order frequency fn can be superimposed on the intensity of the fundamental resonance frequency f1. Theoretically, for example, reducing the second resonance frequency f2 by a factor of 2 results in coincidence with f1, and reducing f3 and f4 by factors of 3 and 4, respectively, also results in coincidence with f1.

f₁ = arg max_f { S_n x (f/n)/m}

d = v / 2f₁

As shown in Equation 2, the fundamental resonance frequency f1 is obtained from the maximum value of the average intensity of the spectrum obtained by superimposing the frequency results obtained by EMAR measurement and then reduced by a factor of n. Then, conversion from sound velocity to specimen thickness is conducted using Equation 3. The maximum spectrum intensity so obtained is treated as the peak SNC value. Here, x(f) is SNC spectrum intensity and m is the number of superimposed resonance orders.

Resonance frequency, as well as the orders of resonance frequency contained within the test frequency range, depends on specimen thickness and sound velocity. For example, the resonance frequency is 324 kHz when the pipe wall specimen thickness is 5.0 mm and sound velocity is 3240 m/s, and the orders of resonance frequency contained within the test frequency range of 1.5 to 3.5 MHz are integrals 5 through 10. By reducing the EMAR frequency measurement results by factors ranging from 5 to 10 and summing them up, the fundamental resonance frequency peak becomes clearly accented. The SNC spectrum intensity is the average value obtained for the added up resonance orders.

In the online monitoring test, sound velocity is determined using a calibration specimen whose material type is the same as that of the pipe under investigation. The estimated wall thickness is necessary to decide the range of resonance frequency orders in each SNC evaluation. Hence, the estimated wall thickness used in the SNC analysis is the nominal thickness for the first round of test and the thickness measured at the (x-1)th round test for the xth round test.

Figure 3 shows the result of SNC analysis applied to EMAR measurements on the simulated wall-thinned area shown in Figure 2. The analysis assumed an estimated thickness of 8 mm and the range of orders n=8 to 17. The spectrum shows multiple fundamental resonance frequencies, indicating a variation in the bottom surface shape. The thickness value of 8.28 mm corresponding to the maximum-peak resonance frequency (195.7 kHz) is regarded as the wall thickness of the specimen.

The EMAR measurement equipment consists of a high-power pulsar receiver, a preamplifier to amplify the received signals, a wide range decade filter to filter the received signals, an oscilloscope to observe the waveform and collect data, and a personal computer to store and analyse waveform data. One measurement point takes around 400 s.

Online monitoring of pipe wall thinning

The pipe wall thinning test equipment allows testing of pipe wall thinning under a single-phase or two-phase flow environment by simulating the water flow conditions of a nuclear power plant [8]. Figure 4 outlines the setup of a test pipe whose features cause fluid turbulence in the test pipe, accelerating wall thinning.

A simulated wall-thinning test that generated a two-phase flow in the test pipe to accelerate wall thinning was conducted. (In a second test, excluded from this article, a single-phase flow with different amounts of dissolved oxygen was generated). The fluid temperature in the pipe was 165°C, which is within the temperature range where pipes are susceptible to flow-accelerated corrosion. The pipe of the test section was wrapped in insulating material and enclosed in a metal protector, and the pipe and the fluid in the pipe had the same temperature. EMATs were installed directly on the pipe surface without peeling off the surface coating. Each EMAT was held by the magnetic force of its permanent magnet and was secured in place with wire so that it would not get misaligned by any accidental contact caused by such as installing and removing insulator material.

The test period was about two months, during which the system was shut down twice for periodic inspection. The probe was placed at about 650 mm from the orifice end, and a biaxial cable was used for the distribution cable. Using a calibration specimen 5 mm thick, the fundamental resonance frequency was measured by EMAR at room temperature to determine the sound velocity as 2340 m/s. Figure 5 compares SNC signals obtained during the rise of pipe temperature, at room temperature before the start-up of equipment, and at 165°C. It is shown that at the higher temperature, SNC peaks tend to be lower in both intensity and resonance frequency. This can be explained by the peak attenuation at the elevated temperature, associated with the decreased flux density of the magnet as well as the decreased sound velocity associated with the decreased elastic modulus (the sound velocity decreased nearly linearly as the pipe temperature rose). The sound velocity of 3180 m/s at the operating temperature was used for evaluating wall thickness.

Figure 6 shows the change of pipe wall thickness with time evaluated by EMAR under a high-temperature environment during the operation of the pipe wall thinning test equipment. The thickness measured by EMAR in the first round was 5.42 mm and that measured in the last round was 5.24 mm, resulting in a thickness reduction of 0.18 mm. During the measurement period of 58 days, the thickness was reduced at a rate of 3.0 µm per day. After the completion of the test, the test pipe was cut out and its cross section was measured under a microscope. The wall thickness at the location where the EMATs were installed was 5.18 mm, which means a difference of 0.06 mm from the final thickness measurement by EMAR. The difference, probably caused by measurement inconsistencies, is within the allowable range, as compared with the error of ± 0.1 mm in the ultrasound thickness gauge.

Figure 7 shows the initial spectrum obtained by SNC analysis and that obtained 30 days later. Compared with the former SNC spectrum, the latter shows a shift of resonance frequency to the higher side, a reduction in amplitude, and an increase in the number of peaks. This indicates the occurrence of unevenness and slope on the inner surfaces of the pipe. It is considered that the change in the bottom surface shape caused ultrasound scattering, which resulted in decreased amplitude and multiple peaks.

Application to piping inspection

In this study [9], we used EMAR to measure pipes of the secondary cooling system of unit 2 of Japan Atomic Power Company's Tsuruga Nuclear Power Plant. The power plant is a pressurized-water reactor that started commercial operations in February 1987. Measurements are done during the reactor outage for periodic inspection. We extracted eight test sections from the plant as targets for EMAR measurements: four straight pipes and four elbows. The purpose of including both elbows and straight sections of piping was to assess the effect of pipe geometry on measurements. In addition, the tested sections include those in long-term service to evaluate effects of advanced thinning. There were a total of 195 measurement points.

The straight pipes are made of STPT38 (carbon steel for high temperature) and have outer diameters of 48.6, 114.3 and 165.2 mm and nominal wall thicknesses of 5.0, 8.6 and 7.1 mm. The elbows are made of STPT38 and SB410 (carbon steel for pressure vessels and a boiler) and have outer diameters of 89.1 and 558.8 mm and nominal wall thicknesses of 5.5 and 10.0 mm. When the EMAT is inclined by the weld padding, the EMAR is measured about 15 mm from the ends of the welds. Therefore, the measuring points of the EMAR in the vicinity of welds are about 10 mm from the points of the UT. Figure 8 shows the setting of the EMAT probe near the weld. The white circles on the pipe are the UT measurement positions. If there is a welding line, the starting position can be measured at a distance of about 20 mm around the downstream side from the welding line. The measurement position of the EMAT is near this position.

UT is performed by a qualified person before the EMAR measurements. We compare the results obtained with EMAR with those obtained in UT.

We use the peak value of the fundamental resonance frequency of the calibration specimen (a STPT38 carbon steel plate with thickness of 5.01 mm) to normalize the value of the SNC peaks. This value is hereafter called the normalized SNC peak value.
Figure 9 compares the thicknesses obtained in the UT and through EMAR with SNC. Closed and open circles indicate the results at base pipes and welded pipes, respectively. The results did not depend on pipe diameters or whether the pipe was straight or an elbow, and the effect of thickness was lower. However, there are differences between the EMAR and UT results near the welding.
Figure 10 shows the relationship between the normalized SNC peak values and the difference between the EMAR and UT results. The result shows that when the normalized SNC peak values are more than 0.15, 0.1, and 0.05, the root mean squares (RMSs) of the differences are 0.18, 0.21, and 0.36, respectively. When the normalized SNC peak value is more than 0.15, the thicknesses obtained with EMAR and UT agree well and are highly reliable. However, if the normalized SNC peak value is less than 0.05, the difference between the UT and EMAR results becomes large and reaches 3 mm. There are discrepancies between EMAR and UT measurements when the probes are close to the welding. As mentioned previously, because of the configuration of the reducer and size of the EMAT, the transducer was put 10 mm away from the measurement points of the UT thickness gauge in the vicinity of welds. The differences in measurement results indicate the difference in measurement position and the influence of the weld. Because the heat near the weld changes the sonic velocity at the entrance, it is possible that UT and EMAR measured the change in sonic velocity. When the normalized SNC peak value was attenuated to less than 0.05, the results of UT and EMAR differed at several measurement points.

In summary, EMAR inspection results have shown good agreement with those of an ultrasonic thickness gauge. In addition, as a non-contact measurement method, EMAR decreases measurement times of pipe inspections in nuclear power plants. Our technique of EMAR with SNC signal processing was validated through two applications in relevant environments, and so is positioned on level 5 of the technology readiness level scale: it still requires prototyping and qualification before it can be used in the field. ¦

References

1. The Japan Society of Mechanical Engineers ed., Codes for Nuclear Power Generation Facilities: Rules on Pipe Wall Thinning Management for BWR Power Plants, (2006), pp. 7-49, The Society of Mechanical Engineers.

2. R. B. Thompson, Physical principles of measurements with EMAT Transducer, Physical Acoustics Vol.XIX, Academic Press Inc., (1990), pp.157-200.

3. N. Yamagata, M. Takahashi and N. Ahiko, Thickness Measuring Technology for Pipes of Thermal Power Plants, Toshiba Review (in Japanese), Vol. 63, (2008), pp.46-49.

4. R. Urayama, T. Uchimoto, T. Takagi, Application of EMAT/EC Dual Probe to Monitoring of Wall Thinning in High Temperature Environment, International Journal of Applied Electromagnetics and Mechanics, Vol.33, (2010), pp.1317-1327.

5. F. Hermandez-Valle and S. Dixon, Pulsed electromagnet EMAT for ultrasonic measurements at elevated temperature, INSIGHT, Vol.53, (2011), pp.96-99.

6. M. Hirao and M. Ogi, EMATS for Science and Industry, Non-contacting Ultrasonic Measurements, Kluwer Academic Publishers, Dordrecht, ISBN-10: 1441953663 (2003).

7. R. Urayama, T. Uchimoto, T. Takagi, S. Kanemoto, Quantitative Evaluation of Pipe Wall Thinning by Electromagnetic Acoustic Resonance, E-Journal of Advanced Maintenance, Vol.2, (2010/2011), pp25-33.

8. R. Urayama, T. Uchimoto, T. Takagi, S. Kanemoto, Online Monitoring of Pipe Wall Thinning with EMAR, Maintenology (Hozengaku), Vol. 11, (2013), pp. 83-89, (in Japanese).

9. R. Urayama, T. Takagi, T. Uchimoto, S. Kanemoto, T. Ohira and T. Kikuchi, Implementation of electromagnetic acoustic resonance in pipe inspection, E-Journal of Advanced Maintenance, Vol. 5, No 1 (2013), pp. 25-33.

This study was performed as part of "Project on enhancement of aging management and maintenance of NPPs" supported by Nuclear and Industrial Safety Agency, METI, JAPAN. This study was also supported by Tohoku University Global COE Program, "World center of education and research for trans-disciplinary flow dynamics". The mock specimen was provided by the Tokyo Electric Power Company. The authors would like to thank Mr. Takeshi Sato and Mr. Tsutomu Watanabe of the Institute of Fluid Science, Tohoku University, for the preparation of the specimens and the fabrication of the probe.